If this blog is of interest to you, we invite you to join us for a one hour live Q&A session focused on chemical characterisation and biocompatibility for medical devices.

Find out more Register for the event here

In January 2020 there was a significant change to the ISO-10993-18 guidance[1] (Chemical characterization of medical device materials within a risk management process). It expanded from 17 to 72 pages and has taken on some important concepts from the medicinal product testing guidances (USP 1663 and USP 1664 as well as PQRI …) such as “includes beneficial and timely information on analytical instruments, quantification methods, reporting requirements, new concepts (AET) etc”.  It also rejected some as well, stating “CDRH does not recognize PQRI recommendations (2006), or various USP chapters, e.g. <1663> and <1664>; some concepts may be scientifically valid, but these documents in whole are not applicable to devices”[2].  These similarities and differences between the testing requirements for medicinal products are worth discussing.  It should also be noted that in July of 2020 the FDA issued a statement[3] covering areas in the guidance that are not recognised.

The ISO 10993-18 document has been expanded to give more guidance and clarity on the expectations on what it means to chemically characterise a device.  The basic premise for testing is to determine if the device/product is sufficiently safe for use. By carrying out the chemical characterisation, a material can be rejected before any biological testing takes place.

In any chemical characterisation/extractable study the basic approach is the same — take the material in question; extract the material with an appropriate range of solvents, using a suitable extraction technique; and analyse the resulting solution with a suite of analytical techniques at a suitable level.

For medical devices, the important concept of the analytical evaluation threshold (AET) was introduced. This threshold (below) determines which chemical species are deemed to not have any safety concerns. The AET is an identification threshold and is an important factor in helping design extractable studies.  The AET is defined as the following:

AET (µg/mL) = DBT (µg/day)* (A/(B*C*D)) / UF* E

Dose-Based Threshold (DBT) = Threshold of Toxicological Concern (ICH M7) = Safety Concern Threshold

A = number of medical devices extracted (No.)

B = extract volume (mL)

C = number of medical devices that contact the body/day[1] (No./day)

D** = dilution factor (D>1), if concentrated (D<1). If not diluted (D=1)

UF = uncertainty factor of analytical methods (UF >=1)

E = number of sequential extractions

For large implantable devices, that require large solvent volumes and have a number of sequential extractions, the AET can easily be in low ng/mL or pg/mL without having the need for significant concentration factors (>100) to be within the ability to carry out screening studies.

The only fixed point in designing extractable/chemical characterisation studies is the finite detectability of the analytical techniques. All other variables such as sample to solvent ration, the amount of solution required, sample introduction approaches, are within the control of the person designing the study. Figure 1 gives an explanation of study design considerations.

Figure 1 – Analytical factors to consider when designing studies

The choice of extraction techniques is completely flexible. A recent series of blogs discussed the various advantages and disadvantages of these techniques[1]. However, ISO 10993-17 does prefer oven extraction at various temperatures. Solvents depend on the use of the device but for permeant implantable devices, extraction is expected by the FDA with 3 solvents: a polar solvent (water), a semi-polar solvent, and a non-polar solvent. Potential solvents are listed in ISO10993-18. The choice of solvents needs to be made to make sure that the material being extracted is not deformed or degraded. The amount of solvent to sample ratio will depend on a number of factors including the ability to see down to the levels required, the amount of solution required for the particular analytical technique. In addition, if extraction is carried out above the glass transition temperature of the material, it is likely more species will be extracted.

Extraction duration depends on the length of contact the device is with the patient. For short term contact, exaggerated conditions can be used e.g. those that are at minimum worst-case clinically relevant conditions or slightly harsher/longer than clinical use:

  • At a temperature that exceeds the clinical use temperature
  • With a duration that exceeds the duration of clinical use
  • With a vehicle whose extraction power exceeds that of the solution that mediates the clinical contact
  • At a surface area/volume ratio that exceeds clinical use exposure

For long term contact (>30days) exhaustive extraction is required. For these prolonged cases of between 1-30 days, exhaustive extraction or exaggerated extraction may be used. Exhaustive extraction is a multistep process where the final extraction level should be less than 10% of the initial extract level. Detecting at this 10% level can be especially challenging.

Once a solution has been prepared it requires analysis using a suite of analytical techniques with the range presented in figure 2. The updated guidance 10993-18 has increased the importance of screening, which is mentioned numerous times.

Figure 2 The analytical techniques required to identify and quantify extractables

Whilst the standard is a great improvement, there will still be a stabilising time while a number of areas are worked on and a better consensus of best practise can be achieved.  This will primarily focus on the uncertainty factor and what levels can and should be accepted.

Contact Us For a Consultation


[1] ISO-10993-18 guidance (Chemical characterization of medical device materials within a risk management process


[3] https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfStandards/detail.cfm?standard__identification_no=41050 (accessed December 2020)

[4] Hall extractable blog reference