In 1989, I suggested that any blending process (powder or liquid) could be followed by simply inserting a (NIR) probe, scanning at regular (short) intervals, and, when there are tiny or no further changes, the mixing process is complete. I was challenged by a manager at NIRSystems to “put up or shut up,” so I did the first blending study. I used a simple four-component system in a small vessel, rotating and stopping at regular intervals. Each time the “blender stopped,” I inserted a NIR probe at several depths and locations, repeating the procedure until the math (Principal Components) showed there to be no further changes. It was blended as well as possible. Clearly, later, wireless NIR units are true process tools, but we had to do some feasibility studies first.
Then, a few years later, I got an opportunity to work on a process, in real time. In 1992, the plant engineer at Merck (West Point, PA) allowed me to chop a hole in the hopper of a production encapsulating machine and insert a NIR probe. I had been trying to monitor, with hopes of controlling, a solid dosage from production line. I was able to monitor the content uniformity of the granulation, enabling the process engineer to stop, should the content drop below a minimum or become enriched.
In 2002, Ajaz Hussain convened a series of meetings/hearings and mini-conferences to present the technical capabilities and methodologies to monitor and control processes in real time. This led to the issuance of the PAT Guidance in 2004. The initial draft and subsequent final version were entirely based on small molecule, solid dosage forms. The U.S. FDA designed the PAT Guidance for small molecules for a number of good reasons:
- There is nearly zero chemistry involved (wetting and drying is hardly “real” chemistry), merely physical mixing. Any complex, chemical reactions all take place at the API synthesis step, at a different location, nearly always before the “process” of dosage form manufacture is performed.
- The use of in-line/at-line instrumentation is largely an extension of current laboratory equipment. Spectroscopy and physical methods are already familiar, only the need for faster sampling and analysis times, so operators/analysts have, at some level, have some familiarity with the newer equipment.
- The personnel roster of a small-molecule manufacturer includes more traditional chemists (organic, R&D, and QC) and numerous types of engineers, so the statistics/algorithms of process chemistry (multi-variate analyses) are not too foreign.
- Since the chemistry of the API was established prior to formulation of the dosage form, nearly all the production problems with solid dosage forms were embedded in the formulation, blending, and formation of the dosage forms: all physical and material issues.
Another pragmatic reason was that the vast majority of marketed drugs (bulk sales = profits) were based on small molecules, while most biopharma companies were either new (start-ups) or still in the “R&D” stages, so there weren’t as many large molecule products in existence. But, as the big-money, super-drug supply line has dried a touch in recent years, there have been a few changes in the weather:
- The consequence of the “good old days” of blockbuster after blockbuster being introduced was the growth of large Pharma, largely through M&A (mergers and acquisitions). Larger facilities were great during growth periods, but, like licking one’s lips when chapped, it felt good, but, in the long run, was deleterious.
- The large overhead of maintaining huge “empires” (I believe, at the peak of growth, Pfizer had 82 sites), could not be maintained with “normal” sales of “merely” several hundred million in sales. Outsourcing was inevitable.
- With fewer small molecule NDEs (new drug entities), most of the big houses have been turning to macromolecules: i.e., biologics.
What does this mean in terms of relationships with CMOs and CROs and their clients (sponsors)? Several things come to mind:
- Since most big houses are in the early stages of producing potential blockbusters (indeed, many are still in the R&D stage, with no marketed products), there is a paucity of contract companies capable of cranking out large numbers of biological products.
- In fact, few big BioPharma companies have true PAT and QbD in play. That makes it ever so difficult to “transfer” a PAT program to a contract site and, for that matter, generics with less available resources will have trouble implementing one.
- The vast majority of CMOs are set up to generate solid dosage forms, so many of them will be starting at zero, with a steep and high learning curve. Let alone having bio-personnel on staff.
- Even parts of big Pharma need intensive training to adapt PAT/QbD to their bioprocess lines.
Now we can address the difficulties in beginning a (true) PAT/QbD program in both large and small companies. First, we need to assume that there is a staff on-site, familiar with the bioprocess (fermentation, clean-up, etc.), namely, biochemists, biologists, molecular biologists and chromatographers. That means that there is a paucity (as from above) of “traditional” personnel in the biopharma divisions/CMOs: analytical chemists, process engineers, material scientists, Chemometricians, etc.
This means the actual training of biotech personnel will have to be modified. For companies rich in analytical chemists (R&D, development, and QC), pharmacists, and process engineers), an introductory course with some examples is sufficient. For a bioprocess site a different approach is needed:
- A deeper explanation of why each step needs to be monitored and controlled needs to be generated.
- Since there is little baseline information about the economics of this “newish” industry, merely using the “saves money” argument (as with tablet/capsule production) rings hollow. Many of these companies either do not make a profit/have a finished product or have one that is raking in money, such that savings are the last thing on their minds.
- Since biopharma companies are heavily weighted to life sciences, simply showing previous applications of PAT instruments will have less than full success. If no one in the audience/class has ever seen, let alone used a NIR, Raman, LIBS (LASER-induced breakdown spectroscopy), or LIF (light-induced fluorescence), merely showing some examples means little or nothing.
- All the FDA, EMA, and ICH Guidances mean little to a company/staff that has never (or seldom) been under Agency scrutiny, so these need to be placed in context; merely reading them will not suffice.
- Concepts such as QRM (quality risk management) and DoE (design of experiment) need more than simple explanations. The best approach, which I have done in Europe, is to choose (have them choose) a process/product as an example. What I did was make a series of PowerPoint slides, featuring their process. Shown as a block diagram, each step shows the chemistry (OK, “biology”), components, and potential monitor/control equipment, as well as software/algorithms used; and don’t forget to show how to validate each step.
- Stress and explain the personnel changes and training that will need to happen to generate a working PAT/QbD program:
b. An enhanced IT department is a need. Granted, biological work requires a lot of statistics for clinical work and reaction overview, but multivariate analyses that will take place in a bioprocess PAT program will need people conversant with Chemometrics and process control algorithms.
c. Hardware and an instrument group will need to be added. Process instrumentation is beyond the comfort zone of lab chemists (especially biochemists) and utterly foreign to the bioprocess people. Just calibration, maintaining, and repairing the hardware will be a new frontier for the company… and the current personnel will not likely be equipped (or emotionally prepared) to deal with utterly different hardware and technologies.
d. AR&D and QC personnel, familiar with these newer technologies. They will need to find the technologies, specify the equipment, determine the location for them, when and how often to sample the process, how to model the data, and generate the equation to control a process.
e. These same people will need to work with the Chemometricians to maintain and update any multivariate equations being used to monitor and control the process. (And, don’t forget that need to explain the work to QA and the Agency.)
So, in short, bringing a successful PAT/QbD program to biopharma is a horse of a different feather. It will be new and different, but very, very worthwhile. This will be a true case for the need of massive cooperation between the innovators and the contract manufacturers.
Emil W. Ciurczak
Emil W. Ciurczak has worked in the pharmaceutical industry since 1970 for companies that include Ciba-Geigy, Sandoz, Berlex, Merck, and Purdue Pharma, where he specialized in performing method development on most types of analytical equipment. In 1983, he introduced NIR spectroscopy to pharmaceutical applications, and is generally credited as one of the first to use process analytical technologies (PAT) in drug manufacturing and development.