The Forgotten Origins of Quality by Design
Transcription
The Forgotten Origins of Quality by Design
[ Analysis and Control of Variation. John McConnell, Coordinator The Forgotten Origins of Quality by Design John McConnell, Brian K. Nunnally, and Bernard McGarvey “Analysis and Control of Variation” is dedicated to revealing weaknesses in existing approaches to understanding, reducing, and controlling variation and to recommend alternatives that are not only based on sound science, but also which demonstrably work. Case studies will be used to illustrate both problems and successful methodologies. The objective of the column is to combine sound science with proven practical advice. Reader comments, questions, and suggestions will help us fulfill our objective for this column. Case studies illustrating the successful reduction or control of variation submitted by readers are most welcome. Please send your comments and suggestions to column coordinator John McConnell at john@ wysowl.com.au or journal coordinating editor Susan Haigney at [email protected]. KEY POINTS DISCUSSED The following key points are discussed: •Quality by design can be traced to the original work of Dr. Joseph M. Juran. •Juran believed that nearly all problems (e.g., deviations, quality defects, etc.) are built into the system, and that they can be traced to inadequate design of the process. Therefore, if addressed during the design phase, they can be largely designed out. •QbD starts with the quality target product profile. Juran calls this defining the customer needs. •QbD then proceeds with identification of critical quality attributes (CQAs). Juran calls this development of product features. These should be reviewed by both customer and producer. •Internal service groups must feel competition from outsourced services to continually improve their services, reduce variation, and positively impact product throughout the product lifecycle. •Once CQAs have been defined, the manufacturing process to meet these CQAs can be designed including technology and human components •Sampling and sample handling variability should be quantified and minimized. •Capability (CpK) is the measure of a process’ ability to meet its targets or specifications. •The final aspect of QbD is designing a control strategy, which Juran called developing process controls. •Understanding the impact of variation, and designing it out wherever possible is an important part of QbD. •The dominant variables of a process can be described as set-up, components, time, and worker elements. •Knowing which is likely to be dominant significantly impacts on design. •Designing a control strategy that ensures a robust process depends on understanding the sources of variation for each step (or assay) and controlling them. Understanding which category a variable helps to determine how best to attack variability. •Juran’s work, the basis for QbD, is now described in ICH Q8 (R2), Pharmaceutical Development. INTRODUCTION The US Food and Drug Administration and the International Conference on Harmonisation (ICH) have [ ABOUT THE AUTHORS For more Author information, go to gxpandjvt.com/bios 30 Journal John McConnell is owner and director at Wysowl Pty Ltd in Queensland, Australia. He may be reached at [email protected]. Brian K. Nunnally, Ph.D., is in charge of process validation at Pfizer in Sanford, NC. He may be reached at [email protected]. Bernard McGarvey, Ph.D., is process modelling group leader, Process Engineering Center at Eli Lilly and Company in Indianapolis, IN. He may be reached at [email protected]. of Validation T echnology [Summer 2010] iv thome.com John McConnell, Coordinator. embraced the concept of building quality into pharmaceutical manufacturing processes. This has been called quality by design (QbD). Many pharmaceutical scientists are not aware of the origins of QbD, nor are they familiar with its originator, Dr. Joseph Juran. The QbD concept was originally published in 1985. The seminal work is Juran on Quality by Design (1). This book outlines the reasons and methodology for planning quality into a manufacturing process. Juran did not use pharmaceuticals or medical devices in his book, but all of the principles are present and have been adapted into ICH Q8 (R2), Pharmaceutical Development (2). Figure 1: The Juran trilogy (adapted from 1). BASICS OF QBD For Juran, quality problems are planned that way (i.e., they are built into the system) (1). He believed that all quality defects were a result of a poorly planned quality system. The Juran trilogy is shown in Figure 1. We have shown this as an interlocking sequence, as we are sure Juran believed it to be. The corollary to this, and the most important aspect for the pharmaceutical and medical device industry, is that the quality problems can be prevented by understanding the process better for the determination of design space, identification of critical quality attributes, and defining an appropriate control strategy. Quality Target Product Profile QbD starts with an understanding of the quality target product profile (QTPP). This is defined as a prospective summary of quality characteristics of a drug product (or medical device, drug substance, etc.) that ideally will be achieved to ensure the desired quality, taking into account safety and efficacy of the drug product (or medical device, drug substance, etc.) (2). Juran would call this defining the customer needs. Juran states, “the goal should be customer satisfaction rather than mere conformance to stated needs” (1). Translated for the pharmaceutical industry (note: medical devices are included but will not be further explicitly discussed), this would mean a focus on producing medicines that improve the lives of the patients we serve. Our industry is not involved in the production of widgets; we produce life-saving and life-enhancing medicines designed to improve, save, and elongate the lives of our customers. In this light, the best way to serve our customers is to be a customer, as Juran would probably say. We need to think like a patient and not forget the patient in all of our work. The quality cannot be built into the drug until the parameters needed are defined (e.g., the route of administration, dosage, strength, etc.) gxpandjv t.com Critical Quality Attributes Once the QTTP is defined, work on defining the critical quality attributes (CQAs) can commence. Juran would call this process the development of product features. His advice rings true with today’s development. “Product development requires not only functional expertise; it also requires the use of a body of quality-related knowhow” (1). Development scientists need a thorough understanding of manufacturing and quality as their data and experiments form the basis of the processes and methods used during commercial manufacture. Juran would be supportive of the efforts made by manufacturing to have influence on the development process and by development to better understand the needs of commercial manufacturing and the product released to the marketplace. It is the job of both to ensure the planned approach includes work to guard against external failures. These include product and process design-related issues, carryover from previous processes, and degradation (2). The carryover of previous processes (think toolbox technologies) can contribute to future quality issues by embedding these into new processes. The CQAs are those product (or drug substance) characteristics having an impact on product quality and should be studied and controlled (2). Juran advises that this criticality needs to be reviewed by both customer and supplier (1). This means both the patient and the business are stakeholders. The patient benefits from a robust process that ensures those attributes central to the safety and efficacy of the drug are predictable. Journal of Validation T echnology [Summer 2010] 31 Analysis and Control of Variation. Figure 2: Pharmaceutical adapted Juran process model (adapted from 1). and controlled to ensure patient and business success. This is not an end, but should be considered to be a process encompassing the entire lifecycle history of the product and process. Critical Process Parameters Likewise, the business benefits through reduction of cost and increase in volume (see Little’s Law [3,4]), lowered frequency of deviations, improved utilization of human capital and fixed assets, and less rework and waste. The CQAs will be dependent on the various inputs to the process—Juran predicted this as well (1). Minimizing the variability in these inputs (including process parameters and raw materials, sampling, and analytical) will improve the ability to determine and maintain the CQAs. Internal Monopolies In his discussions of the development of product features, Juran has an interesting sidebar on internal monopolies. Internal monopolies were once the rule in the pharmaceutical industry. Services such as clinical trial management, active pharmaceutical ingredient (API) manufacture, and testing were all done in-house by the company’s employees. With the increase in outsourcing, there is really no such thing as an internal monopoly. All of the services mentioned previously are being performed by contract organizations. Juran predicted the beginnings of this when he advised that competitive information from outside suppliers of similar services could be acquired to ensure the highest quality, speed, and value are being obtained (1). This put pressure (used in its least pejorative sense) on the internal monopoly to demonstrate its competitiveness. This benchmarking is critical to the survival of the internal monopoly. A word of caution on outsourcing: While it seems to be all benefit with no “cost,” this is not always the case. Part of the agreement between the company and the service providers has to include tough variability reduction (per Deming’s approach to quality) of operating parameters, output characteristics, and assays. Variability should be identified 32 Journal of Validation T echnology [Summer 2010] Once CQAs have been defined, the manufacturing process to meet these CQAs can be selected or designed (2). This process includes both technology as well as the human components (1). The human factor is built into the process and must be considered in the design to error proof the human element. Too many deviations are seen that are blamed on human root causes when, in fact, these are process design elements that could have designed the problem out. We have been told about companies where up to 50% of the root causes are human error. For these situations, we predict these same 50% of the deviations will happen again due to inadequate corrective and preventative actions. Juran offered that process design should be goal oriented, systemic, capable, and legitimate (1). A pharmaceutical adapted Juran model of process development is shown in Figure 2. A process cannot meet a target not specified. The following are examples: •What are the quality goals to be met? •How much variation can be tolerated from the process? •Sampling? •Analytical methods? •What output is needed? •How can the process be scaled to meet demand (higher or lower)? All of these questions (and more) must be discussed to prepare a comprehensive set of goals for the process. The process is a series of interconnected systems; each one having an impact on the other. For instance, variability in analytical results used by the process as part of the manufacture (e.g., used by the process to determine how the next step will be run) have a larger impact on the overall variability than those that are not used to run the process. It is important to remember that the laboratory tests the sample, not the process. Sampling (and sample handling) variability cannot be ignored and should be quantified and minimized as it ultimately affects the overall variability of the process. Capability is crucial to meeting the process goals. Capability is the measure of a process’ ability to meet its targets or specifications. The key aspect of capability (CpK) is variability, as shown in the following equation: iv thome.com John McConnell, Coordinator. where USL is the upper specification limit LSL is the lower specification limit x is the mean σ is a standard deviation. There are only three ways to improve capability, as follows: •Move the specification away from the mean. This is a difficult task to accomplish in the pharmaceutical industry without significant data and lengthy regulatory agency discussions. •Move the mean away from the specification. For double-sided specifications, this only works if the mean is not centered. •Decrease the variability. Given the fact that this is multiplied by three, this has the greatest power to improve capability. In an extreme understatement, Juran suggested this evaluation had “merit” (2). More than this, it is a critical part of process understanding. It is important to remember that it is impossible to know or understand the capability of an unpredictable process. Finally, the process must be legitimate. Juran states, “...if there is the approval of those to whom responsibility has been delegated” (1). This is not a common concern in the pharmaceutical industry where clear accountability is often present. These are critical steps that are important to the determination of the CQAs. The planning associated with these steps should also include planning of the quality system. This plan includes alarm strategy, redundancy, and design to improve robustness and decrease variability. The final process including manufacturing, sampling, and analytical should be designed and demonstrated to be in control (e.g., predictable), appropriate for its intended use, and capable of meeting the defined CQAs (and thus the QTPP). Control Strategy The final aspect is designing a control strategy (2) that Juran called developing process controls (1). This should be done in concert with process development. The concept of design space comes into play at this point. So much has been written and presented on design space that we will not discuss this in great gxpandjv t.com detail in this article. Design space is the relationship of the process inputs to the CQAs (2). Juran did not utilize the terminology, but did describe the different types of variables present in a process. He described the concept of dominant variables, defined as those variables that are more important than others. The variables described include the following: •Set-up dominant variables. These are variables affecting processes that are dependent on the setup and confirmation of the setup to provide stability and reproducibility. Equipment and supplier differences can contribute to variability for these operations. Automated assays and packaging machines fall into the set-up dominant category. •Time dominant variables. These are variables affecting processes that change over time and need to be adjusted based on evaluation of control parameters. An example of this variable is ultra filtration operations. It is important to be vigilant against over-control (e.g., needlessly adjusting the process based on the last result obtained). •Component dominant variables. These are variables affecting processes in which the main variable is the quality of the inputs or materials. The variability in drug substance can be a large source of variability for a drug product process. Many chemical reactions fall in this category as well. In these instances reducing variability upstream is critical (Little’s law). •Worker dominant variables. These are variables affecting processes where the quality attributes are determined by the skill of the worker. Many types of assays fit into this category. Nonautomated processes can be included here as well. Ensuring operators or analysts performing the same functions or on different shifts have identical practices is important to reducing variability in this category. Designing a control strategy that ensures a robust process depends on understanding the sources of variation for each step (or assay) and controlling them. Understanding which category a variable helps to determine how best to attack variability. The closest comparison to the design space concept is what Juran referred to as the “Data Base” (1). He described this as a compendium of lessons learned from human experiences. He referred to the exercise of reviewing the information as “the Santayana Review” named after George Santayana who stated, “those who cannot remember the past are doomed to repeat it” Journal of Validation T echnology [Summer 2010] 33 Analysis and Control of Variation. 34 Journal (1). The purpose of this process is to improve decision-making by using the lessons of the past. For the pharmaceutical industry, the design space is reviewed and documented into process experience reports that are summarized in the development history section of the common technical document format of the dossier. Having this information codified in a single location is critical to resolve process upsets and to continue process improvements post marketing. REFERENCES FINAL THOUGHTS ARTICLE ACRONYM LISTING The concept initially developed by Joseph Juran is the origin of the modern QbD movement in the pharmaceutical industry. All of the relevant parts are there; long before they were published in the ICH. A review of these lessons forms the first principles understanding of QbD. CQA FDA ICH QbD QTPP of Validation T echnology [Summer 2010] 1. Juran, Joseph M., Juran on Quality by Design, The Free Press, 1992. 2. ICH, Q8 (R2), Pharmaceutical Development, 2009. 3. Nunnally, Brian K, and J.S. McConnell, Six Sigma in the Pharmaceutical Industry, Taylor and Francis, 2007. 4. McConnell, John, Brian K. Nunnally, and Bernard McGarvey, “Having It All,” Journal of Validation Technology, Vol. 16, No. 2, Spring 2010. JVT Critical Quality Attributes US Food and Drug Administration International Conference on Harmonisation Quality by Design Quality Target Product Profile iv thome.com