4. Discussion
In a rather short period, various assays for the detection of SARS-CoV-2 antibodies were introduced to the market. A meta-analysis of 38 studies on the performance of different format SARS-CoV-2 antibody tests mainly manufactured from Chinese companies was reported recently [23]. Furthermore, studies on the sensitivity and specificity of SARS-CoV-2 IgG tests, including ELISAs from the Epitope, Euroimmun, and Mikrogen companies, were published previously [20,21,22,24]. A recent prospective study examined the sensitivity of seven antibody tests from Abbott, Euroimmun, Mediagnost, Novatec, Virotech, Roche, and Siemens companies but did not provide data on their specificities or on the correlation of the assay results to the presence of SARS-CoV-2-neutralizing antibodies [25].
Here, we analyzed the diagnostic performance of eight commercially available tests, all of which have been released within the past few months. These were compared to identify appropriate tests to study the kinetics of N- or S-specific SARS-CoV-2 IgG or total antibodies in outpatients with a previous SARS-CoV-2 infection. To our knowledge, such a comprehensive study has not been conducted before. Three assays (Abbott, DiaSorin, and Roche) were applied in a random-access manner, which reduced the needed hands-on-time markedly. Convalescent sera from 26 SARS-CoV-2-infected patients were used to investigate the test sensitivities. These samples were characterized by a previous positive SARS-CoV-2 PCR result of the donors (reference method 1) and may possess SARS-CoV-2-neutralizing capacities (>1:10), as demonstrated in a laboratory-developed PRNT (reference method 2). Furthermore, two versions of an immunoblot, including the determination of SARS-CoV-2 IgG avidities, were applied.
With respect to the previous positive SARS-CoV-2 PCR result considered as the reference of first choice, following a recent suggestion [19], assay sensitivities were found to be relatively high and varied from 80.8% to 96.2% (Table 1). This finding is in accordance to other studies [21,22,24,25,26] and to a meta-analysis [23]. Six (23.1%; five, 19.2%, if the positive retesting result of the initial serum from patient 4 is taken into account; for details, see below) of the 26 SARS-CoV-2-infected patients did not develop SARS-CoV-2-neutralizing antibodies, as demonstrated by a PRNT ≤1:10. Among them were two individuals who were only recognized as IgG-positive in the Abbott and Virion-Serion assays and were partially reactive in the ViraChip® and the IgG immunoblots (Figure 1). This confirms a previous report of a high sensitivity of the Abbott IgG test [27]. The absence of neutralizing antibodies in several COVID-19 patients was also reported by others [28] and may be associated with the mild course of the infection mainly observed in outpatient patients. However, since no clinical data were available to us, a severe underlying disease or a serious course of the SARS-CoV-2 infection cannot be ruled out in the individuals of our study collective. Recently, presence of neutralizing antibodies was found to be associated with protection against reinfection [29]. If the PRNT was chosen as another reference (Figure S1), sensitivities varied from 88.9% to 96.3% (Table 1), which is largely comparable to the results obtained with the first method. These sensitivities can be even higher, as retesting of the initial serum from patient 4 in the second part of our study revealed the presence of low-level neutralizing antibodies (1:10 vs. 1:20–1:40; Figure 5 and Figures S1 and S2). With respect to a sample from patient 1 taken six days after the PCR, two S-based assays failed to detect SARS-CoV-2 IgG, while all assays that used the N as an epitope were reactive (Figure 1). Thus, usage of the latter may lead to the detection of IgG/total antibodies at an earlier stage post-infection. Furthermore, a lack of reactivity in some assays using the S antigen does not necessarily mean that SARS-CoV-2-neutralizing antibodies are missing (Figure 1). Seroconversion was clearly demonstrated in three patients (Figure 1). The majority of SARS-CoV-2 IgG in sera from SARS-CoV-2-infected patients were verified by both versions of an immunoblot (Figure 1). The latter can be used to differentiate the antibody reactivities against individual SARS-CoV-2 antigens. The specificities of the SARS-CoV-2 IgG/total antibody tests are at comparatively high levels between 96.0% and 100.0%, as has also been reported by others [21,22,24,26]. Interestingly, one sample collected in the winter 2018/2019 was found to be reactive in the Mikrogen, Roche, and Viramed assays, as well as in the prototype immunoblot, but could not be confirmed by the PRNT. For all other 99 archived samples, a random pattern of rare, isolated reactivity was demonstrated, and none was confirmed as possessing SARS-CoV-2-neutralizing antibodies by PRNT (Figure 2 and Figure S1). Potential cross-reactivity with the HCoVs responsible for common colds may therefore not be a major problem. The analysis of the immunoblot results shows, however, that sera from SARS-CoV-2-infected patients may simultaneously react with the antigens from HCoVs 229E, HKU-1, NL63, OC43, and SARS-CoV-2 (Figure S3). Due to the currently estimated low prevalence of SARS-CoV-2 IgG in the overall German population, even high specificities of 96–98% would produce a relevant number of false positive results. Thus, such assays, which all show an accuracy over 93%, should be preferentially used for the testing of patients with a history of a probable infection. In the case of doubt, the implementation of the labor-intensive and time-consuming PRNT may be considered.
The kinetics of SARS-CoV-2 IgG was studied up to 165 days after positive SARS-CoV-2 PCR results by the analysis of consecutive samples from 11 patients and samples from five routine patients; four are family members of a SARS-CoV-2-infected patient. The results were obtained with two IgG assays based on the N (Abbott) or the entire S (Virion-Serion) antigens as the epitopes. Both tests turned out to be the most sensitive and specific assays in the validation part of our study (compare in Table 1). Thereby, SARS-CoV-2-specific antibodies were detectable in all individuals. While the N-based IgG assay revealed the existence of two groups of individuals, and one of them showed a rapid decrease of IgG indices within a few months, the IgG reactivity in the S protein-based assay was found to be more stable (Figure 4). This is in-line with a previous preprint [16] and could suggest that the latter test is better suited for epidemiological studies, especially when it comes to questions of prevalence. Accordingly, the majority of SARS-CoV-2-infected patients in our study developed neutralizing antibodies. The titers were stable during the observation period but were in a relatively low range between >1:10 and 1:80 (Figure 5). Only three patients showed conversion from low to high SARS-CoV-2 IgG avidity (Figure S3). The limited number of included SARS-CoV-2 patients, however, limits the informative value of the present study. The evaluability of the simple and reproducible PRNT technique applied here may be improved by overlaying the cells with cellulose or by using specific antibodies to detect the remaining viral antigens in the cells, as reported previously [7,20]. Furthermore, the application of a modified cytopathic effect inhibition assay [30] may be useful to facilitate standardization and quantification and will be valued in future studies. Several SARS-CoV-2-infected patients and routine patients in our study lost their SARS-CoV-2 (N)-specific IgG within a few months or could lose them soon. These individuals can theoretically be prone to reinfection, especially since the S-specific IgG and the neutralizing antibodies are stable but only present in low indices/titers. In a few patients, both were undetectable. However, this view does not take into account the booster effect that can be expected in the event of renewed virus contact. This would most likely lead to a rapid increase in antibody titers. Accordingly, vaccine strategies against SARS-CoV-2 may need boosting to maintain sufficient neutralizing antibody titers over a long time [5]. Waning antibody titers and reinfections are known from various HCoVs mainly responsible for common colds [5,12,31]. Recently a well-documented case of an asymptomatic SARS-CoV-2 reinfection was reported [32]. In our opinion, the importance of such rare events for the progression of the temporary pandemic appears to be minor. Nevertheless, a permanent presence of measurable SARS-CoV-2 IgG in the serum of all patients is unlikely. This is not surprising and has also been observed in patients after infections with SARS-CoV (2002/2003) or Middle East respiratory syndrome (MERS)-CoV [5]. It does not necessarily mean that these individuals will be susceptible to severe reinfection.
Future studies should now address the virus-specific antibody response on the mucous membranes, as well as the cellular immune response. For this, the development of valid routine test procedures is urgently required.