Cytotoxicity testing under ISO 10993-5 is the biocompatibility test you are most likely to fail, and its failures are the hardest to interpret. In all the testing we have supported, no test has a penchant for causing headaches quite like it. Here is how the test actually exposes and scores the cells, why its simplicity makes it so sensitive, what a failure does and does not tell you, and the three sampling principles that make any result, pass or fail, one you can trust.
Given that it is arguably the biggest of the “big three” (cytotoxicity, sensitization and irritation), and the most routinely performed, you would expect cytotoxicity to be among the best understood tests in the suite, and so among the least likely to cause trouble. If that is your expectation, you probably have not managed many of them.
Why is that, and what can be done about it?
How the test actually works
In vitro cytotoxicity is governed by ISO 10993-5:2009. The sample preparation and extraction conditions that feed it are governed by ISO 10993-12:2021. Those two documents, rather than the assay itself, are where most of the trouble starts.
Three ways to expose the cells
ISO 10993-5 recognizes three routes for putting a device in front of a cell layer, and the choice of route matters a good deal more than most sponsors expect:
Extract testing. The device is extracted into culture medium and the extract is applied to the cells. This is by far the most common approach: it suits devices that cannot reasonably be laid on a cell layer, and it tolerates awkward geometries the other methods do not.
Direct contact. A portion of the device is placed directly on the cell layer. It is the closest analogue to real surface contact and, as you would expect, the least forgiving of the three.
Indirect contact. In agar or filter diffusion, a barrier sits between the device and the cells, so only what diffuses through it reaches them. This is useful for materials that would otherwise physically disturb the monolayer.
The right route follows from the device itself: what it is made of, what form it takes and how it contacts the body. Choosing the wrong one is an efficient way to produce a result that is technically valid and practically meaningless.
The routes are not scored on the same scale, either. An extract test is graded on how much of the cell layer is affected; contact and diffusion tests are graded on the size of the zone of reactivity around and under the specimen.
The most common method: MEM elution
The test takes many forms depending on the device, material or substance being evaluated, but the most common is known as MEM elution, described in ISO 10993-5 and, for US submissions, in USP <87> as well:
The device or material is extracted in a cell culture medium called Minimum Essential Medium (MEM), most often for 24 hours at 37 °C.
The extract is applied to a monolayer of L929 mouse fibroblast cells.
The cells are incubated under physiological conditions (37 °C), most commonly for 48 hours.
The cell layer is examined under a microscope and scored.
Traditionally the scoring is qualitative, on a 0–4 reactivity scale:
Grade 0 · None. No cell lysis and no reduction in cell growth.
Grade 1 · Slight. Up to 20% of cells rounded or altered, occasional lysis, only slight growth inhibition.
Grade 2 · Mild. Up to 50% of cells rounded, no extensive lysis, no more than 50% growth inhibition.
Grade 3 · Moderate. Up to 70% of the layer rounded or lysed, more than 50% growth inhibition.
Grade 4 · Severe. Near-total or total destruction of the cell layer.
Grade 2 is where the practical boundary falls: any grade above 2 is considered a cytotoxic effect. That is the origin of the 50% figure people tend to quote, since up to half the layer affected still sits inside a passing grade.
A quantitative assay uses a different cut-off. There, a reduction in cell viability of more than 30% is considered cytotoxic, which is why you will also hear 70% viability quoted. The two numbers come from different methods and do not conflict.
Of all the biocompatibility test methods, this one sounds the most straightforward.
Why a simple test causes so much trouble
In some respects it is straightforward. What makes it a challenge is, in many ways, that very simplicity.
A monolayer of cells has no defenses. Nothing protects it the way tissues and circulation protect cells in the body. A shift in pH, a change in osmolality or a single leachable can cause a significant loss of viability. Oily residues can smother the layer, and antimicrobial materials, including metal ions such as silver, copper and nickel, hit it hard. Some of those materials are biocidal by design, and the assay cannot tell an intended antimicrobial effect from an unintended toxic one.
That sensitivity makes cytotoxicity a valuable test, but it cuts both ways.
Cytotoxicity is the canary in the coal mine, signaling danger before other systems pick it up and before anyone gets hurt. It can also be the boy who cried wolf, signaling danger where there is none.
What a cytotoxicity failure actually tells you
So what do you do with it? In many programs the test is required, and, as above, it can offer real insight into the biocompatibility of a device.
It is also the test you are most likely to fail, and that failure is the hardest to link to a biological risk. Almost every other test evaluates a specific potential for patient harm: irritation, sensitization, hemocompatibility and so on. Cytotoxicity does not. It is an indicator that something is there, not a measure of its real impact on safety, or even of where it came from.
The best practice, then, is to approach cytotoxicity the way we suggest you approach every biocompatibility test: with eyes wide open, and without assuming a pass is the default outcome.
Beyond that, provide samples that represent the true condition of your product. This is the work ISO 10993-12 does (sample preparation, extraction ratios, extraction conditions), and in our experience it decides more outcomes than the choice of assay does. It comes down to three sampling principles.
1. Produce every sample using the true, qualified process
More tests have died on the hill of “close enough” than anyone would care to imagine. The first instinct after a failure is to question the lab's process, but more often than not the failure traces back to an unrepresentative sample.
2. Package every sample properly
“Properly” means the same way the product will be distributed. If that is not possible or appropriate, take real care with the packaging you use instead. If you think that plastic bag from the grocery store is inert or clean, think again.
3. Tell the lab the intended use and the materials of construction
Second only to failures caused by unrepresentative processing are failures caused by an entirely inappropriate test system.
The test system should follow from the biological evaluation itself. Under ISO 10993-1:2025, the nature and duration of body contact drive which endpoints apply and how the device should be extracted, and none of that can reasonably be inferred from a box of parts. Extraction ratio is a good example: ISO 10993-12 sets it by surface area for most devices, typically 6 cm² per mL for thin-walled parts and 3 cm² per mL for thicker ones, or by mass (0.2 g per mL) for irregular shapes, and a lab working from a box of parts has to guess which applies. Labs generally try to catch this kind of mismatch, but in many cases they will test the samples as submitted, under the conditions the sponsor specified.
If you are filling out a test request form and are not certain of the right method or extraction conditions for your product, call or email to find out. You will be glad you did.
A failed test is not the end of the road
None of the above guarantees a pass. It does guarantee that the result you receive truly represents your product, and it gives you a firm foundation for a failure investigation.
That investigation usually starts with questions the lab's own records can already answer: what the extract looked like, whether the pH moved, whether the device swelled, and whether the response held as the extract was diluted. From there it moves to chemistry, the only step in the suite that returns a discrete answer rather than another verdict.
FDA treats cytotoxicity as a baseline endpoint for nearly every device, so the result will be read, not skipped. What reviewers look for after a positive result is the reasoning: what caused it, whether the cause is relevant to patients, and what the rest of the evidence, usually chemical characterization and a toxicological risk assessment, says about it.
A failed cytotoxicity test is not the end of the road for your biological evaluation under ISO 10993-1:2025, either. It is a wrinkle that must be dealt with, and in our experience regulators increasingly favor a clean file. But other evidence of safety, together with an understanding of the failure's root cause, can put the result in its proper context and support the approval and launch of a safe product.
Planning a cytotoxicity test? The test plan builder gives you a reasoned starting point for your device’s endpoints, and you can talk to a scientist about extraction conditions and test route before samples ship, as part of your biological evaluation plan.
May all we do lead to a safer world. Relentlessly Forward.