ELISA detection of SARS-CoV-2 antibodies in saliva

Abstract

To facilitate containment of the COVID-19 pandemic currently active in the United States and across the world, options for easy, non-invasive antibody testing are required. Here we have adapted a commercially available, serum-based enzyme-linked immunosorbent assay (ELISA) for use with saliva samples, achieving 84.2% sensitivity and 100% specificity in a set of 149 clinical samples. This strategy will enable widespread, affordable testing for patients who experienced this disease, whilst minimizing exposure risk for healthcare workers.

Figure 1

Sensitivity and specificity of four serum-based ELISAs: Detection of IgG or IgA antibodies against SARS-CoV-2 in serum samples collected prior to the spread of COVID-19 (pre-Nov. 2019), and people who previously tested negative (−) or positive (+) for COVID-19 using Curative's oral fluid PCR test. The positive cutoff values represented by the dotted lines were provided by the manufacturer. NC = negative control; PC = positive control.

Figure 2

Investigation of protocol changes to enable use of saliva samples: (A) Sample conditions tested for optimized detection of anti-SARS-CoV-2 antibodies in mouthwash saliva samples, in comparison to the original protocol (AUC = 0.946): mucin removed (AUC = 0.935), mucin removed and sample concentrated (AUC = 0.962), and mucin removed. Sample concentrated, and non-specific binding blocked (AUC = 0.965). (B) IgG was detected using the optimized protocol in a large sample set where positive and negative status was determined by both PCR and serum antibody status, providing 100% specificity with 84.2% sensitivity (AUC = 0.979). A Pearson correlation showed a weak but significant correlation between serum and saliva IgG (R = 0.596, p < 0.0001). Line of best fit is displayed with 95% confidence intervals.

Figure 3

Investigating the optimal population for a saliva-based ELISA: (A) Visualization of the gender (n = 149), age (n = 149), and days post symptom onset (n = 85) of our cohort of clinical samples. (B) Stratifying by age impacted the sensitivity and specificity of our optimized saliva protocol. (C) When limited to sampling people over 40 years of age (n = 86), 91.5% sensitivity and 100% specificity (AUC = 0.988) can be achieved.