Comparison of hemagglutination inhibition, single radial hemolysis, virus neutralization assays, and ELISA to detect antibody levels against seasonal influenza viruses

Background The immunological response to influenza vaccine and/or natural infection is evaluated by serological techniques, the most common being hemagglutination inhibition (HI), single radial hemolysis (SRH), and virus neutralization assays, which is commonly used in a micro‐neutralization (MN) format. ELISA is not officially required; however, this assay is able to measure different class‐specific antibodies. The four assays identify different sets or subsets of antibodies. Objectives The aim of this study was to establish the correlation among four serological assays using four seasonal influenza strains. Methods The HI, SRH, MN assays, and ELISA were performed on four seasonal influenza strains. Results A strong positive correlation was found between HI and MN and between SRH and MN assays for influenza A strains. The B strains also showed good correlations among the three assays. A positive correlation was also found between ELISA and the “classical” assays for all strains. Concerning the correlates of protection, as defined by HI ≥ 40 and SRH ≥ 25 mm2, good agreement was observed for the influenza A strains. By contrast, the agreement for the B strains was very low. Conclusions There is a positive strong correlation among the four serological assays for both A and B strains, especially for the HI and MN assays. There is good agreement on correlates of protection between HI and SRH assays for the A strains, but very low agreement for the B strains, suggesting higher sensitivity of SRH than HI assay in detecting antibodies against the influenza B viruses.

The HI assay is considered the gold standard as a correlate of protection for influenza vaccines [2][3][4] and has proved to be simple, rapid, and cost-effective. The aim of the assay is to detect antibodies, capable of inhibiting the agglutination between red blood cells and the viral hemagglutinin (HA). 5 The HI titer is expressed as the reciprocal of the highest serum dilution that shows complete inhibition of agglutination. 6 An antibody titer of 40 is generally regarded as a protective threshold level, beyond which there is less than 50% chance of contracting influenza infection. 7 Despite its wide application, the assay has limitations, including low sensitivity for influenza B and avian viruses, inadequacy in the evaluation of live attenuated vaccines and a high degree of variability among laboratories, due to many factors, including the source of reagents (such as red blood cells and receptor-destroying enzyme) and the lack of standardized protocols. [7][8][9] The SRH assay is a robust technique based on the passive hemolysis of red blood cells, which is mediated by complement and induced by the antibody-antigen complex. The hemolysis produces an easily identifiable "area of hemolysis," which is proportional to the concentration of influenza antibodies, mainly IgG, present in serum samples. 3,10,11 The advantages of the SRH assay are the small quantities of influenza virus and serum required, the ability to simultaneously analyze a large number of serum samples without pre-treatment (apart from complement inactivation) and unbiased results available after overnight incubation. 12 In addition, the assay detects small differences in antibody levels, distinguishes differentiates between closely related influenza strains, and is more sensitive for influenza B strains than the HI assay. [12][13][14][15] A hemolysis area of 25 mm 2 or greater is generally considered to be an immunological correlate of protection. 16 Another widely used serological technique is the virus neutralization assay, which is recommended by the World Health Organization (WHO) for the measurement of functional antibodies against highly pathogenic avian viruses 6 and currently included in the EMA guidelines on influenza vaccines. 1 Commonly used in a micro-neutralization (MN) format, this assay detects antibodies at low titers and distinguishes between pre-and post-vaccination titers in paired sera, especially in the case of small (less than 2-fold) differences in titers. 17 The disadvantage lies in the handling of wildtype viruses, which, in the case of highly pathogenic strains, require high-level facilities. The assay suffers from high interlaboratory variability, owing to the lack of common reference protocols and discrepancies in endpoint determination. Here, MN titer is expressed as the reciprocal of the serum dilution showing at least 50% inhibition of cytopathic effect in mammalian cell culture. 7,18 To date, no correlates of protection have been established for the MN assay.
In addition to the traditional immunological techniques, the enzyme-linked immunosorbent assay (ELISA) detects influenza antibodies. Advantages are its ability to measure different class-specific IgM, IgA, and IgG antibodies in serum samples and nasal wash in response to influenza infection and/or vaccine and to use a wide range of antigen preparations. This assay is particularly suitable for large-scale serological investigations, as it yields unbiased results in a few hours and is cost-effective and amenable to complete automation. In addition, the ELISA assay is reproducible and reagents can be standardized (e.g, coating antigen, conjugate for Ig detection). 7,19,20 ELISA mainly detects anti-HA antibodies, even when the whole virus is used, as HA is immunodominant. 21,22 ELISA can also be used to detect responses to other influenza antigens when purified single antigens are used for coating. The use of purified HA antigens or a recombinant fragment of the HA globular head may considerably improve the specificity of ELISA. 7,19,[23][24][25][26] The four assays identify different sets or subsets of antibodies.
The HI assay detects antibodies that bind to the viral HA and prevent the virus-red blood cells agglutination by blocking the receptor binding site; the MN assay identifies functional neutralizing antibodies, including those that recognize epitopes in the stem region of HA, which are conserved among different subtypes of influenza A viruses. 17,20,27 Consequently, the MN assay could be less specific in adults and elderly subjects with extensive previous exposure to influenza viruses. 27 However, MN is more sensitive than HI, particularly in detecting low-titer seroconversions. A combination of assays could improve the sensitivity and specificity of influenza vaccine evaluation.
The SRH assay may recognize antibodies not only against the surface glycoproteins but also those against the internal antigens, leading to a potential lack of specificity to antibodies against HA. 3 However, the immune response to HA is immunodominant over the response to other viral proteins, such as neuraminidase and internal proteins. 21 Several studies have compared serological techniques for the evaluation of antibody response to influenza viruses or vaccination.
Overall, the HI, SRH, and MN assays have shown significant positive correlations, especially for influenza A strains. 11,14,[27][28][29] A recent study has revealed that the correlation between SRH and MN is greater than that between SRH-HI and HI-MN assays. 30 ELISA shows good agreement between the HI and MN assays; notably, ELISA seems to be more sensitive than HI assay, especially in detecting low levels of antibody, owing to the low background. 24,31,32 Few previous studies have dealt with the correlations among the four assays.
The aim of this study was to establish the correlations among the four immunological assays: HI, SRH, MN, and ELISA using four egggrown seasonal influenza strains.

| Virus antigens
The virus antigens and infectious influenza viruses were seasonal preparations, obtained from NIBSC, were used for HI, SRH, and ELISA. All viruses used were egg-grown. Human serum without IgA, IgM, and IgG was used as negative controls (Sigma-Aldrich, S5393).

| Hemagglutination inhibition assay
All serum samples, including the sheep serum samples and negative control, were pre-treated with receptor-destroying enzyme (RDE) (ratio 1:5) from Vibrio Cholerae (Sigma-Aldrich, Milan, Italy) for 18 hours at 37°C in a water bath and then heat-inactivated for 1 hour at 56°C in a water bath with 8% sodium citrate (ratio 1:4).
Fresh turkey red blood cells were centrifuged twice, washed with a saline solution (0.9%), and adjusted to a final dilution of 0.35%.
From an initial dilution of 1:10, serum samples were 2-fold diluted in duplicate with saline solution (0.9%) in a 96-well plate; 25 μL of standardized viral antigen (4 HA units/25 μL) was added to each well, and the mixture was incubated at room temperature for 1 hour.
Red blood cells were added and, after 1-hour incubation at room temperature, the plates were evaluated for the presence of agglutination inhibition.
The antibody titer was expressed as the reciprocal highest serum dilution that showed complete inhibition of agglutination. As the starting dilution was 1:10, the lower limit of the detectable antibody titer was 10. When the titer was below the detectable threshold, the results were conventionally expressed as 5 for calculation purposes, half the lowest detection threshold.

| Single radial hemolysis assay
Before being used in the SRH assay, serum samples were heatinactivated at 56°C for 30 minutes in a water bath. Fresh turkey red blood cells were centrifuged and washed twice with phosphate buffer saline (PBS). Diluted virus antigen was added to the red blood cell suspension at a concentration of 2000 hemagglutinin units (HAU) per milliliters (mL). In order to allow the adsorption of viral antigen to the red blood cells, the suspension was incubated at 4°C for 20 minutes. A solution of chromium chloride (CrCl 3 ) 2.5 mmol/L was added to the suspension and incubated at room temperature for 10 minutes to increase the binding affinity between the red blood cells and the viral antigen. The suspension was mixed once and subsequently centrifuged. The supernatant was removed, PBS was added, and the pellet was carefully re-suspended. A stock solution of 1.5% agaroseagarose low gelling in PBS containing 0.1% sodium azide was prepared. The agarose stock solution was kept at 45°C in a water bath.
Each SRH plate contained red blood cells-viral antigen suspension and guinea pig complement in the agarose mixture. The final suspension was vigorously shaken and then evenly spread onto each plate and incubated at room temperature for 30 minutes, and the agarose was allowed to set at 4°C for 30-90 minutes. Holes were made in each plate with a calibrated punch, and 6 microliters (μL) of serum samples and controls was added through each hole. The plates were stored in a humid box and incubated at 4°C for 16-18 hours in the dark. After overnight incubation, the plates were incubated in a water bath at 37°C for 90 minutes, after which the diameters of the hemolysis areas were read in millimeters (mm) with a calibrated viewer. 12

| Micro-neutralization assay
Madin Darby Canine Kidney (MDCK) cells were maintained for a maximum of 30 passages in EME medium containing 10% fetal bovine serum (FBS), 2 mmol/L L-glutamine, 1% non-essential amino acid solution, and 100 U/mL penicillin-streptomycin. The MDCK cell cultures were grown at 37°C in 5% CO 2 . Serum samples, previously heat-inactivated at 56°C for 30 minutes, were diluted 2-fold with EMEM culture medium supplemented with 0.5% FBS in a 96-well plate, mixed with an equal volume of virus (100 TCID 50 /well), and incubated for 1 hour at 37°C in 5% CO 2 . At the end of incubation, the MDCK cell suspension (1.5 × 10 5 cells/mL) was added to the plates, which were then stored in an incubator (37°C, 5% CO 2 ) for 5 days.
After incubation, the plates were observed by optical microscopy and evaluated for cytopathic effect. A cytopathic effect higher than 50% indicates infection. The titer was expressed as the inverse of the last dilution that showed inhibition of cytopathic effect.

| ELISA
To evaluate humoral responses, the ELISA was performed on human

| Statistical analysis
All statistical analyses were performed using Microsoft R-Open,

| RE SULTS
The human serum samples (n = 450) were tested by HI, SRH, MN, and ELISA assays to evaluate the assay correlation using   Table 1).
For the A strains, the correlation between HI, MN, and SRH was stronger than that between ELISA and the other assays. By contrast, F I G U R E 1 Continued the correlation between ELISA and HI, and SRH and VN assays was comparable to that of the three traditional assays for the B strains.
In addition, the assay agreement on protection, as defined by  The present study has some limitations, such as missing information on the vaccination status of the subjects involved and the unavailability of paired serum samples. Another limitation is the lack of comparability between ether-treated B viruses and native viruses and between egg-grown and cell-grown viruses.

| D ISCUSS I ON
Overall, this study shows a strong positive correlation among the four serological assays (MN, HI, SRH, and ELISA) for both A and B strains; this is especially true of the HI and MN assays. However, it also highlights the need to further investigate the correlation between the SRH assay and ELISA. Concerning the correlates of protection, as defined by HI ≥ 40 and SRH ≥ 25 mm 2 , we found good agreement regarding protection against A strains between HI and SRH assays, but very low agreement for the B strains, suggesting that SRH is more sensitive than HI in detecting antibodies against the influenza B viruses. As the four serological assays detect different sets or subsets of antibodies, combining all the assays could considerably improve the assessment of the immunogenicity of influenza vaccines and provide a more complete picture of antibody responses. In addition, it could be useful to establish the correlates of protection for the MN assay, in order to compare vaccine assessments based on the three assays. Finally, further research on ELISA could be particularly useful in order to evaluate the immunogenicity of novel universal influenza vaccines.

CO N FLI C T O F I NTE R E S T S
The authors have no competing interests.