Cross-Comparison of a Chemiluminescent Platform and a Commercial Receptor Binding Domain–Based ELISA for Detecting SARS-CoV-2 IgG
Year: 2020
Type: article
Abstract: The coronavirus 2019 (COVID-19) pandemic has rapidly emerged as a defining public health crisis of the 21st century (1,). Not surprisingly, laboratory testing has played a vital role in diagnosing cases and tracking prevalence. Among different techniques, serologic testing detects the body’s antibody response to infection, as opposed to direct detection of the virus (2). Serodiagnostic approaches remain dynamic, with a growing number of products that differ in terms of methodology (e.g., lateral flow, ELISA, chemiluminescence), performance setting (e.g., central laboratory, point of care), antigenic target (e.g., nucleocapsid, total spike protein S1, receptor binding domain), instrumentation requirements, and regulatory status. In addition to products with “research use only” status, serology tests with approval by the US Food and Drug Administration (FDA) carry only emergency use authorization, as the pandemic has precluded premarketing trials. A further complication is how results should be acted on clinically, as it remains unresolved whether the presence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antibodies—even when detected accurately—protect against reinfection. Beyond clinical significance, additional questions surround the analytic performance of the tests themselves. Concerns have been raised regarding unacceptably high rates of false positivity for some assays, potentially because of cross-reactivity with other coronaviruses (3,). Local epidemiology presents more challenges because high-specificity techniques can demonstrate poor positive predictive value in low-prevalence settings (4). Among the first COVID-19 serodiagnostics to receive emergency use authorization was the Abbott Architect IgG assay. This automated chemiluminescent technology has demonstrated high sensitivity and specificity for the analysis of serum or plasma from PCR-diagnosed patients, along with prepandemic specimens (5, 6). In the context of on-site serology implementation, we evaluated the performance of the Architect together with a second candidate assay: the GenScript SARS-CoV-2 Spike S1-RBD IgG ELISA, a research-use-only methodology that would require local validation to meet CLIA and FDA requirements. Parallel Architect and ELISA analyses were conducted for several adult cohorts (with IRB approval). To assess specificity, cryobanked prepandemic specimens were used: 230 serum (from activating serum separator tubes) and 238 plasma (from spun EDTA tubes). To assess sensitivity, specimens were obtained from unique adults who were positive for SARS-CoV-2 by PCR ≥17 days after symptom onset (5). These included 21 paired plasma/serum specimens from recovered donors who never required hospitalization and 25 plasma-only specimens from hospitalized patients (derivatives of care). An additional 31 plasma samples were analyzed from other unique hospitalized patients, but these specimens were considered separately because the patients were <17 days from symptom onset. The Architect assay and ELISA were performed according to manufacturer instructions, including analysis of undiluted serum/plasma by Architect and 1:100 dilutions by ELISA. Categorical positive and negative results were likewise assigned per manufacturer thresholds. For the Architect, seropositivity corresponded to S/C ≥1.4. For ELISA, values for absorbance at 450 nm (A450) were assessed according to both the definition of a “positive” result when GenScript first released the kit (A450 >0.008-µg/mL standard; threshold 1) and the revised definition in current instructions (A450 >0.15;, threshold 2). The results of these experiments are summarized in Fig. 1. The Architect demonstrated 100% specificity for serum (230/230; 95% CI, 98.4%–100%) and 98.8% for plasma (235/238; 95% CI, 96.4%–99.6%), whereas ELISA specificity for these specimens was lower. For the ELISA, applying threshold 1, the specificity of serum and plasma were 81.3% (187/230; 95% CI, 75.8%–85.8%) and 87.2% (212/238; 95% CI, 84.5%–92.4%), respectively. Applying threshold 2, these values were 87.4% (201/230; 95% CI, 82.5%–91.1%) and 92.9% (221/238; 95% CI, 88.9%–95.5%), respectively. The observed ELISA specificity was numerically higher for plasma with both thresholds, but the underlying reason for this observation remains unclear. None of the ELISA cross-reactive specimens demonstrated cross-reactivity on the Architect, whereas cross-reactive Architect specimens were likewise negative by ELISA. Combining hospitalized and nonhospitalized patients who were ≥17 days postsymptomatic, Architect sensitivity was 90.5% (19/21; 95% CI, 71.1%–97.4%) for serum and 93.5% (43/46; 95% CI, 82.5%–97.7%) for plasma. Of note, 2 of 2 Architect false-negative results for serum and 2 of 3 false-negative results for plasma occurred with nonhospitalized donors (i.e., virtually all hospitalized patients demonstrated positivity). By comparison, ELISA sensitivity varied with the positivity threshold: with threshold 1, sensitivity was 85.7% (18/21; 95% CI, 65.4%–95.0%) for serum and 93.5% (43/46; 95% CI, 82.5%–97.7%) for plasma; with threshold 2, it was 66.7% (14/21; 95% CI, 45.4%–82.8%) for serum and 82.6% (38/46; 95% CI, 69.2%–90.9%) for plasma. Again, except for 1 inpatient plasma specimen, all ELISA false-negative results (either threshold) occurred with nonhospitalized donors. In this context, the sensitivity calculations presented for serum are qualified by the fact that these specimens were available only from nonhospitalized donors. As a result, we cannot draw any rigorous conclusions about whether plasma or serum might provide better sensitivity data for the ELISA. Finally, analyses were performed on separate plasma specimens from 31 inpatients who were <17 days postsymptomatic, in whom seropositivity was less predictable, yielding detection rates of 24 of 31 by Architect, 18 of 31 by ELISA at threshold 1, and 16 of 31 by ELISA at threshold 2. Summary of methodologic comparison. The 2 × 2 agreement tables are presented for analysis of serum and plasma on both the Architect IgG Assay and the GenScript IgG Assay with positivity thresholds 1 and 2. The reference comparator represents PCR-defined SARS-CoV-2 infection at ≥17 days after symptom onset (for positive specimens) and prepandemic origins (for negative specimens). For positive specimens, both nonhospitalized and hospitalized sources are combined in the tables. In addition to the number of patients and specimens that fall within each category, each square summarizes the median raw signal value (S/C for Architect, absorbance at 450 nm [A450] for GenScript) for that respective group or, when a limited number of data points precludes a meaningful value, the individual data points themselves. These data support the high sensitivity and specificity of the Architect assay, including for specimens with ELISA cross-reactivity. Additional studies are needed to clarify how observable seropositivity for PCR-confirmed and recovered individuals varies with disease severity (both by Architect and in general). In contrast, our results demonstrated less favorable performance characteristics for GenScript’s receptor binding domain–based IgG ELISA. In particular, institutions considering validation and implementation of this kit (either per the manufacturer’s instructions or with modified protocols or thresholds) may need to consider local COVID-19 prevalence to avoid high false positivity. All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. T.E. Antoine-Reid, statistical analysis, administrative support; J.E. Malone, statistical analysis; J.E. Schmitz, statistical analysis. Authors’ Disclosures or Potential Conflicts of Interest: No authors declared any potential conflicts of interest.
DOI: 10.1093/jalm/jfaa133
Institution: Vanderbilt University Medical Center
Language: English
Subfield: Infectious Diseases
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W3047994802