The challenge of 21st century influenza

Conventional influenza vaccines use an egg-based culture and harvest process. This is slow and inflexible compared with emerging cell culture-based approaches that respond rapidly to the influenza virus's inherent ability to 'drift' or, more dangerously, 'shift' — a critical factor that would arise in the event of a pandemic. ImmunoBodies, created by UK-based ImmunoBiology (ImmBio), is a vaccine under process development at Eden Biodesign (UK) that uses a baculovirus expression system for production. The vaccine is constructed from part of the flu hemagglutinin protein fused to the human dendritic cellbinding Fc region of human immunoglobulin, enabling it to efficiently trigger a broad, protecting response. In the event of a new strain being out of the vaccine's protective range, the construct can be rapidly re-engineered. The baculovirus manufacturing capacity can also be augmented to produce large volumes.

The utility of influenza vaccination is well established, especially in 'at risk' groups. Despite this, infections result in approximately 300000–500000 deaths worldwide every year. Inoculation with a traditional flu vaccine (delivered as an annual trivalent vaccine) leads to a raised antibody response to virus hemagglutinin protein, which protects against anticipated major strains. However, the genome of the influenza virus is encoded as RNA, making it prone to mutation. It frequently evolves into serotypes that evade the immune memory that has been raised in response to previous vaccinations or to formerly prevalent serotypes. In the event of a major 'shift' in the strain, such as from a reassortment of avian influenza capable of human-to-human transmission, the lack of prior immunity in the population could result in a pandemic and a high-mortality rate.

Of the three major pandemics in the last century, the 'Spanish flu' was the most deadly, killing more people than World War I, which had just preceded it. With the current prevalence of avian flu as a source of genetic reassortment, a significant elapsed time since the last major shift and the recognition that egg-based flu vaccines cannot meet anticipated demand, concern has grown. Vaccines against strains originating from avian flu may achieve poor yields in egg-based systems, meaning that such vaccines may require modification, which could potentially erode their efficacy and extend time-to-development. There is a need for alternative systems and many companies have focused development on mammalian cell-based manufacturing approaches, but these also have limitations.

Influenza

The influenza virus belongs to the Orthomyxoviridae family. It is an enveloped RNA virus with a segmented genome consisting of eight single-stranded negative RNA segments. The virus is split into three subtypes — A, B and C — based on antigenic differences in two of its structural proteins: the matrix protein (M2) and the nucleoprotein. As infection with the C subtype is relatively mild, vaccination is typically against the A and B subtypes.

Projecting from the viral envelope is a conglomeration of three proteins: hemagglutinin (HA), neuraminidase (NA) and the matrix (M2) protein. Influenza A viruses are subtyped according to the HA and NA proteins they display on their viral envelope. Although 14 HA proteins and 9 NA proteins have been identified, only a limited number are infectious to humans (H1, H2, H3, N1 and N2). The most common combination in circulation is the subtype H3N2. Distinct strains can be further identified by the virus type, geographic origin and year of isolation.

To cope with this variety, current vaccines are trivalent — with two components against A and one against B. Through the World Health Organization (WHO) surveillance network, which tracks strain changes, manufacturers that supply vaccines to the US and Europe are advised in February which subtypes are expected to be prevalent throughout the following winter when influenza incidence is highest. Typically, at least one component must be altered each year. The antigen composition recommended by WHO may need modification if the strain grows poorly in eggs, a situation that could compromise efficacy. To begin the Northern Hemisphere seasonal vaccination programme, large volumes of vaccine product need to be available by September.

Egg technology

Influenza vaccines are made by inserting the live flu virus into fertilized chicken eggs, then purifying and inactivating the resulting egg-adapted virus to produce trivalent inactivated virus (TIV). TIVs represent the majority of licensed and marketed flu vaccines worldwide. There is also the recently licensed, egg-derived, live attenuated influenza vaccine (LAIV), produced by MedImmune (MD, USA). Given the February-to-September time constraint, vaccine manufacturers have approximately 6 months to develop, manufacture and release millions of doses of trivalent vaccine. This short timeline necessitates manufacturers to speculate each year on likely strains so that they can commence production of at least one or two strain-specific bulk antigens in advance of knowing actual strain requirements.

This tight schedule leaves little time to optimize and validate the production process for the new virus strain. Consequently, the purification stream used for egg-derived flu vaccine involves a trade-off between the need to specifically purify the antigens from the crude egg harvest, and the need to have a process that is not so specific that it no longer works if the properties of the HA and NA antigens change because of drifting or shifting. This trade-off does not lend itself to a stringent purification stream with predictable yields.

The uptake of seasonal vaccination has increased, but although manufacturing capacity has grown, it is still outstripped by demand. Early assumptions were that the global capacity of approximately 300 million doses per annum could be tripled in a pandemic situation by producing only a monovalent rather than a trivalent vaccine. However, the low-natural immunity in the population to a new virus may require a high-dose, possibly in a boost regimen. Switching from seasonal flu capacity, built to meet elective seasonal vaccination in developed countries, is also grossly insufficient for potential global needs. Two of the last three pandemics originated in the Far East, where avian flu is most prevalent and where the cost of vaccines is high. Antigen-sparing approaches, notably through the use of adjuvants, may enable reduced unit dosage, which could increase vaccine coverage from a given manufacturing facility.

New developments

The substantial shortfalls in the current process have initiated development of alternative, non-egg-based manufacturing systems. These are primarily cell culture flu vaccines, such as those developed by Novartis (Switzerland) and Solvay (Belgium), which both produce vaccines in Madin-Darby canine kidney cells; and by Sanofi-Pasteur (France) and Crucell (The Netherlands), which are using Crucell's PerC.6 human cell line.

For these vaccines, live flu virus is used to infect cells in culture. Once the infection has propagated through the cells, the live virus is harvested and inactivated in much the same way as egg-based vaccines. These systems rely on being able to generate a seed stock of live virus to use throughout the production campaign. If an avian influenza pandemic breaks out, it has been predicted that the live strain will require high-containment levels during vaccine production. This could limit available manufacturing capacity to those manufacturers that possess the required contained facilities. The WHO website contains numerous articles on the subject of biosafety risk assessments for avian flu and other pandemic influenzas.¹

Genetic approaches

Alternative approaches have been developed that use the genetic sequences that code for virus surface antigens rather than relying on live virus seed stocks. The genetic sequences are propagated in a form that does not produce infectious virus and does not require high-containment manufacturing facilities. The time required to produce these strain-specific genetic sequences for the commencement of a manufacturing campaign is usually weeks, or even months, shorter than the time required for laying down live viral seed stock. Genetic manipulation should reduce annual time-to-clinic and market. It is also expected to lead to marketed products that will be cheaper to manufacture, while also being potentially easier to transfer and license at additional manufacturing sites, should the need arise.

When a genetic approach is used to produce purified recombinant HA proteins to incorporate into a vaccine (such as Protein Science's (CT, USA) FluBlØk recombinant vaccine, containing trivalent HA antigen made in insect cells), there is a risk that the change in HA strain will lead to altered behaviour of the protein in the purification process. In this regard, the PowderMed vaccine, developed by PowerMed (UK), that involves inoculation with DNA coding for the HA antigens, rather than the HA antigen proteins themselves, has an advantage as the purification of DNA should not vary much with changes in strain. Another approach that could reduce the potential for variability in production quality involves engineering fusion proteins in which only the HA region varies, while the rest of the fusion protein remains the same. This could be exploited as the basis for chromatographic or membrane purification techniques.

One genetic product in development is based on ImmBio's ImmunoBodies platform technology. The product originated from oncology work conducted by Scancell (UK) that ImmBio adapted for anti-infective applications. The resulting HA–Fc fusion protein vaccine uses baculovirus vectors that are genetically engineered to carry the DNA sequence of chosen HA molecules fused to the DNA sequence of the Fc portion of human immunoglobulin.² The vectors are used to infect insect cells in culture, causing the cells to produce a protein containing the combined HA antigen sequence and the Fc protein. The fusion protein can be isolated using the same methods for isolating monoclonal antibodies to yield a highly-purified protein for injection.

The presence of its Fc portion means the fusion protein targets dendritic cells, leading to stimulation of T-cell mediated response, as well as an antibody response to the HA portion. The combined response should provide protection not only at commencement of infection when antibodies can bind to influenza virus outside cells, but also at the subsequent stage of viral replication. Infected cells express some viral antigens on their surface and the T-cell response induced by the fusion protein will stimulate their destruction, halting the virus's ability to replicate further.

Choice of system

Companies developing influenza vaccines should use the following key criteria to select from four main choices for the expression of a recombinant protein (i.e., bacterial cells such as Escherichia coli, yeast, insect cells and mammalian cells):

• Speed: can it meet the annual production cycle or respond to an emergent pandemic?

• Functionality: does the system express an efficacious product? (The HA requires correct folding, and the Fc requires glycosylation for receptor binding.)

• Cost: how does the cost compare to current marketed influenza vaccines?

• Availability: are scaled-up manufacturing facilities and regulatory experience available?

By using this rating process (Table 1), ImmBio identified the baculovirus expression system as the system of choice.

Table 1 The rating process that was used by ImmBio to identify the baculovirus expression system as the system of choice.

The insect cell and baculovirus system have the necessary requirements for the manufacture of HA–Fc influenza vaccine. Recombinant baculoviruses can be constructed quickly and, unlike mammalian cell lines, do not require a period of selection and expansion before production can begin. The yields of similar proteins are also fairly uniform and can accommodate the scale required for real use. Cervarix, GSK's (UK) vaccine for human papillomavirus, is manufactured using baculovirus technology, as is Protein Science's FluBlØk, which consists of three recombinant hemagglutinin (rHA) proteins. Insect cells have been shown to have glycosylation pathways that are similar to those of mammalian cells, so the glycosylation of the HA region is similar to that of mammalian-derived vaccines. A key requirement is to achieve a high-yield of purified HA–Fc vaccine by optimizing culture conditions by proprietary expertise developed for virus-like particle production using baculoviruses. A variety of fermenter technologies, including fed batch, perfusion and 'Wave bag' technology, are amenable to insect cell culture. Additional benefits from an insect cell system include:

• Insect cells grow faster than mammalian cells, reducing production times and costs of batch production.

• Insect cells cannot be infected by mammalian viruses, so the need for viral removal steps and validation is reduced, adding to the safety profile of insect-derived vaccines and reducing time and cost in reaching clinical trials.

Ahead of an emergent strain shift capable of leading to a pandemic, a suitable vaccine cannot be definitively identified. Some national policies are stockpiling H5 vaccine, assuming that the shift from the currently prevalent H3 will be to H5 and that cross-strain protection within H5 assists in at least some degree of protective priming. However, neither of these two assumptions may hold true. To speed up approval of a new vaccine against an emergent strain, both FDA and the European Agency for the Evaluation of Medicinal Products (EMEA) have established approval processes, though with some differences. The EMEA's declared approach is to approve a 'mock dossier' whereby additional approval requirements are modest if prior mock dossier approval has been gained. Working within the emergent regulatory framework is essential throughout product development, and, to date, EMEA approval has been achieved by GSK and Novartis.

Key points

Summary

By using a platform technology, as ImmBio is doing, the variables associated with having to redevelop vaccine production each year to adapt to the evolution of predominant flu strains can be reduced. However, companies seeking to develop a commercially successful flu vaccine need to understand that there are many parameters that add up to making a successful marketed product. A key starting point involves attention to developing and documenting a scalable, rapid and low-cost manufacturing process and its accompanying analytical technology. By dealing early with problems associated with late-phase product development, total product development times and costs can be significantly reduced.

Amanda Shipman is a project manager at Eden Biodesign, National Biomanufacturing Centre in Liverpool (UK).

Graham Clarke is chief executive at ImmBio, Babraham Research Campus in Babraham, Cambridge (UK).

References

1. www.who.int

2. www.immbio.com/platforms/immunobodies