In some respects, the life sciences sector is unique among industries. Every move pharmaceutical and biotech manufacturers make is tightly monitored by the U.S. Food and Drug Administration (FDA), and the quality of the end product is literally a matter of life and death. So, it's no surprise that these manufacturers do everything possible to ensure the highest quality of their end product at all times.
But quality comes at an extremely high cost. Goods are laboriously tested before being released to the market, resulting in an industry average for rework and scrap rates of 50%, according to AMR Research (Boston). The costs of this post-facto approach to quality control are high -- a single scrapped batch can cost a company as much as $3 million to $4 million, according to AMR.
These error rates, which are much higher than those in other process industries such as oil & gas and food & beverage, mean the entire industry suffers not only in terms of cost but in protracted cycle times. Pharma cycle times typically run 30 to 90 days, according to AMR, but when processes go awry, that time can double.
"Big pharma has had the luxury of huge margins," says Alison Smith, senior research analyst at AMR. "The cost of raw materials is small. Unlike virtually every other industry, they have been able to operate remarkably inefficiently. Now they are trying to learn how to fix problems before they have to scrap the product," she says.
Toward that end, the FDA's Center for Drug Evaluation and Research (CDER) began in 2003 putting together guidelines called the Process Analytical Technology (PAT) initiative. The thrust of this non-binding program is to help pharmaceutical and biotech companies figure out how to reduce process variability and build quality into the drug manufacturing process, rather than going through the costly exercise of tacking it on at the end.
PAT can take many different forms, but in general refers to a framework that life sciences manufacturers can use to design, analyze, and control manufacturing processes by measuring quality attributes of raw and in-process materials. Unlike 21 CFR Part 11, it is a voluntary initiative; there are no associated deadlines for compliance.
While other process industries (such as chemicals, oil & gas, and food & beverage) started similar programs years ago, PAT is the first regulatory effort in this sector to focus on process rather than documentation. As such, its potential impact on life sciences is much greater. And while most of the largest pharmaceutical companies in the world have started PAT pilot projects, none is very far along, according to Smith, since the concept is new and requires such sweeping changes in the way these manufacturers do business.
The FDA has taken a strong position with its endorsement of PAT: It's time for pharmas and biotech companies to invest in modern technology. "The whole idea is to bring some science and engineering to the pharma manufacturing process, which has traditionally been very manual," says AMR's Smith. The benefits could be enormous. Chipping away at the scrap rate could add up to huge cost savings for manufacturers. And anything that aims to increase drug quality will benefit consumers by producing safer end products.
To date, the industry has always relied on corrective action. "They manufacture [the product] and then see if there is something out of spec. They find out after the fact and have to make a correction," says Glenn Restivo, marketing manager for Rockwell Automation Inc. (Milwaukee, WI) and the author of a white paper on PAT. "With PAT, the benefit is continuous quality verification. PAT is a paradigm shift. It's not lab-centric."
PAT allows pharma manufacturers to replace some of their incessant testing with detailed explanations of their processes -- a sea change from the way they have done business before. That's one reason companies have been fairly slow to adopt PAT. "We don't today see big projects or corporate deployments," says Frédéric Halley, vice president of marketing for Pertinence, a Paris-based vendor of enterprise manufacturing intelligence software. "Companies are evaluating what they can do. Pharma tends to be conservative and somewhat slow-moving."
Halley is familiar with some PAT pilots in pill tableting that focus on steps such as drying. Rather than employing routine drying processes, which err on the side of more time instead of less, in a PAT application the manufacturer might add an infrared probe to the dryer to see the exact moment when the product is dry. "That allows you to have less variability in the product at the end," and also reduces process cycle time, Halley says.
He calls these PAT pilots "serial-step" initiatives. "There is some benefit if you optimize that part of the process. You output the first step and adapt the next step with the input of the previous one," he says.
The exciting fact for the industry is that there is much automation yet to be done, and therefore a great deal of efficiency to be gained. "There is tremendous room for improvement," Smith says. "Pharmas have tended to think of themselves as different. They have some different paperwork requirements than other industries, but they would greatly benefit by using technology that is used in other industries."
Smith cites the food & beverage sector as a particularly apt comparison. The quality of dairy products and eggs, for example, is paramount -- people can and do die from food infected with salmonella and other harmful organisms. And there is a far greater user base for those goods than for drugs. Smith says pharma should learn to emulate industries like food & beverage and chemicals, where razor-thin margins make process automation an imperative. "There is huge value in cross-industry fertilization," Smith says.
Broad Spectrum of Tools
Though PAT is fairly new, today there is a broad range of tools and technologies -- some long established -- that pharmas can employ to make their product supply more predictable. The list starts with infrared spectroscopy systems, sensors, and historians, and runs to manufacturing execution systems (MES), advanced process simulation (APS) tools, and manufacturing intelligence/ enterprise quality management.
For example, OSIsoft (San Leandro, CA) just announced an agreement to embed its data historian and batch-processing technology in the TruBIO Bioreactor Control System from Finesse LLC (Irvine, CA), which is used for life sciences manufacturing applications. The bioreactor includes a controller and a utility tower with the instrumentation, pumps, motors, sensors, valves, and meters necessary to monitor and control parameters such as temperature, gas flow, and pH. In combination with the OSIsoft technology, the TruBIO System supports the PAT framework, according to the companies.
Similarly, Rockwell Automation's Logix controllers can be used as part of a customized control system that incorporates instruments, controllers, chromatography, and spectroscopy in furtherance of PAT, according to Restivo. Process automation control is not a simple matter in the pharma industry, he adds. "If you made one product on one line, controlling variability would be easy to do. But [pharma companies] are making a lot of different products using different equipment in the same plants. When you take all those differences into account, it becomes difficult to apply the same tools." That's why Rockwell advises that its customers take a highly customized approach to PAT.
ABB (Zurich, Switzerland) is also active in this area, says AMR's Smith. "ABB has a lot of tools in this area. They can measure material distributions in pills. They can manage out-gases during processes. There are all these different ways to tell how your processes are performing," she says.
For its part, Pertinence has two software products that have applications in PAT. One, called Pertinence RuleMaker, is a rule-based modeling technology. The other, Pertinence OperationsAdvisor, allows the manufacturer to put the models into production, according to Halley. Modeling and simulation are among the most sophisticated tools to apply to PAT, Smith says. "Generally once [the pharma] receives FDA approval on a drug there is such a rush to get it from new product development to manufacturing that they don't build a model of the process," she says. Skimping on that critical step is a mistake. "They have to understand how to instrument that process so they can understand what to measure and how to control variability," she says.
Advanced process control vendor Pavilion Technologies (Austin, TX) and analytical instrument maker Thermo Electron Corp. (Waltham, MA) just announced a partnership to provide PAT solutions. The partners will provide best-of-breed technologies to allow life sciences manufacturers to leverage real-time analyzer data to build robust process models, enhance process understanding, and improve production performance.
Processing PAT
Despite the breadth of PAT-related offerings, Smith cautions that it would be a mistake to conceptualize PAT as purely technology. Instead, it is an architecture that incorporates elements of sophisticated instrumentation, real-time data collection, analytics, and advanced simulation and control.
It is not possible for manufacturers to go out and buy a PAT system the way they might buy an ERP system or an MES. "There is no such thing as a PAT vendor, other than as marketing," Smith says. The range of potential technologies that could push a pharmaceutical company toward PAT is so great that putting a "PAT stamp" on a product would be meaningless. Thoughtful vendors are aware of this conundrum and are careful to illustrate how their products can help a manufacturer work toward PAT goals, as opposed to saying that they sell PAT technology.
The bottom-line attraction of PAT is two-fold, according to AMR. First is eliminating error-prone manual intervention from production processes and, thereby, the likelihood that non-conforming products will appear at the end. Second, the ability to accurately produce to demand has the potential to radically reduce scrap and rework. The dollar savings could be simply astronomical.
The pharmaceutical industry is littered with aging manufacturing facilities and their fragmented IT infrastructures, Smith says. Though margins are still generous, the cost of compliance is escalating and will begin to eat away at the bottom line. And traditional pharmas will feel more heat from generic drug makers -- the generics market is expected to reach $50 billion in the next year or so, according to AMR.
So, although PAT compliance is not an immediate threat to these companies, it could help protect their profits. Clearly, the FDA believes it is time that pharmaceutical companies implemented the technology to improve processes that other industries improved long ago. Even if it is not critical to the continued viability of the industry, producing higher quality goods with less variation is a worthy goal. With PAT, Restivo says, "you're building quality in from the get-go. There shouldn't be any mystery at the end."