Meeting report and review: Immunological assays and correlates of protection for next‐generation influenza vaccines

Abstract Background This report summarizes the discussions and conclusions from the “Immunological Assays and Correlates of Protection for Next‐Generation Influenza Vaccines” meeting which took place in Siena, Italy, from March 31, 2019, to April 2, 2019. Conclusions Furthermore, we review current correlates of protection against influenza virus infection and disease and their usefulness for the development of next generation broadly protective and universal influenza virus vaccines.


| INTRODUC TI ON
In 1972, Hobson and colleagues published a seminal paper showing that a hemagglutination inhibition (HI) serum antibody titer in the range of 1:18-1:36 provided 50% protection from influenza A or influenza B virus challenge in adult volunteers. 1 Of note, many subjects in this study had antibody titers due to natural infection, not vaccination and the challenge strain was partially attenuated. 1 Since then, the HI titer has been generally adopted as a correlate of protection against influenza virus infection. 2,3 Importantly, Hobson and colleagues stated in their paper: "…an unusual finding was that volunteers with no detectable pre-challenge antibody often seem to be less susceptible to infection than those with pre-challenge antibody in low titre." 1 Since then, several other immune markers have been reported to correlate with protection, several of them independently of the serum HI antibody titer. These include interferon γ secreting cells, CD8 and CD4 T cells in peripheral blood, 4-8 neuraminidase (NA) inhibition (NI) antibody titers 9,10 antibodies measured by single radial hemolysis (SRH), 11,12 nasal IgA 13 and, more recently, antibodies that target the hemagglutinin (HA) stalk domain. 14 In contrast to HI antibodies, which lead to narrow, strain specific protection, these potential new "correlates of protection" are often based on conserved viral epitopes that are also targeted by next-generation influenza virus vaccines. 15,16 New immunological assays and novel correlates of protection will be needed to assess immunogenicity and protective efficacy of the next generation of influenza vaccines, including broadly protective universal influenza virus vaccine candidates. This was the central theme of the "Immunological Assays and Correlates of Protection for Next-Generation Influenza Vaccines" meeting, orga-

| THE VALUE OF CORREL ATE S OF PROTEC TI ON FOR VACCINE DE VELOPMENT: TARG E T DISCOVERY AND DE-RIS KING OF PRODUC T DE VELOPMENT
According to Plotkin, a correlate of protection is "responsible for and statistically correlated with protection." 17 An absolute correlate of protection is a threshold of a correlate, for example, a certain titer that is protective. Furthermore, a surrogate marker of protection is an immune marker "for the true immunologic correlate of protection, which may be unknown or not easily measurable" but is significantly associated with the true correlate. 17 Correlates of protection can be obtained by studying natural immunity in animal models or humans (eg, in cohort studies or human challenge experiments). Based on these correlates, vaccine targets can be defined. Vaccines can then be designed to induce the immune response that has been identified as a correlate of protection in a (natural) infection setting. This vaccine-induced immunity might then also correlate with protection. However, in some cases differences in immunity induced by natural infection versus vaccination against a certain target might exist and the vaccine-induced protection might not be equal even though the same response is measured.
Correlates or surrogates of protection can also be identified in large efficacy trials of vaccines by statistical analysis of immune readouts that correlate with protection.
When a correlate of protection is available, it simplifies vaccine design and facilitates vaccine development. Vaccines that are based on inducing an immune response that correlates with protection have an advantage since a readout exists that can be used for optimization and evaluation of immunogenicity. Furthermore, having a correlate of protection de-risks vaccine development since one can assess in the pre-clinical and early clinical phase if the vaccine induces the expected immune response. If an absolute correlate of protection exists, early phase clinical studies will provide an indication if the vaccine will be protective. As an example, developing an HI antibody inducing vaccine allows optimization of the vaccine in pre-clinical models to induce high titers of HI antibodies. This can be measured readily and success can be determined by the antibody titers reached. If the same vaccine then induces HI titers above a pre-defined (arbitrary) threshold in early (eg, Phase I) clinical trials, confidence increases that the same vaccine will show protective efficacy in late stage trials.
In addition, influenza virus vaccines have been licensed in some jurisdictions/under some circumstances based on a correlate of protection, whereas in many cases a new vaccine needs to demonstrate clinical efficacy against influenza virus infection/influenza disease.
Once a vaccine has obtained licensure, the existence of a correlate of protection makes changes to the product easier, as comparability can be shown with less expensive immunogenicity trials rather than full efficacy trials.

| E XIS TING AND NOVEL CORREL ATE S OF PROTEC TI ON
In addition to the serum HI antibody titer and SRH zone size, several other potential immune correlates of protection have been identified and are the subject of intense ongoing investigations with the goal of improving the ability to predict vaccine performance.
Updates on several of these investigations were provided at the meeting ( Table 1). Examples of these alternative immune correlates include the serum virus neutralization antibody titer, 18,19 which often, but not always, aligns with the serum HI antibody titer. In addition, serum neuraminidase inhibition (NI) antibody titers have been identified as independent correlates of protection in field studies by Monto et al and Couch et al. 9,10,20 Furthermore, this is supported by recent data from an H1N1 human challenge model. 21 Of note, anti-HA stalk antibodies can interfere with H6NX-based NI assays which have to be taken into account in their interpretation. 22,23 Anti-NA enzyme-linked immunosorbent assays (ELISAs), which do not suffer from this shortcoming, have recently been used to show a negative correlation between anti-NA titers and virus shedding in humans in a cohort study. 24 In the same cohort, anti-HA stalk antibodies have been shown to be an independent correlate of protection against both infection as well as symptomatic disease. 14 Cross reactive CD4+ and CD8+ cells have also been identified as correlates of protection in human challenge and cohort studies. [4][5][6][7] Other immunological markers, including antibody effector functions as measured in antibody-dependent cellular cytotoxicity (ADCC) or antibody-dependent cell-mediated phagocytosis (ADCP) assays, 25-28 complement activation, mucosal antibody levels, entry inhibition titers as measured by pseudotype particle entry inhibition assays, 29 antibodies to the ectodomain of the matrix 2 ion channel (M2e), antibodies to matrix protein 1 (M1) and nucleoprotein (NP), 30 influenza virus protein arrays (IVPM), 31-34 and many others are currently being investigated to assess whether they correlate with protection. Importantly, systems immunology approaches are being used to identify new immunological markers that could then be tested for their potential to predict whether protective immunity was induced through vaccination. 35,36 It is important to note that some immune responses, for example, the priming induced by H5N1 LAIVs, cannot currently be measured with existing assays. However, this priming is clearly the cause of a strong recall response when subjects who received H5N1 LAIV are later boosted with H5N1 IIV. 37

| IMMUNOLOG I C AL A SSAYS: FROM RE S E ARCH TO CLINI C AL TRIAL S
A large number of talks during the meeting focused on reviewing existing assays and the development of novel assays for various immune markers and correlates of protection including HI, MN, NI, ADCC, ADCP, anti-stalk antibody assays, B-cell assays, T-cell assays, pseudoparticle entry inhibition assays, and others. The discovery of new immune markers and correlates of protection often starts in academic laboratories where novel assays are developed. While academic laboratories can be highly innovative, less weight is often put on assay qualification, improvement of reproducibility, and reduction of inter-laboratory variability.
To address these important issues, assay standardization and harmonization were discussed during the meeting. Without standardization, defined thresholds or value ranges for correlates of protection may be questionable due to inter-laboratory variability.
Therefore, it is in the interest of the entire vaccine community to achieve levels of assay standardization that confer confidence in any correlates of protection that may be determined in the future.
While the oldest of the used assays, the HI assay, is still variable from laboratory to laboratory, the use of an international biological standard as well as assay protocol harmonization can substantially reduce variability, as demonstrated in international studies coordinated by the Consortium for the Standardization of Influenza Seroepidemiology (CONSISE) [38][39][40] and others. 41,42 A highly collaborative approach, involving academic, regulatory, and industrial laboratories, toward standardization and harmonization of HI and VN assays, is ongoing and was described at the meeting 40 (http:// www.flucop.eu/). However, standardizing and harmonizing assays can be difficult and resource intensive, especially for complex methods that measure cell-mediated immunity or for quantitative B-cell assays. Some assays, like ELISAs, virus neutralization assays and NI assays may be easier to standardize whether appropriate standards are or become available. 39,42 For instance, for the assessment of a type of immune response that has not been considered as a correlate in the past, the generation and characterization of an international standard for measuring group 1 stalk antibodies are currently underway as a first step (spearheaded by NIBSC). If proven to be useful, this standard will become available to the scientific community. Serum single radial hemolysis antibodies Single radial hemolysis assay 11,12 Nasal IgA Enzyme-linked immunosorbent assay (ELISA) 13 Serum virus neutralization antibody titer Virus neutralization (VN) assay 18,19 Serum neuraminidase inhibition (NI) antibody titers Antibody-dependent cellular cytotoxicity (ADCC) Antibody-dependent cellmediated phagocytosis (ADCP) Antibody-dependent cellular cytotoxicity (ADCC); antibody-dependent cell-mediated phagocytosis (ADCP) reporter and functional assays using specific target proteins [25][26][27][28] Complement activation Complement fixation assay
For pivotal clinical trials (Phase III), assays should be fully validated before initiation of testing. For current seasonal influenza A H3N2 viruses, which do not replicate in mice, a hamster model is now available. 50 The small size of these animals makes them more convenient to work with and allows larger numbers per group than the ferret model (which also supports replication of H3N2 viruses). Of note, recent H3N2

| S TUD IE S TO DEFINE NOVEL CORREL ATE S OF PROTEC TI ON
isolates only amplify in the upper respiratory tract of ferrets but not in the lower respiratory tract. 51 The ferret model has the advantage that these animals show clinical signs similar to humans including sneezing, nasal discharge, lethargy, and fever. This, together with their ability to transmit virus in aerosol-only and con-

| WHERE DO WE G O FROM HERE?
Two recap and discussion sessions were incorporated near the end of the meeting to synthesize the large amount of data presented over the previous two and a half days and to try to reach a consensus on where the field currently stands and what is needed to continue momentum ( emerging influenza virus. Depending on the emerging strain/subtype, stockpiled vaccines could be used, but they are limited in doses and will likely have little impact in disease burden in the general population.
Furthermore, a few countries can afford this approach. The group felt that it is now time to take what has been learned in terms of immune markers and correlates of protection and start to translate this knowledge into broadly protective and universal influenza virus vaccineswhile monitoring and further exploring novel correlates of protection in parallel. The participants also agreed that more work was needed on so called incremental improvements that could, in the short term, enhance the effectiveness of current seasonal vaccines. The meeting attendees were convinced that identifying influenza correlates of protection and the continued development of relevant immunological assays remain extremely important and timely and that there should be a regular series of meetings and workshops to facilitate these efforts.