Development and Validation of the Elecsys Anti-SARS-CoV-2 Immunoassay as a Highly Specific Tool for Determining Past Exposure to SARS-CoV-2

The Elecsys Anti-SARS-CoV-2 immunoassay (Roche Diagnostics) was developed to provide accurate, reliable detection of antibodies to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). We evaluated sensitivity, specificity, cross-reactivity, and agreement with a vesicular stomatitis virus-based pseudoneutralization assay for the Elecsys Anti-SARS-CoV-2 immunoassay. Sensitivity and agreement between Elecsys Anti-SARS-CoV-2 immunoassay and pseudoneutralization assay measurements were evaluated using samples from patients with PCR-confirmed SARS-CoV-2 infection, a majority of whom were hospitalized.

I n December 2019, reports emerged of patients presenting with pneumonia of unknown etiology in Wuhan, China (1,2). This disease was subsequently shown to be caused by a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and was designated coronavirus disease 2019 (COVID-19) (3)(4)(5). Since then, the COVID-19 outbreak has rapidly developed into a pandemic, which has infected millions of people and posed critical challenges for governments and health care systems around the world (6).
SARS-CoV-2 is an enveloped, single-stranded RNA virus of the Coronaviridae family. All coronaviruses share similarities in the organization and expression of their genomes, which encode 16 nonstructural proteins and 4 structural proteins: the spike, envelope, membrane, and nucleocapsid antigens (5,(7)(8)(9). Evidence to date suggests that SARS-CoV-2 is transmitted between people primarily through respiratory droplets and contact routes, although indirect transmission via contaminated surfaces is also possible (10)(11)(12). Infected individuals may exhibit a variety of symptoms, including fever, cough, and breathlessness, and disease severity can range from asymptomatic/mild cases to severe disease and death (13,14).
There is an urgent unmet clinical need to more effectively determine SARS-CoV-2 seroprevalence in the general population in order to improve our understanding of virus circulation dynamics, gain a more accurate estimate of the mortality rate from COVID-19, and identify individuals at risk of infection. Serological assays for SARS-CoV-2 have been suggested as a potential tool to help identify the extent of virus exposure in a given population and thereby indirectly provide information on the appropriate application, enforcement, or relaxation of containment measures (15)(16)(17)(18). Serological assays may also help elucidate a potential correlate for immunity following infection (15,16).
Recent evidence suggests that most SARS-CoV-2 convalescent individuals have detectable neutralizing antibodies (nAbs) for the virus (19,20). Due to affinity maturation, the binding strength of antibodies increases over time following infection or vaccination (21). High-affinity nAbs are critical for the control of infection, since they can recognize and bind to specific viral epitopes, thereby "neutralizing" the virus and rendering it nonpathogenic (20,22). Previous studies involving commercially available anti-SARS-CoV-2 immunoassays have found a positive correlation between antibody titration results from pseudoneutralization assays and SARS-CoV-2 nAbs; however, further investigation is warranted (23,24).
The timing of seroconversion is crucial for determining optimum time points for sample collection for serological testing (25). Although the picture is rapidly developing and robust serology data are not yet available, the kinetics of antibodies to SARS-CoV-2 have begun to be described. Based on current evidence, immunoglobulin M (IgM) antibodies are detectable within 5 days after symptom onset and immunoglobulin G (IgG) antibodies within 5 to 7 days (26)(27)(28). There is a paucity of data on immunoglobulin A (IgA), but it appears to be observable approximately 3 to 6 days after symptom onset (15,27). Depending on the method applied, seroconversion is observed after a median of 10 to 13 days after symptom onset for IgM and 12 to 14 days for IgG; maximum seroconversion occurs at 2 to 3 weeks for IgM, 3 to 6 weeks for IgG, and 2 weeks for total antibodies (20,(28)(29)(30)(31)(32). The levels and chronological order of IgM and IgG antibody appearance are highly variable, supporting the detection of both antibodies simultaneously (17,28,29,31).
The Elecsys Anti-SARS-CoV-2 immunoassay (Roche Diagnostics International Ltd, Rotkreuz, Switzerland) was developed to provide an accurate and reliable method for the detection of antibodies to SARS-CoV-2, in order to facilitate population screening with high specificity and the identification of past infection status as a potential correlate for subsequent immunity. We aimed to evaluate the sensitivity, specificity, and cross-reactivity of the Elecsys Anti-SARS-CoV-2 immunoassay, in addition to agreement with results from a pseudoneutralization assay.
Determinations were performed as single measurements. For sensitivity and specificity, point estimates and 95% CIs were calculated using the Roche Diagnostics bioWARP tool (Roche Diagnostics GmbH, Penzberg, Germany). For contingency between Elecsys Anti-SARS-CoV-2 immunoassay and VSV-based pseudoneutralization assay results, PPA, PNA, and POA were calculated using the Westgard QC 2ϫ2 Contingency Calculator (Westgard QC Inc., WI, USA).
Data availability. Qualified researchers may request access to individual-patient-level data through the clinical study data request platform (https://vivli.org/). Details on Roche's criteria for eligible studies are available at https://vivli.org/members/ourmembers/. Further details on Roche's Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents are available on the company website.

Sensitivity.
A total of 496 samples from 102 patients with prior PCR-confirmed SARS-CoV-2 infection were included in the sensitivity analyses. All positive-sample donors (anonymized and aggregated data) were adults (mean [standard deviation {SD}] age, 66 [16.4] years) and had an average length of stay in hospital of 18 (SD, 14.9) days. Overall, 15.7% of the sample donors were admitted to an intensive care unit during the hospital stay. Sensitivity increased with time after PCR confirmation: 60.2% (95% CI, 52.3 to 67.8%) at 0 to 6 days, 85.3% (95% CI, 78.6 to 90.6%) at 7 to 13 days, and 99.5% (95% CI, 97.0 to 100.0%) at Ն14 days (Table 1; Fig. 1). Figure 2 shows Elecsys Anti-SARS-CoV-2 immunoassay results for 26 consecutive samples from five patients following recovery from PCR-confirmed SARS-CoV-2 infection. Patients had a range of SARS-CoV-2 symptoms, including one patient who reported no symptoms. In all patients with symptoms, cutoff index values were Ն20 following a negative PCR result and increased over time, out to day 40. In the patient with no symptoms, cutoff index values ranged from 0.99 to 1.97 following a negative PCR result.
Agreement between the Elecsys Anti-SARS-CoV-2 immunoassay and neutralization testing (VSV-based pseudoneutralization assay). In total, 47 samples were evaluated using the Elecsys Anti-SARS-CoV-2 immunoassay and a VSV-based pseudoneutralization assay for the purpose of method comparison. One sample was positive on the Elecsys Anti-SARS-CoV-2 immunoassay but indeterminate using both the VSV-based pseudoneutralization assay and in-house immunofluorescence assays, so it was excluded from the analysis. Of the remaining 46 samples, 38 tested positive on both the Elecsys Anti-SARS-CoV-2 immunoassay and the VSV-based pseudoneutralization assay (Table 4). Two samples were negative on both assays, with very low cutoff indexes observed for measurements taken using the Elecsys Anti-SARS-CoV-2 immunoassay (0.067 and 0.082).  Six samples that tested positive using the VSV pseudoneutralization assay were negative on the Elecsys Anti-SARS-CoV-2 immunoassay; these samples had elevated cutoff indexes (0.206 to 0.659) compared with the two samples confirmed negative on both assays. When retested using an in-house immunofluorescence assay, two of these samples were found to be positive and four negative; all results were close to prespecified cutoffs.   The PPA between the Elecsys Anti-SARS-CoV-2 immunoassay and the VSV pseudoneutralization assay was 86.4% (95% CI, 73.3 to 93.6%); the PNA was 100% (95% CI, 34.2 to 100%); and the POA was 87.0% (95% CI, 74.3 to 93.9%).

DISCUSSION
The COVID-19 pandemic has resulted in significant mortality and morbidity and has created major challenges for governments and health care systems. Due to the novelty of the causative agent, SARS-CoV-2, and the rapidity with which it has spread, there is an urgent need for serological assays, which could be used to determine the seroprevalence in a given population and help identify a potential correlate for immunity secondary to exposure (15)(16)(17)(18). This urgency and the lack of existing available tests prompted the U.S. Food and Drug Administration to allow manufacturers of serological assays for SARS-CoV-2 to bypass the normal approval process; as a result, the market has been inundated with tests, many of which have not been sufficiently reviewed by regulatory authorities nor their performance reported in the peer-reviewed literature (16). However, it is vital that any new test be adequately evaluated and validated to demonstrate that it is reliable and accurate for its intended purpose.
To enable population screening to determine the level of exposure and to identify individuals who may be immune following infection, an ideal serological test for SARS-CoV-2 should meet several key requirements: (i) a high general specificity with a small CI, (ii) no cross-reactivity with other endemic coronaviruses, (iii) preference for the detection of mature antibodies to provide the potential for correlation with neutralizing activity, and (iv) a high sample throughput to meet the huge demand for testing. The first three requirements can be achieved through the application of an appropriate assay format and antigen selection, while the last requirement relies on a highthroughput platform and upscaling of laboratories. Although a high assay sensitivity is desirable, it is of secondary importance and should not be pursued at the expense of specificity for past infection, since a serological assay is less likely to be used for the diagnosis of active infection.
The Elecsys Anti-SARS-CoV-2 immunoassay was specifically designed to meet these requirements. It utilizes a double-antigen sandwich test principle for the detection of high-affinity (i.e., late-onset/mature) antibodies to SARS-CoV-2. In the present study, the Elecsys Anti-SARS-CoV-2 immunoassay demonstrated an overall specificity of 99.80% (95% CI, 99.69 to 99.88%) in 10,453 residual samples from routine testing and blood donors. The sensitivity of the Elecsys Anti-SARS-CoV-2 immunoassay in samples from patients with prior PCR-confirmed SARS-CoV-2 infection increased with time after PCR confirmation, reaching 99.5% (95% CI, 97.0 to 100.0%) at Ն14 days. The performance of the Elecsys Anti-SARS-CoV-2 immunoassay is comparable to or better than that observed for other serological SARS-CoV-2 assays (specificity, 94.8 to 99.9%; sensitivity at Ն14 days post-PCR confirmation, 75.0 to 100.0%) (35,36).
The main risk with the use of serological SARS-CoV-2 assays for population screening is the possibility of false-positive results, which could lead to the erroneous assumption of past infection and subsequent putative immunity and thus could put the individual at risk of acquiring or transmitting infection (16). A very high specificity is crucial to reduce the rate of false-positive results, particularly in populations with low seroprevalence, where small differences in assay specificity can result in substantial differences in the positive predictive value (PPV) (15,16). For example, if the seroprevalence of SARS-CoV-2 in a given population was 10%, a serological test with a sensitivity of 83.1% and a specificity of 98.3% would have a PPV of 84.2%; however, if the seroprevalence was 1%, the respective PPV would drop to 32.6% (37). Understanding this relationship between seroprevalence, specificity, and PPV is crucial for population screening using serological testing, and it has been suggested that assay specificity should be Ͼ99% (with a CI of 99.0 to 99.9%) to ensure a sufficient PPV in populations with low seroprevalence (16). Based on a sensitivity of 99.5% and a specificity of 99.8%, the PPV of the Elecsys Anti-SARS-CoV-2 immunoassay in populations of 1%, 5%, 10%, and 20% seroprevalence would be 83.4%, 96.3%, 98.2%, and 99.2%, respectively. A potential cause of false-positive results is cross-reactivity with other analytes. In the present study, only 4/792 samples containing potential cross-reacting analytes showed reactivity with the Elecsys Anti-SARS-CoV-2 immunoassay; importantly, no cross-reactivity was observed for the coronavirus panel (containing strains 229E, NL63, OC43, and HKU1). This resulted in an overall specificity of 99.5%.
Based on our understanding of other respiratory viruses, it has been suggested that the presence of anti-SARS-CoV-2 antibodies will provide some immunity, although it is unclear how long any protection would last after the initial infection (39). Assuming that anti-SARS-CoV-2 antibodies do offer some immunity from further infection, donated plasma from patients who have recovered from COVID-19 may represent a potential therapy by conferring immunity on the recipient (39). In this context, serological assays could play another important role in the identification of potential convalescent plasma donors, particularly individuals with very high anti-SARS-CoV-2 antibody titers (16). Recent studies suggest that both cell-mediated and humoral immune responses are likely to play a protective role in SARS-CoV-2 infection, and the spike and nucleocapsid antigens in particular have been shown to be highly immunogenic and abundantly expressed during infection (7)(8)(9)40). Antibodies targeting these proteins are formed as early as 9 days after symptom onset and have demonstrated a strong neutralizing response, suggesting that seroconversion may lead to immunity for a limited time after infection (15,17,41). However, further research is needed to demonstrate the correlation between the presence of anti-SARS-CoV-2 antibodies and neutralization, as well as, ultimately, the correlation with clinical immunity. Additional work is also required to establish whether there is a correlation between anti-SARS-CoV-2 antibody titers and disease severity and/or prognosis (42).
A major strength of this study is the large number of seronegative samples (n ϭ 10,453) used to determine the specificity of the Elecsys Anti-SARS-CoV-2 immunoassay, which ensures that the study is robustly powered and the reliability of the specificity point estimates is very high (as indicated by the small CIs) (16,43). To our knowledge, this is one of the largest sample sizes used to date to evaluate the specificity of a serological assay for SARS-CoV-2. Another study strength is the inclusion of a high number of confounder samples in the cross-reactivity cohort, which comprised a total of 792 samples across 36 different indications. A limitation is that this was a single-center study, and our results should be confirmed by additional assessments at other study sites. Further clinical data on the samples were not available due to data regulations, and thus, it was not possible to analyze specific subcohorts according to age, disease severity, onset of symptoms, etc. In addition, a comparably small sample size was utilized to measure agreement between the Elecsys Anti-SARS-CoV-2 immunoassay and a pseudoneutralization assay; as such, these data would benefit from further evaluation and validation. We are confident that our results regarding the sensitivity of the Elecsys Anti-SARS-CoV-2 immunoassay and the time course of antibody response are of general value. However, the findings should be interpreted with some caution. Although the samples used in this study were not selected on the basis of patient criteria, the majority of samples were drawn from hospitalized patients and therefore probably represent more-severe cases of COVID-19. Furthermore, the availability of samples depended on the need for consequent routine clinical chemistry diagnosis after PCR confirmation of SARS-CoV-2 infection. Again, it can be assumed that extensive subsequent diagnosis was performed predominantly in cases with moresevere disease. For an asymptomatic patient assessed following recovery from PCRconfirmed SARS-CoV-2 infection in this study, cutoff index values remained close to the positive/negative threshold for anti-SARS-CoV-2 antibodies. Therefore, the validity of our findings in ambulatory settings or for patients with asymptomatic/mildly symptomatic SARS-CoV-2 infection has yet to be shown and requires further study.
Conclusion. The Elecsys Anti-SARS-CoV-2 immunoassay demonstrated a very high specificity of 99.80% and a sensitivity of 99.5% for past infection in patients at Ն14 days after PCR confirmation, supporting its use as a potential tool for the identification of past exposure to SARS-CoV-2 infection. High assay specificity and sensitivity are crucial to ensure a high PPV for population screening, particularly in settings with low disease prevalence.