By Kevin Wall, principal consultant and owner, Cincero Consulting
Pharmaceutical quality by design (QbD) and quality risk management (QRM) principles have become mainstays in pharmaceutical development. However, several myths are prevalent that prevent wider acceptance of the concepts by smaller firms. The lack of understanding of pertinent quality by design methods prevents smaller firms from benefiting from the majority of what QbD offers. The benefits include a meaningful and well-structured chemistry and manufacturing controls (CMC) section of the regulatory submission, a robust process that scales and transfers easily, and effective use of resources during the development process. This article discusses three myths that can be overcome, enabling more companies to capitalize on the benefits of QbD.
As with many myths, elements of truth support their formation and persistence. Most of the misunderstanding comes from the lofty expectations set by the biggest supporters. There is much discussion about real-time monitoring of process parameters and real-time-release (RTR) of product:
“Level 1 control can enable real-time release testing and provides an increased level of quality assurance compared to traditional end-product testing.”1
“establish real-time release mechanisms.”2
I admit these outcomes seem wonderful. However, they overshadow the real benefits of the QbD/QRM process. Unfortunately, many focused on the process analytical technology (PAT) and complex modeling while missing the humble and basic approach of using QRM to capitalize on prior knowledge.
“This understanding can be gained by application of, for example, formal experimental designs, process analytical technology (PAT), and/or prior knowledge. Appropriate use of quality risk management principles can be helpful in prioritizing the additional pharmaceutical development studies to collect such knowledge.”3 (emphasis added)
Myth #1: Quality By Design Is Too Complicated
The basic principles of QbD are straightforward. After defining the quality target product profile (QTPP), quality risk assessments (QRAs) are used to identify the critical quality attributes (CQAs) of the drug product. The next step is to define the critical variability, which has potential to impact the CQAs. These come in the form of critical material attributes (CMAs) of the excipients and drug substance as well as the critical process parameters (CPPs) of the manufacturing process.1, 3, 4
The reality is the CQAs are relatively easy to identify. I have executed dozens of drug substance and drug product assessments, and the CQAs do not vary much. Even the ranges of the CQAs are very predictable. For a drug substance, an assay of 98 to 102 percent is the most probable specification. It doesn’t matter if the drug can be effective and safe at 95 percent; regulatory agencies expect a certain output. It is not a matter of what is safe; it is a matter of what global regulatory authorities expect and will approve. In a global supply chain, the flexibility is only as good as the most restrictive regulatory approval.
Quality by design at its core relies on characterizing and understanding variability. QRAs, properly executed, document prior experience of the scientist and narrow the number of parameters that need in-depth investigation.5 Confirmatory experiments or data from similar products quickly screens some parameters from further consideration. Most of the research time and dollars can be used to characterize the variability most likely to matter. I have found many unit operations are robust enough that only one or two parameters drive quality day to day. Often overlooked is that QbD experts will say DOE is not always necessary.
“The steps taken to gain product understanding may include the following: Design and conduct experiments, using DoE when appropriate … develop a control strategy … for critical parameters, define acceptable ranges.”1 (emphasis added)
QbD has been in the industry for some time. Contract development and manufacturing organizations (CDMOs) are attuned to client needs. While all are not equal, they acknowledge the need for a QbD approach. Design of experiments (DOE) is a common approach. Most scientists I work with understand the need to identify critical variability. Differences arise in a common way of characterizing it and assessing the risk. Companies that miss out on utilizing QRAs and focusing development work on the product and process variability that matter most are missing the biggest benefit of QbD. Resources spent in these activities make processes more robust. They make process scale up smoother and reduce the deviations in commercial production.
Myth #2: Quality By Design Requires Process Analytical Technology (PAT)
PAT has been part of the QbD discussion for some time. During the first part of the century, much was made of the possibility of direct monitoring of CQAs during production online. Proposals ensued on statistical modeling of process data to predict final product quality. It all sounded great, complicated, and expensive. It was great idea for those making instruments and selling software for commercial manufacturing.
I believe the best use of PAT is during development. PAT can cut the cycle time it takes to get data from development process runs. The development data is richer and timely, enabling more experiments in the same time frame. Executing the DOE becomes faster, reducing the amount of time in the plant. Since buying plant time is a significant cost, firms can choose to spend less money for production time or get more data from the budget. Either way, executing the development experiments yields the process understanding needed to make wise conclusions from the QRAs. While I was at Johnson and Johnson, our goal was to make the process robust enough that real-time monitoring using PAT was not an added value. Spending the money in development made the process more robust no matter where in the supply chain we transferred the commercial process.
Myth #3: Quality By Design Means Real-Time Release (RTR)
Much has been made in articles about flexible approaches to product quality and the possibility of RTR.2 For me, RTR is a carrot that is not worth pursuing. For most firms, the cost savings of eliminating end-product testing is not worth the effort expended to get it. For smaller firms with smaller production volumes, the payback would not be there. RTR only saves costs if product testing is the rate-determining step to product being released. If the batch record review process is slower than product testing, then eliminating the testing adds no value. For RTR to be of benefit, all regulatory authorities where the product is approved would need to approve RTR. If not, then companies would have to designate which lots go to which country. The use of bright stock would be severely inhibited. The complication of maintaining separate inventory of product with end testing and product without it would likely result in all product having end-product testing.
With respect to real-time monitoring, most processes would likely not benefit with real-time acquisition of data in a commercial setting. ICH Q8 (R2) defines a control strategy as a “planned set of controls, derived from current product and process understanding that ensures process performance and product quality.”2
In most of the cases, a real-time monitoring control strategy is not needed to meet the stated goal in ICH Q8. I recall a situation a decade ago where real-time and online monitoring of the process seemed like a good idea. We had a hydrogenation reaction where over-hydrogenation would lead to an impurity that could not be removed. The endpoint was critical. Standard practice was to run the reaction dilute enough and slow enough that a sample could be taken for analysis of the endpoint. The cycle time was 90 minutes. The idea was that online near infrared (NIR) would give us the data immediately. The cost of implementing the NIR was in the hundreds of thousands. The solution came from using a bench NIR unit to get the same data at-line in the plant. We were already taking the sample for chromatography, but NIR could get the result in minutes. We cut the analysis cycle time to 5 minutes. We could increase the reaction rate to cut the batch time in half and double the concentration to get twice the product per batch for the cost of a bench NIR sitting unused in the lab.
The payback from QbD is making the product and process more robust. Robust processes are easier to scale up and transfer. They result in fewer deviations and failures, reducing the supply chain risk. QbD allows firms to focus precious development resources on variability that matters most. It screens out the trivial many from the critical few. QbD is the language of the CMC section of the regulatory submission.6 Regulatory authorities expect to see defined CQAs and a control strategy that ensures critical variability stays within defined parameters.7 Winning the QbD game is about the patient first and stockholder second. Keeping it simple is the easiest way to accomplish both.
About The Author:
Kevin Wall is principal consultant and owner of Cincero Consulting. He has worked with five virtual pharmaceutical companies to move eight of their drug substance and drug product projects through development. He applies “fit for phase GMP” from Phase 1 to commercialization. Wall’s process for identifying critical quality attributes (CQAs) and critical process parameters (CPPs) forms the basis for the CMC section of the regulatory submission (ICH Q11). He led the Johnson & Johnson development and global deployment of quality by design (ICH Q8) and quality risk management (ICH Q9) in pharmaceutical development and operations. He helped define and deploy Janssen’s risk-based qualification and process validation systems (ICH Q7). You can contact Wall at 817-915-0822 or firstname.lastname@example.org, or visit his YouTube channel, cGMP Made Easy