Accurate SARS-CoV-2 seroprevalence surveys require robust multi-antigen assays

There is a plethora of severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) serological tests based either on nucleocapsid phosphoprotein (N), S1-subunit of spike glycoprotein (S1) or receptor binding domain (RBD). Although these single-antigen based tests demonstrate high clinical performance, there is growing evidence regarding their limitations in epidemiological serosurveys. To address this, we developed a Luminex-based multiplex immunoassay that detects total antibodies (IgG/IgM/IgA) against the N, S1 and RBD antigens and used it to compare antibody responses in 1225 blood donors across Greece. Seroprevalence based on single-antigen readouts was strongly influenced by both the antigen type and cut-off value and ranged widely [0.8% (95% CI 0.4–1.5%)–7.5% (95% CI 6.0–8.9%)]. A multi-antigen approach requiring partial agreement between RBD and N or S1 readouts (RBD&N|S1 rule) was less affected by cut-off selection, resulting in robust seroprevalence estimation [0.6% (95% CI 0.3–1.1%)–1.2% (95% CI 0.7–2.0%)] and accurate identification of seroconverted individuals.


Supplementary
. Secondary Antibody Dilution Optimization. A SARS-CoV-2 positive serum with high antibody levels (positive high -circles), a SARS-CoV-2 positive serum with low antibody levels (positive low -squares), a negative serum (triangles) and a blank sample (serum diluent only) were tested at four two-fold serial dilutions of the secondary antibody. For each positive and negative sample, the signal-to-noise ratio against the blank sample was calculated for N (black), S1 (orange) and RBD (blue) and plotted against the antibody dilutions. Black lines represent the mean of the three antigens in each condition.

S1.2 Analytical assay performance
For assessment of analytical assay performance (intra-and inter-assay variability) as well as for quality control purposes, we generated two control samples, a positive control made from a pool of sera from PCR-confirmed COVID-19 cases and a negative control made from a pool of negative sera. These samples were tested in quadruplicates in the same and in separate 96-well plates at different days and MFI values were used to calculate the coefficient of variation (CV%) for intra-and inter-assay variability. Also, the ratio of the average MFI values of the positive against the negative control for the SARS-CoV-2 N, S1 and RBD antigens was calculated and used as an indicator for the proper performance of the assay procedure by the end-user. Assay performance was considered successful when a ratio of three or greater for all three antigens was achieved. CV% values for the three SARS-CoV-2 antigens were below 20% across all isotypes (Supplementary Table S1). The IgG assay performed better across all antigens, while the highest variability was observed in the antigens from the endemic coronaviruses in the total IgG/IgM/IgA assay. Table S1. Intra-and Inter-assay variability, measured as % coefficient of variation.

S1.3 Antibody responses to SARS-CoV-2 and endemic human coronaviruses in clinical samples
A total of 155 serum samples from 77 PCR-confirmed COVID-19 cases and 78 pre-epidemic individuals were screened for the existence of reactive antibodies against antigens of the five different coronaviruses (Supplementary  Table S2). SARS-CoV-2 infection was deemed asymptomatic and mild needing no hospitalization in 8% and 60% of the cases, respectively, whereas 32% of the participants required hospitalization. Antibody detection was performed at a median of 46 days (range 13-87) post SARS-CoV-2 infection. Supplementary Figure S3 shows the reactivity of sera to each antigen for the different antibody isotypes. Antibody levels against the SARS-CoV-2 antigens N, S1 and RBD were significantly higher in the SARS-CoV-2 infected population as compared to samples from non-infected individuals (p<0.001, Supplementary Table S4); this finding was true for all isotypes. The analyzed blood samples from SARS-CoV-2 tested positive donors also showed significantly high antibody titers against the HCoV-OC43 S1/S2 antigens in all isotypes and against the HCoV-HKU1 S1 antigen in IgG detection, possibly as a result of antibodies cross-reactivity against S proteins that are conserved between SARS-CoV-2, HCoV-HKU1 and HCoV-OC43 coronaviruses.
Additionally, a power calculation was conducted to investigate the ability of the assay to discriminate between seropositives and sero-negatives, using a t-test with a significance threshold of 0.001. The effect size based on the difference between the means of the negative and positive distributions, was calculated using Cohen's d (= 1 ̅̅̅̅− 2 ̅̅̅̅ ), in which the difference in means is divided by the pooled standard deviation of the two independent samples. As it is expected by observing Supplementary figure S3, there is a large effect size, resulting in a 100% power. This means that the single antigen readouts of the multiplex assay can easily distinguish sero-positive from sero-negative cases given the sample size (Supplementary Table S3).
We further assessed the correlation of the normalized MFI values between antigens in both positive and negative cases (Supplementary Table 5). The SARS-CoV-2 S1 and RBD antigens-related data correlated strongly across all Ig isotypes in COVID-19 positive samples with Pearson's correlation coefficient (r) values > 0.96; a high correlation was also observed between N and S1 or RBD readouts in IgG and total antibody assays (r values between 0.72-0.82) in COVID-19 positive samples. Readouts between the N and S1 or RBD antigens correlated poorly in IgA and IgM detection. Results from negative samples showed no correlation across all antigen comparisons and Ig isotypes with the notable exception of the S1 and RBD antigens in IgM isotype (r=0.78).  Supplementary Table S3. Power calculation regarding the ability of the assay to discriminate between the antibody levels of sero-positives and sero-negatives, using the N, S1 and RBD single antigen readouts. The number of positives is n = 77 and the number of negatives n = 78.  Table S4. Antibody responses to coronavirus antigens in SARS-CoV-2 positive and negative cases.

S1.4 Correlation of SARS-CoV-2 antibody responses to antibody levels from other infectious agents
For assessing interfering factors, we used results from 66 pre-epidemic fully characterized sera for the presence of antibodies against Cytomegalovirus (CMV), Epstein-Barr virus (EBV), Hepatitis A (HAV) and B (HBs) and Toxoplasma (Toxo) (Supplementary Table S6). CMV IgG levels positively correlated with SARS-CoV-2 S1 IgG and total (IgG/IgM/IgA) antibodies (p<0.01) but showed no significant correlation with antibody levels against N or RBD antigens (Supplementary Table S6). No correlation was observed between SARS-CoV-2 reactive antibodies and antibodies against the other infectious agents tested.

S1.5 Receiver operating characteristic curve analysis
For each antibody ROC analysis is performed for the three (N, S1, RBD) antigens. The confidence intervals are calculated using Wilson's method. The AUC (area under the curve) of the ROC curves was calculated and used as a metric to evaluate the diagnostic performance of single antigen readouts (Supplementary Figure S4 & Supplementary  Table S7).

S1.6 Seroprevalence and agreement rates of single and multi-antigen rules in the population screening
Seroprevalence rates calculated using the best performing multi-antigen rules for different cut-off values are presented in Supplementary Figure S5. The agreement rates between all multi-antigen rules and between rules and commercial tests are presented in Supplementary Figures S6 & S7.