On Extraction – the vagaries
In previous articles, some months ago on LinkedIn, I discussed the process of extraction. That is the process of preparing a sample for testing (perhaps biological) from a test article (medical device.) There I discussed the several variables which might affect the outcome, including: time, temperature, extraction ratio, and solvent/vehicle. In those articles, I deliberately held back from a further discussion about extraction for non-biological or analytical sample preparation. I knew that further information would come to light in the preparation of the new edition of ISO 10993-18 Chemical Characterization, and that the recent meeting of ISO/TC 194/WG 14 would be “illuminating.”
Let me report that I was not disappointed! And that the “fun” started before the WG 14 meeting. During the WG 15 Strategic Approaches to Biological Evaluation meeting (held first, at the beginning of the week’s programme), a number of presentations were given which cast uncertainty, if not doubt, on the validity of extraction.
Extraction, as a method of converting a test article to a testable sample, has a very long history and is generally accepted without question. Perhaps that is because: “we have always done it that way.
There are two principle things that I want to discuss in this article:
Repeatability and Reproducibility
Repeatability is when a lab can take identical test articles, process them the same way, conduct the same test, and get the same result. This might also be described as intra-laboratory consistency. There is some evidence that repeatability is achieved. However, this is not nearly as convincing as it might be. Perhaps a lab has done this for their own purposes, and not published it. Equally, a sponsor may have commissioned a lab to do such work and kept the result to themselves. I suppose that there are pretty good reasons not to go public with the results of such a comparison, whether they be good, bad or equivocal. Not the least of such reasons would be economic. It costs money to do tests, and none of us throws away our investment. (Though some of us may be prepared to trade.)
The key requirement of repeatability is identical test articles. This is not necessarily as easy as it sounds. Medical devices are incredibly varied, in terms of both their complexity, and their materials of construction, and their sterilization means (if any.) A “valid” test article is the final, finished medical device, in its saleable condition.
So, if the device is a relatively straightforward surgical steel implant, then all of them will have the same weight % elemental composition, within the specification of that alloy. (e.g. the alloy 1.4441 which is the low Sulfur version of 316L stainless Steel.) But, if the device is made of multiple plastics: moulded or overmoulded, adhesives which cure in place, and coatings, then “all bets are off.” The (molecular) composition of devices will be varied across a range, and perhaps along multiple axis.
This problem is “solved” making all biocompatibility tests: “type” tests. The type is the one which is representative of all those that follow it. Hence, when developing a device, our dear Product Development colleagues produces a lot of “models,” none of which are the “device.” Then they will product the “prototype,” the one before the type. (They may tell you they made many prototypes, but by definition, there is only one. The others are copies of the prototype. Let’s not be too pedantic at this point, as this is already sufficiently confusing.) The one (singular again) that follows the prototype is the type.
Because the type represents every device you make, then the test result for it also represents the reality for all devices. Note that this is: “represents,” not “equals.” Hence repeatability might never exist, depending on the sensitivity of your test method. The medical device industry depends on the validity of the type, and the type test results.
For biological test methods, type tests have always been used, and with good reasons (perhaps to be discussed in a future article.) But recent work by Dr. Bob Przygoda (now retired from J&J) has raised some interesting questions about the sensitivity of some biological tests done on extracts. Dr Bob implies that the test system sensitivity is not really good enough to account for variations in sample/extract preparation. So, the apparent repeatability of biological tests may be an artifact of inadequate sensitivity in the test system, rather than a stunning performance by the lab.
It’s only when we start to look at extracts by analytical chemistry means, that these matters come to light. Analytical instruments like the Gas Chromatograph (GC) and the Mass Spectrograph (MS) can be very powerful tools, with great sensitivity. One can, with an Atomic Absorption Optical Spectrograph, watch the alloy composition of a piece of steel change with successive heating, forming, cooling cycles. (It’s all within the alloy specification, so the change isn’t significant, but it is observable.)
Extracts compared by analytical means are likely to show less repeatability than might be expected. This is particularly so for organic materials (polymers & elastomers) and complex devices. The question that no one has yet adequately considered is whether this is actually significant. And the significance is probably established in a toxicological risk analysis, rather than the analytical lab.
Well, that leaves us with reproducibility. This is where the results from one lab can be duplicated in another lab. This is inter-laboratory consistency. Again, where the test system is fairly simple, and the variables adequately controlled, this can be achieved. Unfortunately, this is not much published either. Some “round robin” tests have been conducted by TC 194 working groups. But the results are not entirely compelling, for all manner of reasons (some better than others!) WG 5 Cytotoxicity is intending to conduct such a study, and it will be very interesting to the outcomes. (Bear in mind that it will apply to cytotoxicity testing, or a subset thereof, and may not apply to other biological endpoint testing.)
BUT inter-laboratory consistency of extraction, when assessed by analytical means, may not be readily established. I intend to discuss this in a future article.
What then is the industry to do in the face of the requirement for chemical information about devices being mandatory for biocompatibility evaluation? (See ISO 10993-1:2018 Table A.1)
Device manufacturers take note.
The following points (and gratuitous advice) are taken from my own experience in this matter. Take from this what you will.
Point 1: Team Play.
I have a degree in Materials Science, so I know a lot about what’s in materials. I don’t know it all, no one does, but you’ll need someone who has this sort of knowledge. (You’ll need them just to ask the right questions of others!)
I know a lot about product development, having spent almost 30 years doing it. Again, you’ll need someone who has this sort of knowledge.
Similarly, with manufacturing engineering. It might just be the lubricant they use to get part A into part B that is spoiling your day.
I’ve also done postgraduate study in toxicology. Because chemical characterization is going to feed into Toxicological Risk Assessment, you’re going to need this sort of knowledge too. Particularly to set some parameters.
Point 2: My Lab
I went looking for an analytical chemistry lab. I ended up with one specializing in environmental contamination. But they taught me what they could do and helped me turn it into something I could both use and trust. Then I kept using them. And over 10 or so years, we developed skill and capabilities together. I solved inter-lab reproducibility by not going to other labs! And now I’m a consultant, I send my clients to them.
In those days, CROs weren’t really offering much. I’m sure they have lifted their game. But the hardest part of using any lab is getting them to do what you want, rather than what they want. And CROs are prone to this problem. Do not, under any circumstances, throw stuff over the wall to them. Take them by the hand and collaborate. If they don’t want to collaborate or listen, move on. You have choices, and reputations have to be earned. But equally, don’t go to them uninformed, or without your team identified, if not together.
Point 3: How Much Extraction
The big point that the new version of ISO 10993-18 is going to make is: only get/do enough chemical characterization to suit your purposes. Don’t go too far if it isn’t required.
Manufacturers usually have a pretty good idea about the risks associated with their devices. After all, we’re all using ISO 14971. So, it should be easy to understand what you need to do.
But if it isn’t easy for you, then there are service providers to help. Consultants, like me, can stand back and have a dispassionate look at the situation. Regulatory affairs consultants can also advise you on the expectations of the authorities having jurisdiction. Money spent in this way is always more efficient than that spent in ignorance.
= = = = =
With Christmas, New Year and Summer holidays here in Australia, my next post won’t be until January. Do watch out for more articles on LinkedIn or come straight back here to our Brandwood Biomedical blog for more info and insight into biological and clinical evaluation.
Looking to fly south for warmer weather? We have the skills in all aspects of Australian regulatory and reimbursement requirements and can assist you to prepare and file regulatory submissions to TGA. We can also hold Australian licenses on behalf of international manufacturers. Talk to us today about how to leverage international reviews to achieve TGA registration. Contact us to discuss your needs and how we can help. You can drop us an Email [email protected] or call 1 888-271-5063 (US toll free) ♦ 400-842 7017 (Beijing – toll free) ♦ +61 2 9906 2984 (Sydney)