For years now, observers have said that pharmaceutical manufacturing must become more agile and that the industry faces a “sea change” in the way that it handles R&D and manufacturing. The last decade, like those that preceded it, has been marked by mergers, and many large, single-product facilities have been closed down. Yet, on the surface, little appears to have changed. Facilities have not become automated, “lights out” plants, continuous manufacturing is still relegated to specific niches, and PAT and QbD have not started a revolution, although they are gradually being applied, or at least studied by more companies.
However, the scene has been set for drastic change. In the U.S., discussions of healthcare reform bring the prospect of higher production levels and lower drug prices, while the likely approval of follow-on biogenerics promises to shave years off the marketable life cycle of biotech drugs. Is this the calm before the post-blockbuster storm? “We talk about personalized medicine but it doesn’t seem like we really believe it’s coming. Is it this year? Is it next year?” Lars Petersen, Genentech’s vice president of automation, asked peers at a life sciences panel discussion at the 2009 Emerson Global Users Exchange earlier this fall. “The most important factor will be the responsiveness of pharma facilities to changing demand,” he continued. “The key question is: How do we create more agility?”
In this article, we’ll try to answer that question, as experts share their views of how drug manufacturing facilities will change to become more agile, what technologies will drive flexibility and how this will affect the industry. We've also made our extended discussions with various experts available via links provided throughout the article, and in the box at right.
Drivers for Flexibility
Several major trends are driving less-expensive, more flexible pharmaceutical plant designs, says Bob Bader, senior manager of technology for pharma and biopharma at Jacobs Engineering (Conshohocken, Pa.). First, we don’t need that much product anymore, he says. “We’re not talking about 2000 kilograms but 200 to 300 kilograms per year,” he says. At the same time, in biotech, titers are going up. As a result, the upstream portion of bio facilities has become smaller and the extent to which single-use equipment can be used has increased to a point where 100% single-use becomes possible. “It gives you a lot of flexibility but also simplifies design, automation and fixstraed capital costs,” he says.
Bader’s colleague Deepak Agarwal, director of pharmaceutical technology at Jacobs, sees pharma engineering projects aiming to:
- minimize the cost of goods and the total installed cost
- further accelerate scheduling for design, build and
- make facilities more flexible and adaptable for a range
- continue to make high-quality product.
But cost containment is only one facet of pharma flexibility. There’s also a very real need to minimize financial risk and liability early in the drug value chain, a need that is driving increased use of disposable process equipment, Bader says.
“If you think of all the curve balls that can come at you when you have drugs in the clinic, plunking down $500 million that you won’t even use or see for the next five years, given the uncertainty that the drug will make it to market, is an incredibly risky proposition,” says Parrish Galliher, CTO of Xcellerex, Inc. (Marlborough, Mass.). Galliher responded to this problem not only by designing a “Biotech Monopoly” board game, but starting up Xcellerex six years ago, to address some of the most frustrating problems he’d seen in three decades of bringing new drugs, and plants, to market. At the heart of the company’s work is the FlexFactory biomanufacturing platform, a system that uses disposable bioreactors and mixers and collapsed cleanrooms around process equipment, and which grew out of a prototype that Galliher developed while at Millennium Pharmaceuticals in 2001. [Click here for more from Galliher.]
Studies have shown that use of FlexFactory can allow companies to reduce plant costs from hundreds of millions to $25 million and project time-frames from five years to under one year. Hedging against risk will be critical, agrees Rakesh Kishan, director of the life sciences practice at UMS Advisory Group (Arlington, Va.). The key is not being tied to long-term ownership of assets that no longer have any use to the company. At the same time, he says, flexibility and scalability will continue to drive building use.
Despite ongoing industry restructuring, the top eight to ten Big Pharma companies still own about half a billion square feet of real estate, Kishan says, and, given a contracting economy and credit market issues, there will be fewer opportunities to shed excess, outdated capacity. Capacity utilization, now at 30% to 40% for the industry, will become more important, Kishan suggests, and more companies will choose to source key functions, such as API manufacturing and some clinical and R&D functions, off shore.
Economic realities should also stimulate interest in Toyota Production System methods such as OEE for both research and manufacturing facilities, and capacity utilization will be closely tracked, Kishan predicts.
Today, agility is thwarted, some observers say, by pharma’s dependenceon documentation. “The industry must come to grips with the paperwork machine that it has built up over the past few years,” says Ray Rogers, a principal with Tunnell Consulting (King of Prussia, Pa.). Paper and documentation issues have not kept up with the way that processes must fl ow, he says. As a result, a pharmaceutical facility “is not just a place that makes product, it’s almost like a documentation machine.”
“As an industry, we’ve messed up the business process around what we do,” says Genentech’s Petersen. “If you make a minor change in a piece of soft ware, for instance, it may take four hours to make the change, but you can count on it taking five weeks to get through the system,” he adds. “I’ve seen flow diagrams for one soft ware change that are over four pages long.”
Consultant Jim Agalloco agrees. “We make our own challenges and create our own barriers to innovation all the time in this industry,” he says, adding, “We’re very comfortable using not just last year’s but last century’s technologies.” Agalloco sees fully automated aseptic processing lines, like those applied in the electronics industry, as a very real possibility for pharma’s future, and a way of eliminating contamination by removing the operator from the process equation. The technology already exists to do this, he says, and regulators understand the benefi ts of automation. ”Someone just needs to have the foresight—or the audacity to say, ‘It’s time,’ ’’ he says. [More from Agalloco in this interview.]
The same conservatism is impeding agile concepts such as Quality by Design, and challenges will only increase as more companies outsource more functions off shore, predicts Prabir Basu, executive director of the National Institute for Pharmaceutical Technology and Education and a professor at Purdue who worked in the pharmaceutical industry for years. “The idea of QbD will resonate with people, but in many instances, applying QbD is like asking people to go from primary school directly into high school. A few students can do it, but most can’t.”
“More incidents like heparin are going to happen before things change,” Basu says. He would like to see equipment being designed for processes rather than the other way around, and to see pharma use more automated equipment. “If we can ever get to a stage like petroleum refining where operations are completely automated without human intervention, we can avoid human error. But right now it’s not happening because it’s difficult to justify the investment in multipurpose plants,” he says. [Read "Musings of a QbD Pessimist."]
Modularizing Construction and Data
Over the past few decades, modularization has allowed plant construction to become much more efficient, says Jacobs Engineering’s Bader. “Builders are getting better at designing equipment so you don’t have to take as much of it apart and the size of modules has increased,” he says. Bader recalls a recent project for Genentech in Singapore. The site was fairly close to the port where the modules had been shipped, so it was possible to put modules that were 25 feet by 100 feet—twice the width limit for transporting them via highway—onto barges to transport to the site.
Modular construction also allows the customer to be much more closely involved in the design phase, says Jacob’s Agarwal, which saves money down the line. He, too, recalls the Genentech Singapore project. The module design and fabrication for the process train was done in the shop in South Carolina before it was shipped. “All the customer input, walkthroughs and other work were done here,” he says.
As modular systems become more common, more vendors will offer pre-engineered projects, although whether they reduce costs or merely move some of the engineering costs into the equipment cost category remains unclear, Agarwal says. [Click here for more from Agarwal on modular construction and single-use equipment.]
Standardization is a double-edged sword, says Tunnell’s Rogers. It’s cheaper on the front end but if you get the wrong system you’re stuck with it. “I like the idea of modular contained equipment because validation and operation issues are worked out, but you can’t force fit such systems to do everything that you may need them to do,” he says.
This modularization also extends to data. Some advocate the use of modules or templates based on the S-88 concept of breaking any process or procedure down into recipes and developing templates for each part of the recipe that can be duplicated easily. For many, the goal is end-toend integration. “We want to take the recipe concept from late-stage development and move it very quickly,” says Genentech’s Petersen.
For Jacob’s Bader, this approach has made plant design much more efficient. “There’s a trend to modularize design aspects, almost taking an S-88 approach to the design of the valves and piping, so that you always fi lter off the tank in the same way, or you do the jacket recirculation system the same way for all tanks in your facility.” For different pieces of the puzzle, he says, “you modularize and use the pieces over and over again.” Bader recalls the design for Amgen’s landmark Enbrel plant in Rhode Island. “When we looked at doing just the process design for that facility, we had over 300 process and instrumentation diagrams (P&ID’s) and a lot of drawings.” There were nine production bioreactors and six smaller seed bioreactors, and another 18 smaller bioreactors, he says.
His team took a modular approach and created S-88 type modules around piping. They did the production bioreactor as one drawing and the rest of the bioreactors were built off that production bioreactor. “We did the most complex thing fi rst then took diff erent parts of that and put it on the seed and smaller bioreactors,” he explains.
As a result, Bader says, most front-end time was spent on 60 drawings and the rest of the 300 drawings were clones. “When you did the control and automation, you were taking these assemblies that were used on multiple drawings and just reproducing that code—all the valves were tagged the same way . . . so the vent valve off the vessel might be valve 10 and that valve 10 is the vent valve on every vessel on the job so you can reuse the code over again,” he says. Although it took more time to organize the project this way, Bader said it saved thousands of hours throughout the design process. In addition, the people operating the plant found it much easier because they always knew what, say, Valve 10 was.
From Data to Information
The plants of tomorrow will also reflect an increased attention to individual user’s data needs, providing the type of information required in the form needed at the right time. Shaping the overall movement will be the shift from a parametric to an attribute approach to data, in which increased focus is given to critical process attributes, says Suroj Patnaik, life sciences industry leader at Emerson Process Management (Austin, Texas). “Fortunately, tools and approaches for collecting, integrating, and managing data from across the operation are emerging and will become more sophisticated in the next 10 years,” Patnaik predicts. “Managers and others will be able to quickly and easily access correlated data on everything from equipment availability to quality monitoring to lot comparisons and use the information to solve production issues—all in a matter of seconds.”
This easy access to integrated information will enable personnel to be more productive. “Imagine a worker on the plant fl oor being able to instantly correlate increasing vessel-fill times with rising pumpbearing temperatures, then identify the best time to schedule bearing replacement so customer deliveries are unaffected,” Patnaik says. “This type of insight has historically required offl ine analysis by experts from multiple domains—but not in the plant of the future.”
Key to this, Patnaik believes, will be human-centered design. In the past, operator displays were oft en based on P&ID’s, a form useful to the engineer. What plant operators are more concerned about is process flow, so the newer user interfaces are starting to refl ect this fact. Instead of deluging workers with data just because sensors and soft ware make it possible, there will be a return to the spirit of the old days when manually operated panel boards alerted operators about process conditions, says Bryan Jones, product marketing manager for Emerson’s DeltaV Human Centered Design program. Such boards had to be easily visible, so they followed ergonomic principles. “It’s back to the future,” he says. At the same time, the idea of “progressive disclosure” of information will allow operators to drill down to the level of detail they need. [For Patnaik's discussion of Human Centered Design and the future of pharma, click here.]
Leaner Plant Layouts
The impacts of Lean and Toyota Production System are starting to be seen, not only in plant operations but in layouts. Lean principles guide all of Pharmatech Associates’ facility design work, says company principal Bikash Chatterjee. However, on the whole, progress has been spotty, says Tunnell’s Rogers. “Some manufacturing processes such as tableting lend themselves more easily to Lean, but other facets don’t,” he says, recalling a sterile facility startup project. “Some features look Lean, but the paperwork issues and underlying control issues are so difficult that the facility can’t really work Lean. They struggle with inventory, moving product around and keeping track of it,” he says.
Breaking down the walls between manufacturing suites and separating them by clear plastic or glass could go a long way to improving process visibility, allowing waste to be more easily identifi ed and eliminated, says Suraj Matthew, head of life sciences consulting for Tefen USA (New York, N.Y.). Deficiencies in pharma facility layouts prompted Joseph Lam, managing director of Beacons Pharmaceuticals, one of Singapore’s largest generic drug manufacturers, to develop a new layout for tableting and solid dosage plants called the Satellite Product Assurity Hub (SPAH).
The design, which was patented two years ago (something unprecedented for plant layouts), aims to improve visibility by including a central monitoring point from which plant operations can be easily viewed, by managers, regulators or even visitors. Modular, it has been designed to interface with standard control equipment and IT, and allows for embedding R&D pilot plants, a feature designed to facilitate technology transfer. SPAH has been eight years in the making so far, Lam says, and is due to be commercialized in another two years. Engineering partners include ABB, Glatt and Foster Wheeler. Lam invented the system, he says, because he believes current plant layouts will not be able to meet regulatory, economic and performance requirements in the near future. Among them, he says, is the need for plants to be Process Analytical Technology (PAT)-ready.
“We will need plants that can cost eff ectively and easily integrate all the technologies to make PAT happen on the production floor, in QC and QA, and even R&D office and computer screens simultaneously,” Lam says. At the same time, Lam says, plants must remain economically viable from the start of capital investment throughout the facility life cycle, given the current global economic environment. “Plants must perform for all stakeholders, the operator, team leader, senior management and shareholders, and offer strategic advantage for the corporate business model.”
Disposables: Blessing But No Panacea
By far the biggest technology shift to hit the pharmaceutical plant has been disposable single-use equipment. Single-use equipment off ers flexibility and greatly simplifies the amount of infrastructure needed in a biopharm plant, says Jacobs’ Bader. For instance, it eliminates the need for CIP skids, clean steam and sterilization, he explains. Currently 90% of the valves on bioreactors are there for steaming and cleaning rather than for the process, so disposables mean a significant reduction in automation requirement and costs. Yet disposables pose layout challenges, Bader says. “It’s hard to pipe materials long distances with single-use equipment, because you’re paying for that tubing every time so adjacencies between media prep and bioreactors and bioreactors and downstream purification equipment become much more critical.”
For instance, one can’t use single use and have cell culture on one side of the hall, and purification on the other, he explains, because that arrangement would require a pipeline to transfer across the hall, and that pipe would need to be cleaned. Observers see single-use technology playing a key role very early in the value chain. Novavax, a vaccine development company, has built a 100% disposable pilot plant using Xcellerex’s single-use bioreactors and mixers, but observers say that the company’s production process, using virus-like particles and insect cell culture, was an unusually good fi t with disposables. [In this article, thoughts from Novavax's James Robinson.]
If one were thinking of going single use, one might run a perfusion rather than a fed batch process, Bader says, because then one could use a smaller disposable bioreactor for perfusion process (500L or 1000 L) and run for a month or more, minimizing the impact of bag costs. With fermentation processes, he says, single use can pose two major problems: Mass transfer is much greater, limiting the size to which you can go with single use; and batch times are short, requiring replacement, so the economics drive users back to stainless steel.
The use of disposables poses some challenges for those who need to adapt sensors to the equipment. Many users work around this problem, in some cases disposing of the sensors along with the equipment. However, disposable sensors are becoming an important technology development area for companies such as SciLog (Middleton, Wisc.), which is marketing precalibrated sensors designed for use in biopharma applications involving single-use equipment. Other issues with single-use equipment, Jacob’s Agarwal points out, are the lack of standardization in size and materials for disposables. Th en, he adds, there’s a need for multiple vendors qualifi ed on the same equipment, in order to have a robust supply chain and minimize manufacturer risk. “Doing one’s homework is critical,” Agarwal says. “We hear of quality issues with some purchased disposable equipment,” he says.
PAT Will Be Mandated
Analytics will continue to play a more important role in improving pharma facility agility, so much so that use of PAT will eventually be required, predicts NIR spectroscopist Emil Ciurczak, our contributing editor. Companies will have to do more and more characterization of non-API raw materials, he says, including such physical parameters common to API as polymorphic form, crystallinity, fl owability, crushability, porosity and surface area, to better understand how solid dosage forms are made. Competitive pressures and business realities will help ensure that PAT plays a prominent place in future plants, Pharmatech’s Chatterjee says, as a tool to help brand and generic companies manage the bottom line, with a profound emphasis on the Cost of Poor Quality. Certain analytical techniques will figure more prominently in future pharmaceutical quality assurance.
Chatterjee predicts that Raman, NIR and inductively coupled plasma mass spectrometry (ICPMS) will all play key roles—Raman, to confi rm laminate structure and integrity in packaging and as a diagnostic aid, NIR, to measure overall potency and provide process verifi cation, while ICPMS will be used on the shop fl oor and in the laboratory, to detect minute quantities of non-metallic substances such as magnesium stearate.
Ubiquitous Analytics, Wireless Plants
The trend to monitor, measure, and analyze every facet of production will also extend to packaging and printing, says Steve Simske, Hewlett-Packard’s chief technologist for security printing. “There’s going to be a lot more integration of inspection and interrogation into manufacturing lines,” he says. With the advent of nano-sensors and other advanced technologies that provide real-time data about your processes, manufacturers will have the ability to control printing and packaging operations like never before. The ability to conduct what Simske calls “ubiquitous analytics” will not necessarily lead to increased complexity of operations, however. Just the opposite, he says. One key to this trend will be the ability to have hybrid sensors, those that, for example, assess label integrity, product weight, package shape, and so on. Whereas manufacturers currently need to install expensive cameras and other inspection devices in various places along their lines, they will now be able to combine them into fewer locations.
“You’re going to see a wider variety of inspection and interrogation devices, but they’re going to be simpler,” Simske says. Another factor leading towards increased printing and packaging simplification is the trend towards integrating myriad safety and brand features into one printable label. From miniaturized RFID tags for security to batteries and LED lights for brand promotion, all will potentially be incorporated into one label and applied via one single printing/production path. [Read "When Manufacturing and Packaging Merge".]
At the same time, vendors such as Emerson and Honeywell report that wireless is proving itself, not just for inventory monitoring but for robust control. [Here is Honeywell's vision.] With handheld devices, experts predict, wireless platforms will allow users to improve transparency and visibility into their processes, and allow for more effi cient information access. Wireless platforms will also help tomorrow’s pharmaceutical facilities improve fl exibility by reducing wiring, installation and verification costs, they say.