Cooking Up Candidates
Safety is key when manufacturing candidate AIDSvaccines for clinical trials, and this involves a costly manufacturing process
By Andreas von Bubnoff
Before every space shuttle launch, all systems and processes are carefully checked and rechecked to prevent anything from going wrong. Failure is unacceptable. The same applies to the manufacture of candidate AIDS vaccines for human clinical trials. “Every step has to be checked and rechecked,” says Hildegund Ertl of the Wistar Institute. Each vaccine is unique and during every step of production the candidates must be inspected and adjusted if necessary to ensure they are safe and that they retain their activity.
For a vaccine candidate to be safe, it has to be pure, and the process of eliminating any potentially harmful substances takes substantial time and money. For example, manufacturing an experimental DNA vaccine in a lab takes just a few days, but regulatory agencies like the US Food and Drug Administration (FDA) will not allow vaccine candidates that are made in a research lab to be tested in humans, even for small Phase I clinical trials, says Eddy Sayeed of IAVI. Vaccine candidates for clinical trials are instead manufactured in specialized facilities by individuals with expertise on the production and purification of vaccines. Producing a DNA vaccine in such a manufacturing facility that is safe enough for human trials can take months. Tomas Hanke of the University of Oxford says that just ensuring a DNA vaccine candidate is sterile can take six weeks. “Checking what you have is the expensive and time consuming bit,” he adds. Sometimes this means just waiting and watching for contaminating bacteria or fungi to grow.
And since time is money, the price tag to manufacture a vaccine candidate for clinical trials is also several magnitudes higher than making one in a research lab. To make enough DNA vaccine for a Phase I trial in an industrial manufacturing facility costs several hundred thousand dollars, says Hanke, who has his DNA vaccines produced by the UK-based manufacturing company Cobra. In a research lab, the same quantity could be produced for only US$100. The Vaccine Research Center (VRC) in Bethesda, Maryland, part of the National Institute of Allergy and Infectious Diseases, paid $12 million to have the company Vical manufacture six different DNA plasmids for the proposed PAVE 100 trial, according to Alan Engbring of Vical. Manufacturing highly-purified viral vector-based candidates is also substantially more expensive than making them in a research lab. A new adenovirus-based AIDS vaccine candidate produced in a research lab for testing in nonhuman primates may cost about $2,000, Ertl says, but a company that manufactures vaccines suitable for clinical trials charges around half a million dollars.
Much of the manufacturing costs for vaccine candidates used in human trials are due to Good Manufacturing Practice (GMP), a set of standards required by regulatory agencies like the FDA or the European Medicines Agency (EMEA) for products that are tested in humans. GMP conditions require, among other things, that the water used to prepare candidates is injection-grade quality and free of any salts. The air in a GMP-certified facility also needs to be highly purified. In addition, everything a person does is double-checked. “One person does the work and another person watches them and they both sign off and follow the protocols exactly,” says Jerald Sadoff, who heads the Aeras Global TB Vaccine Foundation. All measurements are double-checked as well. “When your machine says it’s incubating at 37 degrees Celsius,” Hanke says, “you have to have a different thermometer inside [to check].” GMP manufacturing also involves following Standard Operating Procedures (SOPs), which require that every step is carefully developed and reproducible, according to Sayeed. An SOP might detail the steps required to prepare the media that is used to culture viruses.
And keeping up to snuff on GMP isn’t cheap: Running a compliant facility costs more than $100,000 a week, according to Sadoff, whose organization has its own facility to manufacture vaccines against tuberculosis (TB). In fact, 80% of the expense of manufacturing vaccines is from maintaining GMP conditions, estimates Andreas Neubert, head of vaccine production at IDT Biologika GmbH, a German company that manufactures modified vaccinia Ankara (MVA)-based vaccines for IAVI and Oxford University, among others. Ertl, who will have her adenovirus-based AIDS vaccine candidates made by the California-based company SAFC Pharma, agrees. “Use of a [GMP] facility is a big chunk of the price.”
No pathogens, please
But there is more to vaccine production than GMP. Each type of vaccine—and the cells used to manufacture them—also needs to be free of any pathogens or other potentially harmful substances.
To manufacture DNA vaccines, the plasmids are grown in bacteria whose outer membrane contains endotoxins. These are a concern because they are toxic to humans—they can interact with macrophages that release Tumor Necrosis Factor, a compound that causes the release of nitric oxide, which in turn causes blood vessels to relax. As a result, people go into shock, Sadoff says. To remove endotoxins, DNA vaccines are filtered through substances that bind to them. Tests for remaining endotoxins are done by conducting safety studies in rabbits, which are very sensitive to endotoxins, according to Sadoff.
Some viral vector-based vaccines are grown in cells from chicken eggs, and pathogens found in these eggs, such as avian viruses or bacteria, are also an issue. Sendai virus vectors are grown in cells in the so-called allantoic sac of chicken eggs, and MVA vectors are grown in chicken embryo cells called fibroblasts. The eggs used for such vaccines need to come from chickens held under “specific pathogen-free” conditions. Germany-based MVA vaccine manufacturer IDT buys such eggs at about 20 times the cost of regular eggs, Neubert says. Inactivated flu vaccines are also grown in chicken eggs, but these are much easier to purify because the cells can be treated with the chemical formalin to kill contaminants. This cannot be done for live viruses like MVA because such purification would also inactivate the vector.
Vaccines that use adenoviruses as a vector are typically grown in human cell lines, which also need to fulfill certain safety criteria before getting approved by regulatory agencies such as the FDA, Ertl says. For one thing, the host cell lines need to be free of prions, infectious protein particles that are believed to cause diseases in animals, such as mad cow disease, and a fatal variant, Creutzfeldt-Jakob disease, in humans. To ensure the cell lines used to grow adenovirus vector-based vaccine candidates are free from prions, their entire history must be documented. That history has to show that the cells have always been grown in serum from uninfected animals. “You can’t even have a one- or two-week blank in that record,” Sadoff says. The cell lines and the viruses grown in them also need to be carefully inspected for many other contaminating viruses and pathogens, Sayeed says. Cancer is another concern, since a vaccine candidate contaminated with DNA from human cells might cause tumors. To avoid this, the FDA requires the removal of most of the host cell DNA. Together, these strict requirements are the reason that only a handful of cell lines are available for manufacturing adenovirus-based vaccines.
Keeping it consistent
Consistency is another challenge for vaccine manufacturers. “It’s unethical to do a trial with something that would never be reproducible,” Sadoff says. Ensuring reproducibility requires a series of evaluations. For recombinant proteins, these evaluations include checking their molecular weight, whether the vaccine candidate reacts with the right antibodies, and how potent an immune response the candidate induces. “These potency assays are the most difficult because they have the highest variability and have to be very well characterized,” Sadoff says.
The concern that vaccines may change during production is indeed justified. For example, replication-defective adenoviruses grown in certain cell lines, such as HEK 293, can regain the ability to replicate. This poses safety concerns. Typically, adenoviruses used as vaccine vectors are engineered so they lack certain genes necessary for replication, such as E1. This gene is transferred instead into the cell line used to grow the virus. As a result, the cell line is able to divide indefinitely, making it easier to culture the cells. But sometimes the gene is transferred from the genome of the host cells into the virus genome because of similarities between the DNA sequences of the host cells and the adenovirus. If this transfer occurs, the adenovirus can once again replicate. “If you have too many of such [replicating] viruses, you have to throw away the whole batch,” Hanke says. Other cell lines used to grow adenovirus, such as PER.C6 and HER 96, don’t have this problem because their genomes are not as similar to the adenovirus genome, according to Sayeed.
Viral vectors grown in cell lines can even eject the HIV genes they carry. “They kick the transgene out,” Sadoff says. “Almost all the MVAs that have been tested have had this problem.” One reason is that some HIV immunogens, such as the Env protein, can be toxic to cells and make the viral vector genetically unstable, Sayeed says. Manufacturers therefore have to repeatedly test the vector to verify the HIV inserts are still there.
Developing the process
Along with safety and consistency, the production process also needs to be optimized before larger quantities of vaccine candidates are made. With DNA vaccines, manufacturers must find the most efficient host bacteria and the ideal time to stop bacterial growth before harvesting the DNA. Manufacturers also have to check the ideal growth conditions of cultured cells. Some chicken embryo fibroblasts that are used to grow MVA prefer growing adherent on surfaces, while others grow in suspension, according to Neubert. And growth efficiency drops as soon as HIV transgenes are introduced into the vector, Sayeed says. In the case of MVA, just inserting a single HIV gene renders it 10 times less productive.
Once vaccine production is optimized on a large scale, the price is likely to drop. Making large batches is easier for some vaccines than others, depending on the process. It is relatively easy to make large batches of DNA vaccines that are manufactured in bacteria and large-scale manufacturing could bring the per-dose price of a DNA vaccine down from $1,000, to around $4, Sayeed says. Vaccines that use adenovirus as a vector can also be made rather easily on a large scale because the human cell lines they are grown in can divide indefinitely.
However, scaling up becomes more difficult for MVA-based vaccines. Since the chicken embryo cells they are grown in do not multiply indefinitely, the virus always needs to be harvested from fresh eggs. As a result, manufacturing an MVA-based vaccine for millions of people could require 100,000 eggs per week, Sayeed says. Companies are now developing new avian cell lines for large-scale production to circumvent the dependency on fresh eggs. Another disadvantage of manufacturing MVA-based vaccines is that a rather large volume of vaccine culture is needed to get a small amount of vaccine, according to Sadoff, because it is less concentrated.
Finding a manufacturer that can make a vaccine under GMP conditions is not that easy, says Sayeed, who is in charge of finding companies to manufacture vaccines IAVI has developed with its partners. That’s especially true for vaccines based on viral vectors. “There is a waiting list,” Sayeed says. Only a handful of companies worldwide can do the work, he adds, and some of them are booked for at least nine months.
A few years ago, the manufacturing team at IAVI did a survey and found that of about 100 companies, only a handful could really manufacture vaccines, among them Transgene in France, Cobra in the UK, SAFC Pharma and Vical in the US, IDT in Germany, and Henogen in Belgium. Many companies that used to be involved in manufacturing vaccines have abandoned the business because it’s not profitable enough, Sayeed says. There are also contract manufacturers in countries like India, South Korea, Brazil, and China that can do the job, but researchers are hesitant to go there because they are concerned about protecting intellectual property, according to Sayeed.
Meanwhile, some academic and non-profit organizations have started to make candidate vaccines in their own facilities. The VRC, for example, has its own facility and the University of Oxford may also use its own facility to manufacture adenovirus-based AIDS vaccine candidates in the future. This is generally cheaper than using a commercial manufacturer, according to Pru Bird, head of research at the Oxford facility. AERAS also manufactures vaccines in its own facility, and last year, the Canadian government, in collaboration with the Bill & Melinda Gates Foundation, announced the creation of the Canadian HIV Vaccine Initiative (CHVI). The initiative has proposed building a vaccine manufacturing facility in Canada, according to Ingrid Wellmeier of the Public Health Agency of Canada, in response to a limited global capacity.
There are currently no dedicated large-scale facilities in place that could immediately take over production if an AIDS vaccine was proven to work in efficacy trials, Sayeed says. Manufacturers have to strike a careful balance between building a facility—which can take several years and cost a significant amount—and the risk that it may become useless if a vaccine eventually fails in late stage clinical trials. One strategy adopted by some of the big pharmaceutical companies, with several products in the pipeline, is to build generic facilities that can accommodate different types of vaccine technologies. This way, their construction is flexible enough to switch to the vaccine that is successful, even midway into the building process.
Sayeed, for his part, remains optimistic. “People ask if there is scarcity for large-scale HIV vaccine manufacturing. The answer is yes, but when it comes to crunch time, the capacity will be identified.”