Kinetics and seroprevalence of SARS-CoV-2 antibodies: a comparison of 3 different assays

Comparing seroprevalence and antibody kinetics in three different commercially available assays for SARS-CoV-2. Serostatus of COVID-19 patients was analyzed 5 months and 10 months after their infection, using three different assays: Diasorin LIAISON, Euroimmun, Abbott Diagnostics ARCHITECT. Seropositivity at baseline differed significantly depending on the assay (Diasorin 81%, Euroimmun 83%, Abbott 59%). At follow-up antibody levels detected in the Diasorin assay were stable, while there was a significant loss in seropositivity in the Euroimmun and Abbott assays. There are significant differences in SARS-CoV-2 antibody kinetics based on the specific assay used.

Since the beginning of the SARS-CoV-2-pandemic 1 more than 112 Million cases and almost 2.5 Million 2 deaths have been reported worldwide. Serum antibody testing is becoming a critical tool, both in diagnosis of COVID-19 and in seroprevalence studies.
Most patients develop detectable antibodies within 14 days after their infection 3 . However, there is inconsistent evidence regarding the duration of antibody persistence with studies showing only short lived antibody responses 4 while others showing persistent serum levels 5 .
While differences in patient characteristics and disease course might explain some contradictory findings, assay dependent differences might also play a role.
We analyzed SARS-CoV-2 seroprevalence and antibody kinetics over 9 months in 109 individuals using three different commercially available assays.

Methods
Study design. Patients with PCR confirmed SARS-CoV-2 infection in Dresden (a city in Saxony/Germany with approximately 557,000 inhabitants) were invited via the local health department to participate in the Amb-CoviDD19 study.
After informed consent was obtained, 5 mL of peripheral venous blood was collected from each individual to assess SARS-CoV-2 IgG antibodies at baseline and 9-12 months after their infection (follow-up). Additional information about age, comorbidities, regular medication, COVID-19-symptoms, disease course and test indication were obtained.
The AmbCoviDD19 study was approved by the Ethics Committee of the Technische Universität (TU) Dresden (BO-EK-137042020) and has been assigned clinical trial number DRKS00022549.

Laboratory analysis.
We assessed SARS-CoV-2 IgG antibodies in all samples using three commercially available assays.
First, chemiluminescence immunoassay (CLIA) technology for the quantitative determination of anti-S1 and anti-S2 specific IgG antibodies to SARS-CoV-2 was used: Diasorin LIAISON SARS-CoV-2 S1/S2 IgG Assay. Antibody levels > 15.0 AU/mL were considered positive and levels between 12.0 and 15.0 AU/mL were considered equivocal.
Third, chemiluminescent microparticle immunoassay (CMIA) intended for the qualitative detection of IgG antibodies to the nucleocapsid protein of SARS-CoV-2, Abbott Diagnostics ARCHITECT SARS-CoV-2 IgG, www.nature.com/scientificreports/ was used. This assay relies on an assay-specific calibrator to report a ratio of specimen absorbance to calibrator absorbance. The interpretation of result is determined by an index (S/C) value, which is a ratio over the threshold value. An index (S/C) of < Ethics approval. The AmbCoviDD19 study was approved by the Ethics Committee of the Technische Universität (TU) Dresden (BO-EK-137042020) and has been assigned clinical trial number DRKS00022549.

Consent of participate.
Informed consent was obtained from all individual participants included in the study.

Consent of publication.
Patients signed informed consent regarding publishing their data.

+ Supplemental
Median time between second antibody assessment (follow-up) and infection was 295 days (9.8 months) ( Table 1). 101/109 (93%) participants had detectable antibodies in at least one assay. 89/92 (97%) with detectable antibodies at baseline in at least one assay continued to have detectable antibodies in at least one assay at  (Table 2). However, while 86/88 (98%) of participants with initially detectable antibodies in the Diasorin assay continued to have detectable antibodies in this assay, numbers were significantly lower for the Euroimmun (73/90 (81%) p 0.0004) and the Abbott (19/64 (30%) p = 0.0001) ( Table 4). In addition mean antibody levels were stable in the Diasorin assays over time while they dropped significantly in the Euroimmun and Abbott assays (Fig. 3).  www.nature.com/scientificreports/ At follow-up, 77/109 (71%) had detectable antibodies in at least two different assays ( Table 2); 71/88 (81%) individuals with two positive serological tests at baseline continued to have detectable antibodies in two assays; 15/17 (88%) individuals no longer tested positive in two different assays continued to have detectable antibodies in the Diasorin, while none of these continued to have detectable antibodies in the Euroimmun or Abbott. 24/109 (22%) had detectable antibodies in all three assays at follow-up (Table 2, Fig. 1).
Five out of 21 (24%) individuals without detectable antibodies in the Diasorin assay at baseline had detectable antibodies at follow-up, 7/19 (37%) in the Euroimmun and 7/45 (16%) in the Abbott respectively. Of the 17 individuals without detectable antibodies in any assay at baseline, 12 (71%) had detectable antibodies in at least one assay at follow-up (3 Diasorin/10 Euroimmun/3 Abbott). Two individuals reported a positive SARS-CoV-2 PCR testing between baseline and follow-up. Both had initially no detectable antibodies in any assay, one had antibodies in all three assays at follow-up, one had detectable antibodies in the Diasorin and Euroimmun. Five   www.nature.com/scientificreports/ individuals had no detectable antibodies in any assay at both times. Four out of five were tested by PCR based on symptoms alone, while 1/5 was tested due to travel history.

Discussion
The overall seroconversion rate in our study is comparable to large population based seroprevalence studies 6,7 , however the differences in seroprevalence detected by different assays are striking. While at baseline the Diasorin and Euroimmun provide comparable results (Fig. 1), the seroprevalence in the Abbott is 20% lower compared to the other assays. At follow-up these differences become even more pronounced. While the Diasorin shows stable persistent antibody levels, seropositive rates detected with Euroimmun and Abbott decrease significantly over time, leading to a possible underestimation of seropositivity of up to 60%. These differences might at least partly explain contradicting findings in longitudinal antibody kinetic studies. Given the intra-individual differences, patient characteristics seem unlikely to explain the discrepancies between the different assays. Differences in antibody persistence based on targeted epitopes might play a role comparing Diasorin and Euroimmun with Abbott. However, the significant difference between Diasorin and Euroimmun at follow-up-both detecting antibodies targeting the spike protein of SARS-CoV-2-requires further investigation. The possible role and longevity of IgA antibodies need to addressed in further studies as well.
In addition, antibodies to the nucleocapsid protein of SARS-CoV-2 are thought to be useful in differentiating between seroconversion after infection and vaccination. Given the substantial loss of these antibodies over time in our sample, this approach might be less feasible than expected.
It is remarkable that 71% of individuals without any detectable SARS-CoV-2 antibodies 4-5 months after their infection had measurable levels later at 9-10 months. One possible explanation could be that their initial short lived antibody response was boostered during the study period through repeat exposure. Further immunological studies-including T-cell assays-are needed to investigate this further. The observation that the only PCR confirmed repeat infections occurred in individuals without a detectable antibody response at baseline is somewhat reassuring and might point to a correlation between antibody response and immunity.
The lack of significant associations between seropositivity and patient characteristics in our study compared to previous 5,7 might be explained by the relatively long time period between infection and baseline serological testing as well as the rather homogenous study population with mild clinical courses. The significant differences in seropositivity based on test indication is of particular interest though and most likely due to an increased rate of false positive PCRs in populations with a lower pre-test-probability. While this association is well described in the literature 8 these results are an important reminder that even highly specific test as the SARS-CoV-2 RT-qPCR do have a measurable false positive rate 9 . This observation needs to be given special consideration when implementing population-based screening programs in asymptomatic individuals.

Conclusion
There are significant differences in SARS-CoV-2 antibody kinetics based on the specific assay used. In our cohort the Diasorin LIAISON assay performed best regarding long term detection of seropositivity. Diasorin had the lowest change in seropositivity between baseline and follow up. There is a significant difference in seropositivity in PCR-positive individuals based on the indication for PCR-testing.