Three questions come up on nearly every chemical characterization enquiry, so let us settle them up front.
Can chemical characterization replace animal testing? In part. It does not cover every biological endpoint, but a well-executed characterization package combined with a toxicological risk assessment can justify not running animal studies for a number of endpoints. The converse also holds: a thin characterization package saves nothing downstream, and it usually buys you an extra round of deficiency questions.
When is it mandatory? Whenever a device contacts the body and you are filing for registration or a 510(k), the biological evaluation has to be conducted inside a risk management framework, and chemical characterization is one of the information sources feeding that framework. A change of material, of supplier, or of sterilization method normally triggers a re-evaluation as well.
How long does it take? That depends on how complex the material system is and how the extraction scheme is designed; there is no single number. Rework caused by a poorly designed protocol is the most common reason these projects overrun, and it costs more time than the testing itself.
Where chemical characterization sits in the biological evaluation
Part 1 of the ISO 10993 series sets the overall framework: biological evaluation is carried out within the risk management process, existing information is assessed first, and testing is used only to fill the gaps. Chemical characterization is the piece of that existing information with the greatest potential to reduce test volume.
The underlying logic is simple. First establish what the device can release under the conditions of use and how much of it, then judge whether those constituents are toxicologically meaningful for the intended contact. The first half is the job of ISO 10993-18 (national equivalent GB/T 16886.18-2022); the second half belongs to ISO 10993-17 (national equivalent GB/T 16886.17-2025).
The two parts are used together, and neither is complete on its own. Chemical data without a toxicological assessment is a pile of detections with no conclusion; a toxicological assessment without reliable chemical data has no input to work from.
What ISO 10993-18 asks you to deliver
The title of ISO 10993-18:2020 is chemical characterization of medical device materials within a risk management process, and the phrase within a risk management process is not decoration. The standard does not prescribe a fixed test list. It prescribes a method for deciding how deep to go, based on risk.
In practice the deliverable falls into tiers, from shallow to deep:
| Tier | What it contains | When it is enough |
|---|---|---|
| Material information | Formulation, grade, supplier declarations, list of processing aids | Established materials, short contact duration, substantial history of use |
| Qualitative analysis | Identification of which extractable species are present | You need to establish whether known substances of concern are present |
| Quantitative analysis | Amount of each constituent, compared against the analytical evaluation threshold | Most situations that call for a toxicological risk assessment |
| Simulated-use leachables | Leachables measured under simulated clinical use conditions | Long-term contact, implants, or extractables results that flag a concern |
One concept drives the whole exercise: the analytical evaluation threshold. Constituents detected below it do not each need an individual toxicological assessment. Its purpose is to concentrate limited assessment effort on the constituents that actually matter. How the threshold is set depends on the contact profile of the device and on the assessment strategy, and it has to be fixed during protocol design rather than reverse-engineered once the data is in. Running the analysis first and choosing the threshold afterwards is an easy target for reviewers.
Threshold calculation methods and uncertainty factor values are governed by the text of the standard in its currently effective version. No figures are reproduced here, to avoid misleading readers across editions.
Extractables and leachables: two jobs, not one
This is where protocols most often go wrong.
Extractables studies use exaggerated conditions, with more aggressive solvents, higher temperature and longer duration, the aim being to drive out as much as the material can give up and produce a worst-case inventory of constituents.
Leachables studies simulate real conditions of use and answer a different question: how much is actually released in clinical use.
The relationship between them is sequential. Extractables first, to find out what is there. If everything sits within acceptable limits, a leachables study is often unnecessary. Only when the extractables results flag a concern, or the device falls into a long-term contact or implant category, do you go on to a simulated-use leachables study.
Extraction conditions have to line up with the sample preparation requirements of ISO 10993-12, including extraction vehicle, surface-area-to-volume ratio, and the temperature and duration of extraction. Get these wrong and the entire dataset may be rejected at submission, and redoing it means re-entering the queue.
What changed in ISO 10993-17
This is the part that has moved the most in recent years. The old edition, dated 2002 and mirrored by GB/T 16886.17-2005, was titled establishment of allowable limits for leachable substances, and the thinking was to calculate an allowable limit value.
The current ISO 10993-17:2023 is titled toxicological risk assessment of medical device constituents, and GB/T 16886.17-2025 follows with the same scope. Moving from calculating a limit to performing a risk assessment changes more than the title: the assessment has to weigh the toxicological data for each constituent, the actual contact profile of the device, and an estimate of exposure, instead of checking a result against a single figure.
The practical consequence for manufacturers is that this work needs someone with a toxicology background, and the assessment report is itself part of the submission dossier. It is not finished when the laboratory issues test data. Plenty of projects stall at this step, not because the data fails, but because nobody has translated the data into a risk conclusion.
The deficiency questions we see most often
Across projects, these recur:
- A test report submitted with no toxicological risk assessment conclusion. Complete data with no conclusion does not answer the question of whether the device is safe.
- Extraction conditions that do not match the real contact profile. A short-contact device extracted under conditions intended for long-term contact, for example, or the reverse.
- No disposition for unidentified constituents. Peaks that cannot be identified are normal. What matters is stating how they were handled, not passing over them.
- Legacy data reused after a material change. When a supplier changes grade or masterbatch, the applicability of the existing characterization data has to be re-argued.
- Analytical sensitivity insufficient to support the chosen threshold. Methodology and threshold have to match, otherwise a not-detected result carries no weight.
Samples and documentation
Protocol design needs more input than most manufacturers expect:
- A complete material list: every material on the body-contact path, with grade and supplier
- Processing information: mold release agents, adhesives, inks, coatings and lubricants, the processing aids most often left off the list
- Sterilization method and parameters: sterilization changes how a material releases chemicals, and irradiation and ethylene oxide act through different pathways
- Clinical use information: contact site, duration of a single contact, frequency of use, and whether the device is reused
- Packaging and shelf life: migration during storage can also generate leachables
Sample quantity depends on how many extraction arms the protocol contains, and a definitive list is issued once the protocol is agreed. Talk the protocol through before the design is frozen. Finding a problem while the material can still be changed costs far less than finding it afterwards.
Chemical characterization normally runs in parallel with cytotoxicity, sensitization and irritation testing; for overall scheduling see biocompatibility testing. If the product is also changing its sterilization process, factor in the impact on sterilization validation.
Working with SUNGO Lab
SUNGO Lab (Shanghai Shage Medical Technology Co., Ltd.) is accredited by CNAS, CMA and IAS (USA), with laboratories in Shanghai and Hefei. We take on testing associated with the biological evaluation of medical devices and can issue reports in the formats expected for domestic registration and for 510(k) submissions. Beyond biocompatibility we cover electrical safety, EMC, sterilization validation, packaging and shelf life. The full scope is listed under testing services, comparable projects are described under case studies, and further standard interpretations are collected under technical knowledge. Accreditation marks demonstrate technical competence within the accredited scope; they are not a commitment regarding market access.
If you are not sure how deep your chemical characterization has to go, how to set extraction conditions, or how many samples to reserve, send us the material list and a draft instructions-for-use and we will confirm a protocol first. Call +86 132 4819 8029 or request a quote.