On the Assessment of the Reproductive or Teratogenic Potential of Cosmetic Ingredients

Rim Kaidi
6 hours ago
5 min read

1. Safety of cosmetic ingredients: a legal and ethical requirement

The use of any new cosmetic ingredient must be safe. That is a legal obligation according to Article 3 of the European Cosmetic Regulation and in other cosmetic-related laws and standards throughout the world, but also a moral one, as it would be unethical to place on the market a substance that is toxic, or whose toxicity is unknown and for which safe exposure cannot be demonstrated.

2. Cosmetic safety assessment without animal testing

In Europe first, since 2009 -and in more parts of the world each year- animal testing is no longer authorised to demonstrate the safe use of cosmetic ingredients.

Many toxicological endpoints must be assessed to demonstrate the safe use of an ingredient: some of them have been assessed in vitro for decades, like mutagenicity, for which the Ames test was first described in the 1970s.Others have been greatly improved in vitro since the enforcement of the European Cosmetics Regulation, such as skin sensitization, for which dozens of in vitro and in chemico models have been proposed since 2009, and several now have an OECD-dedicated guideline (DPRA, U-SENS, GARDskin assay, etc.).

3. The challenge of systemic toxicity endpoints

But some endpoints are complex to summarize into a simple in vitro assay, like those related to systemic toxicity (i.e. toxicity that impacts organs not directly exposed to substance, like internal organs). Systemic toxicity includes general toxicity, which can affect organs; carcinogenicity, the induction of cancer by chemicals; and developmental and reproductive toxicity (DART), i.e. toxicity that impact reproductive function or toxicity to the developing foetus.

4. Why DART is particularly complex to assess

The latter is particularly complex to study, because embryonic and foetal developmental is a long process, starting with a single cell and evolving into a million-cell organism, and it includes many intermediate steps with different sensitivities to chemicals at various doses.

5. From molecular events to adverse outcomes: the AOP concept

Therefore, to simplify DART assessment, the first step consists in identifying a Molecular Initiating Event (MIE), i.e. how a chemical may interact with cellular receptors or enzymes. An MIE leads to a Key Event (KE): a change induced by this chemical interaction between the target and the xenobiotic (i.e. the studied chemical substance -from the Greek "ξένος", xenos, "foreign").

A Key Event then leads to an Adverse Outcome (AO), i.e. a "visible" effect, such as a malformation or reduced fertility in the context of DART. The combination of an MIE, KE, and AO defines an Adverse Outcome Pathway (AOP).

6. Integrated Testing Strategies (ITS) and the “2 out of 3” approach

Hence, assays that target a specific AOP may be developed, making DART testing easier to transpose in vitro using simplified models. On the other hand, that also means that several assays, screening several AOPs, are needed to get an overview of the adverse potential of a xenobiotic. Moreover, as any assay always has pitfalls and limitations, combining them reduces the risk of inconclusive results. That is the solution proposed by Burbank et al (2023), which consists in an integrated testing strategy, based on a set of three assays that each show satisfactory predictive capacity, but perform even better when combined, in what is called a "2 out of 3" approach: two tests leading to the same conclusion (positive or negative) determine the overall outcome for the tested substance. The third test is performed if the first two do not show the same conclusion, or to confirm the existing result. If such an ITS leads to a positive conclusion (i.e. that the substance may induce developmental adverse effects), further investigations can subsequently be performed to assess the potential risk, such as PBPK modelling to refine exposure assessment or additional tests to better understand the substance's mode of action.

7. In vitro assays targeting specific AOP steps

7.1 Receptor-based transactivation assays (MIE/KE)

Several in vitro tests exist, most of them focusing on one specific step of one AOP, like the ER and AR transactivation tests, which evaluate the binding of test substances with Estrogen and Androgen Receptor. This is a typical example of MIE/KE-based assay: the test substance interacts with receptors, and this interaction is directly observed via the expression of genes associated with these receptors (the luciferase gene in most case).

7.2 Morphological and biomarker-based assays (AO-oriented)

Other assays such as those developed by ReproTracker focus on Adverse Outcomes: in a nutshell, morphological changes of exposed cells are observed, and abnormal changes are concluded as positive for teratogenicity potential. However, the main results of ReproTracker assays are the measurement of molecular biomarkers related to cells differentiation. So the main step is the identification of key event, adverse outcome observation is a secondary result.

8. In silico approaches and computational prediction

Tests based neither on animal nor in vitro also exist: they are in silico methods, based on computational models. The OECD QSAR Toolbox provide different, free models either based on structure-activity relationship, predicting the effect of a chemical based on its chemical formula and design, or based on its analogy with other known chemicals. These expert models provide various predictivity and are often better at detecting toxic chemicals than detecting non-toxic chemicals, as more data exist, more articles are published, about the toxicity of chemicals than about their non-toxicity !

The development of artificial intelligence and neural networks make those models always more predictive (e.g. Lee et al, 2025), but we can expect a 'glass ceiling', as the predictivity of such models is based on accumulated in vivo data, which will not be infinite.

9. Embryo-based whole organism models still in use

These tests were developed to address the great complexity of embryo and foetus development. To a certain extent though, some assays can still be performed on whole organisms: indeed, the use of animal embryos is still authorised, even in the EU, for the safety assessment of cosmetic ingredients. European directive 2010/63, on the protection of animals used for scientific purposes, applies to "live non-human vertebrate animals, including independently feeding larval forms" and is the legal basis to define "animal testing", even though it is not referenced in the European Cosmetics Regulation. Therefore, it is generally considered that tests performed on embryos, are acceptable for cosmetic ingredient assessment. Moreover, these models have the advantage to be based on complete organisms, authorizing the direct observation of malformations, combined with controlled exposure conditions and great knowledge of the models, as some vertebrate species embryos (e.g. zebrafish), have been reference models in embryology for decades.

10. Ethical debates and possible future regulatory changes

However, some groups, associations, NGOs, and governments have raised ethical concerns on this use of animal embryo models, and an update of European regulations might soon forbid the use of such models, and any animal-based ones (vertebrate or not) for cosmetic testings. Until then, assays such as the Zebrasfish Embryotoxicity Test or even the Frog Embryo Teratogenicity assay on Xenopus are authorised and provide quite predictive results regarding teratogenicity assessment (e.g. Leconte & Mouche, 2013).

11. Conclusion

In conclusion, a significant number of tests are now available to assess some of DART endpoints. A specific set of these performed under appropriate conditions can produce good predictive results. Like many tests, however, most of these results are hazard-based, i.e. they provide a qualitative response (positive or negative), and not a quantitative one that can support the establishment of a reference value. Whether you need assistance in monitoring such assays, performing in silico analyses with our experts from Simply Predict, or designing an assessment strategy, do not hesitate to reach out to Cehtra for DART-related topics!

Our services

Contact our expert