Determining the Efficacy of Saliva in the Detection of SARS-CoV-2 in a Molecular Assay Using Contrived and Clinical Samples ​

The unprecedented burden that the COVID-19 pandemic has placed on individuals and healthcare systems highlights the need for rapid, reliable diagnostic techniques and easy to procure biological matrices to detect the SARS-CoV-2 virus. While saliva contains detectable levels of the SARS-CoV-2 virus in infected individuals, several molecular assays that detect SARS-CoV-2 do not include saliva as an approved sample. We evaluate saliva as a clinical sample in detecting SARS-CoV-2 on the Cepheid Xpert® Xpress SARS-CoV-2/Flu/RSV reverse transcription PCR assay and assess how removal of cellular debris and mucus through centrifugation impacts the performance of SARS-CoV-2 detection in saliva. We further compare sensitivity of uncentrifuged and centrifuged saliva samples by performing limit of detection testing in contrived samples. Cohen’s kappa statistic was 0.78 between the gold-standard, nasopharyngeal swabs in viral transport media (NP-VTM), and saliva indicating significant agreement in SARS-CoV-2 detection between the sample types. The positive percent agreement (PPA) between centrifuged saliva and NP-VTM samples was 0.92 compared to a PPA of 0.85 between uncentrifuged saliva and NP-VTM. In contrived samples, heat inactivated SARS-CoV-2 was reliably detected at a concentration of 51,852 copies/mL in uncentrifuged saliva and 2,074 copies/mL in centrifuged saliva. Ct values increased less consistently in uncentrifuged saliva compared to centrifuged saliva, as indicated by the coefficient of Determination R-squared values of 0.92 and 0.54, respectively. Our study indicates that saliva is a viable clinical sample for detecting SARS-CoV-2 in a reverse transcription PCR assay and removal of cellular debris improves sample performance.