A recent warning letter issued to dietary supplement manufacturer Armada Nutrition LLC highlights the issue. Armada is apparently making an effort to comply with the particulars of the GMP regulations. It has analytical testing apparatuses in place (the exact type of machines were redacted in the letter, as it typical practice). But while Armada is a step ahead of those manufacturers who are simply relying of Certificates of Analysis to do their verification of incoming raw materials, the company has a ways to go to achieve full compliance on this score.
“You failed to establish and follow written procedures for laboratory operations, including written procedures for the tests and examinations that you conduct to determine whether specifications are met, as required by 21 CFR 111.303. Specifically, your firm failed to establish and maintain SOPs for the tests and examinations that you conduct using your (b)(4) and (b)(4), used for identity testing on incoming ingredients, in-process materials, and finished products,”the agency wrote in the letter.
The letter went on to state that, “You failed to calibrate instruments and controls that you use in testing a component or dietary supplement, as required under 21 CFR 111.27(b). Specifically, your (b)(4), used for identity testing on incoming ingredients, in-process materials, and finished products, have not been calibrated since beginning operations in July 2016.”
Instrument is only as good as the user
James Neal-Kababick, director of Flora Research Laboratories, said he runs up against this issue all the time in his training and consulting activities. Companies purchase analytical instruments without an appreciation of what goes in to using them in a fully GMP compliant manner. Even though equipment has become easier to use and more automated over the years, it still takes a trained eye to interpret the results and adequate training to understand how to do the tests in the right way in the first place.
While Neal-Kababick said he couldn’t be sure what instruments are being referred to in the Armanda letter, he said it’s a good bet that they are infrared instruments. He said in the past he has seen similar issues such as those cited in the Armada letter connected with the use of IR machines. IR testing equipment measures the wavelength absorption of materials to identify them. That’s a valid and potentially powerful tool, but not necessarily as easy to use as advertised, he said.
“When we started with this in the ‘90s we had an entry level IR device that would ‘drift’from time to time and would give inaccurate results. Now we have an expensive research-quality machine, and it doesn’t drift at all. We have a service that comes in once a year to calibrate our equipment and to test it. While that machine doesn’t drift, we still check in constantly in the meantime,” Neal-Kababick said.
Using an IR device for identity testing can be a complex task, Neal-Kababick said, especially if that is the sole method used. For single compound analysis, this is relatively straightforward: Measure the wavelength absorption (or put another way, what is reflected), and compare that to a standard. Then the thing is what it says it is or it isn’t. But with botanical ingredients the material is usually complex mixture. Now the issue becomes not just identifying the sound of a single cello playing, so to speak, but listening to a full orchestra and trying to determine how many violins and bassoons are in the ensemble.
Complex chemometrics
Neal-Kababick said in that case such an instrument can be used to parse out this information, but it requires a complicated chemometric statistical analysis that relies on the incorporation of validated standards. And it’s something that needs a certain amount of training and experience.
“The people doing this work at companies are often the low men on the totem pole, with the least amount of training. I have had people come to me for training when someone at the company quit and they were told ‘now you have to do it,’”he said.
And all of that relies on proper instrument calibration, Neal-Kababick said. In some cases, he said he has seen companies run an initial sample, get a positive ID, and then use that result as the ‘standard’ for subsequent tests. He also said that without experience, it can be all too easy to prepare a sample in such a way that it yields invalid results. The infrared radiation penetrates only a tiny distance into the sample, and if the sample does not have a high degree of homogeneity and highly uniform particle size, certain constituents of the material might not show up at all.
“The problem is that these are sold as simple solutions where you load the sample, push the button and get the answer,” he said.