The computation of ROI (return on investment) for a PAT of PAT/QbD project is sometimes not an obvious concept. If you have a “hot” selling product, no matter want you do, you will sell everything you produce, so percent of sales certainly won’t increase. Other concepts such as NPV (net present value) and the like have been applied. But, thinking about the manner in which a PAT project can be bankrolled without breaking the bank, I remembered one favorite project during the summers of my youth.

My family had a shack (really; no hot water or heat) at the New Jersey Shore, where I spent most of my summers as a lad.
Obviously, fishing and crabbing were popular pass times. But, even hungry fish won’t bite on a bare hook and saltwater fish are not partial to worms. Baitfish cost money and we wanted to save our allowances for important things like rides and games at the boardwalk. So, we worked out an investment concept where a small investment led to a large return. All it required was a slice or two of white bread and a cherry bomb, along with all our fishing gear already capitalized.

We would walk to the marina that was carved out of marshland and head for the grassy area. Throwing bits of bread on the surface brought up schools of small baitfish. The cherry bomb, tossed into the feeding frenzy, stunned many of the little guys. We scooped them up and walked to the small pier and proceeded to fish. We would catch some keepers and some small fish that were too much trouble to clean. We placed the keepers on ice and used the small fish as crab bait. At the end of a couple hours we had dinner for the evening: several fish—perch and flounder—and a couple dozen blue claw crabs. All for the cost of some bread and one explosive. That, my friends is ROI on steroids!

“Cute story,” you might say, “but, what the heck does that have to do with a PAT project?” I’m glad you asked. Since management is not in the habit of freely dispersing funds, regardless of how good an idea, you will need to show some fairly rapid returns on any project you get funded. To use a tired cliché, “Go for the low-hanging fruit.” In terms of the Pharma industry, that means shorten the time and effort for some test that is mandated and will be run, anyway.

My first suggestion is to simplify the art of checking incoming raw materials—both actives and inert excipients. As I have often stated, with a worldwide supply chain, merely reading and trusting a certificate of analysis (CoA) is no longer the best idea. As we say in international arms treaties, “Trust, but verify.” In simple terms, a company can no longer simply rely on a piece of paper written by someone, somewhere, at some company.

The first assumption is that the material was actually produced by the company on the label at their own site. Short of hiring someone to follow each lot from manufacture to delivery, the purchasing company relies on trust. Does the term caveat emptor sound familiar? It is, if you ever purchased a “genuine” Gucci bag for $25 from a street vendor in NYC.

The current USP, EP, JP methods can usually tell you what a material is, but seldom do these compendial methods give a clue to the goodness or processability of a material. That is, compendial methods rarely, if ever, give you a true picture of polymorphic form, particle size/range, type of moisture, i.e., part of crystal lattice or surface moisture, or degree of crystallinity. Then there is total system failure. Case in point: heparin. The USP tests totally failed, allowing oversufated chondroitin added at quantities up to 50% to be accepted as pure heparin. No one noticed until people began to die.

Clearly, the 1950s approach to incoming raw materials is outdated and, in some cases, outright dangerous. While the heparin case is not typical, what my 30 years of performing container-wise qualification with NIR has taught me is that all batches are not equal. The very first full-scale test of NIR for qualification I performed was in 1984, and showed two out of 220 bags of lactose failed USP tests. Of course, we would not have known that since the company SOPs called for a combined sample to be run. The two samples—one failing moisture level and the other pH—were buried with 218 good samples and tested as passing. While that may be minor, we discovered, in rather short order, a batch of controlled substance, marked 100 mesh, was seen to be micronized; another batch of API, labeled “anhydrous” was a monohydrate, and so on.

Even with tighter controls, most companies rely on a “square root of N plus one” sampling protocol. That means, referring back to the lactose example that would have meant “14.8 + 1” or 16 bags would have been sampled, leaving 204 bags accepted on faith. That was back when the supply chain was all in the U.S. How much less of a good idea is this sampling plan when the chain is likely 4,000-8,000 miles long? Of course, this same rationale works for assaying lots of 1-5 million where only 20 tablets are taken, but that is for a later column.

It is, of course, true that container-wise testing of incoming materials would be prohibitively time consuming and expensive, were we to still run compendial tests. This is where the ROI comes into the story. By moving to, say, NIR or Raman in lieu of compendial tests, the time saved will allow 100% testing of the incoming materials, not only for identity, but any number of physio-chemical parameters. If we assume that the company will either be using PAT of plans on moving to a PAT environment, these tests will be extremely valuable to adjust the design space of the processes.

If we want to add just a touch more to the pot the containers may be tagged with inexpensive bar code/RFID tags as received and tested. When the immediate results are seen, each pallet may be sent directly to the passed tests storage area or be held awaiting sterility tests. However, since the containers are clearly tagged with electronic signature, they need not be quarantined, saving all sorts of fencing and moving from site to site. When bio tests come back, the materials may either be passed or failed, electronically. Any failing containers may be returned to the vendor or destroyed.

As the materials are used and easily located, since RFID readers may be placed around the warehouse, their electronic records are marked used and automatic re-orders printed, much like supermarkets have been doing for over 20 years. Yes, that’s what that beeping at checkout means.

A second low-hanging fruit is blending. Since all generic companies in the U.S. at least are required to perform blend uniformity testing, they have two choices. Blend like the dickens for a longer time to assure blending, then take a bunch of samples, send them to the lab for testing, and wait for the go/no go result. Or, they can proceed to produce the tablets or capsules on a risk basis, assuming the blend is ok.

Most of the time, this will produce a reasonably good product, but they are wasting time and resources, not to mention adding heat and grinding to the powder blend. The waste in the first instance includes:

1. Tying up the blender for longer periods of time, making more equipment necessary to meet clients’ needs.
2. The sampling, bagging, and all the lab work and chemicals bought, used, and disposed of.
3. If the powder is left in the blender, it cannot be cleaned for the next batch; if it is dumped and sent to risk manufacturing, any later results are moot, since it won’t be re-blended, in any case.
4. The heating caused by constantly pounding hundreds of kilos over and over is not inconsequential. Over-blending may cause degradation of the API or polymorph changes of both the API and excipients. This is potentially very bad with poorly soluble drugs, where the amorphous form is usually preferred.
5. The last negative is the grinding to smaller particles with over-blending, potentially causing stratification and flow problems.

As with the RM ideas, the installation and application of blend uniformity monitors—NIR, Raman, or Light Induced Fluorescence—give several major ROI including:

1. Better utilization of equipment. Ten to twenty minutes of blending per batch, over the course of a year, might be comparable to another blender for less cost than actually buying another unit. If the company is the exception that actually waits for uniformity results before moving forward, these savings are far greater.
2. Less burden on lab personnel, combined with less chemicals, glassware, lab space used all add up to significant savings.
3. Shortened blending times or, occasionally, longer times, if needed could easily add to the quality and reproducibility of the product lots.

These are merely two places in a pharmaceutical process stream where adding PAT tools can have a fast and vast ROI. Remember, using PAT tools doesn’t mean you need to spend lots of money or time or actually commit to a PAT program. They are merely tools to improve quality and save money.

---

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.