Egg-Based Production of Influenza Vaccine: 30 Years of Commercial Experience

Despite improvements in technology and increased automation, egg-based vaccine production still has a long timeline.

The two influenza viruses that are medically and clinically important to the human population, known as Type A and Type B, are segmented, negative-strand genomes with an envelope that contains two major antigens, hemagglutinin (HA) and neuraminadase (NA), and a minor component designated M2. Figure 1(see PDF version for figures) shows the virus envelope with the HA and NA surface antigens. Throughout the vaccine manufacturing process—inactivation, extraction of the virus envelope, and purification—both the quantity and inherent antigenicity, or immunogenicity, of the HA and NA antigens must be preserved.

The HA and NA antigens undergo both mutation and reassortment, and, most significant, both antigens are subject to antigenic drift, subtle mutations that occur continually. Therefore, formulation of the influenza vaccine must be changed on a yearly basis. If two viruses co-infect a single host, the separate gene segments shown in Figure 1 can reassort to produce novel and unexpected combinations, some of which are extremely virulent in humans.

The efficacy of vaccines for Type A and Type B influenza is measured primarily by their ability to stimulate antibodies to the HA antigen. Antibody titers are often expressed in terms of hemagglutinin inhibition (HAI), which becomes a correlate of protection or immunity for influenza vaccines. M2, NP, and other antigenic components, although they probably play a role in cellular immunity, are not part of the yearly evaluation or licensing process for seasonal influenza vaccines.

Types of Vaccine
There are generally two types or classes of influenza vaccines: (1) inactive vaccines and (2) live attenuated vaccines. Inactivated vaccines, which account for more than 90 percent of vaccine sales worldwide, include vaccines made from (1) whole viruses; (2) viruses subject to two different splitting procedures (a simple detergent split with enrichment or a subvirion highly enriched for HA and NA); (3) virosomal vaccines, which are primarily licensed in the European Union (EU); and (4) influenza vaccines formulated with adjuvant to increase their potency. The vaccine primarily discussed in this paper is Fluzone?, a split (subvirion) vaccine manufactured by Sanofi Pasteur.

Live attenuated influenza vaccines represent only a small portion of the licensed influenza vaccines. Examples include FluMist?, made by MedImmune Vaccines Inc., and a similar formulation produced in St. Petersburg for use in Russia.

Seasonal Influenza
Seasonal influenza is the seventh leading cause of death in the United States and the leading cause of death in children ages 1 to 4 years. Ninety percent of deaths in people 65 and older are the result of influenza with associated pneumonia. Every year in the United States, approximately 36,000 people die, 114,000 are hospitalized, $600 million are spent in direct costs to the health care system, and the country incurs $1 billion in economic costs.

Figure 2 shows the seasonal occurrence of influenza; Figure 3 shows the absolute numbers of strains isolated in each season, as well as the proportion of strains isolated over the flu season. Figure 3 also shows that the number of strains isolated varies dramatically from season to season. Therefore, discussions about modifying formulations for the seasonal influenza vaccine by eliminating one strain must include careful considerations of the numbers and types of co-circulating strains. A bivalent vaccine produced for a flu season with three co-circulating subtypes would offer limited protection at best.

Figure 3 also shows that the peak incidence varies from season to season. The 2002–2003 season peaked during the week of February 23, 2003; the 2003–2004 season peaked during the week of November 29, 2003. Because it takes two to four weeks after immunization for an adequate immune response, vaccines must be manufactured and distributed well in advance of November to be effective in protecting against seasonal or pandemic influenza.

History of Influenza Vaccines
The seasonal influenza vaccine of the 1940s was a monovalent preparation; that is, the vaccine was formulated from a single strain of influenza. From the 1960s to early 1970s, seasonal vaccines were bivalent. Since 1978, vaccines have been trivalent with a strain of influenza B and two strains, H1 and H3, of influenza A in each formulation. From 1970 to 2004, the formulation changed approximately 40 times for one strain of the trivalent vaccine. Requests to change two of the strains in the vaccine were made eight times, and additional monovalent preparations to supplement the trivalent vaccine were made several times. On one occasion, all three strains in the vaccine were changed.

Even though the seasonal influenza vaccine is considered a conventional vaccine by the industry, new challenges with respect to timing and availability of strains and the composition of the influenza vaccine are the rule. As we say at Sanofi Pasteur, “when you’ve seen one influenza season, you’ve seen one influenza season.”

Typical Influenza Campaign
Eight discrete activities are associated with a seasonal vaccine campaign; these activities occur concurrently in the northern and southern hemispheres.

Surveillance, supervised by the World Health Organization (WHO), is continuous on a global scale for every month of the year. Strain selection is done twice a year, once for each hemisphere. Reassortant preparation, a step not necessarily apparent to the general public, involves co-infection of the wild-type and egg-adapted strain (PR/8) that have been developed to support egg-based production. The next step, potency reagent preparation, takes place after the strains to be used in a vaccine have been identified. This step involves the development of a release test for any new strain included in the vaccine. Antibodies to the new virus strain must be developed that can be used in a single radial immunodiffusion (SRID) assay for release testing by manufacturers.

The production of vaccines for both the northern and southern hemispheres takes place almost year round. Production is followed by release of the vaccine, which can be a lengthy process (discussed further below). Vaccine distribution must occur early enough for the vaccine to be available well before the flu season. Administration of the vaccine must be done early enough for individuals to accumulate enough antibodies to fight the disease.

The participants in annual influenza campaigns include WHO, national reference centers and contract laboratories responsible for surveillance, the development of reference strains, and reassortants; vaccine manufacturers, who are responsible for preparing the monovalent bulk production and formulation, as well as for filling and distribution; and regulatory agencies/national standardization laboratories, which review yearly license amendments when vaccine formulations change. Regulatory agencies are also responsible for reviewing changes in product labeling, labeling for new products, and changes in packaging and product inserts for new virus strains. In the United States, monovalent bulk vaccine requires a regulatory review and lot release. Standardization laboratories are responsible for the calibration of potency reagents.

Timetable
Figure 4 shows one cycle in the production of a monovalent lot of influenza vaccine. Note that the egg supply must be prepared well in advance of vaccine production dates. Once the egg supply is in hand and the seed is developed from the reassortant, “in-house” seed lots can be developed; this requires passing and expanding the seed lot a number of times until it reaches a level that can support commercial production. Once the seed lot is released, production can begin on a monovalent concentrate bulk batch for one strain to be included in the vaccine. Figure 4 shows one campaign for one strain (e.g., an H1 strain). Additional campaigns are required for the H3 and Type B strains to be included in the trivalent vaccine. After each monovalent batch has been prepared, the formulation stage and filling and packaging stages can move forward.

Figure 4 also shows some of the common constraints to the regulatory pathway, such as clinical studies that must be done with hundreds of healthy volunteers, both young people and elderly adults, to assess the safety of the vaccine and the strength of the HA titers expected from that particular formulation. Such studies are required in the EU, as are market authorization and releases by member states, before the vaccine can be delivered and the immunization campaign begun.

Egg Supply
Until recently, the egg supply was organized to support seasonal vaccine production, and there were often gaps of three or four months when eggs were not readily available. Clearly, this is an unsatisfactory situation in preparing for a pandemic. Working on pandemic preparedness with manufacturers, the U.S. government awarded Sanofi Pasteur a contract to improve the egg supply and develop new technologies and has asked manufacturers to provide expert knowledge and solutions to specific problems in vaccine manufacturing.

To address the variability in the egg supply, Sanofi Pasteur restructured its flock management so that embryonated eggs would be available to support vaccine production at full capacity throughout the year. Because embryonated eggs are themselves potentially susceptible to avian influenza, flocks associated with vaccine production are under strict contract and must be completely housed, monitored by veterinarians, and raised under biosecurity regulations. With government support, Sanofi Pasteur has also established contingency flocks as a backup against avian influenza and other risks.

Automated Process
The typical automated vaccine manufacturing process shown in Figure 5 takes approximately seven days for completion. The upstream process, which lasts for most of the seven days, begins with embryonated eggs brought in on a daily basis from biosecure flocks. A seed ampoule is used to inoculate the chick eggs during the inoculation phase. This is followed by a mandatory three-day incubation period during which the virus grows to ensure that sufficient quantities can support further manufacturing. After three days, all of the eggs are candled to make sure there are no cracks or contamination; the eggs are then chilled to 2? to 8?C to constrict vessels and make harvesting easier. The allantoic fluid is then harvested; a low-speed clarification process follows.

The first step in the downstream process is inactivation, which involves the addition of formalin to inactivate the virus. Extensive filtration and concentration steps yield a concentrate, which is then loaded onto zonal centrifugation equipment. The first purified bulk virus, which is recovered from the centrifugation process in a sucrose band, is split in a fragmentation step by treatment with Triton detergent. The material is then clarified by centrifugation to remove large particulates and treated with formalin in a second inactivation step. An ultrafiltration (UF) step is followed by terminal sterile filtration to generate one monovalent bulk concentrate.

The inoculation and harvesting steps are now 100 percent automated. In addition, although overall this is not a closed system, two distinct operations in the upstream process are closed systems: the harvesting and clarification steps. The first three downstream steps, first inactivation, filtration, and concentration, are also closed; and the final downstream steps: second inactivation, UF diafiltration, and sterile filtration, are closed. Material transported from one step to another in the closed-system portions of the processes is pumped, rather than poured from open containers, to prevent contamination by processing aids and viable or nonviable particulates.

Broad-Scale Manufacturing
The 28-week timeline shown in Figure 6 represents a worst-case scenario, in which one of the strains to be included in the vaccine is changed, making it necessary to create a new reassortant. Preparing high-growth reassortant seed requires a seven-week period to accommodate the preparation of new reagents and antibodies.

A large portion of the time in this timeline is focused not on manufacturing but on regulatory concerns, either regulatory review and release or internal testing, so that when seed lots, monovalent bulk concentrates, or pooled bulks are moved along, the quality of the material is assessed for sterility and potency.

Formulation, Filling, and Labeling and Packaging
Formulation is a manual process consisting of the addition of diluent to achieve the required 15 microgram antigenic component per strain (H1, H3, and B strain) for each dose of vaccine. Quality-control testing of the final container comprises several labor-intensive testing regimens to ensure that all release requirements have been met. Sterile filling of vials or syringes and subsequent inspection of the filled vials or syringes are completely automated. The labeling and packaging process is also automated; individually labeled vials are placed into cartons along with relevant product inserts, leaflets, and syringes. The packing of individual cartons into shipping packs, however, is another labor-intensive manual process primarily because of customer-specific or special orders, which are very difficult to automate.

-----------------------------------------------------------------
A considerable amount of
time is spent on steps beyond
the manufacturer’s control.
_____________________________

Rate-Limiting Steps in the Seasonal Vaccine Process
One of the most time-consuming steps in the seasonal process is the selection of updated reference strains each year. Delays are often justified because the epidemiology or transmission of a particular strain of influenza can change during the course of a flu season, and it is crucial that the best match for all three virus strains be included in the vaccine. Sometimes, one of the strains is not available as an egg-derived isolate. Current licensing requirements stipulate that isolates be passed exclusively in embryonated chicken eggs or cell cultures derived from embryonated chicken cells. If that strain is missing from the vaccine and is widespread during the flu season, the “mismatch” of vaccine and virus can substantially decrease the effectiveness of the vaccine.

A considerable amount of time is spent on steps beyond the manufacturer’s control, such as the creation of reassortants through the WHO Global Influenza Program. Furthermore, the collaborative, highly regulated process of standardizing the antiserum for each strain takes about eight weeks. In addition, a good deal of time is devoted to product-quality assurance or conformance to finished product requirements, as regulatory groups confirm potency values reported by the manufacturer.

Unique Aspects of Manufacturing Influenza Vaccine
The manufacture of influenza vaccine, although considered conventional, is unique in a number of ways. First, it is a global enterprise. The vaccine manufacturer is just one participant in a consortium of regulatory agencies, surveillance laboratories, certified laboratories, hospitals, and clinicians, as well as providers of embryonated eggs, components, equipment, raw materials, and transport agencies.

Second, the composition of the trivalent vaccine changes almost yearly. The vaccine must be licensed yearly (for every amendment), and the license is granted for one year only (July 1 to June 30). This has significant implications for the manufacturer in planning a production campaign:

  • Unused formulated and filled vaccine is usually discarded.
  • Monovalent concentrates are not typically used after 12 months from pool.
  • Yields of the strains are variable and are not known by the manufacturer until commercial scale-up production.

Third, formulations are updated twice a year to reflect the epidemiologies in the northern and southern hemispheres. The manufacturer thus has a very short window of opportunity to respond to changes. If the release of reassortants is delayed because of epidemiological changes during the course of the flu season, this window can be very, very short. The currently licensed influenza vaccine is trivalent; thus if one of the strains cannot be produced, there is no vaccine. Balancing production schedules to provide vaccine in the northern hemisphere for the October 1 immunization campaign can be challenging.

Producing Vaccine for a Pandemic
Seasonal and pandemic influenza vaccines are clearly interrelated. The capacity available for the production of a pandemic influenza vaccine is largely based on current seasonal capacity. Even though a pandemic vaccine is expected to be a monovalent formulation (e.g., H5N1 or H7N7), it will still require using existing processes and capacity because there will be no time either to develop new processes or to expand capacity.

Some 300 million doses of trivalent vaccine are available worldwide. Theoretically, assuming a monovalent pandemic vaccine, 900 million doses could be produced worldwide. However, manufacturing data indicate that the number of doses per egg would be fewer than during typical seasonal production. Clinical data reported on experience to date indicate that significant amounts of antigen, more than for seasonal vaccines, will be necessary to protect even healthy young adults.

Stockpiling is a key element of pandemic preparedness. However, the strain(s) in stockpile vaccines for possible pandemic influenza must be continually updated. Emerging epidemiology indicates that there are two distinct clades in the H5N1 influenza virus (a clade change is indicated when there is a change in the HA gene tree phylogeny). In other words, the ancestral relationship among H5 hemagglutinin (HA) genes from H5N1 avian influenza viruses collected in a specified region has changed (WHO Global Influenza Program Surveillance Network, 2005)

Seasonal influenza vaccination rates are still too low (MMWR, 2001). In the U.S. population, vaccination rates for healthy young adults and children are significantly below the 90 percent target. Figure 7 shows that the vaccination rates for people aged 65 and older are higher but are still below 90 percent. Therefore, meeting the 2010 target rate of 90 percent for each age group might provide an incentive and rationale for expanding manufacturing capacities. This expanded capacity might then be available to meet the demand in the event of a pandemic.

Keep in mind that capacity expansion is a time-consuming process and should not be considered a rapid solution for the vaccine industry. At best, expansion will take four to five years from concept design to validation and licensure.

Engineering Opportunities
Egg-based vaccine technology has been used to produce seasonal vaccine for more than 30 years. Recently, egg-based vaccine technology has also been used to produce more than 30 million conventional doses of H5N1 vaccine and pilot lots of H7N7 vaccine. The technology has been steadily improving since it was introduced and is becoming increasingly automated. However, technological improvements in alternate release and quality assurance assays would be beneficial. In addition, increasing immunization rates could lead to an expansion of manufacturing capacity.

Cell-culture technology is likely to supplement egg-based technology. However, cell-culture technology will have to be very robust and will still focus on the same physical parameters for vaccine purification, extraction, and HA enrichment. In the near term, constraints related to the availability of strains and potency reagents will present constraints for both egg-based and cell-culture technologies.

References
DHHS (Department of Health and Human Services) and Centers for Disease Control and Prevention (CDC). 2006. Available on line at: http://www.cdc.gov/flu/weekly/weeklyarchives2002-2003/02-03summary.htm and http://www.cdc.gov/flu/weekly/weeklyarchives2003-2004/03-04summary.htm.
Ludwig S., O. Planz, S. Pleschka, and T. Wolff. 2003. Influenza-virus-induced signaling cascades: targets for antiviral therapy? Trends in Molecular Medicine 9(2): 46–52.
MMWR (Morbidity and Mortality Weekly Report). 2001. Influenza and pneumococcal vaccination levels among persons aged > or = 65 years—United States, 1999. National Health Interview Survey (NHIS). MMWR 50(25): 532–537 (’01,’03, Jan-Jun ’04).
Reichelderfer, P.S., A.P. Kendal, K.F. Shortridge, and A. Hampson. 1989. Influenza Surveillance in the Pacific Basin; Seasonality of Virus Occurrence: A Preliminary Report. Pp. 412–444 in Current Topics in Medical Virology, edited by Y.C. Chan, S. Doraisingham, and A.E. Ling. Singapore: World Scientific.
WHO Global Influenza Program Surveillance Network. 2005. Evolution of H5N1 avian influenza viruses in Asia. Emerging Infectious Diseases 11(10). Available on line at: http://www.cdc.gov/ncidod/EID/vol11no10/05-0644.htm.
About the Author: James T. Matthews is senior director, Public Policy, Science, and Health Policy, Sanofi Pasteur.