Comparison of two assays to detect IgG antibodies to the receptor binding domain of SARS‑CoV‑2 as a surrogate marker for assessing neutralizing antibodies in COVID-19 patients.

BACKGROUND: Neutralizing antibodies (NAbs) are important for protection against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) reinfection. In this study, two assays that are correlated with NAbs were compared: the haemagglutination test (HAT) and the surrogate virus neutralization test (sVNT).
METHODS: The specificity of the HAT was compared with the sVNT, and the sensitivity and persistence of antibodies in patients with varying severity of illness was assessed in a cohort of 71 patients at 4-6 weeks and 13-16 weeks. The kinetics were assessed in the first, second, and third weeks in patients with varying severity of acute illness.
RESULTS: The specificity of the HAT was >99%, and sensitivity was similar to the sVNT. The levels of HAT were significantly and positively correlated with those of the sVNT (Spearman's r = 0.78, P < 0.0001). Patients with moderate and severe illness had higher HAT titres when compared to those with mild illness. Six of seven patients with severe illness had a titre of >1:640 during the second week of illness, whereas only five of 31 patients with a mild illness had a titre of >1:160 in the second week of illness.
CONCLUSIONS: Since the HAT is a simple and very cheap assay to perform, it would be ideal to use as an indicator of NAbs in resource-poor settings.

Introduction

There are many antibody assays currently in use to determine IgG, IgM, and IgA specific to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus responsible for coronavirus disease 2019 (COVID-19) (Algaissi et al., 2020; Sun et al., 2020; Vogelzang et al., 2020). While some of these assays measure total antibodies to mainly the spike protein receptor binding domain (RBD), others measure IgM or IgG responses to S1, S2, or the nucleocapsid protein (Sun et al., 2020; Vogelzang et al., 2020). These assays are sometimes used in conjunction with PCR assays, and some are used to detect those who have had an asymptomatic infection and in serosurveillance studies (Galipeau et al., 2020). Although RT-PCR is the gold standard to identify individuals who are acutely infected with SARS-CoV-2, serological tests can contribute by providing more accurate estimates of past SARS-CoV-2 infection and the immune status of the population (Ghaffari et al., 2020). However, while different antibody assays can be used to identify those who have been exposed to the virus, only assays that measure, or correlate with, neutralizing antibodies (NAbs), are likely to provide evidence of antibodies that are more likely to protect individuals against reinfection (Galipeau et al., 2020).

NAbs against SARS-CoV-2 are mainly produced against the RBD of the viral spike protein (Kreer et al., 2020; Ni et al., 2020). The gold standard for determining NAbs is the plaque reduction neutralization test (PRNT), which requires biosafety level 3 (BSL-3) facilities and is time-consuming (Galipeau et al., 2020). With COVID-19 spreading at an exponential rate around the world, many assays have been developed to measure NAbs that can be performed within a few hours in a BSL-2 facility (Tan et al., 2020; Townsend et al., 2020). One such assay is the surrogate virus neutralization test (sVNT), which measures the percentage of inhibition of binding of the RBD of the spike protein to recombinant angiotensin-converting enzyme 2 (ACE2) (Tan et al., 2020). Further, Townsend et al. have developed a haemagglutination test (HAT) to measure antibodies to the RBD, where the RBD of the virus is linked to a nanobody IH4, specific for a conserved epitope within glycophorin A on red blood cells (RBCs) (Townsend et al., 2020). In the presence of antibodies to RBD, IH4-RBD-6H bound RBCs will show agglutination. Since the majority of NAbs are directed at the RBD (Kreer et al., 2020; Ni et al., 2020) and the level of antibodies detected by the HAT correlates with the neutralising half maximal inhibitory concentration (IC50)(manuscript in preparation), this assay could be used as an inexpensive test to predict NAbs in research and community settings where high throughput assays are required (Townsend et al., 2020). Therefore, in order to determine the usefulness of the HAT, the performance of the HAT was compared with that of an existing assay that is used as a surrogate to measure the presence of NAbs by blockade of ACE2 binding by the RBD.

Methods

The HAT to detect neutralizing antibodies

The HAT was performed as described previously (Townsend et al., 2020). Briefly, RBCs from an O negative donor were mixed with the IH4-RBD-6H (a nanobody against a conserved glycophorin A epitope on RBCs, linked to the RBD of SARS-CoV-2) and incubated for 1 hour with serum. Phosphate buffered saline (PBS) was used as a negative control. At the end of the incubation, the plate was tilted for 20 seconds and then photographed. The photograph of the plate was read by two independent readers to examine the ‘teardrop’ formation indicative of a negative result. A complete absence of teardrop formation was scored as positive and any flow of the teardrop was scored as negative.

The HAT titration was performed using 11 doubling dilutions of serum from 1:20 to 1:20 480, to determine the NAbs titre. The NAbs titre for the serum sample was defined by the last well in which the complete absence of teardrop formation was observed.

The sVNT to detect neutralizing antibodies

The sVNT (Tan et al., 2020), which measures the percentage of inhibition of binding of the RBD of the S protein to recombinant ACE2 (Genscript Biotech, USA), was performed according to the manufacturer's instructions. An inhibition percentage ≥25% in a sample was considered as positive for NAbs.

Patients

Patients confirmed with a SARS-CoV-2 infection based on a positive RT-PCR were recruited from the National Institute of Infectious Diseases (NIID), Sri Lanka. They were followed throughout their illness while they were in hospital, and the severity grading was based on the worst severity while in hospital. Clinical disease severity was classified as mild, moderate, or severe according to the World Health Organization (WHO) guidance on COVID-19 disease severity (WHO, 2020). Two cohorts of patients were recruited for this study. Serum samples from patient cohort 1 (n = 50) were used to determine the correlation of HAT titres with sVNT levels and longitudinal changes in SARS‑CoV‑2 NAbs levels. Blood samples were obtained from the first cohort of patients during the first and second weeks of illness and again when they were discharged from hospital (4–6 weeks since the onset of illness) (Table 1). The duration of illness was defined as the period from the day of symptom onset and not the day of PCR positivity or admission to hospital. Twenty-one patients showed PCR positivity for more than 25 days since the onset of symptoms and were categorized as prolonged shedders. All of the prolonged shedders were either asymptomatic or had mild COVID-19. Based on the WHO COVID-19 disease classification, six patients had a severe illness, five a moderate illness, 10 a mild illness, and 21 a mild illness but with prolonged viral shedding for >25 days (WHO, 2020).

Table 1

Disease severity information and blood sample collection time point for the two patient cohorts included in the study

	Patient cohort 1n = 50	Patient cohort 2n = 66
Mean age (years)	42	43
Sex: male/female	37/13	32/34
Time point for blood sample collection since the onset of symptoms	First weekSecond week4–6 weeks	4–6 weeks (Time point A)13–16 weeks (Time Point B)
Asymptomatic COVID-19	8	52
Mild COVID-19	31	9
Moderate COVID-19	5	3
Severe COVID-19	6	2

Serum samples from patient cohort 2 (n = 66) were used to study the sensitivity of the HAT compared to the sVNT and to determine the persistence of antibody levels in these patients. Details of both of the patient cohorts and sampling time points are given in Table 1. There were 24 prolonged shedders in this patient cohort and they were either asymptomatic or had mild COVID-19.

Ethical approval was received from the Ethics Review Committee of the Faculty of Medical Sciences, University of Sri Jayewardenepura. Written informed consent was obtained from the patients.

Statistical analysis

The correlation of HAT titres with the percentage inhibition values from the sVNT was assessed with Spearman's correlation using GraphPad Prism version 8.4.2. The sensitivity of the HAT and sVNT in detecting SARS-CoV-2 antibodies at 4–6 weeks and 13–16 weeks following infection was assessed using the Chi-square test.

Results

Determining the specificity of the HAT to measure antibodies against SARS‑CoV‑2

As some antibody assays for SARS-CoV-2 are known to detect cross-reactive antibodies to other seasonal coronaviruses (Galipeau et al., 2020), the specificity of the HAT was initially determined in samples collected during the years 2017 and 2018 from febrile patients in the Sri Lankan population (n = 110). The HAT was performed on serum diluted at 1:20. All of these samples were from patients with confirmed dengue virus infections based on RT-PCR and none of the patients had any respiratory symptoms. With the exception of one sample, all of the serum samples tested negative. Therefore, the specificity of this assay was found to be >99%, as described previously (Townsend et al., 2020). The serum sample that tested positive with the HAT was from a patient who had an acute infection due to dengue virus serotype 3.

Correlation of the HAT with the sVNT in determining SARS‑CoV‑2 neutralizing antibody levels

In order to determine the performance of the HAT in comparison to the sVNT in patients with acute illness (patient cohort 1), the HAT titres were compared with values given by the sVNT. The HAT titre of the sample was determined by the last well in which the complete absence of teardrop formation was observed (Figure 1A). It was found that the titres of the HAT were significantly and positively correlated with levels of ACE2 blocking antibodies obtained from the sVNT (Spearman's r = 0.78, P < 0.0001) (Figure 1B). All of the samples that were negative by HAT were also negative by sVNT. However, the HAT was positive (titre >1:20) while the sVNT was negative in six individuals during the first week of illness.

Figure 1

Correlation of antibody levels determined by haemagglutination test (HAT) titration and surrogate virus neutralization test (sVNT). (A) Example of HAT titration carried out on a sample. The HAT titration was conducted using 11 doubling dilutions of the serum from 1:20 to 1:20 480 (the titres are indicated as ‘20’, instead of ‘1:20’). The HAT titre of the sample was determined according to the last well in which the complete absence of teardrop formation was observed (marked with a black solid circle), indicating a titre of 1:40 for this patient. (B) Correlation of antibody levels determined by HAT titre and the percentage inhibition of binding of the RBD to recombinant ACE2 by the serum sample according to the sVNT assay; This correlation was performed with samples obtained from 50 patients belonging to cohort 1 at three time points during the course of illness. Spearman's r = 0.79, P < 0.0001.

Longitudinal changes in antibodies in COVID-19 detected with the HAT

Next, the longitudinal changes in antibody titres of patients (patient cohort 1) with varying severity of clinical disease were investigated. Patients with moderate and severe illness had higher and more persistent NAbs levels (higher titres in the HAT) (Figure 2A) when compared to those with mild illness with and without prolonged shedding (Figure 2B). With the exception of one patient with severe illness who had a titre of >1:160, all of the other patients with severe illness (six out of seven) had a titre of >1:640 during the second week of illness. In contrast, only five of 31 patients with mild illness had a titre of >1:160 in the second week of illness. A similar pattern of antibody kinetics was observed in a previous study on patients with varying severity of illness using the sVNT assay (Jeewandara C et al., 2021).

Figure 2

Antibody levels were determined by haemagglutination test (HAT) titration with the varying severity of COVID-19. The HAT titration was performed using 11 doubling dilutions of the serum from 1:20 to 1:20 480 (the titres are indicated as ‘20’, instead of ‘1:20’) in patient serum obtained during the first week, second week, and at 4–6 weeks since the onset of illness. (A) Longitudinal changes in HAT titres with the duration of the illness in patients with severe illness (n = 7) and moderate illness (n = 5). (B) Longitudinal changes in HAT titres with the duration of the illness in patients with a mild illness who were prolonged shedders (n = 21) and in patients with a mild illness who were not prolonged shedders (n = 10).

Sensitivity of the HAT with the sVNT in patients with COVID-19

In order to compare the sensitivity of the HAT with the sVNT, these two assays were compared in samples obtained at 4–6 weeks (n = 66) since the onset of illness and in samples obtained at 13–16 weeks (n = 66) since the onset of illness (patient cohort 2). Of the 66 patients, two had severe illness, three moderate illness, nine mild illness, and 52 had completely asymptomatic illness. The HAT was performed at 1:20 dilution, and the complete absence of teardrop formation was scored as a positive result and any flow of the teardrop was scored as a negative result. At 4–6 weeks of illness, 49 of the 66 patients (74.2%) had a positive HAT result, while 48 (72.7%) had a positive sVNT result. Therefore, at 4–6 weeks, the sensitivity of the HAT was comparable to that of the sVNT. At time point B, 37/66 (56%) had a positive HAT result, while 33/66 (50%) had a positive sVNT result. All of the samples that were negative by HAT were also negative by sVNT. Therefore, while the specificity of the HAT was similar to that of the sVNT, the sensitivity appeared to be slightly higher (56% vs 50%).

All of the patients with moderate or severe illness (n = 5) had NAbs detected by HAT at 4–6 weeks of illness, and by 13–16 weeks of illness. Twenty-four of the patients with mild/asymptomatic illness had prolonged shedding (PCR positivity for >25 days). During 4–6 weeks of illness, 16/24 (66.6%) prolonged shedders were positive by HAT and by sVNT (the same 16 individuals were positive by both). Furthermore, 33/42 (79%) of those who cleared the virus early were positive by HAT and 32/42 (76%) by sVNT. By 13–16 weeks of illness, only 12/24 (50%) of the prolonged shedders were positive by HAT and 9/24 (37.5%) by sVNT, compared to 24/42 (57%) who cleared the virus early being positive by HAT and 23/42 (55%) by sVNT. However, these differences were not statistically significant.

These are the first results of HAT measured in asymptomatic patients over such a prolonged period. It will be important to follow up these individuals longitudinally to determine whether asymptomatic infection is fully protective against reinfection in this cohort, and whether the HAT result has any utility in predicting susceptibility.

Discussion

In this study, two assays for assessing the presence of NAbs against SARS-CoV-2, each with a correlation with neutralising titres and that can be performed in a BSL-2 laboratory, were compared. The HAT was found to correlate well with the NAbs levels given by the sVNT and was found to have a comparable sensitivity during acute illness and up to 13–16 weeks since the onset of illness. The assay was also found to have a high specificity (>99%) in the samples that were assessed before any individuals in Sri Lanka would have been exposed to SARS-CoV-2. However, due to the non-availability of a BSL-3 laboratory within the country, it was not possible to compare HAT with virus neutralisation in Sri Lankan patients, and therefore the assay was compared with the sVNT, which evaluates ACE2 blocking antibodies and has been shown to correlate well with neutralizing antibodies (Tan et al., 2020). The titres given by the HAT were higher during early infection in those with severe and moderate illness compared to those with mild illness, as described previously (Jeewandara C et al., 2021; Lynch et al., 2021). Although the reason for the detection of higher levels of NAbs during early infection is not clear, it could be due to increased production by extrafollicular B cells (Galipeau et al., 2020; Lee et al., 2020; Woodruff et al., 2020), or the more efficient ability of IgM to cross-link RBCs.

In Sri Lanka, until early July 2020, all individuals with COVID-19 were discharged from treatment centres/hospitals after two PCR tests performed 24 hours apart were negative. Therefore, repeated PCR tests were conducted on the hospitalized patients to enable them to be discharged once two consecutive PCR tests were negative. This enabled us to detect those who had prolonged shedding of the virus for 4–9 weeks since the onset of illness. Interestingly, these prolonged shedders were less likely to have adequate levels of NAbs at 4–6 weeks and 13–16 weeks compared to those who became PCR-negative early, although this was not statistically significant. It is possible that these individuals continued shedding PCR-detectable viral material for many weeks due to reduced production of NAbs or T cells, which should be investigated further.

Although cross-reactive antibodies specific to other coronaviruses are known to interfere with the specificity of SARS-CoV-2 serological assays (Galipeau et al., 2020), the HAT showed high specificity (99%) and therefore it is unlikely to pick up such cross-reactive antibodies. In addition, it was found to have equal sensitivity to the sVNT in detecting RBD-specific antibodies when compared to the sVNT at 4–6 weeks and 13–16 weeks from the onset of illness, and in asymptomatic cases. These are the first results with the HAT applied to asymptomatic infections, and clearly show a reduced sensitivity of approximately 73% compared to about 1 month after symptomatic infection (approximately 90%) (Townsend et al., 2020). In other work (manuscript in preparation), we have shown a strong correlation between the HAT titre and authentic virus neutralisation titre (IC50). Prospective measurements in adequate cohort studies will be needed to determine the relationship between these measurements and protection from reinfection. In addition, with the worldwide vaccination drive against COVID-19, the HAT could be very useful, especially in resource-poor settings, to measure seroconversion rates and antibody levels after vaccination. We have successfully used the HAT to measure antibody titres against the RBD of SARS-CoV-2 at 1 month after AZD1222/Covishield vaccination in Sri Lanka (Jeewandara Chandima et al., 2021).

In summary, the HAT is a very cheap assay, with high sensitivity and specificity, which is a surrogate test to detect the presence of NAbs to SARS-CoV-2 that correlate with neutralising activity. As the HAT does not require any specific equipment, it would be a valuable and very cheap tool in resource-poor settings. As it appears to have comparable sensitivity to the sVNT assay, it might be suitable for population screening to identify previously exposed individuals.