A total of 2,008 donor samples were collected for this study between April and May of 2020, just as the daily new case rates peaked in RI (https://ri-department-of-health-covid-19-data-rihealth.hub.arcgis.com/). We compared age, sex and race/ethnicity of the sample group to values reported for Rhode Island from the 2010 U.S. Census. The median age of donors was 56 years, significantly older than the Rhode Island median age of 39.4 years (Fig. 1A, Table 1). The sample had ~ 47% female donors compared to 52% statewide (Fig. 1B, Table 1). The distribution of donor Race/ethnicity was 84.7% white, 2.7% Hispanic/Latino and 0.50% Black/African American, compared to the state distribution of 81% white, 12.4% Hispanic/Latino and 5.7% Black/African American. A full comparison appears in Table 1 and Fig. 1. Notably, 9.3% of donors responded to ethnicity as ‘Declined’ or ‘Not Specified’. Finally, geographic location of donors associated with population density, such as Providence and Warwick, with lower representation in the western and coastal regions of Rhode Island (Fig. 1C, 1D). Thirteen donors were identified as convalescent plasma or whole blood donors that were aware of their seroconversion status prior to enrollment in the study and were removed from the analysis, which adjusted the total donors analyzed to 1,996.
To quantify seroprevalence in this sample, donor samples were tested with an HTSA platform (Ortho Clinical Diagnostics VITROS Total Ig Test) and an LFA platform (SD Biosensor IgM/IgG test). The IgM-only LFA assay yielded 68 positive tests for a 2.7% (95% CI 1.7 to 3.8%) seroconversion (Fig. 2A, Table 2). In contrast, the IgG-only LFA assay yielded 13 positive tests for 0.6% seroconversion (95% CI 0.3 to 1.1%) and was in agreement with the Ortho HTSA assay, which had 14 positives for a 0.6% seropositivity (95% CI 0.2 to 1.1) (Fig. 2A, Table 2).
In total, 3.9% of all samples (77 seropositive donors) were reactive for at least one test. To report overlap between test results, we constructed a Venn diagram (Fig. 2B, Table 2). Notably, ~ 76% of seropositive samples (59 of 77) were reactive only with the IgM-only LFA test. The remaining 0.9% of all samples (18 seropositive donors) showed a ~ 62% overlap between Ortho and IgG LFA assays (10 of 18 seropositive donors). Samples that showed at least 2 or more positive reactions was 0.55% (11 seropositive donors).
Donors completed a two-part questionnaire as to whether they had COVID-19 and if so, the results of the diagnostic PCR test. Overall, 76 donors responded that they had received a diagnostic PCR test for COVID-19; of these, 13 donors tested positive while 63 tested negative (Table 3). Of those reporting positive PCR, 4/13 (44%) had positive IgM LFA, 9/13 (69%) had positive IgG, and 11/13 (85%) had positive Ortho test. Of those reporting negative PCR, 59/63 (94%) tested IgM negative, 61/63 (97%) tested IgG negative, and the same number (97%) tested Ortho negative. These limited data are in line with manufacturer-reported estimates of sensitivity and specificity. Importantly, the reliance on self-reported data must be interpreted with caution, and there was no ability to account for the time since infection, which could impact the sensitivity calculations.
The gold-standard in antibody quantification is the ELISA assay for its flexibility in antigen diversity and quantification methodology using monoclonal antibodies to generate standard curves. We designed in-house ELISA assays against S1 and NP specific to SARS-CoV-2 antibodies, since these antigens have been described to elicit the most immunogenic response to infection based on SARS-CoV and MERS research. We analyzed all 77 samples that were positive for any serological assay and 30 random samples that were negative for all serological assays as controls for S1 and NP antibodies. Surprisingly, S1 antibody quantification showed a median value of 73.8µ g/mL for seropositive samples compared to 45.8µ g/mL for seronegative controls (Fig. 2C) indicating moderate antibodies against S1 epitopes. Similarly, NP antibody quantification showed a median value of 46.6 ng/mL for seropositive samples compared to 31.9 ng/mL for seronegative controls, also indicating moderate antibodies against NP epitopes. However, there was ≥ 100-fold range of antibody values for seropositive samples in each ELISA test, suggesting that some of the seropositive samples, but not all, were significantly reactive in S1 and NP ELISA, which is highly predictive of neutralizing activity. Spearman’s correlation analysis of all five tests showed a high degree of positive association between ELISA, HTSAs and IgG LFA tests while IgM LFA test was negatively correlated (Fig. 2E). Thus, we hypothesized that samples reactive for either the IgG LFA and/or the Ortho HTSA may have higher ELISAs values than samples that were reactive only for IgM LFA test.
To investigate this, we evaluated seropositive donors that were reactive to either the Ortho HTSA assay (Ortho+), the IgG LFA (IgG+) or only the IgM LFA (IgM+ only). As expected, the median Ortho HTSA value was 104 higher for both the Ortho+ and IgG+ groups compared to the IgM+-only group (158.5 and 114.8AU vs 0.02AU, respectively) (Fig. 2E). Similarly, both S1 and NP ELISAs showed significantly higher median antibody concentrations for the Ortho+ and IgG+ groups than for the IgM+-only group (S1; 467.2µ g/mL and 363.9µ g/mL versus 60.5µ g/mL and NP; 320.4 ng/mL and 204.8 ng/mL versus 39.4 ng/mL) (Fig. 2F, 2G). Importantly, these results conclude that IgG LFA and Ortho HTSA assays, but not the IgM LFA assay, correlate with immunogenic antibodies specific to SARS-CoV-2 as detected by ELISA.