Saliva-based SARS-CoV-2 serology using at-home collection kits returned via mail

Abstract

Serology provides tools for epidemiologic studies, and may have a role in vaccine prioritization and selection. Automated serologic testing of saliva, especially specimens that are self-collected at home and sent to a laboratory via the mail without refrigeration, could be a highly-scalable strategy for population-wide testing. In this prospective study, non-vaccinated patients were recruited after PCR testing to self-collect saliva and return their specimens via mail. Longitudinal specimens were analyzed in order to monitor seroconversion in the weeks after a diagnostic PCR test for SARS-CoV-2. Diverse users self-collected saliva and returned specimens via mail in compliance with shipping regulations. At our pre-established threshold (0.963 AU/mL), salivary IgG reactivity to full-length spike protein achieved 95.8% sensitivity and 92.4% specificity at 2-4 weeks after diagnostic testing, which is comparable to the typical sensitivity and specificity achieved for serum testing. Reactivity to N antigen also was detected with 92.6% sensitivity and 90.7% specificity at 4-8 weeks after diagnostic testing. Moreover, serologic testing for endemic coronaviruses performed in multiplex with SARS-CoV-2 antigens has the potential to identify samples that may require retesting due to effects of pre-analytical factors. The easy-to-use saliva collection kit, coupled with thresholds for positivity and methods of flagging samples for retest, provides a framework for large-scale serosurveillance of SARS-CoV-2.

Figure 1

Salivary anti-SARS-CoV-2 IgG antibodies in the weeks after a PCR test. Reactivity to (a) full-length spike protein, (b) receptor binding domain (RBD) of the spike protein, and (c) N protein of SARS-CoV-2 were measured in saliva provided by 121 participants who had received a PCR test for SARS-CoV-2. Participants provided up to three samples. Matched samples provided by the same donor are connected with lines. 81 participants tested negative for SARS-CoV-2 (colored purple), and 40 participants tested positive (colored green). The dashed, red line indicates the pre-established cut-point for high versus low antibody levels. (d–f) Box and whisker plots of the same data shown in panels (a–c). The PCR + cohort was significantly different (p < 0.001; see Supplemental Tables) from the PCR- cohort at all timepoints. Asterisks indicate statistical significance (p < 0.05 by Mann–Whitney test) for difference between the < 2 weeks and 2–4 weeks timepoints.

Figure 2

Correlation in IgG concentrations for SARS-CoV-2 antigens. Dotted lines indicate thresholds for classifying high versus low reactivity.

Figure 3

Receiver-operator characteristic (ROC) curves for classifying SARS-CoV-2 PCR results based on salivary anti-SARS-CoV-2 IgG antibodies. Curves are drawn to show performance of assay at < 2 weeks, 2–4 weeks, and 4–8 weeks after a PCR test. The red dot indicates sensitivity and specificity achieved at the 4–8 week time point for the pre-established threshold. The black dot indicates, on the same curve, the threshold that maximizes sum of sensitivity and specificity.

Figure 4

Concentrations of anti-spike and anti-N IgG measured in saliva of patients tested for SARS-CoV-2. Saliva was self-collected and returned via mail to a laboratory for analysis. For indirect serology, recombinant proteins representing the receptor binding domain (RBD) and N-terminal domain (NTD) of spike along with full-length spike and N antigen were multiplexed in wells of a 96-well plate. (a) IgG antibodies to four endemic coronaviruses were readily detected in nearly all samples. Eight samples (highlighted in blue) showed consistently low levels of antibodies to all coronaviruses. (b) 91% of samples were in transit for five days or less, and transit time was not strongly correlated with the mean level of salivary IgG for the four endemic coronaviruses (CoVs). Dotted line indicates geometric mean for the 5th percentiles of endemic CoV salivary IgG measured in a prior study of presumed naïve controls without prior diagnosis, household exposure, or symptoms of COVID-19. (c) Total levels of immunoglobulins (IgG, IgM, and IgA) were also measured in saliva.