Pediatric evaluation of clinical specificity and sensitivity of SARS-CoV-2 IgG and IgM serology assays

To the Editor,

There is urgent interest in the clinical application of serological testing for the detection of antibodies against SARS-CoV-2. However, available SARS-CoV-2 assay evaluations are primarily based on adults, resulting in evidence gaps in pediatric performance [1], [2]. Serology testing is anticipated to play an important role in the clinical evaluation of pediatric patients presenting with potential viral exposure, including delayed-onset conditions such as multi-inflammatory syndrome in children, and in community seroprevalence studies [3], [4]. Evidence suggests SARS-CoV-2 manifests differently in children relative to adults, with recent data reporting distinct antibody responses [5]. Thus, it is essential to characterize the performance of available serological assays in the pediatric population. The main objective of this study was to complete a clinical evaluation of the Abbott Architect and Alinity immunoglobulin M (IgM) and immunoglobulin G (IgG) assays in pediatrics, with a focus on clinical specificity.

A total of 1,496 pediatric samples from three cohorts were evaluated. Cohort 1: 500 sera collected from community children and adolescents as part of the Canadian Laboratory Initiative on Pediatric Reference Intervals (CALIPER) [6] from November 2018–March 2019 (true negative historical controls). Cohort 2: 500 sera collected from community children and adolescents as part of CALIPER from September 2019–February 2020 (specimens collected during the initial phase of the pandemic). Collection periods were selected to coincide with Canadian flu seasons when there is increased community circulation of respiratory viruses. Cohort 3: 496 residual sera collected from patients at The Hospital for Sick Children from May–June 2020 (confirmed RT-PCR negative). In addition to described cohorts, 22 PCR-positive pediatric and adult specimens from Mount Sinai Hospital and The Hospital for Sick Children were assayed for sensitivity assessment.

All specimens were tested on the Architect and Alinity SARS-CoV-2 IgG and IgM assays (Abbott Diagnostics). All assays use chemiluminescent microparticle technology for qualitative antibody detection, reporting as a signal to calibrator ratio (S/C). Alinity and Architect SARS-CoV-2 IgG assays are designed to detect antibodies against the nucleocapsid protein (S/C cut-off: >1.4), while Alinity and Architect SARS-CoV-2 IgM assays are designed to detect antibodies against the spike protein (S/C cut-off: >1.0). Analyzer calibration and quality control for all assays passed specifications prior to testing. Within-laboratory precision for positive control materials, as evaluated by assaying materials five times per day over a five-day period, were: 1.8% (mean S/C: 3.37) and 1.6% (mean S/C: 3.30) for Architect and Alinity IgG assays, respectively, and 1.9% (mean S/C: 3.83) and 2.5% (mean S/C: 2.94) for Architect and Alinity IgM assays, respectively.

All SARS-CoV-2 antibody results are presented in Figure 1. In Cohort 1, negative percent agreement (NPA) was 99.6% (498/500, 95% CI: 98.7–100%) for IgG and 99.8% (499/500, 95% CI: 98.9–100%) for IgM on both platforms. In Cohort 2, NPA was 100% (500/500, 95% CI: 99.3–100%) for IgG and 99.6% (498/500, 95% CI: 98.7–100%) for IgM on both platforms. In Cohort 3, NPA was 99.8% (484/485, 95% CI: 98.9–100%) on Architect and 99.6% (483/485, 95% CI: 98.5–100%) on Alinity IgM assays. One discordant result narrowly straddled the reported S/C cut-off of 1.0. NPA for IgG in Cohort 3 was 99.2% (492/496, 95% CI: 98.0–100%) on both platforms. Details on date of collection, RT-PCR status, and index values for positive results across cohorts are provided in Figure 1. Positive percent agreement (PPA) for all 22 specimens with corresponding positive RT-PCR results was 73% for IgM (95% CI: 49.8–89.3%) and 77% (95% CI: 54.6–92.2%) for IgG. PPA for specimens >10 days post symptom onset was 93% (95% CI: 77.2–99.9%) for IgM and 100% (95% CI: 84.6–100%) for IgG.

Figure 1:

Reported index (S/C) values across study cohorts for (A) Alinity SARS-CoV-2 IgG, (B) Architect SARS-CoV-2 IgG, (C) Alinity SARS-CoV-2 IgM, and (D) Architect SARS-CoV-2 IgM.

Table includes data for all positive results, where bold indicates a positive interpretation for a given assay. DOC, date of collection; N/A, Not applicable.

In our evaluation, minimal antibody positivity was observed across cohorts, suggesting both SARS-CoV-2 IgG and IgM assays have comparably high specificity in the detection of SARS-CoV-2 antibodies on both systems. Positive results observed for both immunoglobulins in Cohort 1 can be definitively considered false positives as they were collected a year prior to the pandemic. Positive results are likely due to minor cross-reactivity which has been previously reported in the literature and by Abbott Diagnostics (e.g., cytomegalovirus, coronaviruses 229E) [7]. While there is a possibility for true positives in Cohort 2, we suspect observed positive IgM results are false due to low index value and negative status for IgG. Further, known circulation of SARS-CoV-2 infection in Ontario, Canada during this period is reported to be very low, suggesting true positivity as unlikely. Conversely, we suspect some of the positive results in Cohort 3 are true positives as negative RT-PCR does not preclude the possibility of antibody positivity and reported IgG index values are quite high in relation to the cut-off. It is likely that negative IgM results in individuals with positive IgG results in this cohort is resultant from the narrower diagnostic window and expected physiological decrease in IgM titres following SARS-CoV-2 exposure [4]. In terms of sensitivity, PPA was similar across assays, with IgM demonstrating slightly lower performance. Assay performance significantly increased when excluding samples from individuals less than 10 days post symptom onset. These findings are supported by previous literature, suggesting seroconversion occurs at approximately 7–10 days post symptom onset [4].

This study is the largest pediatric evaluation of clinical specificity of any available SARS-CoV-2 serology assay to date. Our findings demonstrate assay specificities reported in adults are comparable to findings in pediatrics [1], [2]. Considering low global seroprevalence, particularly in children and adolescents, these findings are encouraging towards the application of serology tests to seroprevalence studies where high specificity is needed to achieve appropriate positive predictive value. However, further work is necessary to characterize assay sensitivity in larger cohorts of children and adolescents. Additionally, the correlation between antibody positivity, as assessed via commercially available assays, to neutralization activity is not yet understood. Furthermore, increasing reports have suggested neutralizing antibodies predominately target the S1, S2, and RBD domains of the SARS-CoV-2 spike protein, particularly in children [5]. Thus, commercial assays targeting the S protein rather than the nucleocapsid have been estimated to better associate with neutralization activity. However, recent data reports modest correlation between SARS-CoV-2 neutralizing titer and antibody levels, regardless of antigenic target [8], [9]. An additional consideration in data interpretation is the need for better harmonization between available SARS-CoV-2 antibody assay cut-offs, particularly if antibody testing is implemented towards widespread public health monitoring [10]. Taken together, further work is necessary to characterize the immunological significance of positive antibody results across commercial assays in both children and adults. As effective vaccination becomes a possibility, understanding the role of serological testing in monitoring immunity in the pediatric population will be essential.