Whenever the U.S. Environmental Protection Agency (EPA) sets limits on trace contaminants, it is based on currently available technology. For example, when plasma emission spectroscopy became commercially available, the EPA was able to set far lower limits for heavy metals in food and water. Unfortunately, the U.S. FDA did not have science behind all of its Congressional oversight. The Food Additives Amendment of 1958 to the Pure Food and Drug Act, known as the Delaney amendment was intended to proscribe any proven cancer-causing additive form food and drugs sold in the U.S.

Starting in 1959, the amendment was invoked for food colors, etc. as well as any chemical in drug products. It was later expanded to cover by-products and trace materials in a dosage form. Eventually, the law was interpreted to mean “zero” levels were allowed to be present, even though we can never prove the total absence of anything, merely the material was below our levels of detection.

When I first started at Sandoz, we had a product in clinical trials. I noticed that one of the starting materials for the API resembled benzidine (4, 4’-diaminobiphenyl), a known carcinogen. I asked for and received an Ames test for potential carcinogens or mutagens; it came back as a mutagen. When I asked how we tested for its presence, I was told gas chromatography. Fine, only the limit for our test—mirroring the USP standard for unknown substances—was 0.1% or 1000 ppm. Despite the fact that the Delaney Amendment says “zero”, I knew we could do better. After a few weeks, I developed an ion exchange method with a LOD of <1ppm.
Upon testing all the lots in clinic trials, we found between 10 and 50 ppm of the mutagen; the lots were pulled, of course. Since the synthesis and clean-up could not do better than 10ppm residual, the product was scrapped. What is the point of this story? Simply that we are still being asked to make measurements or control processes, often without the tools being available or so expensive as to not make a product worthwhile.

Another example occurred in the early 1980s when the EMEA, now the EMA, or the European Medicines Agency, suggested that all incoming raw materials should be tested for identity; every container of every lot was to be opened, sampled, and its ID verified. Since the EP or USP tests were chemical and, sometimes a UV and/or IR spectrum, the tests could be quite involved and time-consuming. Well, in Europe the bulk of raw materials were delivered by rail car while, in the U.S., by trucks. That meant that, for our parent company in Basel, Switzerland, 200 kilograms of aspirin would require, at most, four containers; for us, in the U.S., it could mean at least 20 fiber drums. A shipment of lactose could easily contain 200-250 bags, and so on.

Since we had just undergone a massive expansion of the QC department, my suggestion to add a second shift to perform the USP tests was met, to make it family-friendly, with a touch of hostility. I was told to find another way. Fortunately, there was a coming-together of several technical factors:

1. I was invited to a near-infrared (mini) conference at a nearby location. Technicon was the company in Tarrytown, NY.
2. I was struck by the similarity of the agricultural, textile, and food samples to materials used in Pharma—starch, lactose, etc.
3. Mini-computers were beginning to become available. We used an HP-1000 with an external memory.
4. The statistician/chemist/chemometrician at Technicon had just written a program to perform discriminant analyses using Mahalanobis distances.¹

These factors came together and allowed me to solve the problem of container-wise testing of all containers of all batches of all raw materials. We had FDA permission by 1984 and were using it full-time by 1985. It is now, 30 years later, that NIR is recognized, along with Raman, as a quick and easy way to conduct the IDs.^2-5

As I pointed out in another publication, da Vinci designed helicopters and computers, but couldn’t build them because neither internal combustion engines nor semiconductors had been invented yet. In a similar manner, several suggestions in recent years have seemed to predate our ability to implement them.

Take the Process Analysis Technology (PAT) Guidance in 2004; the FDA had the best interests of both industry and patients in mind when they posted it and I was on the validation sub-committee when they were consulting with industry workers on the form it should take. It outlined and encouraged bold testing and actions, including using Risk Analysis software, on-line measurements, and even real-time-release of products, bypassing QC final testing.

Unfortunately, several things hadn’t been developed as yet:

1. The yearlong series of meetings, showcasing the successes of PAT, mostly by Pfizer, was attended primarily by engineers, analysts and instrument companies. There was almost zero attendance by QA and compliance personnel. Thus, when the draft Guidance was published in 2002, the normal 6-8 month Q&A period took over 1.5 years, because the QA people were hung up on ideas such as best scientific judgment in lieu of strict adherence to cGMP.
2. While Pfizer worked with Carl Zeiss AG, Switzerland, to develop a wireless NIR unit for real-time blend analysis, there were minimum pieces of monitoring units available and certainly not for low prices. There were few published articles or applications and a large number of FDA inspectors had no idea of how to evaluate real-time data.
3. Since the massive amounts of data generated by on-line and in-line instruments required a working knowledge of statistical process control (SPC) and multivariate analysis (MVA) software, few FDA or EMA inspectors were trained to advise or judge PAT applications.

The answer to these problems was education of the FDA/EMA personnel. A MVA/SPC curriculum was developed with the cooperation of academics and industry experts and is being given to all FDA inspectors and reviewers. This paves the way for money-saving approaches to process control and elimination of OOS problems and investigations.

Another problem, largely caused by the dependence on 50-year old USP methods, was the heparin disaster of several years ago, where a number of people were sickened and died due to adulterated materials from China. There was a diminished supply of pigs due to illness that year, thereby lessening the amount of intestinal linings available. To make up the difference it is believed the suppliers used over-sulfated chondroitin (OSC). While OSC, when taken internally, is considered an aid to relieving deterioration of joints due to arthritis, taken intravenously, it is toxic.

The USP tests, based on 50-year old spot tests and analyses, cannot differentiate between heparin and OSC. A series of meetings and investigations were held and one suggestion by the FDA and USP is that either expanded H1 or two-dimensional (2-D) NMR (nuclear magnetic resonance) spectroscopy be used to clearly delineate the difference between the two materials.⁶ One difficulty with this technique was the availability of proper instrumentation at pharma companies in most countries. The second problem was the cost. Standard 2-D NMR instruments easily can cost into the high six figures to purchase and build specialized rooms for their use. In addition, the magnets need to be cooled with liquid helium and liquid nitrogen to slow the loss of the He; this cost is also quite large and continuous, lest the magnets warm and warp. It was also suggested that anion HPLC replace capillary electrophoresis (CE), but that was a lot less expensive to institute.

The suggestion, had it been implemented with existing equipment, would have kept all but the largest pharma companies from importing and selling heparin with the accompanying increase in price and lessening of availability. However, this past month, I came across a benchtop NMR instrument⁷ at the Eastern Analytical Symposium in Somerset, NJ. Costing well below $100,000, the instrument can generate NMR spectra of 13C, 1H and 19F containing samples and perform 2D methods such as HMQC, HETCOR, COSY and 2D JRES. The best part is that it was a fixed magnet, not needing external cooling. Now, almost any size company can assure the safety of its heparin at a modest cost, not needing to resort to third-party labs for analyses.

This is simply a case of technology catching up with regulations. As the PAT and QbD (Quality by Design) Guidances were released and modified, the impetus for instrument companies to design and provide proper instruments was amplified. No longer satisfied with off-the-shelf lab-based analyzers, Pharma companies have been asking for smaller, more robust, less expensive monitors for process control work and compatible with production conditions. Many instrument manufacturers have answered with small NIR and Raman, hand-held and in-place, although only a few have come down significantly in price so far. However, these hand-held instruments are making immediate qualification, both ID and physical qualities, of 100% of raw material containers. This is a boon to large and smaller companies, saving money and time, yet allowing smaller companies to begin PAT-related assays.

A number of other breakthroughs have streamlined all companies’ ability to comply with cGMPs. One remarkable instrument is a direct descendant of an instrument used by the TSA (Transportation Security Administration) at airports: the ion mobility spectrometer (IMS). Designed to detect explosives and illicit drugs on travelers, it vaporized the materials on cotton swabs used to swab hands and luggage and separates the components, much like a time-of-flight mass spectrometer, while the units designed for airports (Smiths Detection)⁸ use thermal desorption, breaking down the chemicals discovered on surfaces. This makes it problematic for cleaning validation. A more recent incarnation of the instrument⁹ uses evaporation of an injected solution to vaporize the samples. Thus, the initial portion of the procedure currently used for cleaning validation need not be varied. The current approach usually consists of a cotton swab, dipped in water or a solvent, then used to wipe a designated area of a process instrument, usually 10 cm2. The swab is then placed in a vial and eluted with solvent.

Normally, this solution is sent to a lab for GC or HPLC analysis. This part of the process may take days to a week to generate a go-no go answer, meaning the equipment being tested for cleanliness must remain out of commission and unusable for this time. In fact, a large proportion of equipment will be idle at any given time, adding to capital expenses for the facility.

With the IMS, samples may be injected and levels and identities of potential contaminants may be gotten in 20 milliseconds and the equipment either re-cleaned or put back into use immediately. The end result is as if the company had a larger inventory of process equipment.

The take-away from this is that, with newer equipment, products may be generated with less cost and effort, while being made more efficiently. This allows all sizes of pharma companies to compete on a level playing field. 

References

1. Anal. Chem. 57, p.1149 (1985)
2. Spectroscopy,  1 (1), 36 (1986).
3. Appl. Spectrosc. Rev.,  23 (1-2), (1987)
4. 24th Pharmaceutical Analysis Conference, Dells, WI, August (1984).
5. 7th International NIRA Symposium, Tarrytown, NY, July (1984).
6. “Heparin Stage Two Monograph Revisions”, Open Microphone Web Meeting, March 3, 2009
7. Magritek, San Diego, CA
8. Smiths Detection, Inc., Edgewood, MD
9. Excellims Corporation, Acton, MA

---

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.