I was impressed that there was at least one full session dedicated to the role of excipients in Pharma manufacturing at the recent IFPAC meeting held March 3-6, 2019 in North Bethesda, MD. For many decades, too many formulators viewed excipients as needed, but not glamorous ingredients for their solid dosage forms. Over the last few decades, a number of things happened allowing us to shed light on their chemical and physical properties.

First, with the introduction of Near-Infrared Spectroscopy to Pharma (c.1984), physical parameters, such as mean particle size, moisture, degree of crystallinity, and polymorphic forms, were easily discerned with a short, non-destructive scan. The speed of NIRS vs. “traditional” (compendial) methods brought them from “occasionally performed” on combined samples (from numerous parts of a single lot) into the “easily performed on each container” for every lot of incoming material.

Second, mega-products happened. When a typical lot of dosage forms was a few thousand to a few hundred thousand, production personnel experienced fewer problems with flowability or particle size distribution. As batch sizes increased to the millions, hoppers and tablet presses became larger and the granulation/mixture became used/dispensed more quickly. This meant the powders/granules not only had to flow more smoothly at a faster speed, but were exposed to higher vibrations, causing stratification, cavitation, and other physical problems.

Third, in-line measurements started being made. While real-time measurements were not new to the rest of the world—e.g., mid-range IR in paper mills to control water in paper rolls, NIR for controlling nicotine in cigarettes—they weren’t formally introduced until Pfizer underwrote the creation of a wireless NIR unit for blend uniformity testing and control. In the 1980s, NIRSystems hardened their lab-based unit (model 6500) in a NEMA enclosure for production applications. While the NIRSystems unit was not used for in-line solids measurements, the unit did have fiber-optic probes, capable of following chemical (polymerization, chemical syntheses) and biological reactions (fermentations).

Fourth, as both instruments and the available software (and computers) became faster and better, faster and better spectra were being taken. Blend uniformity only required one measurement every 10-15 seconds; moisture measurements in a fluid bed drier could take several seconds, since the average is good enough; and following a pan-coating process is also an average measurement, so several seconds per measurement is sufficient. But, real-time measurements of solid dosage forms were not happening, yet.

Fifth, as diode-array and acousto-optic tunable filter instruments were introduced, measurements in the reflectance more became faster and faster. It was possible, as early as 1992, to measure the moisture content in freeze-dried vials as they came down the line from the lyophilizer. Tablets and capsules could be analyzed, but only more slowly than generated, in real time.

Sixth, one breakthrough that enabled fast, fast measurements was the ability to make rapid optical measurements. A few years back Innopharma introduced its instrument for evaluating particles as they are produced or as fast as they fall past the optical window. Two years ago, Redstart was introduced. This device was designed to measure the growth of pellets as they are coated in a dynamic process (e.g., a Wurster coater) at a rate of at least one million particles per minute. Both devices show size and shape distributions in real time. Since so many granulations are milled and bulk raw materials sieved, these are good ways to follow the size of both granulations or raw materials size parameters.

Lastly, thanks to some excellent R&D by the Finnish research institute VTT, a methodology initially introduced as an imaging technique, allows for exceedingly rapid measurements. Of course, the “flaw” was that it was a line-measuring device and the head needed to move. There was a brief time lapse until some engineer remembered that product came off a line moving. It was, after the fact, a simple deduction to have the line move under the device, then add fiber optics and multiple heads (Indatech, Montpelier, France) and you can analyze moving materials in real time (for tablets, 100,000 units per minute per probe).

What’s the point?
What does all this have to do with raw materials and their properties? In short, with PAT and QbD becoming more common, there is a great need to precisely characterize all the chemical and physical properties of all the materials used in manufacturing.

Why now?
It has nothing to do with suddenly caring, just that now, more things matter to a production engineer. There are several very critical differences between batch manufacturing and QbD, and eventually continuous manufacturing (CM):

1. Batch processing begins with simply dumping (excuse me: “loading”) all the materials into the device used for blending. The most critical parts of the GMP approach are to:
a. Assure the correct weights/volumes are added; and
b. Closely follow the order of addition, as outlined in the MMF (master manufacturing formula).
2. In batch processing, the materials are physically moved from step one to step two, etc., by hand, so characteristics, such as flowability, are not critical. When we get to the tableting, the mass of ingredients often has been granulated—API spread across the agglomerated mass—and sized to a pre-determined diameter. This lubricated, flowing powder is usually nearly identical and usually flows properly into the tablet press or encapsulating machine.
3. When there are problems with tablets in a batch process, in theory, you are not allowed to make changes in the MMF. I was part of a study, several years ago, where an independent consulting firm was following the source of failed batches at a company. It was seen that the greatest number of failed lots occurred when the newest operators were on duty. Despite an initial thought that these people needed training, it was uncovered that more experienced personnel made “needed” changes in the printed method, while newer people followed the “menu” as written. Under GMP, you cannot make the changes encouraged under QbD.
4. If we follow a QbD program, the parameters are allowed to change, but only within what we call Design Space (DS). DS, determined experimentally, consists of a series of allowable changes to operating and chemical parameters, changed to assure that variations in raw materials and instrument variability are noted and the process changed to compensate.

Moving forward
If we are to move to QbD (powered by PAT) and, eventually CM, we need to massively change the way in which we evaluate incoming raw materials—excipients and APIs. Currently, we are tied to compendial methods for “qualifying” incoming raw materials. Why is this a problem?

1. Compendial methods were mostly written to ensure identity and purity of raw materials. There was never an effort to place tests of processability into, for example, the USP.
a. One test for the ID of lactose is ‘boil in a copper salt solution; if the color turns red, that is proof of a reducing sugar.’ There are also refractive index and heavy metal tests, but nothing to describe useful physical parameters.
b. The particle size is determined by sieve analysis. Years ago, we had a problem with content uniformity of caffeine. While it was sold as 80, 100, or 120 mesh, it was eventually shown to all be micronized, only holding together through static cling. Just imagine what this would do to a continuous manufacturing process.
2. In short, all compendial methods were written assuming that raw materials were either generated by the company, itself (API synthesis) or purchased in the same country, under the auspices of the EMA, FDA, etc. All I will say on this point is: multi-country supply chain.

So, what can be done?
Basically, we need to change to meaningful tests for:
1. Particle size: reliable methods such as LASER light scattering or newer rapid optical measurements.
2. Flowability: now, with CM, a critical parameter. One very nice technique is practiced at Rutgers School of Engineering. They designed a cylinder (diameter ~1/3 m; length ~1 meter) with transparent end windows. This is filled to about one quarter volume and rotated along its long axis (horizontally). With a series of detectors at one end and LEDs at the other, the slow rotation causes the powder to display its ability to cascade, a measure of flowability.
3. Crystallinity and polymorphic state: not considered necessary under most GMP applications, but critical to QbD. Simple Near-Infrared or Raman scans can inform the production staff of how to modify the feeders for CM or the speed of a blender, etc.

In short, excipients, or “innocent bystanders,” are neither. Since more and more companies are moving to using PAT to control a QbD, there is and will be a need for more updating of information gleaned from raw materials. Since most raw material suppliers are not likely to change their manufacturing procedures, one answer could be to characterize incoming materials and mark them as to which products and processes they might be assigned. That is a “patch,” and ultimately cooperation with raw material suppliers is needed. But, that’s a topic for another day.

---

Emil W. Ciurczak
DoraMaxx Consulting

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.