Tools and Approaches
Unilever’s safety assessments are exposure-led, which means we use our deep understanding of how our products are used, as well as the level of each ingredient in the product to calculate the typical exposure to each ingredient that the consumer might encounter.
To ensure we also consider the environmental impact of products that go ‘down the drain’ after use, we also calculate the amount of each ingredient that enters the environment. We consider water use in the home and the connection to any type of wastewater treatment. We then apply our expertise to evaluate all these data to generate a full understanding of exposure for the overall safety assessment.
We consider it essential that we use internationally recognised approaches and authoritative scientific evidence in our safety assessments. This may be through using existing data alone, either directly or indirectly (through read-across approaches), or these data can also form part of a weight of evidence, as can a history of safe use and exposure-based waiving. If we consider there is insufficient data for the safety assessment, we generate new approach methodology (NAM) data to address the gaps and increase the robustness of our safety assessments.
Animal-Free In Vitro Tools
There is an increasing acceptance of the role in vitro assays play in assuring the safety of consumer goods, particularly as part of next-generation risk assessment (NGRA). In addition to any ethical considerations, there is a growing desire to remove any animal products from these in vitro assays to make them more scientifically robust & human-relevant. For example, the use of foetal bovine serum (FBS) and animal-derived antibodies can introduce a lot of batch-to-batch variabilities, potentially resulting in experimental quality issues (e.g., contamination of FBS; specificity of antibodies) and reproducibility issues. Our longer-term aim is the replacement of all animal-derived products within SEAC’s current and future assays where scientifically feasible and to collaborate with other groups across the wider scientific community to meet these goals together.
Human Safety
Exposure Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
Applied Dose Estimate | An estimate of the externally applied dose for single personal care product types | As an initial starting point for an exposure-led safety assessment | Colipa and CTFA studies |
Absorption, Distribution, Metabolism and Excretion (ADME) Parameters | A series of in vitro tests, including plasma protein binding and hepatocyte clearance assays, were used to parameterise physiologically based kinetic (PBK) models, as well as additional tests to further refine exposure such as commercially available human skin models, in vitro human liver S9 incubation | The data from the assays are used to inform PBK modelling to describe the ADME properties of an ingredient following exposure | Assays can be performed by contract research organisations (CROs) |
Crème Care and Cosmetics | The software which enables a probabilistic estimate of the applied dose. It allows the examination of the likely use pattern of a variety of different products | To refine aggregate exposure assessments, where lower tier approaches are insufficient | |
DaDiet | Software tool used to estimate of exposure when performing assessments for food products | Realistic survey-based information | |
Exposure-Based Waiving Approach | This approach uses a database (e.g. COSMOS) along with decision trees to understand whether exposures to an ingredient of interest are so low they are of no concern | To waive the toxicity testing of an ingredient based on exposure assessment scenarios | |
Particle Size Distribution (PSD) Testing | This approach uses a MalvernTech to characterise the PSD of an aerosol-producing product, such as personal care, beauty and wellbeing, and home care products | To provide exposure metrics in the form of PSD data to be used within the inhalation screening assessment | Guidance on information requirements and chemical safety assessment (ECHA, 2012) |
PBK Modelling: Gastroplus | A physiologically based simulation tool that predicts the internal concentrations of ingredients from applied external doses from different routes of exposure. It uses ADME and transportation parameters that are either predicted or user specified | To estimate internal concentrations based on external doses | |
RIFM 2-Box Indoor Air Dispersion Model | Originally developed to model the exposure to fragrance ingredients, this approach is a two-compartment model that determines the concentration of an ingredient in air and therefore what is possible to be inhaled by humans | It is used as part of a tiered approach to inhalation exposure modelling | |
RIVM ConsExpo Web | A mechanistic modelling approach that uses PSD data combined with other facts, including spray rate, duration of product usage, room volume, etc. to predict the amount of an ingredient inhaled and deposited in the lung. Factsheets are available for various product types, such as personal care products, and custom scenarios can also be modelled | As part of a tiered approach to inhalation exposure modelling | |
Simulated Use Exposure Test (SUET) | This approach uses mannequins connected to an Aerodynamic Particle Sizer (APS) to measure particles (diameter < 20 µm) produced by an aerosol product, such as personal care, beauty and wellbeing, and home care products. This approach is used to mimic human exposure scenarios | To provide exposure metrics for inhalation safety assessments, such as the derived no-effect level (DNEL) approach, inhalation exposure-based waiving, and respiratory sensitisation | FEA, 2009. Guide on Particle Size Measurement from Aerosol Products. First Edition: August. FEA, 2013. Guide on Inhalation Safety Assessment for Spray Products. Correction: November. Appendix B. SCCS Notes of Guidance for the testing of cosmetic ingredients and their safety evaluation |
Predictive Chemistry Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
ADMET Predictor | A machine learning software tool that quickly and accurately predicts over 175 properties, including the site of metabolism (SOM) models for nine CYP isoforms | To identify ingredients` metabolites and potential sites of metabolism together with enzymes. | |
Derek Nexus | A knowledge-based expert system and contains a set of structural alerts for a variety of toxicological endpoints | For a general screen to predict the potential toxicity hazards of an ingredient. To identify read-across candidates for skin sensitisation safety assessment | |
Meteor Nexus | A knowledge-based expert system for predicting the metabolic fate of ingredients (based primarily on liver metabolism data) | To identify potential metabolites based on the site of metabolism | |
Molecular Initiating Events (MIEs) Atlas | A tool which predicts MIEs based on ingredient structure for a number of safety targets (based on Bowes et al 2012 ). Data from ChEMBL has been used to identify characteristics of ingredients which hit these targets. These characteristics are stored as chemical fragments in a database | When an in silico screen against a known set of safety targets may help to define a testing strategy (for receptor binding assays) or add to a safety assessment | |
OECD QSAR Toolbox | Used to fill gaps in toxicity data needed for assessing the hazards of ingredients. It consists of a toxicity database and several predictive methods covering different toxicity endpoints | To identify relevant structural features or the mode of action (MoA) of an ingredient, to identify other ingredients with similar characteristics (to use as read-across candidates), to fill data gaps with existing experimental data | |
TIMES | Tissue Metabolism Simulator – Skin Sensitisation is an expert tool which is a knowledge and statistical-based system. Skin metabolism is simulated, and predictions are made for the ingredient and its metabolites | To predict the skin sensitisation potential (and potency) of an ingredient and its metabolites | |
ToxTree | Estimates toxic hazards by applying a decision tree approach. It contains several models to predict different toxicity endpoints. Four models are used in SEAC: Cramer Rules, Benigni/ Bossa rulebase (for mutagenicity and carcinogenicity), skin sensitisation reactivity domains, and Verhaar scheme for predicting toxicity mode of action (MoA) | To fill a data gap in a safety assessment or to classify ingredients before applying the dermal sensitisation threshold (DST) or threshold of toxicological concern (TTC) approaches | |
VEGA | A platform containing several QSAR models for toxicity, ecotoxicity, environmental fate and physchem properties | To predict the potential toxicity hazard, especially for developmental and reproductive toxicology (DART) related endpoints |
Bioactivity Assessment Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
Direct Peptide Reactivity Assay (DPRA) | Quantifies the reactivity of ingredients towards synthetic peptides (cysteine and lysine) as a surrogate for covalent binding of an ingredient to skin proteins (haptenation) which represents the MIE or key event 1 of the skin sensitisation adverse outcome pathway (AOP) | To identify skin sensitisation potential and can be used as a SARA Model input to determine skin sensitisation potency and safety assessment for a given exposure | |
Peptide reactivity profiling | This in chemico assay was developed by SEAC. It uses conditions tailored to maximise reactivity towards seven peptides. Similar to DPRA, it addresses the MIE of the skin sensitisation adverse outcome pathway (AOP) | To predict the skin sensitisation potential of an ingredient and its metabolites | |
KeratinoSens™ | Measures the activation of the Keap1-Nrf2-ARE pathway for the purpose of identifying skin sensitisers. This measure represents key event 2 of the skin sensitisation adverse outcome pathway (AOP) | Can be used as a SARA Model input to determine skin sensitisation potency and safety assessment for a given exposure | |
Human Cell Line Activation Test (h-CLAT) | Quantifies changes in the expression of cell surface markers (CD54 & CD86) associated with the process of activation of monocytes and dendritic cells for the purpose of identifying skin sensitisation potential. The method addresses the key event 3 of the skin sensitisation adverse outcome pathway (AOP) | Can be used as a SARA Model input to determine skin sensitisation potency and safety assessment for a given exposure | |
USENS™ | Quantifies change in the expression of a cell surface marker (CD86) associated with the process of activation of monocytes and dendritic cells for the purpose of identifying skin sensitisation potential. The method addresses the key event 3 of the skin sensitisation adverse outcome pathway (AOP) | Can be used as a SARA Model input to determine skin sensitisation potency and safety assessment for a given exposure | |
Skin Allergy Risk Assessment (SARA) Model | A Bayesian statistical model was developed to aid skin allergy safety assessment. The model infers human-relevant points of departure (PoDs) by integrating skin sensitisation data from multiple sources in a weight-of-evidence manner. The PoD is the concentration which would induce sensitisation in 1% of the entire human repeated insult patch test (HRIPT) eligible population (ED01) and is expressed in µg cm-2. The model calculates a margin of exposure which is compared to defined benchmarks for risk decision-making. Data sources: historical in vivo data (HRIPT, and local lymph node assay, LLNA), DPRA, KeratinoSens™, h-CLAT, USENS™, and benchmark consumer exposures | Can utilise appropriate historical in vivo and/or in vitro data to give a prediction of skin sensitiser potency and can also give a low risk of sensitisation probability for a given exposure | |
In vitro Genotoxicity | A battery of in vitro tests to understand the genotoxic potential of ingredients: bacterial reverse mutation test; in vitro mammalian cell micronucleus test; and in vitro mammalian cell gene mutation tests using the thymidine kinase gene | To determine the genotoxic potential of an ingredient | Various techniques can be performed by CROs. |
ToxTracker | An in vitro tool that investigates the ability of an ingredient to damage the DNA. It can provide information about the mechanism by which an ingredient is toxic to the DNA. It consists of 6 reporter cell lines each linked to a biomarker that helps evaluate the potential to induce DNA damage, oxidative stress, protein damage, and general cellular stress | To determine the genotoxic potential and MoA of an ingredient. | |
In vitro Phototoxicity | A battery of in vitro tests to understand the phototoxic potential of ingredients: assessment of the UV/VIS absorption; the 3T3 neutral red uptake phototoxicity test; and reconstructed human epidermis phototoxicity test | To determine the phototoxicity potential of an ingredient. | Various techniques can be performed |
In Vitro Pharmacological Profiling | An in vitro tool that investigates the ability of an ingredient to specifically interact with pharmacologically significant receptors. A dose-response assay is performed where signs of agonist or antagonist activity are observed in the screening. Additional receptors beyond the initial 44 can be added. | Can contribute to the weight of evidence in a systemic safety assessment, either through the absence of observed effects or a determined point of departure (PoD) | |
Cell Stress Panel | A broad-spectrum in vitro tool that generates dose response data across a range of biomarkers that are representative of 10 different stress pathways in 2D and 3D hepatic cell systems (e.g. HepG2 cells and HepaRG 3D cultures): oxidative stress, metal stress, hypoxia, endoplasmic reticulum stress, mitochondrial toxicity, cell health markers, DNA damage, inflammation, osmotic stress, and heat shock. This assay was developed by SEAC in partnership with Cyprotex Discovery Ltd | Adverse effects can be caused by cellular stress. This panel, therefore, provides important information for the safety assessment of novel ingredients | |
High Throughput Transcriptomics (HTTr) | A tool which looks at the effect of an ingredient on the whole gene expression levels of in vitro cell systems (e.g. 2D or 3D cell systems). Dose response data are generated to identify a no-observed transcriptional effect level (NOTEL). The data can also be used to provide mechanistic information | The broad biological screen which provides points of departure (PODs) that can be used to calculate bioactivity exposure ratios (BERs) in a safety assessment | |
ReproTracker® | In vitro tool which can identify the developmental toxicity hazard of ingredients by measuring the disturbance or early embryonic development biomarkers of induced pluripotent stem cells differentiating into heart, liver, and neuronal cell lineages. The assay was developed by Toxys to provide a binary response of teratogenic versus non-teratogenic substances. SEAC is collaborating with Toxys to develop the assay further to allow point of departure (PoD) calculations | To provide assurance that exposure will not compromise embryonic development | |
devTox quickPredict™ | In vitro tool which measures changes in cell metabolism (ornithine/cysteine ratio) in induced pluripotent stem cells (iPSCs) to predict developmental toxicity potential of an ingredient | To provide assurance that exposure will not compromise embryonic development | |
Tcpl/tcpl2 | Dose-response modelling of high-throughput and high-content ingredient response data | Provides points of departures (PoDs) that can be used to calculate bioactivity exposure ratios (BERs) in a safety assessment | |
Point of Departure (PoD) Modelling | Bayesian statistical approaches developed in SEAC that can interpret dose response data from a variety of assays and determine the point at which an ingredient-induced effect is observed (PoD) | Provides PoDs that can be used to calculate bioactivity exposure ratios (BERs) in a safety assessment | |
Literature review database | Different databases (e.g. PubChem, ToxCast, Scopus) are used to extract existing data related to an ingredient | Provides credibility and supports rationale for conclusions by literature review. It can also provide points of departure (PODs) that can be used to calculate bioactivity exposure ratios (BERs) in a safety assessment | e.g. |
Other Safety Assessment Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
Antimicrobial Resistance (AMR) Protocol | An experimental in vitro protocol that can be adapted to reflect realistic use criteria such as in-use concentrations and contact times, for antimicrobial ingredient and full product formulas | To predict bacterial resistance to microbiocides and cross-resistance to antibiotics | |
History of Safe Use (HoSU) | A multi-criteria-decision-analysis tool which can be used to assess the safety of natural ingredients, usually plant-derived, based on their similarity to a comparator material which has a HoSU either in foods or, sometimes, traditional medicines | To put together a robust safety assessment of natural ingredients using the HoSU. This covers both systemic and local endpoints and so is applicable across all product types | |
Micro-biological Risk Assessment Approach for Microbiome Perturbations from Use of Products Providing Hygiene, Care and Skin Benefits | A decision framework based on using considerations of history of safe use (HoSU), stability of the microbiome, and functions to be protected | To qualitatively assess the safety of ingredients targeting the human microbiome | |
Micro- biological Risk Assessment (MRA) Approach for Personal Care Products | Application of the 4 step Codex Alimentarius MRA framework (frequently used in the Food industry) to the safety assessment of products providing hygiene, care and skin benefits | Risk-based, approaches that enable more robust decision-making to ensure microbiological safety |
Environmental Safety
Exposure Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
EUSES | A tool designed to facilitate quantitative safety assessment of ingredients in accordance with the EU Technical Guidance Documents | The model is used to prepare environmental safety assessments under the REACH (Registration, Evaluation, Authorisation, and Restriction of Chemicals) regulation. | EUSES - European Union System for the Evaluation of Substances - ECHA |
Pangea-EcoHope | Multi-scale multimedia model based on the more spatially resolved, catchment-based hydrological HydroBASINS dataset covering the entire globe | This tool is used to inform research and understand spatial differences in exposure | |
ScenAT | Global GIS-based model to estimate environmental emissions and concentrations of home and personal care ingredients | Calculates risk characterisation ratios (RCRs) allowing an environmental safety assessment for an ingredient | |
Simple Treat | Estimates the fate and emission of ingredients in sewage treatment plants (STPs) | This tool predicts the removal efficiency of chemicals in STPs which is used in environmental exposure assessment |
Physical-Chemical Properties Tools
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
ACD Labs | Predictive chemistry tool used to predict a variety of physical-chemical properties of ingredients, e.g., pKa, logKow, solubility | To determine physical-chemical properties as the behaviour of an ingredient in the environment is often driven largely by its physical-chemical properties | |
Bioloom | Predicts octanol-water partitioning coefficient (logKow) according to the widely accepted Leo and Hansch fragment approach | Aquatic toxicity is driven by the ability of an ingredient l to partition into biological membranes. LogKow can be used as a descriptor for such behaviour in quantitative structure-activity relationships (QSAR) approaches | |
Coarse-grained modelling | Method to predict membrane-water partitioning using coarse-grained simulations and the MARTINI force field | To predict membrane-water partitioning of ingredients to inform physiologically based kinetic (PBK) modelling and bioaccumulation potential | |
COSMOmic | An extension of the COnductor-like Screening MOdel for Realistic Solvation (COSMO-RS) for the study of anisotropic systems. COSMOmic can be used to predict membrane-water partitioning | To predict membrane-water partitioning of ingredients to inform physiologically based kinetic (PBK) modelling and bioaccumulation potential | |
Episuite | Suite of in silico tools developed by the United States Environmental Protection Agency (US EPA) for predicting properties and behaviour of ingredients in the environment | The software contains several models which can be used to predict a variety of properties including physical-chemical properties, aquatic toxicity, biodegradation and bioaccumulation |
Mode of Action (MoA) Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
ChemProp | Suite of models for predicting properties, fate, and effects of ingredients in the environment | Used to predict toxic mode of action (MoA) and guide quantitative structure-activity relationships (QSAR) selection for aquatic toxicity predictions | |
MechoA | A tool for understanding mechanisms of toxic action based on structural alerts of ingredients | Used to assess mechanisms of action (MechoAs) for the purposes of predicting toxicity (through the application of quantitative structure-activity relationships, QSARs) or chemical grouping for supporting read-across | Available in the OECD toolbox. |
MIE profiler | Mechanistic scheme for the classification of ingredients | Used to assess mechanisms of action (MechoAs) of ingredients to support the prediction of toxicity or to support chemical grouping/ read-across | |
OECD Toolbox | A suite of in silico tools used to fill gaps in toxicity data needed for assessing the hazards of ingredients as well as support mode of action (MoA) identification. It consists of a toxicity database and several predictive methods covering different toxicity endpoints | To identify relevant structural features or MoA of an ingredient, to identify others with similar characteristics (to use as read-across candidates), to support arguments to fill data gaps with existing experimental data | |
Toxtree | Estimates toxic hazard of an ingredient by applying a decision tree approach | Used to predict toxic mode of action (MoA) and guide quantitative structure-activity relationships (QSAR) selection for aquatic toxicity predictions |
Ecotoxicity Data Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
ECOSAR | Part of the US EPA Episuite software focused on aquatic toxicity | Used to predict acute and chronic toxicity in aquatic species | |
Ecotoxicological Threshold of Concern (EcoTTC) | A concept of applying a pre-derived chemical class or mode of action (MoA) dependent threshold value in place of the predicted-no effect concentration (PNEC) value to assess the hazard of an ingredient | Used to fill data gaps, to potentially waive certain hazard identification studies when exposure to an ingredient is unlikely to exceed a level of concern, to apply in cases where it is difficult to predict physical-chemical properties | |
Literature review | Different databases (e.g. ECHA, EPA Dashboard) are used to extract existing hazard data of an ingredient | Provides credibility and supports the rationale for conclusions by literature review. It can also provide points of departure (PoDs) that can be applied to environmental safety assessments | e.g. OECD-SIDS |
Quantitative Structure-Activity Rel-ationships (QSAR) Model | QSARs for predicting toxicity | Used to fill gaps in toxicity data needed for assessing the hazards of ingredients |
Biodegradation, Fate, and Toxicokinetics Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
BIONIC | Model for estimating the bioconcentration factor (BCF) in fish | Used as an in-silico method for supporting the estimation of BCF of neutral and ionisable organic ingredients in fish | |
BIOWIN | Part of the US EPA Episuite software focused on biodegradation | Used to predict if an ingredient will be readily biodegradable in the environment | |
Catalogic Models | Simulate the sequence of catabolic reactions responsible for the mineralization of ingredients | Predicting biodegradation of ingredients in the environment | |
qIVIVE-Kinetic mass transfer models | Computer coded models developed externally and publicly available to support the conversion of external to internal concentrations of ingredients in fish and vice versa | Used to convert the concentration of the point of departure (PoD) at target site to the external concentration required to achieve PoD internally |
Other Safety Assessment Tools and Approaches
Tool/ Approach | What is it? | Why use it? | Where to find it? |
|---|---|---|---|
Euro-Monitor | Market research database providing detailed data and analysis on industries and consumers across 210 countries | To understand the volumes and sales data for consumer products cross industry | Euromonitor International | Strategic Market Research, Data & Analysis |
SeqAPASS | A web-based application to assess protein sequence/structural similarity across taxonomic groups to predict relative intrinsic susceptibility | Allows for evaluation of any potential protein target and its conservation across taxonomic groups toward rapid, screening-level assessments of probable cross species susceptibility |