A modern pharmaceutical manufacturing plant looks a lot like a classical chemistry lab. As has been the case for more than a century, most drugs that come out of these factories are made via stepwise reactions, purifications, and final product formulation. In a handful of plants, however, continuous-flow processes are starting to creep in.
Continuous processing offers several advantages over start-and-stop batch production. A continuous-flow reactor requires less space and less investment. It can provide better control over reaction conditions and productivity. And, equally important to the lab chemist and the plant engineer, continuous methods can often be used to rapidly scale up reactions, including ones that involve hazardous reagents.
Even regulators are encouraging. Coupled with real-time monitoring and control of product variability, continuous manufacturing has the potential to “not only meet current expectations, but to really give enhanced product quality over what some of the current batch technology provides,” says Christine Moore, acting director of the Food & Drug Administration’s Office of New Drug Quality Assessment.
Although continuous processing isn’t mainstream in drug production, many people working in the area see it as inevitable. To make the shift, academic researchers, equipment suppliers, and end users are creating needed tools. Some are even looking beyond converting single reaction steps to achieving ambitious end-to-end processes that flow from raw materials to final pill.
But making the switch has been slow. Although DSM, Lonza, and other fine chemicals firms are developing equipment and processes, and pharma companies such as GlaxoSmithKline and Novartis are putting systems in place, the new technology competes against entrenched batch capacity. Even when corporate resistance is overcome, continuous processes have yet to test the regulatory waters in any great number.
Mind-set has made the move toward continuous processing easier for some than for others. Jaeyon Yoon, vice president of marketing at SK Life Science, the contract manufacturing business of South Korean petrochemical producer SK, says it was easy for his firm. “When we started 15 to 20 years ago, our team was originally SK engineers and chemists,” he explains. They began making pharmaceutical chemicals with the continuous bulk chemical processes with which they were familiar.
They started with flow hydrogenations using fixed-bed catalysts. Today, SK operates a wide range of continuous-flow chemistries, as well as distillations, extractions, and crystallizations, from the lab to commercial scale. In deciding whether to use a flow or a batch process, SK scientists weigh economics, timing, and the suitability for a specific chemistry, Yoon explains. About 60–70% of what SK runs are still batch reactions.
Another sign of the technology’s value, Yoon says, is the conclusion by the American Chemical Society Green Chemistry Institute’s Pharmaceutical Roundtable that continuous processing is the leading green engineering research area.
To assess how the technology is being practiced, the group surveyed its members—12 major drug firms and two manufacturing services providers—on the level of adoption (Org. Process Res. Dev. 2012, DOI: 10.1021/op300159y) and on business drivers (Org. Process Res. Dev. 2013, DOI: 10.1021/op400245s). Most large pharma companies have formed groups to develop and implement continuous processing and serve as internal advocates, the surveys found.
Nevertheless, there is a general lack of experience in using the technology and reviewing it with regulators. For example, Yoon says SK must work to educate customers on how its processes work and how it will “manage a project, deliver a product with the right quality, and meet regulatory requirements.”
Although continuous processes are cost-competitive, drug manufacturers are risk-averse and thus apt to stick with familiar batch methods, says James R. Bruno, president of the consulting firm Chemical & Pharmaceutical Solutions. The high attrition rate in drug development makes companies reluctant to invest in new technology, especially if they view it as unnecessary because of existing batch capacity. “The biggest thing holding us back is that we have assets in the ground,” Bruno says.
Pharma firms are typically reluctant to change processes for approved products. Thus, industry watchers expect companies to adopt continuous reactions mostly for new products or possibly to lower the cost of making generic drugs. To capture this business, “just about every major contract manufacturing organization (CMO) has some kind of a program in place to look at continuous-flow chemistry,” Bruno says.
Continuous processes have advantages over batch methods, but they have challenges as well.
◾ Low capital investment
◾ Less space required
◾ Safer with hazardous reactions
◾ Shorter processing times
◾ Possible novel chemistries
◾ Straightforward scale-up
◾ Need for less inventory
◾ Potential cost savings
◾ Better product quality
◾ Improved environmental impact
◾ Competes with existing investments
◾ Mind-set change needed to shift
◾ Perception of higher risk
◾ New engineering/operating skills
◾ Lack of adequately trained people
◾ Equipment availability at all scales
◾ Up-front development demands
◾ Limitations with solids
◾ Mastery of start-up and shutdown
◾Relatively untested regulatory path
At Chemtrix, a Dutch flow-chemistry equipment supplier, CMO customers have far exceeded pharma customers over the past 18 months, according to Charlotte Wiles, the firm’s chief executive officer. The drug companies already had equipment for discovery and some pilot-scale work, and many larger firms even engineer systems themselves, she adds.
Gauging the level of adoption within pharma companies is difficult for outsiders because the firms don’t share information widely, Wiles acknowledges. “Only a few case studies that have been done on the industrial scale actually come out into the public domain,” she says. For example, Chemtrix worked on a few dozen contract projects last year and is only able to discuss a handful publicly. “The market is crying out for precompetitive sharing of both success and failures so it can learn from itself.”
“What you do see is a much broader adoption among contract and fine chemicals manufacturers all the way up to production scale,” Wiles says. CMOs want to have novel chemistry capabilities to “hook a project that they may not have gotten any other way,” she adds.
Last year, Chemtrix joined with the pharma products arm of DSM, now part of the pharmaceutical services firm DPx, to provide flow-chemistry process development and equipment design services. Peter Poechlauer, a DSM principal scientist, says his firm is teaming up with equipment suppliers to lower the hurdle for pharma firms that intend to enter the space. These firms want guarantees that the equipment exists to quickly and predictably scale up, he adds.
“The pharma industry has adopted continuous-process technology whenever it seems to pay off immediately, reliably, and safely,” Poechlauer points out.
For the French drug developer Nicox, DSM developed an industrial-scale flow process to conduct a nitration step that was not feasible in batch for safety and cost reasons. It used Corning’s glass reactor technology to build a regulatory-compliant system that could produce a few hundred tons per year of the active pharmaceutical ingredient (API). DSM achieved this by “numbering up,” that is, running tens, if not hundreds, of microscale reactors in parallel.
Although numbering up microreactors allows a user to increase output without any process changes, the approach can be economically impractical, Wiles says. Microreactors are still widely used for research and process development, but for production purposes the shift has been to larger plate-based or tubular systems. With larger channel sizes, these systems can process bigger quantities while maintaining the material-handling benefits of the smaller systems as they relate to mixing and heat transfer. In addition, they may also be able to handle particulates. Both Chemtrix and Corning now have reactor classes that range from the milliliter to liter scale.
But even such larger units have limits, and running them in parallel again becomes an option. “You will come to a point where you can’t increase the features without losing critical performance,” Wiles says. “At that point, you would replicate the units rather than lose the performance.”
Beside the economics, technical issues arose with numbering up microreactors. “The early numbering up strategy apparently was an avenue that was quite more complex than expected,” says Dominique Roberge, leader of a manufacturing technology group at Lonza. Problems included controlling equal flow to all the reactors and ensuring all channels were clean.
Be it for microreactors or larger plate reactors, development is more complex for a flow-based process than for a batch one, Roberge says. It requires knowledge of process control, surface phenomena, reactor designs, and mechanics. After working with microreactors since 2003, Lonza created a line of larger plate reactors and licensed them to Ehrfeld Mikrotechnik BTS, an equipment supplier owned by Bayer. Lonza uses plate reactors of various sizes in a facility that supports process development and scale-up of pharma and specialty chemicals.
Roberge figures that all of the top 20 pharmaceutical companies have flow systems. “Some of the larger players are even ready to invest double-digit millions of euros.” Meanwhile, small to midsize firms with more limited budgets are exploring flow technology through outsourcing to CMOs such as Lonza.
Estimates by drug firm executives and their academic collaborators suggest that the pharma industry has invested more than $1 billion in continuous-manufacturing development over the past decade. Investments, for the most part, were made quietly by companies on their own and didn’t involve much collaboration.
GlaxoSmithKline is among a handful of firms that have been public about their commitment to continuous processing. It is building a $50 million plant in Singapore where it will manufacture, among other things, the asthma and allergy drug fluticasone propionate. The multistage synthesis involves some continuous steps to make the API, with isolation of the product in batch to integrate with existing downstream infrastructure, explains Toby Broom, GSK senior investigator.
“Technological challenges for continuous processes remain in the areas of separation, purification, and the integration of biocatalysis,” Broom points out. Still, GSK has high hopes for the technology. It expects to achieve a 75% reduction in cycle time; significant reductions in waste, carbon emissions, and raw materials; tighter impurity control; higher consistency in downstream processes; and greater overall yield.
GSK has developed continuous-manufacturing methods for a number of new chemical entities and has used them in pilot campaigns and for manufacturing material in support of regulatory filings. Although the company has done most of the development work in-house, it has used third parties for equipment and facilities.
“In GSK, there are highly experienced scientists working on flow processes, but they make up a small percentage of the overall staff, and we are currently trying to expand this,” Broom says. “In academia, there is limited experience too, which presents a challenge for GSK and the industry.”
To tackle the challenge, GSK joined with AstraZeneca and Novartis to help found the U.K.’s National Centre for Innovative Manufacturing in Continuous Manufacturing & Crystallisation, or CMAC. The center has about 10 equipment supplier partners and nearly 20 other collaborators that include drug firms and CMOs. Projects under way at several CMAC-associated universities are investigating continuous crystallization and particle formation.
Continuous crystallization will help bridge the gap between upstream flow synthesis and downstream product formulation, where continuous manufacturing is already an established method. In the absence of that bridge, many manufacturers flow the output of a continuous reaction into a batch vessel to hold it for processing.
On the plus side, this approach makes use of existing equipment, and by measuring the quantity of material collected or duration of reactions, it often is the means to define a “batch” for regulatory testing purposes.
On the minus side, the downstream isolation and purification—or workup—process is more than 50% of fixed costs, Yoon points out. “If you could do efficient continuous extraction or filtration or crystallization, it would be a good cost savings.”
Making the flow step synthesis itself more efficient—for example, by generating fewer impurities—can simplify downstream steps such as purification. But optimization can do only so much.
“We’ve had a couple of feasibility cases at the lab level where we have proven very fast reaction kinetics, but the bottleneck would be filtration and isolation,” Chemtrix’s Wiles says. “And there’s no need to accelerate the process beyond the weakest point.” Adding more continuous steps might improve the situation or simply move the bottleneck, she adds.
Ultimately, the goal is to make the entire process continuous. “We’ve had pieces of the technology but not necessarily all the technology,” Bruno, the consultant, points out. “We’re just doing it step-by-step and then slowly will start to integrate that more into a continuous process.”
Developing a continuous end-to-end plant was an initial goal of the Novartis-Massachusetts Institute of Technology Center for Continuous Manufacturing. In 2007, Novartis agreed to invest $65 million over 10 years to support R&D at the center. By 2012, a 13-member team had succeeded in creating a prototype to make aliskiren hemifumarate, the API in the cardiovascular drug Tekturna (Angew. Chem. Int. Ed. 2013, DOI: 10.1002/ange.201305429).
The 2.4- by 7.3-meter bench-scale plant reduced the number of operations from 21 in batch mode to 14, mainly through improvements in formulation steps. Starting with an advanced intermediate, the process includes reactions, separations, purification, crystallization, drying, formulation, and tableting in one stream (Org. Process Res. Dev. 2014, DOI: 10.1021/op400294z). Having faster reaction times and higher yield compared with batch, it has an output of 41 g/hour, or about 3 million tablets per year.
Because effective flow processes are not just reoptimized batch reactions, they present a lot of opportunity for new approaches, says Bernhardt L. Trout, an MIT chemical engineering professor and director of the center. For example, the starting material for the aliskiren reaction is a solid, and the MIT team conceived of an initial step using neat, molten material.
Ingenuity is important because the toolbox of flow reactions, equipment, analytics, and control systems is incomplete. “Although commercial equipment is available, you can’t buy an end-to-end process,” Trout points out. And even finding individual pieces of the desired scale and specifications was “a big gap,” he says. As a result, the MIT team largely built its plant in-house.
The biggest up-front investment, however, is not in the equipment or the process. “From the beginning, you have to put more effort into understanding and modeling your system,” Trout says. “A big part of continuous manufacturing is having a systems approach and accurate models.”
In 2013, the MIT center spun off a company called Continuus Pharmaceuticals to design, build, and run manufacturing processes at clients’ sites. Novartis, meanwhile, has been transferring the technology to its own plants. The drug company has said it expects to have a continuous-manufacturing technology and business plan by 2015. For its part, the MIT center continues its work on equipment, reactions, processes, and dosage forms.
MIT professors Timothy F. Jamison, Klavs F. Jensen, and Allan S. Myerson are also building a portable unit capable of making about 1,000 doses per day under a contract from the Defense Advanced Research Projects Agency. The system could operate in the field or serve humanitarian and small-scale needs such as orphan drugs, Jensen explains. The reconfigurable design has been demonstrated with processes for making diphenhydramine, lidocaine, diazepam, and fluoxetine.
Last week, MIT and CMAC sponsored the International Symposium on Continuous Manufacturing of Pharmaceuticals to bring together individuals from academia, pharma companies, suppliers, and regulatory bodies. An outcome of the much-anticipated symposium is a series of white papers covering all aspects of continuous manufacturing from technical issues through regulation. The keynote speaker was Janet Woodcock, director of FDA’s Center for Drug Evaluation & Research and a continuous manufacturing supporter.
Participants were keen to hear what Woodcock had to say since regulation remains a largely uncharted territory for continuous manufacturing. In general, guidelines from FDA and other international drug agencies do not specifically address continuous processing but at the same time do nothing to inhibit its use. Bruno and others believe that this situation provides enough flexibility to the manufacturing sector to take the lead in implementing new technologies.
Like Woodcock, FDA’s Moore has been championing flow processes. “FDA definitely is supportive of continuous manufacturing, as it is in line with our quality initiatives,” she says. “We believe that our current guidelines and regulations are consistent with the implementation of continuous manufacturing, and we are working with industry and academia, and with other regulators internationally, to determine what, if any, additional documents might be helpful to clarify expectations.”
The agency is preparing itself to handle a shift to new manufacturing processes, according to Moore. She and others from FDA “have been very active on the speaking circuit,” regularly attending workshops and conferences. For several years, the agency has been training its staff internally and externally, including visits to company and academic sites. “We are also very active in supporting academic research and collaboration with at least a half-dozen different universities,” she adds.
Although continuous manufacturing has been practiced for decades in chemical manufacturing and food processing, the pharmaceutical industry’s needs are unique, Moore points out. “At least one major difference between those operations and pharmaceuticals is that you have very small unit doses in pharmaceuticals with a tight requirement for consistency and uniformity,” she explains.
As a result, drugmakers have increased their focus on measurements, sensors, control systems, and the corresponding mathematical models. Academic research in these areas has “exploded” in the past few years, Moore says. “These areas are rapidly progressing,” she adds. “I think that the community as a whole could use more experience in integration of all the different technologies that are currently available.”
Industry managers and regulators alike have experience to gain as new continuous-manufacturing processes face regulatory review. Many drug developers wonder what questions might arise, for example, around process validation or product sampling and testing. If any drug firms have been through the review process, they are being characteristically tight-lipped, but the general belief is that the number of related regulatory filings has reached the double digits.
“We have had some approvals, but I’m not going to go into details,” Moore says. “We have seen a little bit of everything related to continuous manufacturing of both drug substance and drug product, both single-unit operations and end-to-end processing, and we have seen companies looking to put continuous manufacturing in for both existing products and for new products. It is being applied broadly.”