Presentation of the SEURAT-1 COSMOS Project: Prediction of Systemic Toxicity Following Dermal Exposure Mark Cronin 1, Elena Fioravanzo 2, Judith Madden 1, Andrea Richarz 1, Lothar Terfloth 3, Fabian Steinmetz 1, Faith Williams 4, Chihae Yang 5 1 Liverpool John Moores University, England 2 Soluzioni Informatiche srl, Italy 3 Molecular Networks GmbH Computerchemie, Germany 4 Newcastle University, England 5 Altamira LLC, USA
Project: Development of Computational Models Threshold of Toxicological Concern (TTC) Cosmetics Inventory COSMOS TTC v1.0 Munro New Toxicological Databases PBPK and In Vitro In Vivo Extrapolation In Silico Models
Role of Metabolism Prediction in the Prediction of Chronic Toxicity Prediction of detoxification Identification of toxic metabolites Prediction of clearance Supporting in vitro-in vivo extrapolation PBPK modelling for route-to-route extrapolation
TTC is Derived from Oral NOEL Values: Is the Oral Route Protective of Dermal Exposure? Scenario 1 Oral bioavailability high Dermal bioavailability high Scenario 3 Oral bioavailability low Dermal bioavailability high Scenario 2 Oral bioavailability high Dermal bioavailability low Scenario 4 Oral bioavailability low Dermal bioavailability low absorption/permeability via dermal and oral routes metabolism differences between skin and liver
COSMOS Dermal Absorption Database Data for 380+ compounds 2400+ in vitro studies (rat, mouse, pig, human) 1000+ in vivo studies (rat, mouse, pig, human, monkey) Thanks to
COSMOS Dermal Absorption Database Data for 380+ compounds 2400+ in vitro studies (rat, mouse, pig, human) 1000+ in vivo studies (rat, mouse, pig, human, monkey) Very few or no data for metabolism e.g. kinetics Very little systematic information on metabolites Thanks to
Existing Software for Metabolism Prediction [Summarised from Kirchmair J et al (2012) J. Chem. Inf. Model. 52: 617] Prediction of Metabolites OASIS TIMES Sites of Metabolism MetaPrint2D Prediction of Kinetics Prediction of CYP Binding, Affinity, Induction and Inhibition Molecular Networks Virtual ToxLab
Existing Software for Metabolism Prediction [Summarised from Kirchmair J et al (2012) J. Chem. Inf. Model. 52: 617] Prediction of Metabolites Prediction of Kinetics OASIS TIMES Molecular Networks Sites of Metabolism MetaPrint2D Software Optimised for Skin Metabolism Prediction of CYP Binding, Affinity, Induction and Inhibition Prediction Virtual ToxLab
Skin Metabolism Rules - Details Metabolism type Classification (EC enzyme nomenclature) Enzyme (skin) Enzyme activity type (liver) Specification Xenobiotic Lipid (essential ingredient in cosmetics) Steroid Protein Carbohydrate Phase I Oxidoreductase Hydrolase Isomerase Ligase Lyase Phase II Transferase Enzyme activity type Alcohol dehydrogenase Aldo-keto reductase Monoamine oxidase Carboxylesterase... Enzyme + isoform Skin metabolism ADH1B = Short chain ADH5 = Long chain alcohols Compound class Primary amine Aliph. alcohol Ketone in ring... Thanks to
Skin Metabolism Parameters Rule set enhanced by metabolism parameters Metabolism probability Kinetics (Michaelis-Menten constants for enzymatic reactions) Skin/liver ratio (low ratio low probability of transformation in skin) Detection rate of enzyme in skin (population variance) Thanks to
KNIME Nodes for Human Skin Metabolism Prediction Metabolism Rules Compounds Property Profile i.e. Probability of Metabolite Formation Thanks to
QSARs for Metabolic Clearance: Global Models Dataset of 670 compounds reported by Obach et al. Based on molecular descriptors including probability of metabolism Reasonably good model performance Similar study for metabolic clearance (ongoing) using ToxCast Data Thanks to
QSARs for Metabolic Clearance: Are Local Models a Better Approach? Local (QSAR) models usually provide more reliable prediction but are very restricted Limited read-across may be possible given: Reliable data for a (small) number of compounds An understanding of effects of physico-chemical properties on clearance
Conclusions: Future Needs for Skin Metabolism Prediction Better and more systematic data collation Development of improved models with greater coverage Possibility of adapting current liver metabolism prediction software Develop using knowledge of the differences between liver and skin metabolism Identification of most probable stable metabolite(s) Better prediction of rate and extent of clearance
Acknowledgements The European Community s Seventh Framework Program (FP7/2007-2013) COSMOS Project under grant agreement n 266835 and Cosmetics Europe http://www.cosmostox.eu The contributions of a number of experts through ILSI-EU Expert Groups