In Vitro Safety Testing for EU Cosmetics: Which Methods SCCS Accepts Under Regulation 1223/2009

The last cosmetic ingredient legally tested on animals and placed on the EU market crossed the threshold in March 2013. That’s more than a decade of absolute prohibition — yet we still receive Product Information Files with safety reports that read as if the 7th Amendment to the old Cosmetics Directive was a suggestion rather than law.

It was not. It is not.

Article 18 of Regulation (EC) No 1223/2009 does two things that brands frequently conflate. First, it prohibits animal testing for cosmetic products and their ingredients within EU territory. Second — and this is what catches importers — it bans placing on the EU market any product whose safety has been established using animal data, regardless of where that testing was performed. The Court of Justice of the European Union affirmed this extraterritorial scope in 2016, explicitly ruling that the marketing ban applies even when no validated alternative exists for the endpoint in question. If your US or Chinese supplier’s dossier rests primarily on in vivo rodent studies, that data cannot anchor a compliant Cosmetic Product Safety Report (CPSR) under Regulation (EC) No 1223/2009.

So the question becomes practical: which in vitro methods does the Scientific Committee on Consumer Safety (SCCS) actually accept, and for which toxicological endpoints are validated alternatives still missing?

The Validated Methods That Work — and the Gap Nobody Advertises

As of 2026, EURL ECVAM (the EU Reference Laboratory for Alternatives to Animal Testing) has formally validated or issued positive scientific opinion on more than 50 non-animal test methods across chemistry and toxicology. That sounds reassuring until you map it against the 11 core endpoints required for a complete cosmetic ingredient safety assessment. Fully validated, regulatory-accepted in vitro equivalents exist for perhaps five or six of those endpoints. For the rest, brands must construct their safety case using read-across, computational modelling, or existing human data — none of which is a substitute for robust in vitro evidence, but all of which are accepted when properly documented.

Skin irritation is the endpoint where in vitro science is most mature. OECD Test Guideline (TG) 439 — the reconstructed human epidermis (RhE) model — uses three-dimensional epidermal constructs such as EpiDerm™ (MatTek) or EpiSkin™ (L’Oréal) to assess cytotoxicity following topical application. Both models are formally accepted by the SCCS for skin irritation under cosmetic regulation 1223/2009. Turn-around is typically 10 to 14 days, and per-ingredient cost at a GLP-accredited contract laboratory runs approximately €1,200 to €2,800 — a fraction of what historical Draize patch tests involved.

Phototoxicity is nearly as settled. OECD TG 432, the 3T3 NRU Phototoxicity Test, measures cell viability of mouse fibroblasts before and after UVA irradiation. It has been accepted by the SCCS for decades and remains the standard for ingredients with photosensitising structural alerts: citrus-derived terpenes, coal tar derivatives, certain UV filters, and polycyclic aromatic compounds, for instance. If your formula contains bergamot essential oil, furocoumarins, or a novel organic UV absorber, this test is non-negotiable.

Skin sensitisation is where things become substantially more complex. No single in vitro method captures the complete adverse outcome pathway (AOP) for skin sensitisation, which unfolds across four mechanistic key events: hapten formation (protein binding), keratinocyte activation, dendritic cell activation, and T-cell proliferation. The SCCS — in its current Notes of Guidance, SCCS/1628/21 (11th Revision, 2021) — requires a battery of at least three mechanistically complementary methods:

• OECD TG 442C — the Direct Peptide Reactivity Assay (DPRA): addresses Key Event 1, covalent protein binding
• OECD TG 442D — KeratinoSens™ (ARE-Nrf2 luciferase assay): addresses Key Event 2, keratinocyte activation
• OECD TG 442E — the Human Cell Line Activation Test (h-CLAT): addresses Key Event 3, dendritic cell activation

Two additional validated methods — OECD TG 442F (U-SENS™) and OECD TG 442G (GARDskin) — can substitute or supplement within a defined approach. The OECD’s 2021 document on Defined Approaches for Skin Sensitisation consolidates how these combinations are scored and interpreted, giving safety assessors a structured framework for translating multi-assay data into a regulatory conclusion.

The full battery, including data integration and toxicological justification, typically costs €5,000 to €10,000 per ingredient at a qualified laboratory. That figure surprises clients who assumed in vitro would always be cheaper than in vivo. For sensitisation specifically, the battery requirement means multiple assays, potentially across different labs, plus qualified toxicologist time to synthesise the outputs. Build this into your product development budget, not your post-launch contingency.

Where Brands Routinely Run Into Trouble

Eye irritation has a more fragmented picture. OECD has published several alternative methods — TG 437 (Bovine Corneal Opacity and Permeability, BCOP), TG 460 (Fluorescein Leakage), and TG 492 (Short Time Exposure, STE) — but none functions as a full stand-alone replacement for all chemical classes. The SCCS accepts these methods for identifying severe eye irritants but notes their limitations for borderline substances. For eye-area products — eye creams, mascaras, micellar formulas — the safety dossier needs explicit justification of method selection and applicability domain, not just a test result.

Genotoxicity is better served. OECD TG 471 (the Ames bacterial reverse mutation test) and OECD TG 487 (the in vitro micronucleus test using human lymphocytes or TK6 cells) are both accepted by the SCCS for cosmetic ingredients. Together they cover two complementary genotoxicity mechanisms — gene mutation and chromosomal damage — and are generally included as a pair in any complete ingredient safety package. The Ames test in particular is inexpensive, running under €800 per compound, and a negative result for both assays provides meaningful comfort even when systemic toxicity data is thin.

Systemic toxicity — repeated-dose toxicity, reproductive and developmental toxicity, and toxicokinetics — is the genuinely unresolved frontier. No fully validated, regulatory-accepted in vitro equivalents exist for these endpoints in the cosmetics context. The SCCS acknowledged this plainly in its 2023 update to the safety evaluation framework. For these endpoints, SCCS/1628/21 (Section 3-6) accepts three approaches, in descending order of scientific weight:

1. Human clinical and epidemiological data — for ingredients with a long and well-documented history of safe use, this constitutes the primary evidence base. Traditional cosmetic raw materials such as glycerin, cetyl alcohol, and hyaluronic acid are assessed largely this way.

2. Read-across from structurally analogous substances — formally documented using the OECD QSAR Toolbox or equivalent, with a scientifically justified analogue selection. The SCCS’s 2021 Note of Guidance on Read-Across specifies what “justified” means: matched structural features, comparable metabolic pathways, similar molecular weight range, and source data generated via the same route of exposure as the cosmetic use.

3. Integrated Approaches to Testing and Assessment (IATA) — combining in vitro data, in silico predictions (DEREK Nexus, VEGA, TIMES-SS, and similar tools), mechanistic reasoning, and existing human data into a structured weight-of-evidence argument reviewed by the safety assessor.

The read-across documentation quality is where files most frequently fail regulatory review. We have seen dossiers where analogue selection was done informally — a structurally similar compound identified without explaining why that structural similarity implies equivalent systemic behaviour or metabolic fate. That is not a read-across argument; it is an assumption. National Competent Authorities in France (ANSM) and Germany (BfR) have both issued non-compliance notices in recent years for exactly this category of documentation failure, and neither authority has shown particular patience for remediation requests once a product is already on shelf.

What This Means When You’re Launching a Novel Active

If your product contains an ingredient not listed in established EU inventories, not covered by an existing SCCS opinion, and without substantial published human safety data, expect a full in vitro testing programme to take four to six months from compound submission to final study reports. A reasonable endpoint battery — skin irritation (OECD TG 439), skin sensitisation (three-method approach under SCCS/1628/21), phototoxicity (OECD TG 432), genotoxicity (TG 471 + TG 487), plus a read-across-supported argument for systemic endpoints — typically costs €18,000 to €45,000 before your safety assessor drafts Part B of the CPSR. That range widens depending on the ingredient’s novelty and the strength of the available analogue data.

One assumption we encounter consistently: brands believe that safety data generated under the US FDA’s framework, Japan’s quasi-drug system, or Chinese GB/T standards transfers cleanly into the EU regulatory context. It rarely does. The SCCS does not automatically recognise testing conducted under protocols it hasn’t formally reviewed, and a US GRAS designation or Japanese cosmetics ingredient approval carries no legal weight under Regulation (EC) No 1223/2009. Supplier dossiers arriving from outside the EU almost always need supplementary in vitro work before an EU-qualified toxicologist will countersign Part B.

In our experience reviewing incoming dossiers from Asian and North American manufacturers, around 60 to 65% require at least one additional study before the safety file is adequate for CPSR completion. That is a planning assumption, not an anomaly — and it should sit in your product development timeline long before your launch date, not surface as a crisis six weeks out.

One last practical point worth stating plainly: in vitro data intended for regulatory use should be generated under GLP (Good Laboratory Practice) conditions. The SCCS position is that non-GLP data can be included in a safety file but must be explicitly justified — meaning your safety assessor must document in writing why the study’s reliability is not compromised by the absence of GLP oversight. That justification adds cost and risk. Verify GLP accreditation status with your testing partners before signing a study order.

The in vitro science for EU cosmetic safety has advanced substantially since validated alternatives first appeared in the late 1990s. But “advanced substantially” is not “fully solved.” Understanding exactly where validated methods are robust, where they must be combined into defined approaches, and where read-across is still doing the structural heavy lifting — that distinction is what separates a CPSR that passes a National Competent Authority review from one that earns a non-compliance notice and a product withdrawal.

---

Written by Nour Abochama, Quality & Regulatory Advisor, Care Europe | VP Operations, Qalitex. Learn more about our team

Talk to our team about EU market entry. Contact us

Related from our network

• GLP-Accredited In Vitro and Analytical Testing for Cosmetic Ingredients — Qalitex Laboratories provides ISO 17025-accredited testing services under FDA-accepted and OECD protocols for US and EU regulatory submissions
• Cosmetic and NHP Safety Testing for Canadian Market Entry — Androxa supports European brands navigating Health Canada GMP and ingredient safety requirements when expanding beyond the EU