Getting quality down PAT
Batch standards build in quality, timeliness in life sciences processes
- Market pressures mean drugs need faster production time.
- Building in quality essential to life sciences manufacturing processes.
- Batch standard improves processes, quality; reduces time to market.
By Baha Korkmaz, Arnold Martin, and Cenk Undey
Life sciences industry companies are under increased pressure to introduce new drugs to the market more frequently than ever. But while getting drugs to market quickly is important, keeping quality inside the process is paramount. Companies that proactively approach the new technological trends and invest in automation have a better chance of success. Using the concepts behind process analytical technology (PAT) is one effective tool to achieve this success, especially when incorporating batch standards, such as ISA88. (See accompanying article on process analytical technology on page 26.)
When those in the life sciences industry discover a new drug, the corporation applies for a patent. During the patent approval process, scientists plan to introduce the new drug. The Food and Drug Administration, or FDA, approval cycle activates, and clinical trial phases begin. The pharmaceutical or biotechnology business has plenty of incentives to decrease the introduction time and maximize the growth and maturity time for the product. Once the FDA approves the patent, the product must hit the market as soon as possible.
Batch standard builds maturity
Product sales are at their maximum levels when drugs reach their maturity stage during the life cycle. The objective is to get to the maturity level as soon as possible and continue at that level as long as possible. But when the patent expiration time nears, the product sales begin dropping. If it was a blockbuster drug, it is likely other life sciences companies introduced similar drugs to the market in the mean time. These competitive pressures encourage life sciences corporations to invest in product lifecycle management with ISA88 guidelines, manufacturing execution systems for electronic work instructions, site-specific automation above the process cell specific automation, and other enterprise resource management initiatives.
Since the early 1990s, manufacturers have applied the ISA88 batch control standards successfully. These standards became an expected model and terminology for all the batch process control and automation projects. These applications are often limited to the process cell domain and mostly deployed at commercial manufacturing facilities and pilot plants. In a large pharmaceutical corporation with hundreds of process cells, most are automated by now under the ISA88 guidelines.
Embedding PAT into ISA88 components will help in gaining quality and safety. These results probably will come from reducing production cycle times by using on-, in-, and/or at-line measurements and controls. This process could prevent rejects and re-processing (considering the possibility of real-time release), increase automation, and facilitate continuous processing to improve efficiency and manage variability. There are at least two ways of applying ISA88 to PAT. One is using the PAT process with ISA88 modularity, model, and terminology to start, execute, end, and report PAT results. The other is integrating PAT to master recipes, control recipes, and advanced process control supervised by the ISA88-based unit supervision.
Other control activity model elements such as production information management, process management (executing PAT-specific small recipes), unit supervision, and process control see use with PAT. The ISA88 architecture and modular approach allows building the quality into products via PAT. This approach supports the understanding and use of the relevant multi-factorial relationships among material, manufacturing process, and environmental variables and their effects on quality. You can obtain the data and information to help understand these relationships through pre-formulation programs, development, and scale-up studies, and from data collected over the life cycle of the product. Using ISA88 is essential to organize and execute all these activities, including the data acquisition, calculation, and reporting.
Better batch in biomanufacturing
Commercial bulk biomanufacturing processes are typically comprised of a series of unit procedures operated in a batch mode to produce therapeutic proteins. Batch biopharmaceutical processes have complex reaction mechanisms and non-linear, time-variant process dynamics that make their modeling, monitoring, and control challenging. Mammalian cell culture processes, in particular, are characterized by prescribed growing and processing of cells with nutrient additions for a finite duration to produce a target protein as a product. Purification processes run to obtain specified purity and efficacy, to complete the production yielding into the bulk drug substance. It is optimal to have a high degree of reproducibility in cell culture and purification to obtain successful batches. Besides the process complexity, biomanufacturers can measure quite a few process variables during the course of a production run. It is important to efficiently monitor and diagnose deviations from the in-control space for troubleshooting and process improvement purposes.
Statistical monitoring more efficient
Multivariate statistical process monitoring (MSPM) technology allows for an efficient means of monitoring many variables simultaneously and is able to explain how they are changing in correlation with performance variables. It is aligned with FDA's 21st Century cGMP guidelines around how the pharmaceutical industry designs, monitors, and controls its processes by implementing quality-by-design PAT tools. And it is one of the key technologies needed to achieve these aspirations and expectations from the agency.
By enabling MSPM in real time, you can further increase the process understanding and supervision, since you can monitor more variables together while a batch is in progress. This technology also provides a proactive and comprehensive monitoring framework towards design and control space of processes. This enables process engineers to proactively monitor the process performance and allows earlier detection (risk reduction) of any developing deviations within a batch or across batches and efficient identification of improvement opportunities.
One approach uses ISA88 hierarchical models (both procedural and physical) in automation and data integration systems to better enable multivariate process monitoring and control. The initial methodology is to take advantage of MSPM by simply making good use of abundant data generated by the manufacturing process to gain a better understanding of the process, and to quickly identify variables that are most likely causes of deviations from the desired process trajectory. Manufacturers can evaluate capability and availability of the current sensors in a non-intrusive manner. This allows an assessment of the potentials of generating soft sensors that are process-model driven.
This knowledge can help in making decisions regarding the acquisition and installation of new analytical measurement and sensor technologies (such as in-situ, in-line or at-line). Real-time MSPM also minimizes resource time requirements due to its ability to quickly and efficiently analyze large and complex data sets. Current solutions under evaluation provide a non-invasive, platform-independent approach offering a quick deployment opportunity to fully enable use of the batch manufacturing data towards proactive monitoring and to facilitate process operational excellence.
ABOUT THE AUTHORS
Baha Korkmaz is a principal consultant at Invensys in Warwick, R.I. Contact him at firstname.lastname@example.org. Arnold Martin is an automation engineer at Amgen, a biotechnology human thera-peutics company in Coventry, R.I. Contact him at email@example.com. Cenk Undey, Ph.D. is a principal engineer at Amgen in West Greenwich, R.I. Contact him at firstname.lastname@example.org.
Four tools of PAT initiative
By Alex Habib
The Food and Drug Administration's (FDA) process analytical technology (PAT) initiative is an idea whose time has come. Instead of depending on validation protocols, reports, and other documentation that still leave us with processes we do not understand and cannot control, in a PAT framework you can monitor and control manufacturing processes using on-line process instrumentation and control systems.
This will lead the industry into continuous improvements in quality and productivity.
Most pharmaceutical and biotech manufacturing facilities have an established general quality system, product-specific standard operating procedures, and equipment qualification and validation procedures to ensure a product's quality before it ships to customers.
The FDA now recognizes PAT has the added advantage of being a real-time monitoring and controlling tool and can facilitate continuous improvement of the process and hit the best quality target.
The FDA recognizes four tools for PAT: process and endpoint monitoring and control systems; multivariate data acquisition and analysis tools; modern process analyzers and process analytical chemistry tools; and continuous improvement and knowledge management tools.
Most pharmaceutical and biotech manufacturing companies are now aware of the importance of PAT to keep up with the FDA's initiative and ensure good and consistent product quality. The first step was to implement PAT on a small scale in the research and development labs. Several pharmaceutical companies have successfully implemented this step. The next important and urgent step is transferring the PAT concept from the laboratory and pilot plant to the full- scale manufacturing factory floor.
One obstacle is developing ways to justify the economics of deploying PAT projects on a large scale. Another is designing and building in-line systems to get representative samples from the process equipment to the analyzers in a timely manner. Other challenges include training and certifying plant personnel to calibrate and maintain online analyzers and adapting the current multivariate statistical process control techniques in a batch environment. (See multivariate statistical process control in accompanying article.)
In a fermentation process, monitoring starts from the seed stage with mostly off-line measurements. As the fermentation moves to the production stage, online measurements begin with the fundamentals, such as temperature, pressure, oxygen flow, and then the more complex analyzers that deal with gases and liquid chromatography.
Distributed control systems (DCSs) and programmable logic controllers (PLCs) play an important role in creating the optimum fermentation conditions by automating the oxygen addition and controlling the temperature, pH, dissolved oxygen, and others. These PLCs and DCSs coupled with batch-sequence-control software further ensure the product consistency and quality from batch to batch.
One of the most important aspects of optimizing the fermentation process is determining the end point. This is possible using multivariate statistical process control software tools that take several signals from the process and create a fingerprint of the ideal batch.
By ending the fermentation process at the perfect time, optimal and maximum product is possible. Of course, off-spec product is at a minimum. Additional benefits to the fermentation process are possible by using modern alarm-management software tools to predict sensor failure.
The pH and dissolved oxygen sensors are subject to several heating and cooling conditions as the fermenter goes through the sterilization cycles.
By predicting a sensor's failure accurately using PAT techniques, a valuable batch of product will not become waste product.
ABOUT THE AUTHOR
Alex Habib, P.E., is an automation and validation consultant working with process control systems for pharmaceutical and food facilities. He also chairs the ISA5.6 Control Software Documentation standards committee, and is ISA's District 2 vice president. Contact him at email@example.com.
Process analytical technology: A primer
Process analytical technology (PAT) is a process for designing, analyzing, and controlling manufacturing through real-time measurements of critical quality and performance attributes of materials and processes. The Food and Drug Administration promotes PAT as a risk-based framework for innovative pharmaceutical development, manufacturing, and quality assurance, while maintaining and improving the current level of product quality assurance. The goal of the PAT framework is "to design and develop processes that can consistently ensure a predefined quality at the end of the manufacturing process. Such procedures would be consistent with the basic tenet of quality by design and could reduce risks to quality and regulatory concerns while improving efficiency" (http://www.fda.gov).
Today's technology allows the life sciences industry to incorporate PAT into the drug discovery and scale-up process, manufacturing process, and quality assurance. Applying ISA88 to this new path improves the efficiency and quality within the product life cycle. The new drug discovery process and the type of drugs in discovery today are requiring cutting-edge scientific and engineering knowledge. Since PAT is an integrated (even embedded) part of the process control, you can also handle it by using the ISA88 standard. An ISA88 recipe will contain PAT elements, such as formula and algorithms.
PAT in biofuels
Process analytical technologies (PAT) are not just seeing use in food and pharmaceuticals. The biofuels industry is using analyzers online and in process testing samples in biodiesel and bioethanol. "For online analyzers they have density mass flowmeters calibrated to either provide percent solids or moisture content," said Michael Tay, director of sales engineering at Pavilion Technologies, an Austin, Tex., model-based software company specializing in biofuel processes. "The online analyzers' real-time collecting samples automatically form the process without manual intervention. And in process testing, they're collecting samples and analyzing them in an in-process laboratory," he said.
Implementing PAT helps one energy company "control the distillation columns and molecular sieves processes, which are challenging to manage optimally and yet critical to achieve our operating objectives-more, high quality ethanol for less," said David Culver, director of operations at Glacial Lakes Energy in Watertown, S.Dak. "Furthermore, we are able to continuously reduce our energy usage. And in today's marketplace, being able to produce more ethanol at higher yields equals real money."
In fact, the basic concepts and principles of PAT in biofuels are closest to those used in brewing. "Both are fermenting grain into an alcohol," Tay said. "Bioethanol plants are new compared to most brewing facilities, so their automation level is higher. And the scale of their fermenters is larger," he said.
One of the biggest challenges of using PAT in biofuels is the consistency in process analyzer calibrations. The first step in PAT is the analytical technology-"how you're collecting samples and measuring quality," Tay said. "And there are best practices in how you collect samples and calibrate instruments so you get reliable measurements. You have labs and people taking samples, but you have to make sure the way they take samples and measure quality is consistent."
Some manufacturers are seeing a 5% increase in yield using PAT. "They look at gallons of ethanol produced per bushel of corn used. So if they're using a full-blown process analytic technology, in process testing, they're analyzing the data to model the processes relationships. And they use that in a closed-loop control system. They see a 5% increase in yield. So they get 5% more ethanol per bushel of corn," Tay said.
Now with PAT, manufacturers are taking consistent samples. "If you take the analytics and measure with modeling software, you can get new insights into the process, and that enables optimal control," Tay said. One way to do that is using artificial neural networks and fundamental modeling toolsets to develop relationships between the data. "If the pressure is x and the temperature is going up, how much am I going to change quality? It's all about understanding relationships and developing a model-based controller," he said. "Here's what quality is, now I can control temperature to maintain quality."