Scaling up biologics production from the bench to clinical and commercial scale is a central consideration of process design and development. Issues, challenges, and new technologies in scale-up were all discussed at “BioProcess International Europe” held in Vienna recently. The meeting began with a case study of scale-up in action, when Russell Thirsk, Ph.D., vp, global head of influenza cell culture at Novartis Vaccines and Diagnostics, described large-scale manufacture of H1N1 influenza vaccine.
Global capacity for vaccine manufacture allows for just 20–30% of the population to be vaccinated for seasonal influenza and would not be sufficient for dealing with a pandemic of H1N1. The reality is that it would not be cost effective to have such a large infrastructure merely waiting for an event that might not occur.
One solution, Dr. Thirsk noted, is to use an adjuvant. “This allows expansion of the amount of vaccine available by reducing the dose needed.” Adjuvant is also useful because it allows some cross-protection of a vaccine against strains of flu it was not designed for, which is important in a pandemic scenario because of the likely antigenic drift of the virus.
Novartis has made three vaccines against H1N1—two of them manufactured in eggs and one in mammalian cell culture. One of the egg-manufactured vaccines includes an adjuvant and the other does not, while the cell-manufactured vaccine contains an adjuvant. It took Novartis four months, from registration to supply of the product.
“This is an incredibly short time frame, but it is what is required for response to a pandemic.” The cell-based vaccine was faster to produce than the egg-based vaccines. So far, Novartis has made 120 million doses of vaccine overall.
Dr. Thirsk noted that one egg can produce just one to two doses of vaccine, so the logistics of this route do not make sense in terms of scale-up. Furthermore, many strains of influenza do not grow in eggs, and the involvement of hens in producing eggs for manufacturing vaccines against avian flu strains may invite contamination.
In the H1N1 project, Novartis used Madin-Darby canine kidney cell culture in a closed bioreactor with no animal components. The resulting vaccine, Celtura®, is approved in Germany and Switzerland, and Novartis is seeking approvals elsewhere. “Close cooperation with universities and other institutions were key success factors in the speed of the project,” Dr. Thirsk said. “We think this is the way the influenza vaccine business is going to move in the future.”
Novartis has a big stake in this future, with its new facility at Holly Springs, NC, which produces flu vaccines with cell-based rather than egg-based culture. The company received nearly $500 million from the Department of Health and Human Services to help build the $1 billion, 130,000 sq. ft. plant, which Novartis believes is capable of producing 150 million doses of vaccine within six months of a pandemic declaration.
Bo Kara, Ph.D., director of science and technology at MSD Biologics (formerly Avecia Biologics), which was recently acquired by Merck & Co., talked about how to build scalability into biologics process development through good design. He pointed out that this is worthwhile as a robust manufacturing process because it can decrease the time from launch to peak sales by two years, which may represent revenue of up to $600 million.
One element in the company’s success in this respect is its investment in process platforms, where scale-up issues can be better understood and managed. To this end, MSD Biologics has a range of optimized platforms that are well understood and scalable. These are all off the shelf but include some custom optimization.
MSD Biologics is using advanced tools for process design including predictive models such as neural nets, statistical experimental design, and other chemometric and multivariate techniques to aid in acquiring process understanding. Also important is actual manufacturing experience. Carrying out a process on a large scale builds up manufacturing understanding that, in combination with good small-scale data derived using statistically designed experiments, allows an effective control strategy to be developed. The result is good process robustness and consistency. This ultimately supports the establishment of a design space.
Dr. Kara listed some useful tools for process characterization such as failure mode and effects analysis (FMEA), design of experiment (DoE), and lab models, as well as experimental studies at lab-, pilot-, and full-scale. Risk-assessment tools such as FMEA are an important element in process characterization to focus efforts where required. They can and should be used during the whole life cycle of the product including preclinical, clinical, and commercial manufacture.
Dr. Kara noted that during the product life cycle, risk assessments may be done for different reasons but with a common emphasis on reflecting current understanding and process robustness. He concluded that good design happens early on and looks at scalability, using well-understood platforms.