A focus on quality will enable transformation

By Justin O. Neway, Ph.D.

In most discrete manufacturing industries, quality initiatives have long played a critical role in improving production outputs while saving money. Billions of dollars have been invested in Six Sigma and other continuous process improvement projects that leverage technology and manufacturing data. Why, then, has the pharmaceutical industry been slower to embrace such progress and the bottom line benefits that come with it?

A new trend that may lead to more widespread adoption of Process Analytical Technology (PAT) is the industry's focus on "Quality by Design" (QbD) in the new regulatory environment. Leveraging accessible, real-time and historical process data for process understanding and quality in PAT efforts will change the positive correlation between PAT implementation and bottom line improvements. In 2003, the U.S. Food and Drug Administration (FDA) really started emphasizing process understanding as a way to improve quality. Through its 2004 "Guidance for Industry: PAT - A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance," the FDA challenged pharmaceutical manufacturers to achieve a level of process understanding consistent with controlling process variability and assuring product quality in "real-time" while the batch is being manufactured (Real Time Quality Assurance or RTQA). Ideally, the ability to achieve the appropriate quality outcome must be designed into the process itself rather than relying on final product testing. This increased emphasis on QbD requires pharmaceutical manufacturers to make larger investments earlier in the product life cycle - during process development in advance of approved commercial operations. The goal is to develop a sound scientific basis for a manufacturing "Control Space" that accommodates a range of defined variability in the commercial process materials and operations and still produces the desired product quality outcomes.
Quality Requires On-Demand Access to Data

(Click image for larger version.)
When it comes to leveraging data, accessing and using it for collaboration become essential. Many pharmaceutical companies have been standardizing the desktop and back office environments in order to gain better systems control and consistency. As the industry moves toward collecting and analyzing more process data (including continuous or on-line data), standardization requires new thinking. One example is the data warehouse. Its value has been based on the premise that if data is hard to get from disparate data sources, then putting it in one place will solve the issue. While data warehouses typically contain certain data that is useful for identifying process trends or the causes of variability, warehouses are generally not real time as they do not manage continuous data well because of complex synchronization requirements. To be effective, process improvement initiatives must leverage real-time data from newer PAT instruments and other on-line measurements as well as data from offline measurements and historical data all at the same time. Process improvement and PAT need a framework for managing the manufacturing process and enabling collaborative investigational analysis of the resulting data to improve the predictability and quality of operations and products. The key is to provide direct on-demand access to end users not only to the summary production data but also to the individual underlying data elements in a context that is natural to users who are (non-IT) process experts. This improves the speed with which they can identify and understand underlying cause-and-effect relationships.
The Role of PAT in Quality by Design
Achieving QbD may involve the use of instruments more sophisticated than those currently used in pharmaceutical manufacturing processes. Some of these instruments have been used for decades in other industries, but have not yet been applied to pharmaceutical production processes. Some of the newer instruments available to life science manufacturers make relatively simple measurements like effusivity. Other instruments make much more complex measurements like Near Infrared (NIR) absorption. In many cases, these instruments are capable of measuring the Critical Process Parameters (CPPs) and Critical Quality Attributes (CQAs) in real-time. Such instruments generate large amounts of data that must be understood if the measurements are to be useful. The usefulness of any PAT or other process improvement initiative in QbD depends on all the data (discrete, replicate, continuous and paper-based) and the right process trending, reporting, descriptive analysis, univariate and multivariate cause-and-effect analysis, and parameter relationship modeling capabilities all being easily available on-demand to users in the same integrated environment. Users must be able to work with continuous, discrete and replicate data together for quantitative analysis. As understanding of the cause-and-effect relationships CPPs and CQAs improves, it may be desirable to adjust process control systems in real-time by interfacing them directly with process models derived from examination of process development and/or manufacturing process data. This is consistent with the real-time quality assurance goals of QbD that a PAT approach can offer. Such models derived from manufacturing data are generally the best models to use for control of full-scale operations because the effects of scale have been accounted for due to the fact that they were derived using full-scale data.
The Regulatory Perspective
An understanding of the regulatory perspective and a fundamental desire to realize the many benefits of process improvement are the keys to wider adoption of PAT and achievement of QbD. An important section of the FDA's PAT guideline points to the value of continuous learning that comes from the analysis of process data when coupled with systems that support the acquisition of knowledge from that data: "Continuous learning through data collection and analysis over the life cycle of a product is important. These data can contribute to justifying proposals for post-approval changes. Approaches and information technology systems that support knowledge acquisition from such databases are valuable for the manufacturers and can also facilitate scientific communication with the Agency." (From "PAT - A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance. Pharmaceutical CGMPs," September 2004) In fact, the data to which the FDA refers comes not only from process instruments making real-time measurements on the current batch, but also from off-line measurements of the current batch and on-line and off-line measurements of previous batches and process development work. All these different types of data from the current and previous batches are essential components of the knowledge base that can be tapped by manufacturers to achieve QbD. Furthermore, these data are fragmented because they have accumulated in the operational data stores of many different systems. In many pharmaceutical manufacturing companies and contract manufacturing organizations (CMOs), a lot of important data can be found on paper records. The team members who need to use this data are normally trained in disciplines other than IT, and today they rely on others with IT skills to extract their data for them. They also depend on highly skilled statisticians to do basic analyses and prepare basic reports for them using general statistical packages that lack the functionality needed to correlate and generate actionable information from all data types. These realities have to be taken into account in the process of "knowledge acquisition" to which the FDA has referred above. In short, the most critical aspect of any process improvement initiative may be a single point of on-demand access directly by end users to all the relevant data (including that on paper) in a context that is meaningful to diverse groups of users and fully integrated with a collaborative, graphical data analysis and reporting environment for identifying and understanding cause-and-effect relationships in process data. Process improvement and QbD become practical realities only when the barriers to easy access to, and correlation of, all the process data together are removed, and the team can spend its time instead on productive science-based collaboration. This is the best way to undertake the Design Space Development, Design for Manufacturing and PAT efforts needed to achieve manufacturing process excellence using the principles of QbD.

About the Author Justin Neway, Ph.D. is executive vice president and chief science officer at Aegis Analytical Corporation, 1380 Forest Park Circle, Suite 200, Lafayette, CO 80026. Tel. 303-625-2102,,