Influenza vaccination plays an important role in reducing the severity of flu epidemics and reducing the disease burden caused by seasonal influenza. However, due to the way in which the predominant strains drift and shift from one flu season to the next, the candidates to be included in the vaccine(s) need to be reassessed each year and one or more strain may have to be replaced in those doses prepared for the coming year’s immunization program.
The historical method for checking whether a novel vaccine is likely to be effective involve community-based trials where many thousands of people need to be immunized, in the hope that sufficient numbers will encounter the target virus, to enable the vaccine’s protective index to be calculated. During flu seasons there is no guarantee that enough subjects will come into contact with the virus for the trial to achieve statistical significance, even with a very large pool of people being immunized. This is particularly likely if the amount of circulating virus is low at the time the trial is carried out. The presence of other circulating viruses can also complicate the results.
In addition, owing to the complex program required for the testing and licensure of new flu vaccines, seasonal changes in strains are not tested for efficacy in community trials but manufactured to Good Manufacturing Standards (GMP) and released after safety checks to ensure timely distribution for the forthcoming season. This raises the potential for mismatched or egg-adapted strains to be included reducing the effectiveness of the vaccines.
These problems have led to the growing acceptance of the human challenge trial (HCT), where studies involve healthy volunteers being infected with the target virus within a controlled environment. This allows the effectiveness of an influenza vaccine at preventing disease to be assessed and also any potential to cause unwanted side-effects. Such a study is generally guaranteed to achieve statistical significance unlike a community-based trial and additional parameters may more easily be measured and monitored, including whether the vaccine stops infection completely, reduces the duration of infection, or attenuates the symptoms experienced by the patient.
Running a Human Challenge Study
Cohorts of anywhere between 25 and 50 healthy volunteers are enrolled into each arm of a human challenge study (vaccine and placebo). However, before they can be enrolled, they have to be checked for the presence of other infections that might impact the study, and also whether they already have any immunity to the challenge agent that will be used. If they meet the above conditions as well as the inclusion/exclusion criteria, then they will be immunized with the test vaccine (prime) three or four weeks before the challenge part of the study takes place.
During the challenge phase, subjects are initially quarantined for two days so that any recently acquired infections can manifest themselves, and if there are none, they will be admitted to an isolation ward. Within the ward the subjects are placed in individual containment (rooms with ensuite facilities) and are exposed to the challenge virus, which is normally administered using an intranasal spray. This controlled inoculation typically has an infection rate of 80–90%, compared to the 5–10% that is normally seen during a typical seasonal influenza epidemic.
If the subject becomes infected, the virus will be replicating rapidly after about 12 hours, and will reach peak shedding in two days with maximum symptoms observed at day three. After about the fifth day, the viral load will subside, and the subject should start to improve with full recovery after approximately a week.
The signs and symptoms that develop in the subjects are monitored throughout, including: body temperature, blood-pressure, pulse, lung function and mucus production. Host immune factors and viral fitness are also determined e.g. blood antibody levels and viral shedding (a count of both genomic copies and live virus present in the nasal passages). A symptom scorecard is filled out by the patient to give a sense of how unwell they feel. All of these measurements give a broad picture of the vaccine’s efficacy, and enable it to be quantified.
Because of the nature of challenge trials with a live virus, active infections and the requirement to prevent other infectious agents entering the ward, a challenge trial should be run in a dedicated unit within a biosafety level 2 compliant quarantine facility in a hospital and must obviously be conducted according to Good Clinical Practices (GCP). This unit should also have a large pool of potential (healthy) volunteers to draw upon for recruitment purposes.
Selection of a Challenge Agent
The two main options for an influenza virus to use as a challenge agent are the predominant Influenza A strains circulating globally: H1N1 and H3N2. cGMP manufacturing processes are available to manufacture such strains as safe challenge agents, either in eggs or in cell lines. These strains first emerged in the late nineteenth century, and have been responsible for multiple pandemics in the intervening years. The severity of disease associated with viruses of pandemic origin attenuates over time as a result of accumulated mutations in the genes associated with attachment; for example, the H1N1 pandemic of 1918 had a fatality rate of 2%, whereas the reappearance of a related strain, H1N1pdm in 2009 demonstrated a much lower fatality rate of 0.03%. Associated reductions in morbidity renders recent seasonal strains of H1 and H3 appropriate for use as challenge agents.
Signs and symptoms in community infections are similar for H1 and H3, typified by pyrexia, with a temperature of 39–40°C, alongside a sore throat, muscle aches, lethargy and a headache. Vomiting and diarrhea can develop, but occurrence is rare. The cytokines (cascades or storms) that are the cause of the most severe reactions in pandemic infection are not expressed in strains chosen as challenge agents as the genes responsible have been lost over time.
Several characteristics differ between the H1 and H3 strains and these may impact on choice of agent. Mean ages of infection with H1N1 tend to be a decade older than with H3N2, and hospitalization is more common with the latter. H3 infections usually result in a higher fever, leucopenia and raised levels of inflammatory biomarkers such as C-reactive protein. The death rate is usually higher in a regular influenza season where H3N2 predominates.
However, more moderate markers associated with H1 disease do not automatically mean that H1N1 is the ideal choice. The detection and measurement of any impact of interventions on symptoms is easier when they are more severe, and thus H3N2 may make the trial results easier to interpret.
It is critical that subjects enrolled in a challenge trial have not previously been exposed to the chosen challenge agent. If they have already developed an immunity to the strain of challenge virus that is to be used in the trial, it will make it impossible to assess the efficacy of the vaccine, as infection rates will naturally be suppressed.
Careful selection of the challenge agent is therefore essential. As an example, approximately 95% of the population may have previously been exposed to the H1N1 2009pdm strain owing to its widespread distribution and historically high attack rates, so recruiting healthy, sero-sensitive subjects can be extremely difficult. In contrast, closer to 50% of the population have been exposed to recent, nonhemagglutinating H3N2 strains since 2013, and therefore a subject cohort for a trial involving this virus should be simpler to recruit to (95 vs 50% drop-out rates).
Development of a New Challenge Agent
In the 2014–15 influenza season, the most common circulating H3N2 serotype in NW Belgium was an A/Switzerland/9715293/2013 [H3N2]-like virus (A/Switzerland is one of the strains used in the seasonal influenza vaccine). SGS isolated a related strain as part of a community surveillance program to develop as a new challenge agent. This new virus was named A/Belgium/4217/2015. The challenge agent is a drifted strain of the Swiss-lineage virus grown in eggs.
With the virus now manufactured to a high titer, the next stage was to carry out in vitro testing for contaminating, adventitious agents (viruses, bacteria and fungi) to establish purity. The presence of any contaminating agents could affect the quality of the inoculum and lead to potential co-infections, adverse events or other issues affecting the accuracy of human challenge trial results. The manufactured product was therefore tested for rhinoviruses (the common cold), RSV and other respiratory tract viruses, as well as less likely contaminants e.g. hepatitis A, B and C, HIV, measles, mumps and rubella plus other families of infectious agents such as the mycoplasmas.
The sheer number of potential contaminating agents make purity and safety testing by far the most time-consuming and challenging aspect of the development program for any new challenge agent – more than 30 potential pathogenic contaminants must be tested for, plus a whole range of toxins and allergens.
A/Belgium 2015 in Human Trials
The new strain, A/Belgium/4217/2015 H3N2, has now been successfully used in human challenge trials, and top-line data was presented at the World Vaccine Conference in May 2017. When objectively measured, its performance characteristics represent a significant improvement on previous GMP-manufactured influenza viruses relative to viral shedding and attack rates for example.
Approval for use in clinical trials was granted in October 2016, and recruitment for the first-in-human study began the next month. The proof-of-concept or characterization trial commenced in December 2016 with three cohorts, each comprising 12 subjects, being challenged with increasing doses of the virus within the SGS human challenge unit. Over the course of the 13-day study, doses of 105, 106 and 6.78 x 106 were tested and virological, clinical laboratory and symptomological measurements collated.
The aim of the open label, ascending dose study was to determine the optimal dose of live virus that would have an acceptable safety profile while giving an attack rate of at least 60%. Its secondary objectives were to determine the safety and tolerability of intranasal inoculation, the etiology of influenza signs and symptoms over time, and the number and severity of respiratory symptoms occurring.
To enable the study, subjects were pre-screened for prior infection with virus similar to the challenge strain. Instead of the traditional hemagglutination-inhibition (HAI) assay, a novel microneutralization test (MNT) was employed to determine the presence or absence of A/ Belgium/4217/2015 neutralizing antibodies ahead of the challenge test. The new MNT assay was developed in response to a loss of ability to hemagglutinate noted in almost all circulating H3N2 strains since 2010. MNT cut-offs have not yet been established, however it was decided subjects would be included if they had a titer of <20 (lower limit = 10).
In this initial FIH study a pre-screening failure rate of 30–40%, (MNT <20 vs <10)was observed. A later, commercial study carried out after the 2016–17 H3N2 influenza season showed a pre-screening failure rate of approximately 50%. This failure rate still compares favorably to the failure rates in recent H1N1/2009pdm challenge agent studies, which are reported as being between 75% and 90%.
Once enrolled, subjects were isolated prior to being inoculated or dosed using a VaxINator device. This instrument delivers a nebulized charge of challenge agent into the upper airways, with a particle range of 30–100μm. The highest dose gave a 100% attack rate across all subjects irrespective of pre-screening MNT titer below 20.
In viral shedding studies, an increasing dose was associated with a steeper curve and higher peak viral load, as can be seen in Figure 1. This was reflected in larger vAUC values for both qRT-PCR and TCID50 measurements. The consistency of shedding within the cohorts was shown not only by the vAUC but also by the absence of significant blipping (intermittent shedding).
Figure 2. Viral Shedding: Mean Plot of Viral Load over Time
Signs and symptoms of individual subjects were also recorded via a symptom scorecard to assess the impact of viral infection upon the host. Symptoms were consistent across all cohorts with 11 developing typical symptoms of influenza in the lowest dose cohort, 10 with the intermediate dose, and 11 with the highest dose. Symptoms began about 30 hours after RNA shedding started. Overall, the signs and symptoms were consistent with an influenza-like infection.
Overall, the etiology of symptoms, changes in vital signs and deviations in peripheral blood markers (humoral and cellular), indicate that the novel A/Belgium challenge agent behaves like a wild-type influenza H3N2 strain.
As a human challenge agent, A/Belgium challenge results in high attack rates and measurable disease-related changes within subjects, allied to consistent variations in virological, hematological, clinical laboratory and host parameters.
Every cohort gave an attack rate well above the 60% that may be considered the minimum measure for a human challenge agent to be effective. A/Belgium, therefore, may represent a new generation of challenge agents, as it emulates community disease more closely, while retaining a strong fidelity to circulating, community strains of wild-type influenza virus.