Re-evaluating positive serum samples for SARS-CoV-2-specific IgA and IgG antibodies using an in-house serological assay

To the Editor,

We read the recent article by Caruana et al. exploring the current landscape of diagnostic tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which causes coronavirus disease 2019 (COVID-19), and signalling interpretive issues of test results [1]. We were particularly interested in serological testing, which may fill the gap between negative results of RT-PCR—the reference standard for SARS-CoV-2 diagnosis [2]—and clinical (and radiological) findings suggestive of COVID-19 [3,4]. Like molecular testing [5], targeting the SARS-CoV-2 spike (S) protein (or the subunit S1 thereof) rather than the nucleocapsid (N) protein with an ELISA to detect virus-specific antibodies in patient serum may be crucial for diagnostic yield [6]. Sensitivity of ELISAs based on the N or S protein varies depending on the infection timing [1]. Additionally, testing for only IgM and IgG [[7], [8], [9]] may be limited in samples taken around symptom onset [10]. In this context, individuals who present within the first week after symptom onset could benefit from IgA testing [11]. In a recent study [11], the S1-based IgA Euroimmun (Lübeck, Germany) assay revealed good sensitivity compared with an S (or S1) -based IgG Wantai test (Beijing, China) or Euroimmnun assays with individuals sampled at early infection times. Consistently, Caruana et al. experienced a 96% sensitivity with samples collected 15–30 days post infection, using an N-based ELISA (Epitope Diagnostics, San Diego, CA, USA) [1]. Finally, mild (non-hospitalized), moderate (hospitalized) or severe (admitted to the intensive care unit) illness may affect antibody responses in individuals with COVID-19 [8,9].

Using in-house ELISA targeting the SARS-CoV-2 N protein [7], we re-evaluated positive results from the Euroimmnun ELISA for SARS-CoV-2-specific IgA and IgG detection for 122 serum samples of individuals admitted to the emergency department of our institution for suspicion of COVID-19. The institutional ethics committee approved the study (no. 27015/20), and informed consent was obtained from all individuals. Except for 105 individuals with RT-PCR-confirmed SARS-CoV-2 infection, COVID-19 diagnosis in 17 RT-PCR-negative individuals was based on both abnormal radiological findings and positive serology results. Initially, reproducibility of in-house ELISA was assessed testing 30 serum samples from individuals with COVID-19 with different levels of IgA or IgG antibodies. We found that the coefficients of variation were 1.38%–32.22% and 2.06%–21.05% for IgA and IgG, respectively, whereas intra-class correlation coefficients were 0.88 and 0.98 for IgA and IgG, respectively.

As shown in Table 1 and depicted in Fig. 1 , all samples with positive IgA/IgG results by Euroimmnun ELISA included samples positive for IgA (n = 119) and IgG (n = 113); of these samples, 110 were positive for both IgA and IgG, nine for only IgA and three for only IgG. In parallel, samples with positive IgA/IgG results by in-house ELISA included samples positive for IgA (n = 98) and IgG (n = 111); of these samples, 95 were positive for both IgA and IgG, 3 for only IgA and 16 for only IgG. The in-house assay detected 96/119 IgA-positive samples and 109/113 IgG-positive samples, corresponding to a positive per cent agreement of 80.7% (95% CI 72.4%–87.3%) and 96.5% (95% CI 91.2%–99.0%), respectively. Discrepancies between the two assays mainly involved samples that tested negative for IgA by the in-house assay (Table 1). These samples were from individuals with mild (11/30 samples) or moderate (12/62 samples) disease, as well as those collected within the first 5 days (9/30 samples) or after 40 days (9/56 samples) of admission. Although N-based serological correlates of protection from SARS-CoV-2 infection are not fully understood [12], similar to us, other investigators emphasized the role of anti-SARS-CoV-2 IgA in the current serodiagnostic arsenal for SARS-CoV-2 [13,14], especially in the early phase of infection [15].

Table 1.

Summary of serological SARS-CoV-2 antibody testing results for 122 symptomatic COVID-19 patients sampled at different days from the emergency department admission

Patient group (no. of tested)	No. (%) of samples with positive results for:
	Immunoglobulin A detected with:		Immunoglobulin G detected with:
	N-based in-house assay	S-based Euroimmun assay	N-based in-house assay	S-based Euroimmun assay
SARS-CoV-2 infectiona
Confirmed (n = 105)	88 (83.8)	104 (99.0)	101 (96.2)	100 (95.2)
Unconfirmed (n = 17)	10 (58.8)	15 (88.2)	10 (58.8)	13 (76.5)
Severity on admissionb
Mild (n = 31)	19 (61.3)	30 (96.8)	26 (83.9)	27 (87.1)
Moderate (n = 86)	74 (86.1)	84 (97.7)	80 (93.0)	81 (94.2)
Severe (n = 5)c	5 (100.0)	5 (100.0)	5 (100.0)	5 (100.0)
Testing from admission, days
0–5 (n = 32)	23 (71.9)	30 (93.8)	25 (78.1)	26 (81.3)
6–20 (n = 8)	7 (87.5)	8 (100.0)	6 (75.0)	7 (87.5)
21–40 (n = 26)	21 (80.8)	25 (96.2)	24 (92.3)	25 (96.2)
>40 (n = 56)	47 (83.9)	56 (100.0)	56 (100.0)	55 (98.2)

Abbreviations: COVID-19, coronavirus disease 2019; N, nucleocapsid; S, spike; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2.

^aAccording to positive (confirmed) or negative (unconfirmed) results for SARS-CoV-2 RNA detection by RT-PCR. Except for 105 patients with confirmed SARS-CoV-2 infection, diagnosis of SARS-CoV-2 infection in 17 individuals with negative RT-PCR results was based on both clinical/radiological presentation and positive serology (by Euroimmun assay) findings.

^bAccording to the individuals' requirement for non-hospitalization (mild), hospitalization (moderate) or intensive care (severe).

^cSamples from these individuals also tested positive for IgM by the indicated N-based in-house assay. However, IgM results for all the 122 samples included in the study were not reported because these results were beyond the comparison purposes between in-house and Euroimmun assays.

Fig. 1.

Agreement of results for 122 serum samples obtained with the Euroimmun and the in-house ELISA tests. Unlike the commercial Euroimmun assay, the in-house assay for IgA and IgG detection was developed based on the use of a recombinant nucleocapsid protein of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) as described elsewhere [7]. For both assays, the antibody levels are shown expressed as spectrometrically measured values divided by the cut-off (S/CO), as are the percentage between-assay agreement values calculated for IgA and IgG antibodies, respectively. The cut-offs for IgA (0.08 and 1.10) and IgG (0.45 and 1.10) antibodies in both assays are marked with vertical blue (in-house assay) or green (Euroimmun assay) lines. The Cohen's κ values indicate fair (range 0.21–0.40) or substantial (range 0.61–0.80) agreement for IgA and IgG results, respectively. Among five samples that tested positive with the in-house assay but negative with the Euroimmnun assay, two were positive for IgA antibodies and three for IgG antibodies, respectively.

We also determined the specificity of N-based serological testing using sera from 85 healthy blood donors or from 15 individuals with non-SARS-CoV-2 respiratory infection and we found that no sera were positive with the N-specific IgA (and IgG) assay. Furthermore, we observed that IgG antibodies detected in two individuals who tested positive—one with the in-house ELISA only and one with both the in-house and Euroimmun ELISAs—were able to neutralize the Vero E6 cell-cultured SARS-CoV-2 (titres were 1 : 80 in both individuals). Likewise, IgA antibodies detected in two other individuals who tested positive—one with the in-house ELISA only and one with both in-house and Euroimmun ELISAs—were able to neutralize the Vero E6 cell-cultured SARS-CoV-2 (titres were 1 : 20 and 1 : 640, respectively). Although these observations are consistent with recently published data [16,17], for reasons of comparability, we did not include data regarding the detection of IgM antibodies by the assay.

In conclusion, we suggest that serology targeting the SARS-CoV-2 S protein, such as the Euroimmun ELISA, should be preferable. We recorded the highest sample positivity rates with S-based testing for IgA antibodies in individuals tested early or in individuals with mild COVID-19 (not requiring hospitalization) on admission (Table 1). Hence, we propose considering IgA testing in all situations where serology is the solely practicable diagnostic strategy for SARS-CoV-2 infection. Future studies will help to decide on the deployment of serological assays for specific contexts in COVID-19 diagnostics.

Transparency declaration

The authors declare that they have no conflicts of interest. No external funding was received for this study.

Editor: L. Leibovici