New Methods for the PAT Toolbox

Oct. 19, 2004
Instrument suppliers are moving a growing number of analytical techniques out of the lab and onto the manufacturing floor. As healthcare economics change, drugmakers that once rejected process analytics are embracing it as they look to reduce costs.
By Angelo De Palma, Ph.D., Contributing EditorTop drugmakers have always sought--at least in principle--some sort of process analytic capability. However, vendors and manufacturers agree that without FDA's prodding, nothing remotely like the current push toward process analytical technology (PAT) would ever get off the ground. "Manufacturing was always viewed as drug discovery's ugly stepchild," says Andrew Malcolmson, director of business development at Malvern Process Systems (Southborough, Mass.). As a developer of particle size measurement equipment, Malvern has offered in-line laser diffraction analyzers for thirteen years, with more than 400 installations worldwide. "But until recently pharmaceutical companies wouldn't touch them," says Malcolmson.Now, as healthcare economics change, drugmakers that had once rejected the idea of process analytics are embracing it as they look for new ways to reduce manufacturing costs. FDA's endorsement of analytics provides a much greater incentive.Not only is the Agency advocating PAT, but it's also suggesting that users take a systems approach to implementation. "PAT is not just in-line analyzers, it's about process understanding," explains Chris Watts, Ph.D., one of FDA's PAT point people at the Center for Drug Evaluation and Research (CDER). "PAT is about designing, understanding, and controlling operations," he says. "It's a thought process."So far, pharmaceutical manufacturers' responses to FDA's PAT initiative range from doubtful to enthusiastic. Watts dismisses doubters' concerns. "PAT is receiving top-down support, right from the Commissioner," he promises. Companies should no longer be thinking strictly in terms of process validation, he adds. "Rather, they will have active controls on the process and validate those controls," he says.Light Induced Fluoresence, used to determine powder blend uniformity, will soon be available commercially. Shown here, the technique is used to analyze blending and tableting operations in a CAMP pilot facility.More Knowledge RequiredDespite FDA's pronouncements and the urgings of process analytics experts, a knowledge gap remains among manufacturers regarding PAT and its goals, says G. Patrick Stahly, Ph.D., chief operating officer at SSCI, a consulting and analytical services provider in West Lafayette, Ind. A key goal is quality by design. Another often-stated goal is deeper process understanding. However, a good grasp of process fundamentals is essential at the offset of any PAT project, Stahly says. "Manufacturers need to understand which process factors are critical to obtaining quality product and then control and/or monitor those factors."SSCI provides services for studying solid materials' crystallization and polymorph behavior. The company employs the usual spectroscopic tools for monitoring nucleation, crystal growth rates, and polymorph composition--understanding which factors are critical to crystallization, picking the right modality for monitoring them, and deploying that technology within the process. "Unless you understand the process, all the monitoring in world won't do you any good," he says.The (Un)usual SuspectsFDA's open-mindedness about specific analytics may be liberating, but presents processors with choices they may not be used to making. On the technology front, several PAT-worthy technologies already are established, particularly infrared (IR) and near-infrared (NIR) spectroscopies. Materials processors have used IR/NIR, usually through fiberoptic probes, since at least the late 1980s.Although use of IR/NIR is expected to grow, alternative process analytic developments are emerging from techniques that are less familiar to bench organic chemists, plus methods that haven't hit big-time real-time analytics markets (box).Ionics (Boulder, Colo.) has been offering process analytics, under the rubric "Process Analytical Instrumentation," for about twenty years. The firm's specialty, total organic carbon (TOC) measurement, is employed in wastewater, semiconductor, and chemical industries in real-time but in pharmaceuticals only remotely, off-line. "Pharmaceuticals are lab-centric," explains Nissan Cohen, environmental marketing manager. "Back in 1993, when we began in this business, pharmaceutical companies didn't even want to look at on-line or in-line analyzers."So Ionics found a way to turn an off-line test into a "fail-safe on-line method," according to Cohen. Fitting an existing process with a TOC analyzer requires a quarter-inch takeoff line and "an electrical box," although, according to Cohen "pharmaceutical manufacturers will make a bigger project out of it because of their need for installation qualification, operation qualification, and performance qualification."Ionics also developed a methodology it calls EWPS, for Exponentially Weighted Process Statistics, which provides a data filter based on the age of the information. For example, a measurement taken every second will be considered extremely relevant to current process conditions, but one taken an hour ago will be considered much less relevant. "The more frequent the information, the more reliable it is for controlling processes."Raman Makes InroadsLike most instrument companies active in pharmaceutical PAT today, Kaiser Optical Systems' (Ann Arbor, Mich.) activity in process analytics predates FDA's interest by many years. Kaiser claims to have built the first Raman process analyzer in 1993.Although less used than IR and NMR, in a process setting Raman does some things quite well. Unlike IR techniques, Raman is non-invasive, requires no sample preparation, and may be combined with microscopy for studying small particles or differentiating among crystalline polymorphs. Raman also complements IR: Where the latter detects changes in molecular dipole moment, Raman observes changes in polarizability--for example, the symmetric stretch in carbon dioxide. Raman's non-invasive, non-contact analysis makes it ideal for processes where sampling is impossible or inconvenient.A majority of pharmaceutical companies are using Raman in one form or another during discovery and compound screening," notes Ian Lewis, Ph.D., marketing manager at Kaiser. "It's also picking up during manufacturing," he says.Kaiser Optical's experience with Raman cuts across pharmaceutical processes, for example, to perform polymorph screening, or to monitor catalytic hydrogenation and Grignard reactions, crystallization, and general reactions.Bruker Optics (Billerica, Mass.) has a similar story. It had been active in vibrational spectroscopies (IR, NIR, and Raman) for more than a decade and was among the first companies to participate in PAT concept meetings. Instead of a new set of regulations, PAT will be provide a framework for overcoming real or perceived regulatory hurdles to new technology adoption, says company vice president Dan Klevisha.NMR Poised At-LineProcess nuclear magnetic resonance (NMR) spectroscopy, while not as convenient as NIR, is a technology worth watching. NMR gives detailed, quantitative information about chemical environment and is familiar to every pharmaceutical chemist. NMR is versatile, analyzing more than a dozen nuclei of interest to pharmaceutical processors in addition to 13C and protons.NMR's drawbacks--expensive instrumentation, sample preparation that supports, at best (at least for now), at-line analysis--could hold it back in PAT settings, however. Another NMR drawback, that signals are easily overwhelmed by solvent, can be compensated for through frequency-lock techniques and solvent suppression. A workaround for high cost is to use inexpensive low-field instruments in process settings.John C. Edwards, Ph.D., manager for analytical and process NMR services at Process NMR Associates (Danbury, CT), claims that 60-MHz instruments work quite well for routine chemical processing. One of Process NMR's goals is a fully-automated, low-priced 300 MHz NMR that a technician can run easily. The company has teamed up with the process control specialist, Invensys Process Systems (Foxboro, Mass.), to reach the pharmaceutical manufacturing market. Up-and-Coming PAT Techniques
A number of other traditionally laboratory-bound techniques are being adapted to process manufacturing applications.Light-induced fluoresence is one option. The technique uses a portable sensor incorporating a light source, solid-state photomultiplier tube and optical filter, using the degree of fluorsence to determine the homogeneity of powder blends. This nondestructive technique was developed at the Massachusetts Institute of Technology, and further refined by the Consortium for the Advancement of Manufacturing of Pharmaceuticals (CAMP), based at PurdueUniversity. It is now being offered by Light Pharma (Cambridge, Mass.). Studies have indicated that the technique could be used to analyze up to 3,000 tablets per minute.Ultrasonic spectroscopy, meanwhile, shows promise for analyzing such challenging materials as microemulsions. Unlike NIR, which can only provide composition data, the technique provides composition as well as microstructure information, according to Ultrasonic Scientific Ltd.'s (Dublin) president Vitaly Buckin. The technique, which is nondestructive and can be used with opaque materials, measures the velocity and attenuation of high-frequency sound waves as they propagate through materials, allowing rapid analysis of formulation consistency, ingredient and intermediates. Ultrasonic Scientific is currently working with a number of research institutes and companies on PAT applications, Buckin says. The technique can be used with samples as small as 0.03 mL. Thermal effusivity does not enjoy the popularity of NIR or Raman as a PAT technique, but Mathis Instruments (Fredericton, New Brunswick) makes a good case for effusivity sensing. Based on the flow of heat from high-temperature to low-temperature objects, effusivity is proportional to the object's thermal conductivity, density, and heat capacity. Mathis president Nancy Mathis, Ph.D., likens effusivity to the sensation that a piece of metal is "colder" than a piece of wood or foam at the same temperature. "The reason is that the metal draws heat from your fingers more rapidly than the foam." Mathis sensors supply a small amount of heat to objects with which they come in contact, and instantly measure how much heat reflects back to them. Like fingertips, effusivity sensors are interfacial, supplying both the heat and measurement. Since they measure an intensive property rather than an extensive property (e.g. total heat) samples need not be a specific size. In addition, the analysis is nondestructive, and takes from one to sixty seconds. Thermal effusivity serves as a proxy for several parameters of interest in pharmaceutical manufacturing and formulation. In blending, effusivity correlates with blend uniformity and homogeneity of multi-component powders. The technique's ability to "see" through layers allows it to provide insight into a tablet's core ingredients, or to detect components within controlled-release structures or packaging. Since effusivity depends on density and composition, the technique can quickly alert tabletters that blends are low in a key ingredient or high in moisture. In fact, moisture detection is a no-brainer using effusivity. Water's effusivity is 1600 whereas a powder's is typically about 300, Mathis explains. "This massive difference allows you to detect down to 0.25% water content," she says. Thermal effusivity does have some limitations. Powders close in effusivity appear blended when they are not, and the technique requires contact with product, which slows down the analysis. "You can't use it to measure material that's flying by," Mathis admits. "So it's more useful for on-line, rather than in-line, testing."