The world of physics has a foundation built on beautiful universal constants, things like π and ε0, which work their way steadfastly into virtually every aspect of modern life. Ironically, universal constants form the foundation of life for which it is popularly said that the only constant is change. Over the past 5 years, the medical device community has swung from nearly full ignorance of the potential power of chemistry testing to full adoration and acceptance, and now back—in a sense—to a state of scrutiny and skepticism. Regulators, in response to an influx of medical device submissions centered on supporting chemistry data, have increased their knowledge and finesse with the science and have been asking tough questions. In response to this feedback, as medical device chemistry for toxicology (ChemTox) has matured, the overall strategy has changed dramatically on some points.
In this environment, with a new ISO 10993-18, EU Medical Device Regulation (MDR), and the new ISO 21726, it is common for us to hear, “So, what are you guys doing for medical device chemistry testing?” This article provides a high-level answer to that question, which is, in short “everything we can.” The primary goal of ChemTox is to provide data useful for an unambiguous toxicological risk assessment. To meet that requirement, the study must be sufficiently broad in scope and sensitive enough to avoid missing potentially toxic compounds as well as to provide positive identifications. In addition to the foundational scientific requirements, the study must also meet regulatory expectations of completeness on points that labs might have a professional disagreement regarding scientific validity.
Sufficient Breadth in ChemTox
Sufficient breadth in study design is essential to ensure that important classes of compound that might migrate from a medical device aren’t missed in an extractables study. Therefore, extraction solvents covering a range of polarities are required. The static dielectric constant is a good measure of solvent polarity and is often used in pharmaceutical compounding to formulate liquids suitable to dissolve drugs. We know that the bulk polarity of human tissue ranges from that of water (dielectric constant, δ= 80) to slightly less polar than water (in fatty tissue like the brain δ= 43-58).From a clinical perspective, we would expect extraction solvents in this range to be sufficient. In practice, for all devices with prolonged or permanent contact, three solvents are requested by FDA: water (δ= 80), a mid-polar like isopropanol (IPA) (δ= 18.3), and a non-polar like hexane (δ= 2.0).
In addition to ensuring a broad range of compounds are soluble in the extraction matrix, breadth of analytical methods and instrumentation is also required. At a minimum, a study should include a method for volatile organic compounds (VOCs), semi-volatiles (SVOCs), non-volatiles (NVOCs), and inorganic/elemental compounds. A typical extraction and analysis matrix providing sufficient breadth (and meeting current regulatory expectations) is shown below in Table 1.
ICP/MS and ICP/OES
Direct Injection GC/MS
HRAM-UPLC/MS, APCI ±
HRAM-UPLC/MS, ESI ±
Above: Table 1: Typical Analytical Test Matrix
The Question of Extraction Duration – Is it Exhaustive?
The term “extractables and leachables” for medical devices is an adaptation or extension of the same term that has been applied to pharmaceutical container/closure systems. For patient-contacting medical devices, a true leachables study is impossible because the device leaches into the body—not a drug. Therefore, we seek to conduct a single-phase study that both uses aggressive extraction conditions for identification of all hazards as well as has an acceptable level of quantification. Using a definition commonly applied in ISO 10993 for other tests, regulators have required that extractables studies be conducted exhaustively for devices with prolonged or long-term contact. This involves serially extracting a device until the amount of material extracted is less than 10% of the initial extraction. In case non-volatile residue (NVR) is used to provide evidence of exhaustiveness as was required in the previous version of ISO 10993-18, the submitter must be prepared for questions regarding why the amounts detected by mass spectroscopy do not align. A better, more complete, picture is given by serially extracting at 24-hour intervals and analyzing each extract with the full suite of analytical methods.
Sufficient Sensitivity in ChemTox
Sufficient sensitivity in ChemTox is required to ensure that chemicals not reported or not identified are at such low concentrations that they would not be concerning from a toxicological perspective. Therefore, during the design phase of a study, toxicological information is needed to determine what that toxicological threshold is. ISO 21726:2019 “Biological evaluation of medical devices - Application of the threshold of toxicological concern (TTC) for assessing biocompatibility of medical device constituents” provides baseline thresholds that can be used to understand the sensitivity required for identification and reporting (also called the analytical evaluation threshold or AET).
After identifying the appropriate TTC for toxicological assessment, application of uncertainty factors in setting an AET can be tricky. Historically, labs have relied on two key papers that measured variance in response factors; more recently, regulators have been requesting lab-specific data to support an appropriate uncertainty factor in screening and that the AET be clearly justified.
Providing an Identification of Sufficient Quality
After ensuring the necessary breath in the analytical approach and applying a scientifically sound analytical evaluation threshold, the next step in ChemTox is to identify the detected compounds with a sufficient level of confidence. These identifications, together with the detected concentration of the compounds, define the starting point of the subsequent toxicological assessment. Correct identification of a compound is absolutely critical, as a misidentification might lead to incorrect conclusions on safety and biocompatibility of the tested device.
Historically it was common practice to solely rely on mass spectral matching. Using this approach, the mass spectrum of investigation is compared with mass spectra stored in mass spectral libraries. The higher the similarity between the two spectra is, the higher the resulting match factor. For those unfamiliar with MS, this may sound like a solid and reliable approach. However, when more experienced, it becomes obvious match spectral matching is only the first step to a confident identification and unreliable when used on its own.
One of the problems with spectral matching is the availability of libraries containing compounds that are expected to be present in and on medical devices. While for VOCs and SVOCs commercially available libraries such as NIST and Wiley exist, no such library is available for NVOCs. Authorities now expect significant effort on unknown compounds. These efforts may rely on a proprietary database and/or on the experience of mass spectrometrists that manually interpret the obtained spectra. Nelson Labs has invested for decades in building its own library that was gradually built by purchasing analytical standards that were measured and recorded. To date this library contains 1000 VOCs, 3500 SVOCs and 2000 NVOCs, avoiding reporting unidentified compounds as much as possible.
To help a toxicologist understand the reliability of reported compounds, ISO 10993-18:2020 asks to indicate the identification status in chemical characterization reports. Proposed levels for high to low confidence are confirmed, confident, tentative, speculative, and unidentified. The higher the confidence level, the lower the uncertainty to be used in a toxicological assessment. To gain confidence, additional evidence of the identification is needed. Examples are manual interpretation of the mass spectrum by a mass spectrometrist, detection of the compound by another analytical technique, or known use of the compound in the manufacturing of the medical device. Confirmed identifications are especially important in case the reported concentration is close to the permitted daily dose of the compound.
Medical device ChemTox has been a moving target as it has matured. While shifting expectations can be frustrating, studies conducted today provide much more thorough and protective data than just 2 to 3 years ago. It can be expected that things will continue to shift until regulatory bodies reach a consensus on their expectations of these studies and provide guidance on the same. Nelson Labs remains committed to frequent and transparent communication and guidance to both FDA and sponsors we help through this process.