Back in 1983, in the midst of a huge lab expansion encouraged by a FDA form 483, I encountered the VP who signed the capital requests I was writing and sending for signature (e.g., 20 new HPLC’s in eight months). I didn’t know he knew me, but when he stopped me, his comment was a simple, “I like the way you document your requests.” When I asked why, he responded, “All the other labs go on about ‘better resolution’ and ‘peak shapes,’ while you explain how many more samples per night the new instrument can analyze.”
Taking this to heart, I henceforth only described an instrument in terms of how much time and/or money it can save. Not upgrading a lab might cost far more than spending the money needed to move into the future.
Replacing analytical units one-for-one or updating?
In order to save on overhead, many companies choose to keep existing units alive with continuous repairs and maintenance. This has some drawbacks.
The turnover in the lab of samples for release and/or stability testing is lowered as older instruments need more time to actually run the samples and to clean and re-equilibrate. Older instruments also require more downtime for maintenance; parts become difficult to obtain with older models. In phamra lab terms, older could be just five years.
It’s also difficult to attract new business without being able to dazzle clients with the latest and greatest technology.
I never recommend buying a new instrument or technology just to have a new, shiny tool. The best approach is not to simply replace an instrument with a similar tool, but to use the opportunity for improvement. For example, replacing an HPLC with a UPLC which provides several advantages.
A typical UPLC unit increases throughput. A typical 20-minute chromatogram may be generated in, perhaps, 3-4 minutes. That amounts to between a five to seven-fold increase.
Along with the speed of analyses, there is a huge decrease in the amount of organic solvents needed—both being purchased and disposed of, legally. One method I worked on went from a flow or 2.5 mL/min for 25 minutes to 0.6 mL/min for 5 minutes. This saved (62.5 mL – 3.0 mL) = 59.5 mL of solvent per sample injection. Thus, a typical run of 40-50 samples a day saved 2.4 – 3.0 liters per day per instrument. This does not including equilibration and time between samples.
One option, exercised by the lab I worked for, was to sell the excess HPLCs in order to purchase a state-of-the-art near-infrared instrument to replace even more tests. This approach has a three-fold benefit. For one, funds are not taken from the general lab budget; they are generated by the savings in solvent costs and can be used for new equipment in lieu of diverting existing capital funding; and a plethora of bench space is now available, without having to build or expand facilities.
Now, all these data are based on simply continuing to perform the same previous tests—USP or client-supplied.
Try something completely new
Several difficulties arise when innovations to analytical methods are suggested.
Most contract labs are expected to use the exact analysis methodology as the client company. There is almost always pushback when changes are suggested by the contracted facility, and the first hurdle is to convince management that change is good.
In addition, it takes staff and time to research and implement any new technique. Cross-validation also takes time and client permission, who may not be familiar with the suggested technique.
Of course, the elephant in the room is the FDA, or EMA; any new technique needs to be validated and approved by the relevant agency. This adds paperwork and further lag time to implementation.
What do we mean by new and improved?
Unlike the U.S. Environmental Protection Agency (EPA), the FDA sometimes passes guidances before there are technologies mature enough to support them. As an example, the EPA sets levels of acceptable contaminants in drinking water, based on actual equipment available to test them. The FDA’s 2004 PAT Guidance assumed technologies would follow the demand for them. While the response has been good, the conversion to process analyses has been slow.
There was one existing wireless near-infrared (NIR) monitor, designed to follow blend uniformity in real time sponsored by Pfizer, built by Zeiss. An excellent idea that worked well, but, at the time it was presented to audiences there were possibly only three or four in existence. So, while a good idea, the odds of many or all pharma companies immediately converting from sample thieves to real-time blend monitoring were quite long.
One technology that has been used for rapid in-place analyses since the 1980s is NIR for raw materials qualification. Considering that the first validated use for this application was 30 years ago, it’s hard to explain why companies are still taking physical samples and running compendial tests—tests that show impurities, but are not designed to measure how well a material will perform in the process stream (e.g., heavy metals and residue on ignition tell us nothing about how well an excipient will tablet).
When we first adapted NIR in 1984, the compelling reason was the vast number of chemical and physical tests that would be required for 100% testing as per the EMEA recommendation, at that time. The 220 bags of lactose per shipment alone would have taken hundreds of analyst hours to perform USP tests on all bags including IR, sieve analysis, heavy metals, optical rotation, LOD, etc. And that would be repeated nearly every month. Add to that all the other incoming raw materials and we couldn’t hire enough people or build enough lab space to practically perform the testing.
By comparison, a batch of lactose could be sampled and brought to the lab—newer instruments are portable and do not require physical sampling—analyzed, and a report issued in less than one day. The analyses could also, using the same sample scan, show the average particle size, percentage of moisture, degree of crystallinity, and other physical parameters. So, one single replacement technology can more than show a ROI in one or two months. In addition, since so many raw material NIR methods have already been accepted by the FDA and EMA, the pathway is quite clear. USP chapter <1119> is well-written and gives almost all the information one needs to build a RMID, based on NIR.
Another case of science and hardware catching up with regulatory recommendations is heparin. After the disastrous adulteration problem of a few years ago where over-sulfated chondroitin was used to bulk up a lower-than-expected harvest of heparin, the USP and FDA held meetings and brought in experts to comment on the proper manner in which to analyze heparin, considering that compendial methods could not spot this adulteration.
The experts agreed on two methods: capillary electrophoresis and two-dimensional NMR (nuclear magnetic resonance). The former is not commonplace in many labs and takes a fair amount of training to develop and validate and run methods. The latter, at least at that time, was not common, hugely expensive, and required liquid nitrogen to slow the evaporation of the liquid helium needed to cool the magnets. These maintenance costs could run thousands of dollars per day.
This past November at EAS (Eastern Analytical Symposium), I discovered an instrument that excited me. It was a benchtop NMR (nuclear magnetic resonance) spectrometer by Magritek. Its heart was a room-temperature, permanent magnet capable of running 13C, 1H and 19F as well as performing 2-D methods such as HMQC (hetero-nuclear multiple-quantum correlation), HETCOR (hetero-nuclear multiple-quantum correlation), COSY (correlation spectroscopy) and 2-D JRES (J-resolved spectroscopy). This was what was recommended by the USP gathering, and the cost was well below $100K, far below the millions needed for the previous models suggested.
One really good thing would be for smaller companies to be able to produce and analyze heparin as a product. Prior to such an instrument, the best scenario was for a smaller company to produce heparin then pay for the analysis at a contract lab.
While we are talking about totally replacing a technique, there is one that we are all familiar with already: ion mobility spectroscopy (IMS). The earlier units are ubiquitous at airports as protection against drugs and explosives, where the sample is heated and vaporized, the gas passes through an ionizing grid, which imparts a charge on each one. The ions are repulsed by the grid and drift towards a detector, arriving according to mass and shape. In short, it’s a mini-mass spectrometer.
A newer unit by Excellims Corp. uses a softer vaporization process and preserves the molecules in whole form for easier mass identification. The unit was designed to be used for cleaning validation, allowing fast 20 ms analyses. It was suggested that it would make a wonderful detector for GC or LC. In fact, it can be used in lieu of chromatography. Of course, the drudgery of sample prep for dosage forms would still need to be performed.
That’s where another new instrument comes in to play. It’s a not so simple sample prep unit. A unit, shown by another EAS exhibitor Gerstel makes sample prep and injection instrumentation for GC and HPLC units. When I casually mentioned that the unit would be nice to attach to the aforementioned IMS, the gentleman at the booth informed me that they were already working with Excellims to do chromatography.
This is a case where many solid dosage forms could be analyzed in a fraction of their former time. The current sample preparation might involve shaking, grinding, sonification, filtering, and final volume adjustments and dilutions. The above-mentioned tool may be able to do sample prep in just a few minutes instead of hours and inject directly into an IMS, which can produce an answer in 20 milliseconds, not minutes. Plus, no solvents or columns are required.
In short, you really do need to spend some money to save some or lots of money. All you need is the will and some imagination, and go back to sending your lab people to meetings.
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.