Chemical characterization is the analytical and documentary process of identifying what a medical device is made of and what it can release into the body. Under EN ISO 10993-1:2025 and ISO 10993-18, it is the starting point of every biological evaluation, not an optional add-on. For MDR technical documentation, it feeds directly into Annex I Section 10 on material safety and Annex II Section 6 on pre-clinical and clinical evidence. When material composition is well characterised and the release of substances under intended conditions of use is understood, many biological endpoints can be justified without additional animal or clinical testing. When it is not, the biological evaluation has no foundation and a Notified Body will say so.

By Tibor Zechmeister and Felix Lenhard. Last updated 10 April 2026.

TL;DR

  • Chemical characterization under ISO 10993-18 is the analytical backbone of biological evaluation under EN ISO 10993-1:2025 — it identifies materials, processing residues, and substances that can migrate out of the device.
  • MDR Annex I Section 10 requires that devices be designed and manufactured so that material properties, including toxicity and, where relevant, flammability, are fit for the intended purpose. Chemical characterization is how manufacturers show they know what those materials actually are.
  • The chemical characterization file feeds MDR Annex II Section 6, which requires pre-clinical and clinical evidence to demonstrate conformity with the general safety and performance requirements.
  • A full extractables and leachables (E/L) study is not required for every device. It is required when material composition data, supplier information, and prior evidence cannot close the gap to a safety conclusion under the device's contact type and duration.
  • Starting with material composition and risk-based gap analysis — rather than jumping straight to E/L testing — is the Subtract to Ship move. It prevents six-figure testing costs on substances that were already adequately characterised.

Why chemical characterization matters before anything else

A startup we worked with had commissioned a full extractables and leachables study from a specialised contract lab before anyone had written a biological evaluation plan. The quote was in the high five figures. The work had already started when we were brought in. The first question we asked was simple: what are the materials actually made of, and what does the supplier already know about them? The second question was: what is the contact type and duration per EN ISO 10993-1:2025? Neither question had a written answer.

The testing went ahead because it had been paid for. But the study generated a list of extracted substances that nobody knew how to interpret, because the baseline — what the device was made of in the first place — had never been documented. The company eventually rebuilt the biological evaluation from the correct starting point, which was chemical characterization of the materials as supplied, not a black-box extractables run on finished devices. The first attempt was not wasted entirely. It was expensive homework on the real order of operations.

That order matters. EN ISO 10993-1:2025 is explicit that biological evaluation is a risk management process and that material characterization is the first analytical step. ISO 10993-18 then specifies how that characterization is done — from compositional information through to, when needed, quantitative chemical analysis. Skipping the first step and going straight to the most expensive test in the series is a pattern we see repeatedly. It is also the most preventable form of regulatory waste in the entire biocompatibility domain.

What chemical characterization actually is

Chemical characterization is not a single test. It is a tiered process defined in ISO 10993-18 that begins with compositional information — what the device is made of, at the raw material level, including polymers, additives, processing aids, colourants, and any residues from manufacturing. It proceeds, where necessary, through identification of constituents that could be released under the conditions of use, and, where the risk assessment requires it, through quantitative determination of those constituents.

The hierarchy is intentional. The lowest tier is compositional data from suppliers, certificates of analysis, and the manufacturer's own process knowledge. The middle tier is qualitative analysis — what substances come out of the device under conditions that simulate or exaggerate clinical use. The top tier is quantitative extractables and leachables analysis, where the amounts of specific substances are measured and compared against toxicological thresholds.

Not every device needs every tier. The decision of how far to go is driven by the risk assessment required under EN ISO 10993-1:2025 and by the contact category of the device — surface-contacting, externally communicating, or implant, combined with contact duration (limited, prolonged, or long-term). A short-duration surface-contact device made of a well-characterised, pharma-grade polymer may need very little beyond compositional data. A long-term implant made of a novel polymer blend will need the full chemical characterization treatment and more.

How ISO 10993-18 frames the work

ISO 10993-18 is the part of the ISO 10993 series that specifies the framework for identification and quantification of constituents of medical device materials. It sits under EN ISO 10993-1:2025, which is the overall standard for biological evaluation within a risk management process. EN ISO 10993-1:2025 tells you that biological evaluation must address the risks, identifies the endpoints (cytotoxicity, sensitization, irritation, systemic toxicity, and so on), and requires a documented rationale for which endpoints need experimental evaluation and which can be addressed by existing data. ISO 10993-18 tells you how the chemical side of that rationale is built.

The standard asks for three things in sequence. First, establish the material composition — what is in the device, down to substances present in known quantities in raw materials and substances potentially introduced during processing. Second, determine whether there are substances of toxicological concern whose presence and release need to be understood. Third, where experimental work is required, conduct it under conditions representative of clinical use, with documented methodology, analytical limits, and interpretation.

Each step produces records that belong in the technical documentation. Compositional information links to raw material specifications and supplier declarations. The risk-based decision on whether to proceed to experimental characterization links to the biological evaluation plan. The experimental work, when performed, links to test reports and the toxicological risk assessment. The whole chain must be coherent — a Notified Body assessor should be able to trace any conclusion about biological safety back to either published data, supplier data, or a specific analytical result.

Extractables and leachables — concept and distinction

Extractables and leachables are two related but distinct concepts that startups often confuse, and the confusion is expensive.

Extractables are substances that can be pulled out of a device material using exaggerated extraction conditions — typically solvents and temperatures more aggressive than clinical use — to establish a worst-case profile of what could theoretically migrate. Extractables testing answers the question: what is chemically present in the material that could come out under any plausible condition?

Leachables are the substances that actually migrate out of the device under the real, intended conditions of clinical use. They are a subset of extractables. Leachables testing answers a narrower, more clinically relevant question: what does the patient actually encounter when the device is used as intended?

The distinction matters for design of studies and for toxicological interpretation. Extractables data provide a worst-case bound that is useful when the clinical exposure is hard to model directly. Leachables data provide a realistic exposure estimate that, for long-term and implant devices, is often what the toxicological risk assessment ultimately relies on. Some devices need both. Some need only one. Some need neither, because compositional data combined with prior evidence are sufficient.

When a full extractables and leachables study is needed

A full E/L study is needed when the existing evidence — compositional data, supplier information, published literature on the same materials, prior biological testing — cannot close the gap between what is known about the materials and what the biological evaluation needs to conclude. That is the test. It is not "do it because the consultant said so" and it is not "skip it because it is expensive."

In practice, the triggers look like this. The contact category is implant or externally communicating with long-term contact. The material is novel, or is a blend, or has been processed in a way that could introduce substances not present in the raw material. The supplier data are incomplete and cannot be supplemented by additional documentation. The published literature does not cover the specific formulation. The toxicological endpoints that would otherwise need animal testing can be addressed more efficiently by chemical characterization combined with a toxicological risk assessment — this is a legitimate and increasingly accepted pathway.

When those triggers are absent, the right move is usually to build the biological evaluation on compositional data and existing evidence, and to document the rationale for not performing E/L studies. That rationale is itself a piece of regulatory evidence. Notified Bodies accept reasoned exclusions when the reasoning is traceable to the risk assessment required by EN ISO 10993-1:2025. They reject silent exclusions.

How chemical characterization feeds the biological evaluation

The biological evaluation plan is the document that decides which of the biological endpoints listed in EN ISO 10993-1:2025 need experimental evaluation and which can be addressed by existing data or by chemical characterization. Chemical characterization is often the single input that reduces the experimental burden most effectively, because it can replace entire categories of animal testing when the analytical data and the toxicological interpretation are strong enough.

The logic is that many of the biological endpoints — systemic toxicity, subchronic toxicity, genotoxicity, carcinogenicity — are ultimately about exposure to specific substances. If you know exactly what substances the device can release and in what quantities, and if those quantities fall below toxicological thresholds of concern, the biological endpoint can be addressed without an animal study. This is not a loophole. It is the explicit direction of EN ISO 10993-1:2025, which encourages replacement of animal testing wherever scientifically justified.

The practical consequence is that a well-executed chemical characterization file — even a relatively modest one — can change the entire shape of the biological evaluation. Studies that would otherwise be required drop out. The evidence becomes cheaper, faster, and arguably more scientifically robust. The chemical characterization file is where this leverage lives.

Documentation structure

The chemical characterization documentation in the technical file belongs under the biological evaluation section, which in turn sits under MDR Annex II Section 6 on pre-clinical and clinical evidence. Structure it so an assessor can follow the chain in one reading.

At the top, the biological evaluation plan states the contact category, the contact duration, the biological endpoints to be addressed, and the decisions made for each endpoint. For endpoints that rely on chemical characterization, the plan references the chemical characterization report directly.

The chemical characterization report itself contains the compositional information, the risk assessment that determined how far characterization needed to go, the experimental methodology if experimental work was performed, the analytical results, the identification and quantification of substances of toxicological concern, and the toxicological risk assessment that interprets the results against established thresholds.

Supporting records live in separate sub-sections: supplier declarations and material specifications, certificates of analysis for each production lot, process validation records where processing can introduce substances, and any published literature relied upon. Every claim in the top-level report should be traceable to one of these supporting records.

Finally, the biological evaluation report closes the loop — it references the chemical characterization report, explains which biological endpoints were addressed by it, which required additional experimental work, and why the combined evidence demonstrates conformity with MDR Annex I Section 10 on material safety.

Common gaps we see

Incomplete supplier data. A startup builds its biological evaluation on the assumption that the raw material supplier has full compositional data, then discovers at the Notified Body review stage that the supplier's data are limited to regulatory declarations and do not cover specific substances the assessor wants to know about. Fix this early by requesting full compositional information in writing from every material supplier before the biological evaluation plan is finalised.

No documented rationale for not performing E/L. The file contains compositional data and jumps straight to biological evaluation conclusions, with no explicit argument for why E/L testing was not performed. The rationale may be good, but if it is not written down, the assessor cannot accept it.

Processing residues ignored. The biological evaluation considers the raw material but not the substances introduced during manufacturing — mould release agents, cleaning residues, sterilant residues, adhesives. These belong in the chemical characterization and in the process validation, and the two must be cross-referenced.

Mismatched contact categories. The chemical characterization is performed under one contact category assumption, and the biological evaluation plan uses a different one. Usually this happens because the intended purpose was refined after characterization started. Re-open the chemical characterization when the intended purpose changes.

Test reports without interpretation. An E/L study is performed, produces a long list of identified substances, and sits in the file unaccompanied by a toxicological risk assessment that interprets the results against thresholds of concern. Raw data are not evidence. Data plus interpretation are.

The Subtract to Ship angle

The biggest subtraction opportunity in biocompatibility is the studies you do not need to run because the chemical characterization closed the gap first. That is the order-of-operations discipline — start with composition, run the risk assessment, and only then decide which experimental work is actually required. Startups that reverse this order spend money on tests that were never needed and still end up redoing the characterization at the end.

The second subtraction is the reasoned exclusion. For every biological endpoint in EN ISO 10993-1:2025 that does not require experimental evaluation, write down why — and make the why traceable to the chemical characterization, the published literature, or the prior evidence. A documented exclusion that a Notified Body accepts is worth the same as an expensive study that produces the same conclusion, and costs orders of magnitude less.

The third subtraction is the parallel-chain cut. If compositional data plus supplier declarations plus published literature already close the gap, do not commission a confirmatory E/L study "to be safe." Two weak chains of evidence are not stronger than one strong chain; they are more expensive and more confusing to an assessor.

Each of these subtractions traces to a specific provision. MDR Annex I Section 10 is the material safety requirement. MDR Annex II Section 6 is the documentation requirement. EN ISO 10993-1:2025 is the methodology. ISO 10993-18 is the analytical framework. The framework for applying all of this under startup constraints is the Subtract to Ship framework for MDR compliance.

Reality Check — Where do you stand?

  1. Do you have written compositional information for every material in your device, and did it come from the supplier or from a guess?
  2. Has the contact category and duration of your device been documented against EN ISO 10993-1:2025 before any biological testing was commissioned?
  3. Is there a written biological evaluation plan that lists each endpoint and states how it will be addressed — experimental, chemical characterization, literature, or exclusion?
  4. For every endpoint you are excluding from experimental evaluation, is the rationale written down and traceable to a specific piece of evidence?
  5. Have processing residues — mould release agents, cleaning agents, sterilant residues, adhesives — been considered explicitly in the chemical characterization?
  6. If an extractables or leachables study has been performed, is it accompanied by a toxicological risk assessment that interprets the results against thresholds of concern?
  7. Can a Notified Body assessor trace every biological safety conclusion in your file back to either a supplier document, a published reference, a chemical characterization result, or a test report?

Frequently Asked Questions

Is a full extractables and leachables study always required under MDR? No. MDR Annex I Section 10 requires material safety to be demonstrated, and EN ISO 10993-1:2025 defines the process for doing so within a risk management framework. ISO 10993-18 provides the tiered chemical characterization methodology. A full E/L study is required only when compositional data, supplier information, and prior evidence cannot close the gap to a safety conclusion for the device's contact category and duration. For many lower-risk devices, compositional characterization and reasoned exclusions are sufficient.

What is the difference between extractables and leachables? Extractables are substances that can be pulled from a material under exaggerated extraction conditions, providing a worst-case profile of what could theoretically migrate. Leachables are substances that actually migrate out under real clinical conditions of use. Leachables are a subset of extractables. Long-term implants and devices with prolonged tissue contact typically need both; short-duration surface-contact devices often need neither if compositional data are strong.

Can chemical characterization replace animal testing for biocompatibility? In many cases, yes. EN ISO 10993-1:2025 explicitly encourages replacement of experimental biological testing with chemical characterization combined with toxicological risk assessment, where scientifically justified. The substitution works best for systemic, subchronic, genotoxicity, and carcinogenicity endpoints when the characterization is thorough and the toxicological interpretation is rigorous. It is less applicable to local effects such as irritation and sensitization, which typically still require experimental confirmation.

Where does chemical characterization live in the technical documentation? Under MDR Annex II Section 6 on pre-clinical and clinical evidence, inside the biological evaluation section. The chemical characterization report is referenced from the biological evaluation plan and from the biological evaluation report, and is supported by raw material specifications, supplier declarations, certificates of analysis, and process validation records held elsewhere in the technical file.

What are the most common reasons Notified Bodies reject biocompatibility files? Missing or incomplete compositional data from suppliers, no documented rationale for which biological endpoints require experimental evaluation and which do not, processing residues not considered, E/L test reports without accompanying toxicological risk assessment, and mismatches between the contact category assumed in the chemical characterization and the intended purpose stated in the clinical evaluation.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Annex I Section 10 (chemical, physical and biological properties) and Annex II Section 6 (pre-clinical and clinical data). Official Journal L 117, 5.5.2017.
  2. EN ISO 10993-1:2025 — Biological evaluation of medical devices — Part 1: Requirements and general principles for the evaluation of biological safety within a risk management process. Current harmonised edition, published December 2025, superseding EN ISO 10993-1:2020.
  3. ISO 10993-18 — Biological evaluation of medical devices — Part 18: Chemical characterization of medical device materials within a risk management process. Referenced under EN ISO 10993-1:2025 as the methodology for material and extractables/leachables characterization.

This post is part of the Technical Documentation series in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. Chemical characterization is one of the highest-leverage decisions in the biological evaluation file — start with composition, run the risk assessment, and let the evidence drive the testing, not the other way around.