Search

Drug-induced liver injury (DILI) remains a challenge when translating knowledge from the preclinical stage to human use cases. Attempts to model human DILI directly based on the information from drug labels have had some success; however, the approach falls short of providing insights or addressing uncertainty due to the difficulty of decoupling the idiosyncratic nature of human DILI outcomes. Our approach in this comparative analysis is to leverage existing preclinical and clinical data as well as information on metabolism to better translate mammalian to human DILI. The human DILI knowledge base from the United States Food and Drug Administration (U.S. FDA) National Center for Toxicology Research contains 1036 pharmaceuticals from diverse therapeutic categories. A human DILI training set of 305 oral marketed drugs was prepared and a binary classification scheme applied. The second knowledge base consists of mammalian repeated dose toxicity with liver toxicity data from various regulatory sources. Within this knowledge base, we identified 278 pharmaceuticals containing 198 marketed or withdrawn oral drugs with data from the U.S. FDA new drug application and 98 active pharmaceutical ingredients from ToxCast. From this collection, a set of 225 oral drugs was prepared as the mammalian hepatotoxicity training set with particular end points of pathology findings in the liver and bile duct. Both human and mammalian data sets were processed using various learning algorithms, including artificial intelligence approaches. The external validations for both models were comparable to the training statistics. These data sets were also used to extract species-differentiating chemotypes that differentiate DILI effects on humans from mammals. A systematic workflow was devised to predict human DILI and provide mechanistic insights. For a given query molecule, both human and mammalian models are run. If the predictions are discordant, both metabolites and parents are investigated for quantitative structure-activity relationship and species-differentiating chemotypes. Their results are combined using the Dempster-Shafer decision theory to yield a final outcome prediction for human DILI with estimated uncertainty. Finally, these tools are implementable within an

Determination of the no observed adverse effect level (NOAEL) of a substance is an important step in safety and regulatory assessments. Application of conventional in silico strategies, for example, quantitative structure-activity relationship (QSAR) models, to predict NOAEL values is inherently problematic. Whereas QSAR models for well-defined toxicity endpoints such as Ames mutagenicity or skin sensitization can be developed from mechanistic knowledge of molecular initiating events and adverse outcome pathways, QSAR is not appropriate for predicting a NOAEL value, a concentration at which "no effect" is observed. This paper presents a chemoinformatics approach and explores how it can be further refined through the incorporation of toxicity endpoint-specific information to estimate confidence bounds for the NOAEL of a target substance, given experimentally determined NOAEL values for one or more suitable analogues. With a sufficiently large NOAEL database, we analyze how a difference in NOAEL values for pairs of structures depends on their pairwise similarity, where similarity takes both structural features and physicochemical properties into account. The width of the estimate NOAEL confidence interval is proportional to the uncertainty. Using the new threshold of toxicological concern (TTC) database enriched with antimicrobials, examples are presented to illustrate how uncertainty decreases with increasing analogue quality and also how NOAEL bounds estimation can be significantly improved by filtering the full database to include only substances that are in structure categories relevant to the target and analogue.

California's Office of Environmental Health Hazard Assessment was tasked with conducting risk assessments for United States Food and Drug Administration-approved food dyes relative to neurobehavioral concerns. The purpose of this assessment was to evaluate the evidence for neurodevelopment effects based on three streams of evidence: 1) studies identified by OEHHA for consideration in a quantitative risk assessment; 2) studies relevant to understanding mechanisms of neurobehavioral effects; 3) an in silico assessment of the bioavailability of USFDA-approved food dyes. The results indicate a lack of adequate or consistent evidence of neurological effects, supported by a lack of bioavailability and brain penetration predicted by the in silico assessment. Further, the mechanistic evidence supports a lack of activity from in vitro neurotransmitter assays, and a lack of evidence to support molecular initiating events or key events in adverse outcome pathways associated with neurodevelopmental effects, supporting a lack of biological plausibility for neurobehavioral effects following food exposures to colors. These conclusions are consistent with other authoritative bodies, such as JECFA and EFSA, that have determined (i) other effects are more appropriate for estimating acceptable daily intakes and (ii) evidence from the neurobehavioral studies lack the strength to be relied upon for quantitative risk assessment.

A new database of antimicrobial-enriched chemicals for the Threshold of Toxicological Concern (TTC) approach has been compiled, comprising 1357 chemicals with 276, 54, and 1027 substances in Cramer Classes I, II, and III, respectively. To enrich the chemical space of the No-/Lowest-Observed-Adverse Effect Level (NOAEL/LOAEL) database, a reference Antimicrobial (AM) Inventory (681) was established for chemical inclusion. To this database, the three existing TTC datasets were combined via robust data fusion process. From the final AM TTC Dataset, the fifth percentiles were derived to be 2.7, 0.43, and 0.12 mg/kg-bw/day for Cramer Classes I, II, and III, respectively. Considering the high percentage of AMs being Cramer Class III, the thresholds are remarkably stable across various TTC datasets. Based on the AM-enriched database, a set of AM categories stratified across potency were developed to classify AMs beyond the capability of the conventional Cramer Tree approach. Grouping the query chemical within the AM category, further distribution analyses were conducted to identify subclasses and differentiate potency. This study proposes a new framework for potential assessment of chronic toxicity made possible with the power of modern reliable databases and chemoinformatic methods.

To facilitate the practical implementation of the guidance on the residue definition for dietary risk assessment, EFSA has organized an evaluation of applicability of existing in silico models for predicting the genotoxicity of pesticides and their metabolites, including literature survey, application of QSARs and development of Read Across methodologies. This paper summarizes the main results. For the Ames test, all (Q)SAR models generated statistically significant predictions, comparable with the experimental variability of the test. The reliability of the models for other assays/endpoints appears to be still far from optimality. Two new Read Across approaches were evaluated: Read Across was largely successful for predicting the Ames test results, but less for in vitro Chromosomal Aberrations. The worse results for non-Ames endpoints may be attributable to the several revisions of experimental protocols and evaluation criteria of results, that have made the databases qualitatively non-homogeneous and poorly suitable for modeling. Last, Parent/Metabolite structural differences (besides known Structural Alerts) that may, or may not cause changes in the Ames mutagenicity were identified and catalogued. The findings from this work are suitable for being integrated into Weight-of-Evidence and Tiered evaluation schemes. Areas needing further developments are pointed out.

Since its creation in 2002, the European Food Safety Authority (EFSA) has produced risk assessments for over 5000 substances in >2000 Scientific Opinions, Statements and Conclusions through the work of its Scientific Panels, Units and Scientific Committee. OpenFoodTox is an open source toxicological database, available both for download and data visualisation which provides data for all substances evaluated by EFSA including substance characterisation, links to EFSA’s outputs, applicable legislations regulations, and a summary of hazard identification and hazard characterisation data for human health, animal health and ecological assessments. The database has been structured using OECD harmonised templates for reporting chemical test summaries (OHTs) to facilitate data sharing with stakeholders with an interest in chemical risk assessment, such as sister agencies, international scientific advisory bodies, and others. This manuscript provides a description of OpenFoodTox including data model, content and tools to download and search the database. Examples of applications of OpenFoodTox in chemical risk assessment are discussed including new quantitative structure–activity relationship (QSAR) models, integration into tools (OECD QSAR Toolbox and AMBIT-2.0), assessment of environmental footprints and testing of threshold of toxicological concern (TTC) values for food related compounds. Finally, future developments for OpenFoodTox 2.0 include the integration of new properties, such as physico-chemical properties, exposure data, toxicokinetic information; and the future integration within in silico modelling platforms such as QSAR models and physiologically-based kinetic models. Such structured in vivo, in vitro and in silico hazard data provide different lines of evidence which can be assembled, weighed and integrated using harmonised Weight of Evidence approaches to support the use of New Approach Methodologies (NAMs) in chemical risk assessment and the reduction of animal testing.

The existing Carcinogenicity Potency Database (CPDB) has been curated and extended with new data to facilitate a re-evaluation ofthe Threshold of Toxicological Concern (TTC) for carcinogenicity. New data have been added from the National Toxicology Program (NTP) and other open sources. All data were subject to a thorough review to ensure accuracy of the chemical structure as well as toxicological information. The new CPDB now comprises data for more than 650 chemicals with acceptable studies, according to defined criteria, of which >570 were carcinogens with >450 associated with reliable genotoxicity data. The CPDB was subjected to various analyses to determine points of departure, notably TD25 and TD50 as well as Benchmark Dose Levels. These analyses were compared and demonstrated the need for appropriate dose response data. A strategy was also implemented to identify mode of action with regard to genotoxicity. In vivo, in vitro and in silico data were applied through a defined strategy to identify genotoxic and non-genotoxic carcinogens. Specifically, where available, experimental genotoxicity (mutagenicity or clastogenicity) data or information from computational models (QSARs and structural alerts) for DNA reactivity, in vivo micronucleus effects and chromosomal aberration were utilised. Analysis of the new CPDB confirmed the conservative nature of the current TTC values for carcinogens. The new CPDB is publicly available as an Access file and searchable via COSMOS DB (cosmosdb.eu) and is intended to support further TTC evaluations for carcinogenicity. The funding of the CEFIC LRI-B18 Project is gratefully acknowledged.

Pharmaceuticals are considered as a new, important group of pollutants. These compounds can enter the environment via several routes and can disturb the natural balance of ecosystems. Factors affecting the environmental fate of medical substances can be determined with computational modelling. The routine application of the modelling methodology in the environmental risk assessment for newly designed pharmaceuticals would enable prediction of their important physical/chemical properties and forecasting their long-range transport and fate. In this contribution, we present the existing state-of-the-art and review the currently available modelling tools of two groups: Quantitative Structure-Property Relationship techniques and Multimedia Mass-balance models. We discuss the current research needs in the context of extending the applicability of the existing tools onto pharmaceuticals, being a more structurally diversified group of contaminants than persistence organic pollutants, for which the majority of the existing models have been originally developed. Multimedia fate models applied for pharmaceuticals are reviewed.QSPR approach to predict physical/chemical properties of pharmaceuticals is reviewed.QSPR-MM combined model is suggested to predict Pov and LRTP of chemical pollutants.

This paper reviews in silico models currently available for the prediction of skin permeability. A comprehensive discussion on the developed methods is presented, focusing on quantitative structure-permeability relationships. In addition, the mechanistic models and comparative studies that analyse different models are discussed. Limitations and strengths of the different approaches are highlighted together with the emergent issues and perspectives.

The aim of this paper was to provide a proof of concept demonstrating that molecular modelling methodologies can be employed as a part of an integrated strategy to support toxicity prediction consistent with the mode of action/adverse outcome pathway (MoA/AOP) framework. To illustrate the role of molecular modelling in predictive toxicology, a case study was undertaken in which molecular modelling methodologies were employed to predict the activation of the peroxisome proliferator-activated nuclear receptor γ (PPARγ) as a potential molecular initiating event (MIE) for liver steatosis. A stepwise procedure combining different in silico approaches (virtual screening based on docking and pharmacophore filtering, and molecular field analysis) was developed to screen for PPARγ full agonists and to predict their transactivation activity (EC50). The performance metrics of the classification model to predict PPARγ full agonists were balanced accuracy=81%, sensitivity=85% and specificity=76%. The 3D QSAR model developed to predict EC50 of PPARγ full agonists had the following statistical parameters: q(2)cv=0.610, Nopt=7, SEPcv=0.505, r(2)pr=0.552. To support the linkage of PPARγ agonism predictions to prosteatotic potential, molecular modelling was combined with independently performed mechanistic mining of available in vivo toxicity data followed by ToxPrint chemotypes analysis. The approaches investigated demonstrated a potential to predict the MIE, to facilitate the process of MoA/AOP elaboration, to increase the scientific confidence in AOP, and to become a basis for 3D chemotype development.

All 19 possible chlorophenol (CPh) and 12 chlorobenzene (CBz) congeners as potential impurities or additive were quantified in a relatively highly chlorinated type of technical chlorobiphenyl (CB) mixtures of Aroclor 1268, Chlorofen, Clophen T 64, Kanechlor 600, and Kanechlor 1000 using isotope dilution technique and HRGC/HRMS. The total CBzs content of Aroclor 1268, Chlorofen and Clophen T 64, Kanechlor 600, and Kanechlor 1000 was 0.039, 0.5, 230, 0.068, and 400 mg/g, respectively, while of CPhs was0.007, 0.48, 10, 0.093, and 0.98 microg/g. All 12 congeners of chlorobenzene could be quantified in all the formulations examined, but their proportions varied largely. It seems that stockpiles of technical chlorobiphenyl formulations and hazardous wastes containing CBs, both with added and/or by-side CBzs are also a somehow forgotten source of environmental contamination with those environmentally relevant compounds. No CPhs were found in Aroclor 1268 (7 ng/g). 2-MoCPh, 2,6-DiCPh, 3,5-DiCPh, 3,4,5-TrCPh, and 2,3,4,5,6-PeCPh were absent (1-20 ng/g) in Chlorofen and Clophen T 64, while other chlorophenol congeners were found at concentration from 7.4 to 130 and from 9.9 to 8,800 ng/g, respectively. Then, 3- and 4-chlorophenol, which co-eluted, were main contributors (88%) to the total CPhs content of Clophen T 64, while 2,3,4,6-TeCPh with 27% abundance was a major congener among CPhs in Chlorofen. Then, 2,4,6-TrCPh was the most abundant congener in Kanechlor 600, while 2,4,5-TrCPh was the most abundant congener in Kanechlor 1000.

The toxicological assessment of genotoxic impurities is important in the regulatory framework for pharmaceuticals. In this context, the application of promising computational methods (e.g. Quantitative Structure-Activity Relationships (QSARs), Structure-Activity Relationships (SARs) and/or expert systems) for the evaluation of genotoxicity is needed, especially when very limited information on impurities is available. To gain an overview of how computational methods are used internationally in the regulatory assessment of pharmaceutical impurities, the current regulatory documents were reviewed. The software recommended in the guidelines (e.g. MCASE, MC4PC, Derek for Windows) or used practically by various regulatory agencies (e.g. US Food and Drug Administration, US and Danish Environmental Protection Agencies), as well as other existing programs were analysed. Both statistically based and knowledge-based (expert system) tools were analysed. The overall conclusions on the available in silico tools for genotoxicity and carcinogenicity prediction are quite optimistic, and the regulatory application of QSAR methods is constantly growing. For regulatory purposes, it is recommended that predictions of genotoxicity/carcinogenicity should be based on a battery of models, combining high-sensitivity models (low rate of false negatives) with high-specificity ones (low rate of false positives) and in vitro assays in an integrated manner.

In the regulatory assessment of chemicals, the use of non-testing methods such as (quantitative) structure–activity relationship models ([Q]SARs), is increasingly required or encouraged, in order to increase the efficiency and effectiveness of the risk assessment process, and to minimise the reliance on animal testing. The main question for the assessor concerns the usefulness of the non-testing approach, which can be broken down into the practical applicability of the method and the adequacy of the predictions. A framework for assessing and documenting (Q)SAR models and their predictions has been established at the European and international levels. Exactly how the framework is applied in practice will depend on the provisions of the specific legislation and the context in which the non-testing data are being used. The framework leaves largely open the question of how to determine the adequacy of predicted data. In fact, there is a considerable need to develop clear guidance on how the predictions generated by non-testing methods can be translated into regulatory conclusions and decisions. This chapter describes the current framework for documenting (Q)SAR models and their predictions, and discuses how it might be built upon to provide more detailed guidance on how to use (Q)SAR predictions in regulatory decision making. Some of the scientific issues that need to be considered, as well the difficulties encountered, are illustrated with respect to some widely used software tools (Toxtree, Caesar, ToxBoxes, and DEREK for Windows) and their predictions of genotoxicity for two case study compounds, sodium nitroguaiacolate and methylparathion.

The methods and tools of computational toxicology form an essential and integrating pillar in the new paradigm of predictive toxicology, which seeks to develop more efficient and effective means of assessing chemical toxicity, while also reducing animal testing. The increasingly prominent role of computational toxicology in the implementation of European chemicals' legislation is described, along with initiatives by the European Commission's Joint Research Centre to promote the acceptance and use of computational methods. Outstanding needs and scientific challenges are also outlined. In recent years, there have been impressive scientific and technological advances in computational toxicology. However, considerable progress is still needed to increase the acceptance of computational methods, and in particular to develop a deeper and common understanding of how to apply computational toxicology in regulatory decision making.

A major source of uncertainty when assessing the human health and environmental risks of chemicals is the paucity of experimental information on the metabolic and (bio)degradation pathways of parent compounds and the toxicological properties of their metabolites and (bio)degradation products. Taking into account animal welfare and cost-effectiveness considerations, the only practical means of obtaining the information needed to reduce this uncertainty, is to use alternative (non-animal) methods, such as in vitro tests and in silico models. In this report, we explore the usefulness of in silico metabolic simulation tools (expert systems) as a means of supporting the regulatory assessment of chemicals. In particular, we investigate the use of selected in silico tools to: (i) simulate microbial and mammalian metabolic pathways; (ii) identify potential metabolites resulting from biotransformation; and (iii) gain insights into the mechanistic rationale of simulated metabolic reactions and the likelihood of their occurrence. For illustrative purposes, the microbial and mammalian biotransformation pathways of a case study compound, the fungicide carbendazim, were generated by using the CRAFT Explorer 1.0 (Molecular Networks GmbH) and Meteor 12.0.0 (Lhasa Ltd.) software tools. Additionally, the set of potential metabolites resulting from microbial and mammalian metabolism was predicted with the OECD QSAR Application Toolbox 2.0 (beta version). Comparison of the in silico predictions with existing experimental data on carbendazim metabolism showed the potential usefulness of using software tools for metabolite prediction. However, the results are strongly dependent on the software constraints specified by the user, and require careful interpretation, taking into account the needs of the exercise and the availability of existing information. Further efforts are needed to develop guidance on the use of in silico metabolic simulation tools for the purposes of regulatory risk assessments., JRC.DG.I.6-Systems toxicology

A model-based prioritization exercise that considers two aspects, i.e. the hazard of a certain chemical and its exposure levels, and focuses on aquatic ecosystems has been carried out for the Water Framework Directive (WFD) implementation. The exercise also takes into account hazards due to secondary poisoning, bioaccumulation through the food chain and potential human health effects, e.g. due to fish or drinking water consumption. First a list provided by Member States, Stakeholders and NGOs comprising 2034 compounds was evaluated according to hazard and exposure criteria. Then 78 compounds classified as ¿of high concern¿ where analysed and ranked in terms of PEC/PNEC risk ratio (Predicted Environmental Concentration/Predicted No-Effect Concentration). This exercise would need an expert assessment to analyze the applied methodology and the results obtained. As far as possible, one of the main requirements for the tools employed in the model-based prioritization has been that they are freely accessible to interested parties; however this has not always been possible due to the tight schedule of the process and the fact that some estimation/calculation procedures were not available and needed to be developed at JRC. The proposed approach constitutes the first step in setting the basis for an open modular tool that could be used for the next prioritization exercises., JRC.DDG.I.6-Systems toxicology

This report presents research results obtained in the framework of a project on the Applicability of Quantitative Structure-Activity Relationship (QSAR) analysis in the evaluation of the toxicological relevance of metabolites and degradates of pesticide active substances. During this project, which was funded by the European Food Safety Authority (EFSA), the Joint Research Centre (JRC) performed several investigations to evaluate the comparative performance of selected software tools for genotoxicity and carcinogenicity prediction, and to develop a number of case studies to illustrate the opportunities and difficulties arising in the computational assessment of pesticides. This exercise also included an investigation of the chemical space of several pesticides datasets. The results indicate that different software tools have different advantages and disadvantages, depending on the specific requirements of the user / risk assessor. It is concluded that further work is needed to develop acceptance criteria for specific regulatory applications (e.g. evaluation of pesticide metabolites) and to develop batteries of models fulfilling such criteria., JRC.DG.I.6-Systems toxicology

In the assessment of industrial chemicals, cosmetic ingredients, and active substances in pesticides and biocides, metabolites and degradates are rarely tested for their toxicologcal effects in mammals. In the interests of animal welfare and cost-effectiveness, alternatives to animal testing are needed in the evaluation of these types of chemicals. In this report we review the current status of various types of in silico estimation methods for Absorption, Distribution, Metabolism and Excretion (ADME) properties, which are often important in discriminating between the toxicological profiles of parent compounds and their metabolites/degradation products. The review was performed in a broad sense, with emphasis on QSARs and rule-based approaches and their applicability to estimation of oral bioavailability, human intestinal absorption, blood-brain barrier penetration, plasma protein binding, metabolism and. This revealed a vast and rapidly growing literature and a range of software tools. While it is difficult to give firm conclusions on the applicability of such tools, it is clear that many have been developed with pharmaceutical applications in mind, and as such may not be applicable to other types of chemicals (this would require further research investigation). On the other hand, a range of predictive methodologies have been explored and found promising, so there is merit in pursuing their applicability in the assessment of other types of chemicals and products. Many of the software tools are not transparent in terms of their predictive algorithms or underlying datasets. However, the literature identifies a set of commonly used descriptors that have been found useful in ADME prediction, so further research and model development activities could be based on such studies., JRC.DG.I.6-Systems toxicology

Some data on production, toxicity, properties, uses, analytics as well as an environmental occurrence of PFCs in Poland are reviewed. In total 16 fluorochemicals were detected in surface water (Radunia River and Gulf of Gdańsk), beaver's liver (Warmia and Mazury region), cod and eider duck blood (Gulf of Gdańsk), young cattle blood (County of Stezyca) and human blood (Gdańsk Coast; donors which declared elevated Baltic fish intake) in Poland. In blood of the Gdańsk Coast inhabitants PFHxS, PFOS, PFOSA, PFHxA, PFHpA, PFOA, PFNA, PFDA, PFUnDA and PFDo-DA were found. In surface water for the first time were found fluorochemicals such as PFBuS, PFOcDA, PFBA and PFPeA, while in beavers' liver also PFTeA and N-Ethyl FOSA.

Aroclor 1268, Chlorofen, and Clophen T 64 technical chlorobiphenyl formulations were examined for 75 congeners of chlorodibenzo-p-dioxin (CDD) and 135 congeners of chlorodibenzofuran (CDF) using isotope dilution technique, separation, and enrichment on silica gel impregnated with activated carbon and final high resolution gas chromatography (HRGC)/high resolution mass spectrometry (HRMS) quantification. Three the most highly chlorinated congeners of CDD were found in Aroclor 1268, Chlorofen, and Clophen T 64. In the case of CDF, the number of congeners identified was 108 with 44 coeluting in pairs and 3 in triplicate in Aroclor 1268, 16 with 4 coeluting in pairs in Chlorofen, and 88 with 46 coeluting in pairs and 3 in triplicate in Clophen T 64. The total CDD and CDF concentrations of Aroclor 1268, Chlorofen, and Clophen T 64 were 24, 160, and 8.5 ng/g and 1600,270,000, and 4000 ng/g, respectively. No mono- to hexa-CDDs could be quantified in Aroclor 1268 (

 To identify key pro-fibrotic events downstream from PPARγ dysregulation that lead to fibrosis Background • Liver fibrosis is a wound healing response to a variety of chronic injuries including toxic injury from chemicals. • Hepatic stellate cells (HSCs) activation represents a key cellular event in the development of liver fibrosis that requires reprogramming of HSCs gene expression , orchestrated by the changes in the expression and/or the activity of key transcription regulators. • Peroxisome proliferator-activated receptor γ (PPAR γ) is a member of the nuclear hormone receptor family that has been shown to function as a key transcription regulator in the liver. • PPAR γ protein level is high in the quiescent HSCs and its expression and activity are reduced during HSCs activation. • While increased PPARγ expression in HSCs is essential for protection against liver fibrosis, its increased activity in hepatocytes is one of the key MIEs leading to steatosis (Fig. 1)

The European Union (EU) Cosmetics Regulation established the ban on animal testing for cosmetics ingredients. This ban does not assume that all cosmetics ingredients are safe, but that the non-testing procedures (in vitro and in silico) have to be applied for their safety assessment. To this end, the SEURAT-1 cluster was funded by EU 7th Framework Programme and Cosmetics Europe. The COSMOS (Integrated In Silico Models for the Prediction of Human Repeated Dose Toxicity of COSMetics to Optimise Safety) project was initiated as one of the seven consortia of the cluster, with the purpose of facilitating the prediction of human repeated dose toxicity associated with exposure to cosmetics-related compounds through in silico approaches. A critical objective of COSMOS was to address the paucity of publicly available data for cosmetics ingredients and related chemicals. Therefore a database was established containing (i) an inventory of cosmetics ingredients and related structures; (ii) skin permeability/absorption data (route of exposure relevant to cosmetics); and (iii) repeated dose toxicity data. This thesis describes the process of “knowledge discovery from the data”, including collation of the content of the COSMOS database and its subsequent application for developing tools to support the prediction of repeated dose toxicity of cosmetics and related compounds. A rigorous strategy of curation and quality control of chemical records was applied in developing the database (as documented in the Standard Operating Procedure, chapter 2). The chemical space of the cosmetics-related compounds was compared to food-related compounds from the U.S. FDA CFSAN PAFA database using the novel approach combining the analysis of structural features (ToxPrint chemotypes) and physicochemical properties. The cosmetics- and food- specific structural classes related to particular use functions and manifested by distinct physicochemical properties were identified (chapter 3). The novel COSMOS Skin Permeability Database containing in vivo and in vitro skin permeability/absorption data was developed by integrating existing databases and enriching them with new data for cosmetics harvested from regulatory documents and scientific literature (chapter 4). Compounds with available data on human in vitro maximal flux (JMAX) were subsequently extracted from the developed database and analysed in terms of their structural features (ToxPrint chemotypes) and physicochemical properties. The profile of compounds exhibiting low or high skin permeability potential was determined. The results of this analysis can support rapid screening and classification of the compounds without experimental data (chapter 5). The new COSMOS oral repeated dose toxicity database was established through consolidation of existing data sources and harvesting new regulatory documents and scientific literature. The unique data structure of the COSMOS oRepeatToxDB allows capturing all toxicological effects observed at particular dose levels and sites, which are hierarchically differentiated as organs, tissues, and cells (chapter 6). Such design of this database enabled the development of liver toxicity ontology, followed by mechanistic mining of in vivo data (chapter 7). As a result, compounds associated with liver steatosis, steatohepatitis and fibrosis phenotypic effects were identified and further analysed. The probable mechanistic reasoning for toxicity (Peroxisome Proliferator-Activated Receptor gamma (PPAR ) activation) was formulated for two hepatotoxicants, namely 1,3-bis-(2,4-diaminophenoxy)-propane and piperonyl butoxide. Key outcomes of this thesis include an extensive curated database, Standard Operating Procedures, skin permeability potential classification rules, and the set of structural features associated with liver steatosis. Such knowledge is particularly important in the light of the 21st Century Toxicology (NRC, 2007) and the ongoing need to move away from animal toxicity testing to non-testing alternatives.

Highlights • The historical development of the public COSMOS database is provided. • COSMOS NG is a knowledge hub to share toxicity data and in silico tools. • COSMOS NG has broad chemical coverage, with a focus on cosmetics. • Chemical and toxicological data are quality assured through inclusion criteria. • In silico TTC, profiling and read-across workflows are illustrated., The COSMOS Database (DB) was originally established to provide reliable data for cosmetics-related chemicals within the COSMOS Project funded as part of the SEURAT-1 Research Initiative. The database has subsequently been maintained and developed further into COSMOS Next Generation (NG), a combination of database and in silico tools, essential components of a knowledge base. COSMOS DB provided a cosmetics inventory as well as other regulatory inventories, accompanied by assessment results and in vitro and in vivo toxicity data. In addition to data content curation, much effort was dedicated to data governance – data authorisation, characterisation of quality, documentation of meta information, and control of data use. Through this effort, COSMOS DB was able to merge and fuse data of various types from different sources. Building on the previous effort, the COSMOS Minimum Inclusion (MINIS) criteria for a toxicity database were further expanded to quantify the reliability of studies. COSMOS NG features multiple fingerprints for analysing structure similarity, and new tools to calculate molecular properties and screen chemicals with endpoint-related public profilers, such as DNA and protein binders, liver alerts and genotoxic alerts. The publicly available COSMOS NG enables users to compile information and execute analyses such as category formation and read-across. This paper provides a step-by-step guided workflow for a simple read-across case, starting from a target structure and culminating in an estimation of a NOAEL confidence interval. Given its strong technical foundation, inclusion of quality-reviewed data, and provision of tools designed to facilitate communication between users, COSMOS NG is a first step towards building a toxicological knowledge hub leveraging many public data systems for chemical safety evaluation. We continue to monitor the feedback from the user community at support@mn-am.com.

This paper reviews in silico models currently available for the prediction of skin permeability with the main focus on the quantitative structure-permeability relationship (QSPR) models. A comprehensive analysis of the main achievements in the field in the last decade is provided. In addition, the mechanistic models are discussed and comparative studies that analyse different models are discussed.