A 2006 Georgetown University Report on Pharmaceutical Manufacturing stated, “If the FDA could change the way it regulated . . . the industry could save 10% to 50% of the cost of goods sold.” Well, the FDA has changed, most recently with its revised Process Validation guidance, which embodies modern concepts such as product lifecycle, Quality by Design, risk-based approaches and statistical process control.
Most impressive is how FDA has clearly defined the three major production lifecycle areas and integrated current risk- and science-based approaches. The three stages focus on:
1. designing the process itself
2. proving the process works
3. monitoring the process, to verify that it is well understood and that one can continually learn from it via “continued process verification.”
The major difference from previous guidance (circa 1987) is that the technical expectations for Stages 1, 2 and 3 are now more clearly delineated—there is far less ambiguity in what the Agency expects. There are other significant differences as well. Two of the most important are that worst-case testing during the qualification lots is not necessarily expected, and the term revalidation is not used.
Fit for Everyone
How much does the new guidance raise the bar? And just as importantly, does it raise it out of reach for small companies?
The expectation for Stage 1, Process Design, is to employ principles of Quality by Design and the Design Space—that is, to use lab- and pilot-scale testing to gather data, and set the permissible operating ranges for the process.
Every quality manufacturer wants to know why its process runs where it does. This requires that scientists document their work, following good documentation practices. It requires solid technical reports describing why the process operates where documents say it should, and also provides a way of keeping track of these reports. These are basic, fundamental expectations, and well within reach of even the smallest company that wants to manufacture drugs.
Stage 2, Process Qualification, is largely the same as practiced today, and perhaps less cumbersome than before since it formally removes the need for commercial-scale worst-case testing if sufficient process data are available. A key change is that while industry largely relies on three qualification lots to prove the process, the FDA is focusing on the data, not the number of lots. For a robust, well characterized process with a good clinical manufacturing history, this could mean fewer than three lots. Conversely, if the data is less compelling, more than three lots will be needed to prove its performance.
With Stage 3, Continued Process Verification, a firm should identify what to monitor, and why it’s being monitored—for example, it’s a high risk or critical process parameter. The firm then tracks and statistically analyzes the data in a formal program for monitoring the process. Other industries call this Statistical Process Control. It will require identifying what you need to monitor, gathering the data and analyzing it. SPC methods are well within reach of any manufacturing company.
Dealing with Risk
The concepts outlined in the revised guidance embrace a risk-based approach, which is not surprising, given what we have been hearing from the FDA in recent years (e.g. FDA Guidance for Industry - Q9 Quality Risk Management). Yet it’s been said that doing an in-depth risk-analysis might take longer than just doing all the work in the first place.
The beauty of risk analysis is that it forces one to employ structured thinking. Identify what could go wrong, how it can go wrong and what the impact is to product quality and patient safety. Once risk analysis is well understood, and done properly (for example, with process experts), it’s really not difficult. ICH Q9 lists several structured risk analysis methods, and the guidance spells out clearly what to work on and why, focusing attention on the more important issues. As the saying goes, “Anything is possible, but not everything is equally probable.” Risk analysis allows one to spend your resources on the likely problems.
True risk analysis was formally introduced to our industry just a couple of years ago, so this is still relatively new territory for pharma. A period of confusion and adjustment is to be expected, but the best preparation is to study simple risk assessment methods such as Preliminary Hazard Analysis and Fail Modes and Effects Analysis.
The guidance makes a point of clearly referencing a number of CFR’s which require statistics. For example, the revision states, “The data should be statistically trended and reviewed by trained personnel.” Such statements requiring a knowledge of statistical process control are consistent with employing better science. Will the increased emphasis on statistics be too much for some companies?
When the FDA makes statements such as, “Procedures should guard against overreaction,” it indicates that they have a solid understanding of the realities of SPC—that firms should avoid the temptation to overadjust processes. Most engineering and many life science degree programs have statistics as a core course, so there is already a good knowledge base in industry. Some firms may supplement that training with additional in-house statistics courses. While not every company can afford that, there are many continuing education courses as well as service providers available to bolster the use of statistics. The American Association for Quality (ASQ.org) is an excellent resource.
Hence, it really shouldn’t be a challenge.
Another matter that draws a lot of attention in the guidance is ongoing process monitoring. In the many comments submitted to the FDA during the official review period, a number of big companies questioned whether the amount of ongoing process monitoring needs to be the same as it was during process development, (which is what the guidance implies). Firms feel that they should be allowed to scale monitoring back once a process is well established.
Take a closer look at what the FDA is asking for. The revision states, “We recommend continued monitoring and/or sampling at the level established during the process qualification stage until sufficient data is available to generate significant variability estimates. Once the variability is known, sampling and/or monitoring should be adjusted to a statistically appropriate and representative level.” This is saying that once your process is well established, you can scale back the monitoring to the parameters that exhibit variability or are critical. You will need some production history to acquire the data to enable you to determine what’s well controlled and what’s really important at commercial scale.
FDA has pointed out that additional sampling and testing is expected during the process qualification lots. It in your best interest to do additional analytical testing to prove that your commercial-scale process in fact performs the way your development studies indicated. Additional process characterization testing makes a lot of sense at Stage 2, since you’re providing additional measurements to prove your process at scale.
That said, while it pays to continue to monitor process parameters, continued intensive analytical testing does seem unnecessary and actually seems to contradict the intent of Stage 2. Therefore, it’s reasonable to assume that FDA will move to clarify this and eliminate this ambiguity.
Ongoing monitoring is a very pragmatic and science-based approach to keeping your process in controls. A few methods that have proven effective for our clients:
- Gather substantial data initially—with today’s equipment, most of the data gathering is built-in, and essentially free.
- Apply a technique called “Short-Run SPC” which was pioneered by the aerospace industry for cases where very few, maybe only one, systems are built. Short-Run SPC normalizes and combines data. If you’re interested in how well chromatography flowrate is controlled, you could easily extract 15 discreet data points from a single batch by using Short-Run SPC techniques. You don’t need to wait until you have 15 batches to judge control.
- With some history, identify the key and critical parameters. Keep a close eye on these parameters by moving them into a formal monitoring program with both Specification Limits and Control Limits. This should be looked at during or immediately after every lot, so that you know that what’s important is in control.
- The rest of the data can go into a database that is used for troubleshooting, should an important parameter move out of control. Keep in mind that with SPC, an out of control result is designed to occur before it impacts product. It’s an alarm that something is not right. Go find out what, and fix it before it really becomes a problem and you have to rework or fail lots. This is good for the industry and good for the patient.
A number of companies also submitted comments raising concerns about the Agency’s expectations regarding legacy products. Should companies be required to revalidate their processes for products already on the market?
As stated before, much of what is expected in the revised guidance was already expected in the old guidance, only now the expectations are clearer. For example, a manufacturer was always expected to have data on operating ranges. And a manufacturer was always expected to monitor the process.
Of course, some manufacturers have done a better job at this than others. Therefore, it is wise for a manufacturer to perform a risk assessment of their process to determine what they know about it and how they monitor and control the manufacturing process. This risk assessment would point out gaps in what they have with what is expected.
Let’s say you’ve been making a biologic product for the last 20 years. And perhaps the scientists who developed the process in the 1980’s didn’t document their work as clearly as today. But, over the course of the last two decades, you’ve done many studies on the process, and you have 20 years of valuable production data. This would certainly lower your risk, and the combined body of knowledge you have may well exceed what would be available for a new product. Legacy products go into the same human beings as new products, and they unquestionably should be held to the same standards.
Will this new guidance make it any easier for companies to carry out continuous improvement work? The guidance begins by recognizing the process and product lifecycle. It clearly states: “Data gathered (during the Continued Process Verification stage) might suggest ways to improve and/or optimize the process by altering operating ranges and set-points, process controls, components, or in-process materials.” So the FDA is clearly stating that it expects manufacturers to study, learn and improve their processes. Ultimately, better control of the process, more efficient processes, more consistent processes, and better product quality will be good for the industry and good for the patient.
As proof of the FDA acting on product lifecycle and process improvement concepts, in April 2009, the Agency approved a higher titer process improvement which gave Biogen Idec a fourfold yield increase for its Tysabri manufacturing process. With its new guidance, FDA is betting that better science and regulation will mean better results for patients, and the industry.
Can It Change An Industry?
Initiatives start with the top echelon of the FDA and understandably take some time to trickle down to the inspection cadre and reviewers. Therefore, if manufacturers follow the guidance, can they be assured that the inspectorate is in tune with it? It will take time for the inspectorate to get fully trained on the guidance and the science behind it. Expect a relatively short transitional period as the FDA has been giving presentations and issuing guidance on these topics for several years.
On the plus side, the guidance is more specific than before, so this will minimize vague or subjective interpretations. The FDA also has a Formal Dispute Resolution appeals process which can be used as a last resort (Guidance for Industry—Formal Dispute Resolution).
Will the industry change as a direct result of the new guidance? Maybe it already has. Some companies are already practicing the 1, 2, 3 concepts outlined in the draft guidance. When I was at Amgen in the 1990’s, we explored parameter ranges and characterized processes before transferring them to manufacturing. We went into manufacturing equipped with reports and process databases, and SPC was then applied to follow the process performance once in manufacturing. Some firms are practicing principles such as PAT, risk analysis, SPC and Quality by Design and have a head start. The new guidance will aid them in organizing and standardizing their approaches. Recently, an industry consortium of Pfizer, Genentech, Abbott, GSK, Amgen, Lilly and MedImmune published “A-Mab, a Case Study in Bioprocess Development” which details approaches which directly address the expectations in the draft guidance. For some firms, the guidance will provide a clear roadmap of what is now expected. The guidance will establish the minimum expectations for good process science and good manufacturing science. It will result in more clear and more comprehensive validation and a better understanding of our manufacturing process. This is ultimately good for the industry and good for the patient.
About the Author
Peter K. Watler, Ph.D, is currently the Chief Technical Officer for Hyde Engineering + Consulting (Hyde), where he is responsible for the strategic direction of Hyde and its expansion into new technical markets. Dr. Watler has 22 years of process engineering and GMP manufacturing experience in the biotechnology industry. His areas of expertise include process validation, centrifugation, filtration, chromatography, process modeling and economic/COGs analysis. He has prepared 3 IND and BLA submissions and participated in four FDA inspections and numerous GMP audits.