Workshop report: Immunoassay standardisation for “universal” influenza vaccines

The development of broadly reactive influenza vaccines raises the need to identify the most appropriate immunoassays that can be used for the evaluation of so‐called universal influenza vaccines and to explore a path towards the standardisation of such assays. More than fifty experts from the global influenza vaccine research and development field met to initiate such discussion at a workshop co‐organised by the EDUFLUVAC consortium, a European Union funded project coordinated by the European Vaccine Initiative, and the National Institutes of Health/National Institute of Allergy and Infectious Diseases, USA. The workshop audience agreed that it was not possible to establish a single immunoassay for “universal” influenza vaccines because the current approaches differ in the vaccines' nature and immunogenicity properties. Therefore, different scientific rationales for the immunoassay selection are required. To avoid dilution of efforts, the choice of the primary evaluation criteria (eg serological assays or T‐cell assays) should drive the effort of harmonisation. However, at an early phase of clinical development, more efforts on exploratory assessments should be undertaken to better define the immune profile in response to immunisation with new vaccines. The workshop concluded that each laboratory should aim towards validation of the appropriate immunoassays used during the entire process of vaccine development from antigen discovery up to establishment of correlates of protection, including the different steps of quality control (eg potency assays), animal studies and human clinical development. Standardisation of the immunoassays is the ultimate goal, and there is a long way to go.


| Haemagglutination inhibition and single-radial haemolysis assays
The immune response induced by influenza vaccines has been measured traditionally by three classical immunoassays: haemagglutination inhibition (HI), single-radial haemolysis (SRH) and virus neutralisation (VN) assays. The HI assay detects antibodies that bind to the viral haemagglutinin and prevent the virus-mediated agglutination of erythrocytes. There is a consensus, although not universally accepted, that an HI antibody titre of 40 correlates with a 50% reduction in the risk of contracting influenza 1 . This correlate of protection has been established from studies in healthy adults and is not appropriate for children 2 . The HI assay faces various technical challenges, such as variations when using erythrocytes from different species and variability between batches of erythrocytes from the same species. The antibody titres measured by HI and VN assays usually correlate somewhat for seasonal influenza viruses, but correlation may not be observed when testing non-human subtypes such as H5N1 viruses.
The SRH assay measures antibodies that bind to the influenza virus and fix complement (usually guinea pig complement) 3,4 . The SRH assay displays higher sensitivity for H5N1 viruses than the HI assay and has been shown to be relatively reproducible between laboratories. A correlate of protection has also been defined for SRH, that is a zone of 25 mm 2 5 . While correlates of protection have been widely used for the HI and SRH assays, their relevance has been questioned and they are not valid in Europe any more 6 .

| Virus neutralisation assay
Neutralising antibodies are generally accepted as primary mediators of immunity against influenza virus 7 . The VN assay mainly detects functional antibodies that bind the HA globular head and block receptor binding but can also measure antibodies that block entry at the stage of membrane fusion. There are various formats of the VN assay (Table 1).
Collaborative studies have shown that interlaboratory variability of VN assays can be considerable; the use of antibody standards can reduce it 8 ; however, antibody standards are not usually available for influenza serology studies.
As opposed to the HI assay, no correlate of protection has been established for the VN assay. In a recent household transmission study, microneutralisation titres ≥ 40 were associated with 49% protection against H3N2 infection 9 .
As the three main serologic assays HI, VN and SRH detect different populations of antibodies, different degrees of correlation between these assays can be observed.

| Automated imaging of hemagglutination assays
Reading of haemagglutination and HI plates by operators is one cause of the variability of the HI assay. Different automated imaging analysis systems are currently under development and could standardise the reading of HI plates. The plate reader Cypher One ™ was shown to achieve good correlation with expected titres in validation experiments. Moreover, there was good agreement between expert human readers and the automated reader. A high-throughput machine is currently under development.

| ANALYSIS OF ANTIBODY RESPONSES TO THE INFLUENZA VIRUS HAEMAGGLUTININ STALK DOMAIN
Although antistalk antibodies can be produced after seasonal influenza vaccination, they comprise only a small proportion of total anti-HA antibodies and are not efficiently boosted even after routine annual seasonal vaccination. However, anti-HA stalk antibodies that have been isolated from humans (and mice) have been found to be cross-reactive and neutralising between different subtypes and occasionally between group 1 and group 2 haemagglutinins 10 . There are a number of ways that antistalk antibodies may mediate immunity, including inhibiting viral fusion during entry, inhibiting new virion budding and mediating macrophage killing via Fc receptor (FcR) binding to antibody-coated infected cells 11 . These different mechanisms inform the different types of serological assays that can be developed.
ELISAs can be used to detect total antistalk antibodies as well as heterosubtypic cross-reactivity. The laboratory at Icahn School of Medicine at Mount Sinai has developed ELISAs using chimeric haemagglutinins 12 . These chimeras contain the stalk and globular head Viruses generated via reverse genetics, containing chimeric haemagglutinins with mismatched neuraminidase genes, can also be used in anti-HA-stalk-specific neutralisation assays. Plaque reduction neutralisation assays are the preferred format currently. Other microneutralisation assays are carried out in 96-well format using nucleoprotein (NP) staining, haemagglutination assay and cytopathic effect end point read-outs.

| ELISA for influenza serology
An advantage of the ELISA is that a wide range of antigen preparations can be used as target antigens. Such preparations can be whole-inactivated virus, subunit vaccines or purified HA1; therefore, antibodies towards specific regions or formulations of interest can be detected. In addition, IgG subclasses and immunoglobulin isotypes can be quantified in serum by the use of subclass/isotype-specific conjugated secondary antibodies. ELISAs allow measuring the actual amounts of immunoglobulin, and the use of standards allows good intra-and interlaboratory standardisation. The ELISA method is also suitable for high throughput and is cost effective.
Comparison between HI assays and ELISAs has shown some correlation; however, the mean fold increase in HI titres is usually higher than in ELISA IgG titres. It should be noted that antibodies detected in HI assays comprise only a minor fraction of the total IgG antibody response. ELISA data were also compared to VN assay results in a clinical study of a non-adjuvanted H5N1 vaccine candidate, showing correlation between HA-specific IgG levels and VN titres 14 .

| APPROACHES FOR THE EVALUATION OF ANTIBODY-MEDIATED ENHANCED INFECTION
Some reports indicate that vaccine-induced immune mediators may These results have relevance for assessing "universal" vaccines that have been designed to target the HA2 region. Studies should include screening for possible vaccine-induced disease enhancement. Relevant in vitro assays include those measuring Fc-mediated influenza virus uptake using relevant cell lines (eg human epithelial cells, monocytes). Proposed read-outs are plaque assays or anti-NP immunostaining, or real-time PCR for viral RNA quantification in highthroughput assays. In vivo studies in pigs can also be used to assess vaccine-induced disease enhancement.

| Influenza antibody profiling by protein microarray
An alternative approach to current serological assays is the use of protein microarrays that can simultaneously measure antibody re- The protein array was successfully applied using pre-and post-H1N1pdm09 human sera 15 . There was good agreement with HI assay in terms of positive responders. However, the two assays varied when serum was assayed over a time course whereby the protein array assay appeared to detect positive responses earlier. This may be because the protein assay can detect broader antibody responses which are present following multiple previous exposures. Interestingly, antibodies against H9 were also detected in some sera in this study even though it is unlikely these subjects would have been naturally exposed 15 . This is perhaps due to the increasing breadth of antibodies made against conserved epitopes over a person's lifetime that can cross-react with multiple subtypes. Whether these antibodies offer any protection remains to be determined.
Future work should include more comparative experiments between the protein array and biological assays to assess the functional relevance of the protein array data.

| MEASUREMENT OF ANTINEURAMINIDASE INHIBITION TITRES BY ENZYME-LINKED LECTIN ASSAY
Traditionally, anti-NA assays have used the thiobarbituric acid (TBA) method. To reduce the use of highly hazardous chemicals, a mini-TBA assay has been established 18 . A more practical alternative assay for measuring human anti-NA antibody titres is the enzyme-linked lectin assay (ELLA) 19 .
The ELLA method consists of coating plates with fetuin, a substrate for NA, and adding test serum and virus and/or NA. Active NA cleaves the terminal sialic acid residue from the fetuin, leaving an exposed galactose molecule. Peanut agglutinin (PNA) conjugated to horseradish peroxidase is then added whereupon the PNA binds

| ELISpot
The enzyme-linked immunosorbent spot (ELISpot) assay is used to Although the ELISpot assay performance is straightforward, there are a number of variables that must be considered for assay standardisation both within and between laboratories, for example peripheral blood mononuclear cell (PBMC) preparation procedures, reagents, incubation periods and the stimulating antigen used. It is also necessary to use clinical trial data and human subjects in order to ascertain correlates of protection.

| Intracellular cytokine staining assay
The intracellular cytokine staining (ICS) assay can be used to evaluate

T-cell immunity and correlates of protection for influenza vaccines.
This assay is currently being standardised through the European Union funded UNISEC project.
The ICS assay measures multiple cytokines and lymphocyte sub-

| Cytotoxicity assays
"Universal" and novel influenza vaccines frequently aim to target the cytotoxic T-cell (CTL) response, and therefore, relevant assays are needed. It has been shown that CTL responses can be a valid correlate of protection against symptomatic influenza disease 26 .

CTLs recognise infected cells through T-cell receptor interaction
with major histocompatibility complex (MHCI) molecules coated with peptides, followed by target cell killing (by lysis and/or apoptosis mediated by release of perforin, granzyme and FasL expression). Activated CTLs also produce cytokines such as IFNγ and TNFα. A summary of established CTL assays is presented in Table 3.

| Antibody-dependent cellular cytotoxicity assay
The antibody-dependent cellular cytotoxicity (ADCC) assay is a rel-

| Antibody-dependent respiratory burst assay
Antibody-dependent respiratory burst (ADRB) assay is an assay to measure the combined (antiviral) effects of antibodies and neutrophils 28 . The assay was first described for malaria and was well correlated with protection 28 . The ADRB assay is a relatively simple technique to assess antibody functionality (similar to ADCC) requiring only luminescence equipment. The ADRB assay can be performed in high throughput, and it can be standardised. Two potential assay setups 29 are described in Table 5.
The relevance of the ADRB for influenza is still being explored, and a number of issues remain to be investigated; these include the following: correlations between formats using plate-bound antigen T A B L E 5 ADRB assays and those using infected cells and the use of a reference standard on every plate and correlation with other (ADCC, ELISA IgG) assays.

| DISCUSSION AND CONCLUSION
In summary, establishing "universal" immunoassays for "universal" Steps towards standardisation of immunoassays between laboratories are the harmonisation of assay protocols, use of common standards and regular proficiency assessment of laboratories. Although the harmonisation of assay procedures might be feasible, standardisation of critical reagents will be more difficult, if not impossible, primarily due to local sourcing of reagents. Harmonisation of methods does not always lead to improved agreement between laboratories, as demonstrated for the HI assay 30 .
Each laboratory should aim to appropriately validate the immunoassays used during the entire process of vaccine development from antigen discovery up to establishment of correlates of protection, including the different steps of quality control (eg potency assays), animal studies and clinical development. Global standardisation of the immunoassays is the ultimate goal; however, there is still a long way to go.
The CONSISE consortium, where more than 100 members from 40 countries share study and laboratory assay protocols and other information on the Internet with free access, is a good example of efforts towards the standardisation of immunoassays. CONSISE developed new consensus protocols for the HI and VN assays, and member laboratories have aligned their assay parameters to the consensus protocols.
There are different levels/types of standards: -WHO International Standards-highest level -National/regional/pharmacopoeial standards -Working reagents -In-house standards -Run controls Possible approaches towards harmonisation of immunoassays could be the establishment of antibody/serum standards in case the vaccine candidate targets antibody responses. However, they might need to be product-specific if particular epitopes are targeted. Standards for specific biological analytes (eg cytokines) can also be used. Cell standards for cellmediated immunity assays have been made in the past and more could be generated.
Proficiency testing and external quality assessment are important for the evaluation of the process of harmonisation and standardisation. Testing a laboratory's quality system with samples of known, but undisclosed content will give an overview of the laboratory's performance and will provide a comparison with that of other laboratories.
A clear need for further collaboration, networking and international communication regarding the harmonisation of immuno-assays was identified, and CONSISE could be the venue for these activities.
The standardisation should be initiated early, before methods drift too much. The European-American partnership established during the organisation of this workshop was recognised as a clear asset in this direction and will be sustained.