Credit: Novartis | Novartis's continuous manufacturing center in Basel, Switzerland, replicates an end-to-end manufacturing process developed at the Novartis-MIT Center for Continuous Manufacturing.
While most industries have employed continuous manufacturing processes for decades, if not centuries, the pharmaceutical industry has remained staunchly committed to batch operations. Drugmakers have steered clear of making major changes to validated processes, despite the advantages to be gained in cost, process time, and quality. But new drugs coming to market—chemically complex and targeted therapies—call for just the benefits that continuous processes offer. With the US Food and Drug Administration’s blessing, drug companies are finally implementing technologies that have been in development for 20 years.
“It’s called a BHAG,” says Hayden Thomas, divulging a term popular among chemical engineers at Vertex Pharmaceuticals, where Thomas is vice president of formulation development. “A big, hairy, audacious goal.” In his case it was committing to continuous process manufacturing for all new products in development.
As Thomas prepared in 2011 to pitch the idea to the top brass at Vertex, he knew it would be a long shot. After all, the pharmaceutical industry has resisted continuous processes, even though virtually all other industries have replaced step-by-step batch manufacturing over the past 2 centuries.
Highly profitable and ultraconservative, the drug industry has traditionally held back on investing in any transformative production or information technology despite solid evidence of benefits. In the case of continuous process manufacturing, benefits include lower costs, lower waste, higher yield, and higher quality. Such processes also allow ready application of complex and hazardous chemistries and much smaller—often benchtop—facilities than those traditionally employed to make drugs.
Thomas had a sense that the timing was right. Concerns about regulatory validation of continuous processes have slowly evaporated since the US Food and Drug Administration began encouraging them 15 years ago. And the industry has been experimenting with continuous techniques in engineering research for at least that long; Vertex began in 2005.
“I was kind of nervous leading up to the event,” Thomas recalls about his meeting with management. “We were making a pitch for a lot of money and commitment of resources. But they looked at it and said, ‘This is where Vertex needs to be.’ ”
Vertex’s commitment to the project paid off in 2015 when the FDA approved a continuous process for Orkambi, its cystic fibrosis drug. It was the first time the agency approved a finished-dose drug—called a drug product in industry parlance—made on an entirely continuous basis. Since then, the company scored a second time with a tableting plant for Symdeko, also for cystic fibrosis.
Janssen, Eli Lilly and Company, and Pfizer subsequently received approvals for drug products employing continuous production. Janssen’s Prezista treats HIV, Lilly’s Verzenio treats breast cancer, and Pfizer’s Daurismo treats acute myeloid leukemia. Indeed, every major drug company is either testing continuous technologies or beginning to use them to produce pharmaceutical chemicals and finished-dose drugs. Several have made aspirational statements regarding implementation of continuous techniques. GlaxoSmithKline’s manufacturing technology road map, a manifesto against traditional batch manufacturing, looks toward widespread use of continuous techniques. Janssen has said it expects to manufacture 70% of its drug product via continuous processes.
Yet most of the drug industry’s progress has been in the continuous manufacture of finished-dose tablets. To date, no active pharmaceutical ingredient—also known as the API or drug substance—has been fully synthesized in a continuous process in a facility that conforms to the FDA’s current good manufacturing practice (cGMP) standards. At best, proponents can point to hybrid processes involving both batch and continuous systems.
Drug companies and the contract development and manufacturing organizations (CDMOs) that serve them insist that continuous manufacturing of APIs will happen, though they acknowledge that the delay, coming after decades of development, is frustrating.
Alessandra Vizza, commercial manager of Corning’s reactor technologies unit, which launched continuous-flow reactors in 2007, notes an irony in the current interest among pharmaceutical makers. “When we started the business, our only focus was on pharma,” she says. But pharma lagged. “We decided we had to cover other markets that might go a bit quicker to industrial steps.”
Another irony is noted by Bernhardt L. Trout, a professor of chemical engineering at the Massachusetts Institute of Technology and director of the Novartis-MIT Center for Continuous Manufacturing, a partnership launched to develop continuous technology. “What is kind of ironic is that the major benefit of continuous manufacturing is with drug substance,” he says. “It’s best in terms of streamlined processes, cost savings, footprint savings, and quality improvement.”
Designing continuous processes for drug substance is technically more challenging, however, than designing them for drug product, Trout says. And the industry moves slowly. “It tends to start with the low-hanging fruit,” he says, “but drug substance is coming.”
The center, in fact, has proved it can be achieved, having built the first end-to-end continuously automated drug manufacturing process in 2012. In the second phase of the 12-year partnership, which ends next month, the center developed a range of continuous process technologies—including flow chemistry, crystallization, and dosage manufacturing—transferring several to Novartis.
The project also spawned a systems integration firm, Continuus Pharmaceuticals, headed by Salvatore Mascia, a former project manager at the Novartis-MIT center.
Snapdragon Chemistry, a process design and consulting firm spun out separately from MIT, is focused on API manufacturing, using flow-based retrosynthetic analysis and design techniques in areas such as photochemistry, liquid-gas reactions, high-temperature and high-pressure reactions, and electrochemistry. The company was formed in 2014 in response to drug companies’ requests for MIT’s assistance designing continuous processes.
As Snapdragon CEO Matthew M. Bio acknowledges, one of batch manufacturing’s advantages is that it is not dedicated to a particular product, whereas continuous systems are. Batch plants can switch from one chemistry to another when compounds fail after a short time in the clinic, or they can remain workhorses for drugs that are commercialized.
But the technical demands of producing emerging therapies has upset the manufacturing status quo.
“Some of the molecular complexity coming forward from the discovery groups has kind of focused people on putting together molecules and technologies that don’t scale very well in batch production because they were hazardous,” Bio says. “You could look at the antibody conjugates or the protein conjugates that were relying on azide click chemistry. Making the azides, a high-energy material—nobody wants to do that in a big batch reactor.”
Photochemistry is another growing application for continuous systems. “Light doesn’t penetrate reactors very far,” Bio says. “Only a few millimeters before it’s fully absorbed. It doesn’t make sense to put a lamp in a big batch reactor.” Snapdragon develops technologies for customers, mostly drug companies, that have molecules they want to make continuously. “We develop the technology and build the reactor system,” Bio says, noting that the company recently shipped systems to North Carolina and Milan. “These are full-scale reactor systems that do one or more steps on the order of a kilo per hour of material.”
He expects Snapdragon to have its own manufacturing capability by the end of the year.
Snapdragon has also partnered with nearby Johnson Matthey, one of many CDMOs moving into continuous process manufacturing services in response to growing customer demand.
Drugmakers can deploy a number of different reactors and reactions in continuous processes.
▸ Plug flow reactor: Plate or tube configurations enabling a continuous flow of fluid material during progressive chemical reactions. They are generally configured with multiple units of comparable capacity.
▸ Continuous stir tank: Small-volume stir-tank reactors operating constantly, often configured as cascading reactors pumping from one to the next with materials added and product pumped out continuously. They handle solids better than flow reactors.
▸ Packed-bed reactor: Tubular reactors with immobilized catalyst in the tube. They can be used, for example, for hydrogenation with an immobilized palladium catalyst or an enzymatic reaction with an immobilized enzyme.
▸ Photochemistry: Light-catalyzed reactions in fluid channels limited to a few millimeters in depth as opposed to immersion of lamps in large batch reactors.
▸ Electrochemistry: Reactions catalyzed by electricity. Close placement of electrodes in flow channels allows efficient electron transfer during reactions.
▸ Cryogenic/exothermic chemistry: Reactions characterized by low temperature or high heat. Efficient removal of heat in flow reactors allows reactions to take place at lower or higher temperatures than with batch reactors.
Sources: Snapdragon Chemistry, C&EN research.
The addition of continuous process services—for both drug product and drug substance—at CDMOs reflects increased interest among the drug companies that are their customers. Hovione, for example, moved quickly to establish a unit at its R&D and manufacturing center in East Windsor, New Jersey, after landing a contract with Vertex to produce Orkambi tablets. The company plans to put similar assets in place at its headquarters in Portugal. Italian CDMO Flamma says it is experimenting with a continuous unit at its plant in Chignolo d’Isola near Milan.
Others have been at it much longer. Thermo Fisher Scientific, for example, has continuous process experience dating back to DSM’s introduction of micro flow reactors in 2004. DSM’s drug substance unit merged with the drug product firm Patheon in 2013; Thermo Fisher acquired the combination in 2017.
“We started to develop continuous processes for active pharmaceutical ingredients in DSM’s custom pharma manufacturing division,” recalls Peter Pöchlauer, now head of innovation management for Thermo Fisher’s pharma services group. “That dates back to the early 2000s when we did our first experiment in small-structured flow rectors in nondrug applications.”
In 2007 DSM took on its first drug project: an intermediate that could only be made using a flow process. “It was a nitrate ester formation,” Pöchlauer says, “that simply couldn’t be done in batches at the scale we wanted to run it.”
As DSM got started in the area, the FDA was beginning to push continuous processes, Pöchlauer says, noting that the agency began recommending them in 2004. The agency further advocated continuous processing in its 21st Century Cures Act, signed by President Barack Obama in 2016, before issuing formal guidance in February of this year.
Industry started coming around as price pressure and tighter delivery schedules focused firms on the cost and efficiency of production. Major drug companies began outsourcing, and continuous process manufacturing became an asset for CDMOs. DSM advanced its initial R&D program, where it developed microreactors the size of cell phones, to a pilot plant in Ravensburg, Germany, and a commercial manufacturing plant in Linz, Austria.
Flow chemistry and other continuous technologies are developed in modular fashion, Pöchlauer says, allowing a plant to be disassembled and its components repurposed in the event—frequent enough—of drug failures in the clinic. In pharmaceutical chemical production, he notes, continuous technology is likely to be a step in a longer process addressing chemical reactions that cannot be accomplished safely or efficiently with batch equipment.
Today, a showcase example of continuous technology for Thermo Fisher is the development of a low-temperature methylation that accounts for about a third of the full process for synthesizing vaborbactam, a β-lactamase inhibitor initially developed by Rempex Pharmaceuticals and now marketed by Melinta Therapeutics. The drug won FDA approval in 2017.
Thermo Fisher is building continuous process research capacity at its plants in South Carolina. “Our strategy is to roll out our knowledge over the small-molecule API operation at Thermo Fisher,” Pöchlauer says. “We’re doing it now because we are seeing an increased demand for flow process in the United States.”
Cambrex also moved continuous-flow chemistry know-how from Europe to the US. Drawing on expertise at its facility in Karlskoga, Sweden, the firm has invested over $1 million to develop flow chemistry services in High Point, North Carolina. The company, like Thermo Fisher, is responding to a “pull from the market,” says Shawn Conway, director of engineering R&D at Cambrex High Point.
“A number of drug companies have expressed an interest in flow chemistry and want to know about our capabilities,” Conway says. “Their thinking is that if the industry is moving in this direction, starting to pursue this type of technology, there needs to be a network of providers able to work with them in developing process and manufacturing material as well.”
Cambrex is positioning High Point to provide flow chemistry services through late-stage clinical trials, with the goal of expanding operations to commercial scale. The Karlskoga site already offers commercial-scale continuous-flow manufacturing.
SK Biotek began developing flow chemistry for pharmaceuticals in the mid-1980s, according to Seongho “Ryan” Oh, head of R&D. The firm used know-how from its parent company, SK Group, which owns the largest refining and petrochemical firm in South Korea. SK Biotek has offered flow chemistry at the pilot scale for nearly 20 years, Oh says. Its latest R&D development is a continuous triazole process.
“These days customers are accumulating knowledge and come by asking if we can develop a continuous process on the basis of their ideas,” Oh says. SK Biotek is currently applying continuous techniques at its plant in Daejeon, South Korea, to an AstraZeneca diabetes drug being developed at SK Biotek’s Swords, Ireland, facility, he adds. The Korean plant was approved for ton-scale advanced intermediates production in 2014.
Lonza also has a history of developing continuous processes. “We have done a lot of work in microreactors over the years, but more recently we have been focusing on different types of flow reactor,” says Lee Newton, head of the company’s API business. “We now have a couple of operational units big enough to produce ton scales in our non-cGMP plant in flow reactors, and we want to move that into cGMP very soon.”
Newton cites major drug companies’ broad statements of intent to convert large swaths of manufacturing to flow and other continuous technologies as the incentive for expanding contract services. The industry as a whole is following pioneers like Novartis and Lilly, he says. It is also looking ahead at the nature of new drugs, the cost of producing them, and the feasibility of doing so in traditional batch operations. “Today, it’s very much an economic driver,” he says.
But there are competing economic arguments, Newton observes. Some industry watchers insist that batch operations are less expensive over time given that vessels can easily be cleaned and repurposed, as they have been for decades.
“And then there is this kind of mythical Lego set,” he says, referring to modular design in flow chemistry. “The idea that you can take all these bits and pieces, build a plant, run it for a few weeks, take it apart, clean it, put it back together, and make a few molecules. But all these reactions have to be fast enough to work in flow. Not all reactions are.” And regulatory considerations persist, despite the FDA’s encouragement of continuous process techniques, Newton says. “You can build it, but if it takes 6 weeks to qualify it, you’ve lost another advantage.”
Other CDMOs are holding back altogether. “Companies have been talking about it for a long time,” says Vivek Sharma, CEO of Piramal Pharma Solutions. “For us it’s early. We are talking to customers, exploring, seeing how it fits into new investments—whether it makes sense.” Although customers have inquired about Piramal’s capabilities, none have said they require continuous process manufacturing, Sharma says.
“If we invest the capital, we have to find the right buyer,” he says. “I think current processes are good enough to service customers for the near future.”
And views diverge on how continuous processes should be approached. Ian Shott, who has been advocating for process acceleration centered on continuous methods for more than 20 years, is critical of much of the current activity.
“Over the last 2 decades, the high ground on continuous process has been taken by what I call the gadget makers rather than the process owners,” says Shott, who is now CEO of the British CDMO Arcinova. “What I am trying to do is get back to fundamentals and design processes based on thermodynamics and kinetics, not picking up a particular continuous reactor and trying to force fit the process.”
Shott describes an approach that takes modular engineering a step beyond configuring flow reactor plates and tubes like Lego pieces. He advocates mixing technologies and in some cases using batch-oriented gear in a continuous process.
For example, as an alternative to conventional flow manufacturing, Arcinova has modeled a cascade of continuous stir-tank reactors. The advantage, Shott contends, is that the stir tanks can be typical glass reactors used for batch chemistry. “You put them together in a string.”
The ideal approach is a mix of technologies determined by reaction modeling. Shott says Arcinova has developed a process for one of its clients that turned a 55-batch campaign into a 5-batch campaign, employing continuous techniques at some of the stages.
Advantages of continuous processes
▸ Reduced cost of production
▸ Reduced factory space
▸ Reduced processing time
▸ In-line/real-time quality monitoring
▸ Rapid scale-up
▸ Faster technology transfer
▸ Lower rate of product deviation
▸ Access to difficult chemistries
Pharmaceutical industry hesitance
▸ Low product volume
▸ Regulatory uncertainty
▸ Quality uncertainty
▸ Commitment to current manufacturing assets
▸ Concerns about handling slurries and other nonliquid material
▸ Concerns about versatility of manufacturing assets
As a relatively new company without large fixed assets, Shott says, Arcinova is well positioned to bring continuous process manufacturing to bear on the high-tech, low-volume drugs that are taking up an increasing amount of the development pipeline.
Big drug companies have lots of fixed assets, though, dampening the incentive to embrace the new. Still, several of them have established centers for the development of continuous technologies.
Lilly and GlaxoSmithKline invested $40 million and $50 million, respectively, in plants dedicated to continuous manufacturing. Lilly opened its site in Kinsale, Ireland, in 2016 and expanded it the following year with a plant producing clinical-scale cGMP material. The company also has continuous capability at facilities in Indianapolis and Puerto Rico.
GSK declared its commitment to continuous processing in 2013 with the announcement of a plant in Singapore where it will develop continuous technologies. Then-CEO Andrew Witty hailed the move as “a really signicant technology leap for the company.” He said the firm was undertaking a “shift from synthetic chemical reactions to enzymatic reactions and a whole reframing of how we do analytical testing in all of our facilities.”
GSK says it has a filing pending with the FDA for commercialization of a drug stubstance manufactured with a mix of continuous and batch processing at the plant. And the company is building an additional continuous manufacturing facility in Singapore, where it plans to produce a potential new treatment that would be the first new chemical entity developed at the Singapore site with continuous chemistry.
In 2011, Novartis opened a center for continuous manufacturing in Basel, Switzerland, where it has established an end-to-end process based on work done at the Novartis-MIT center. Though it is operational, it has yet to find practical application, largely because of the difficulty of converting API production to an entirely continuous process.
“We have a road map but no end-to-end production for a molecule at this time,” says Markus Krumme, head of continuous manufacturing.
Pfizer, meanwhile, has incorporated continuous processing into its pharmaceutical chemical plant in Freiburg, Germany. Kevin Nepveux, vice president of manufacturing, defines the company’s approach as hybrid. Rather than strive for end-to-end processes, the company looks for individual reactions where continuous production offers a significant advantage over batch, he says.
For those reactions where continuous manufacturing makes sense, Pfizer prefers to employ standard technologies with minimum customization, Nepveux explains. Off-the-shelf technology is more effective than custom designs in achieving cost and time savings, he says. “Our real competitive advantage, as with other innovative pharmaceutical companies, is our molecules.”
If the quest for that advantage is tipping the balance toward continuous processes, it is also vindicating early efforts and bold moves in engineering that entailed a high level of the industry’s greatest aversion: risk. Even if continuous drug substance manufacturing still hasn’t truly arrived, the fear of it is gone.
“We were not the first drug company studying continuous processes when we started our initiative with MIT,” Novartis’s Krumme says. “But we were taking a very brave approach. We decided to forget almost everything we knew about making drugs and conceive a different approach.”