Accounting for data variability, a key factor in in vivo/in vitro relationships: application to the skin sensitization potency (in vivo LLNA versus in vitro DPRA) example.

When searching for alternative methods to animal testing, confidently rescaling an in vitro result to the corresponding in vivo classification is still a challenging problem. Although one of the most important factors affecting good correlation is sample characteristics, they are very rarely integrated into correlation studies. Usually, in these studies, it is implicitly assumed that both compared values are error-free numbers, which they are not. In this work, we propose a general methodology to analyze and integrate data variability and thus confidence estimation when rescaling from one test to another. The methodology is demonstrated through the case study of rescaling the in vitro Direct Peptide Reactivity Assay (DPRA) reactivity to the in vivo Local Lymph Node Assay (LLNA) skin sensitization potency classifications. In a first step, a comprehensive statistical analysis evaluating the reliability and variability of LLNA and DPRA as such was done. These results allowed us to link the concept of gray zones and confidence probability, which in turn represents a new perspective for a more precise knowledge of the classification of chemicals within their in vivo OR in vitro test. Next, the novelty and practical value of our methodology introducing variability into the threshold optimization between the in vitro AND in vivo test resides in the fact that it attributes a confidence probability to the predicted classification. The methodology, classification and screening approach presented in this study are not restricted to skin sensitization only. They could be helpful also for fate, toxicity and health hazard assessment where plenty of in vitro and in chemico assays and/or QSARs models are available. Copyright © 2016 John Wiley & Sons, Ltd.

Evaluation of QSAR models for predicting mutagenicity: outcome of the Second Ames/QSAR international challenge project.

Quantitative structure-activity relationship (QSAR) models are powerful in silico tools for predicting the mutagenicity of unstable compounds, impurities and metabolites that are difficult to examine using the Ames test. Ideally, Ames/QSAR models for regulatory use should demonstrate high sensitivity, low false-negative rate and wide coverage of chemical space. To promote superior model development, the Division of Genetics and Mutagenesis, National Institute of Health Sciences, Japan (DGM/NIHS), conducted the Second Ames/QSAR International Challenge Project (2020-2022) as a successor to the First Project (2014-2017), with 21 teams from 11 countries participating. The DGM/NIHS provided a curated training dataset of approximately 12,000 chemicals and a trial dataset of approximately 1,600 chemicals, and each participating team predicted the Ames mutagenicity of each trial chemical using various Ames/QSAR models. The DGM/NIHS then provided the Ames test results for trial chemicals to assist in model improvement. Although overall model performance on the Second Project was not superior to that on the First, models from the eight teams participating in both projects achieved higher sensitivity than models from teams participating in only the Second Project. Thus, these evaluations have facilitated the development of QSAR models.

The use of metabolising systems for in vitro testing of endocrine disruptors.

Legislation and prospective legislative proposals in for instance the USA, Europe, and Japan require, or may require that chemicals are tested for their ability to disrupt the hormonal systems of mammals. Chemicals found to test positive are considered to be endocrine active substances (EAS) and may be putative endocrine disruptors (EDs). To date, there is still little or no experience with incorporating metabolic and toxicokinetic aspects into in vitro tests for EAS. This is a situation in sharp contrast to genotoxicity testing, where in vitro tests are routinely conducted with and without metabolic capacity. Originally prepared for the Organisation of Economic Cooperation and Development (OECD), this detailed review paper reviews why in vitro assays for EAS should incorporate mammalian systems of metabolism and metabolic enzyme systems, and indicates how this could be done. The background to ED testing, the available test methods, and the role of mammalian metabolism in the activation and the inactivation of both endogenous and exogenous steroids are described. The available types of systems are compared, and the potential problems in incorporating systems in in vitro tests for EAS, and how these might be overcome, are discussed. Lastly, some recommendations for future activities are made.

QSAR and mechanistic interpretation of estrogen receptor binding.

A multi-dimensional formulation of the COmmon REactivity PAttern (COREPA) modeling approach has been used to investigate chemical binding to the human estrogen receptor (hER). A training set of 645 chemicals included 497 steroid and environmental chemicals (database of the Chemical Evaluation and Research Institute, Japan - CERI) and 148 chemicals to further explore hER-structure interactions (selected J. Katzenellenbogen references). Upgrades of modeling approaches were introduced for multivariate COREPA analysis, optimal conformational generation and description of the local hydrophobicity of chemicals. Analysis of reactivity patterns based on the distance between nucleophilic sites resulted in identification of distinct interaction types: a steroid-like A-B type described by frontier orbital energies and distance between nucleophilic sites with specific charge requirements; an A-C type where local hydrophobic effects are combined with electronic interactions to modulate binding; and mixed A-B-C (AD) type. Chemicals were grouped by type, then COREPA models were developed for within specific relative binding affinity ranges of >10%, 10 > RBA > or = 0.1%, and 0.1 > RBA > 0.0%. The derived models for each interaction type and affinity range combined specific prefiltering requirements (interatomic distances) and a COREPA classification node using no more than 2 discriminating parameters. The interaction types are becoming less distinct in the lowest activity range for each chemicals of each type; here, the modeling was performed within chemical classes (phenols, phthalates, etc.). The ultimate model was organized as a battery of local models associated to interaction type and mechanism.

A kinetic model for predicting biodegradation.

Biodegradation plays a key role in the environmental risk assessment of organic chemicals. The need to assess biodegradability of a chemical for regulatory purposes supports the development of a model for predicting the extent of biodegradation at different time frames, in particular the extent of ultimate biodegradation within a '10 day window' criterion as well as estimating biodegradation half-lives. Conceptually this implies expressing the rate of catabolic transformations as a function of time. An attempt to correlate the kinetics of biodegradation with molecular structure of chemicals is presented. A simplified biodegradation kinetic model was formulated by combining the probabilistic approach of the original formulation of the CATABOL model with the assumption of first order kinetics of catabolic transformations. Nonlinear regression analysis was used to fit the model parameters to OECD 301F biodegradation kinetic data for a set of 208 chemicals. The new model allows the prediction of biodegradation multi-pathways, primary and ultimate half-lives and simulation of related kinetic biodegradation parameters such as biological oxygen demand (BOD), carbon dioxide production, and the nature and amount of metabolites as a function of time. The model may also be used for evaluating the OECD ready biodegradability potential of a chemical within the '10-day window' criterion.

Model-predicted occurrence of multiple pharmaceuticals in Swedish surface waters and their flushing to the Baltic Sea.

An exposure assessment for multiple pharmaceuticals in Swedish surface waters was made using the STREAM-EU model. Results indicate that Metformin (27 ton/y), Paracetamol (6.9 ton/y) and Ibuprofen (2.33 ton/y) were the drugs with higher amounts reaching the Baltic Sea in 2011. 35 of the studied substances had more than 1 kg/y of predicted flush to the sea. Exposure potential given by the ratio amount of the drug exported to the sea/amount emitted to the environment was higher than 50% for 7 drugs (Piperacillin, Lorazepam, Metformin, Hydroxycarbamide, Hydrochlorothiazide, Furosemide and Cetirizine), implying that a high proportion of them will reach the sea, and below 10% for 27 drugs, implying high catchment attenuation. Exposure potentials were found to be dependent of persistency and hydrophobicity of the drugs. Chemicals with Log D > 2 had exposure potentials <10% regardless of their persistence. Chemicals with Log D  <  -2 had exposure potentials >35% with higher ratios typically achieved for longer half-lives. For Stockholm urban area, 17 of the 54 pharmaceuticals studied had calculated concentrations higher than 10 ng/L. Model agreement with monitored values had an r2 = 0.62 for predicted concentrations and an r2 = 0.95 for predicted disposed amounts to sea.

CATALOGIC 301C model - validation and improvement.

In Europe, REACH legislation encourages the use of alternative in silico methods such as (Q)SAR models. According to the recent progress of Chemical Substances Control Law (CSCL) in Japan, (Q)SAR predictions are also utilized as supporting evidence for the assessment of bioaccumulation potential of chemicals along with read across. Currently, the effective use of read across and QSARs is examined for other hazards, including biodegradability. This paper describes the results of external validation and improvement of CATALOGIC 301C model based on more than 1000 tested new chemical substances of the publication schedule under CSCL. CATALOGIC 301C model meets all REACH requirements to be used for biodegradability assessment. The model formalism built on scientific understanding for the microbial degradation of chemicals has a well-defined and transparent applicability domain. The model predictions are adequate for the evaluation of the ready degradability of chemicals.

Metabolic activation of chemicals: in-silico simulation.

The role of metabolism in prioritising chemicals according to their potential adverse health effects is extremely important given the fact that innocuous parents can be transformed into toxic metabolites. Our recent efforts in simulating metabolic activation of chemicals are reviewed in this work. The application of metabolic simulators to predict biodegradation (microbial degradation pathways), bioaccumulation (fish liver metabolism), skin sensitisation (skin metabolism), mutagenicity (rat liver S-9 metabolism) are discussed. The ability of OASIS approach to predict metabolism (toxicokinetics) and toxicity (toxicodynamics) of chemicals resulting from their metabolic activation in a single modelling platform is an important advantage of the method. It allows prioritisation of chemicals due to predicted toxicity of their metabolites.

Evaluation of human pharmaceutical emissions and concentrations in Swedish river basins.

An emissions inventory for top consumed human pharmaceuticals in Sweden was done based on national consumption data, human metabolic rates and wastewater treatment removal rates. Concentrations of pharmaceuticals in surface waters in Swedish river basins were predicted using estimated emissions from the inventory and river discharges. Our findings indicate that the top ten emitted pharmaceuticals in our study set of 54 substances are all emitted in amounts above 0.5ton/y to both surface waters and soils. The highest emissions to water were in decreasing order for Metformin, Furosemide, Gabapentin, Atenolol and Tramadol. Predicted emissions to soils calculated with the knowledge that in Sweden sludge is mostly disposed to soil, point to the highest emissions among the studied drugs coming from, in decreasing order, Metformin, Paracetamol, Ibuprofen, Gabapentin and Atenolol. Surface water concentrations in Sweden's largest rivers, all located in low density population zones, were found to be below 10ng/L for all substances studied. In contrast, concentrations in surface waters in Stockholm's metropolitan area, the most populous in Sweden, surpassed 100ng/L for four substances: Atenolol, Metformin, Furosemide and Gabapentin.

QSAR Toolbox - workflow and major functionalities.

The OECD QSAR Toolbox is a software application intended to be used by governments, the chemical industry and other stakeholders in filling gaps in (eco)toxicity data needed for assessing the hazards of chemicals. The development and release of the Toolbox is a cornerstone in the computerization of hazard assessment, providing an 'all inclusive' tool for the application of category approaches, such as read-across and trend analysis, in a single software application, free of charge. The Toolbox incorporates theoretical knowledge, experimental data and computational tools from various sources into a logical workflow. The main steps of this workflow are substance identification, identification of relevant structural characteristics and potential toxic mechanisms of interaction (i.e. profiling), identification of other chemicals that have the same structural characteristics and/or mechanism (i.e. building a category), data collection for the chemicals in the category and use of the existing experimental data to fill the data gap(s). The description of the Toolbox workflow and its main functionalities is the scope of the present article.

Base-line model for identifying the bioaccumulation potential of chemicals.

The base-line modeling concept presented in this work is based on the assumption of a maximum bioconcentration factor (BCF) with mitigating factors that reduce the BCF. The maximum bioconcentration potential was described by the multi-compartment partitioning model for passive diffusion. The significance of different mitigating factors associated either with interactions with an organism or bioavailability were investigated. The most important mitigating factor was found to be metabolism. Accordingly, a simulator for fish liver was used in the model, which has been trained to reproduce fish metabolism based on related mammalian metabolic pathways. Other significant mitigating factors, depending on the chemical structure, e.g. molecular size and ionization were also taken into account in the model. The results (r(2)=0.84) obtained for a training set of 511 chemicals demonstrate the usefulness of the BCF base line concept. The predictability of the model was evaluated on the basis of 176 chemicals not used in the model building. The correctness of predictions (abs(logBSF(Obs)-logBCF(Calc))=0.75)) for 59 chemicals included within the model applicability domain was 80%.

POPs: a QSAR system for developing categories for persistent, bioaccumulative and toxic chemicals and their metabolites.

This paper presents the framework of a QSAR-based decision support system which provides a rapid screening of potential hazards, classification of chemicals with respect to risk management thresholds, and estimation of missing data for the early stages of risk assessment. At the simplest level, the framework is designed to rank hundreds of chemicals according to their profile of persistence, bioaccumulation potential and toxicity often called the persistent organic pollutant (POP) profile or the PBT (persistent bioaccumulative toxicant) profile. The only input data are the chemical structure. The POPs framework enables decision makers to introduce the risk management thresholds used in the classification of chemicals under various authorities. Finally, the POPs framework advances hazard identification by integrating a metabolic simulator that generates metabolic map for each parent chemical. Both the parent chemicals and plausible metabolites are systematically evaluated for metabolic activation and POPs profile.

A QSAR evaluation of Ah receptor binding of halogenated aromatic xenobiotics.

Because of their widespread occurrence and substantial biological activity, halogenated aromatic hydrocarbons such as polychlorinated biphenyls (PCBs), polychlorinated dibenzofurans (PCDFs), and polychlorinated dibenzo-p-dioxins (PCDDs) comprise one of the more important classes of contaminants in the environment. Some chemicals in this class cause adverse biological effects after binding to an intracellular cytosolic protein called the aryl hydrocarbon receptor (AhR). Toxic responses such as thymic atrophy, weight loss, immunotoxicity, and acute lethality, as well as induction of cytochrome P4501A1, have been correlated with the relative affinity of PCBs, PCDFs, and PCDDs for the AhR. Therefore, an important step in predicting the effects of these chemicals is the estimation of their binding to the receptor. To date, however, the use of quantitative structure activity relationship (QSAR) models to estimate binding affinity across multiple chemical classes has shown only modest success possibly due, in part, to a focus on minimum energy chemical structures as the active molecules. In this study, we evaluated the use of structural conformations other than those of minimum energy for the purpose of developing a model for AhR binding affinity that encompasses more of the halogenated aromatic chemicals known to interact with the receptor. Resultant QSAR models were robust, showing good utility across multiple classes of halogenated aromatic compounds.

A computationally based identification algorithm for estrogen receptor ligands: part 1. Predicting hERalpha binding affinity.

The common reactivity pattern (COREPA) approach is a 3-dimensional, quantitative structure activity relationship (3-D QSAR) technique that permits identification and quantification of specific global and local stereoelectronic characteristics associated with a chemical's biological activity. It goes beyond conventional 3-D QSAR approaches by incorporating dynamic chemical conformational flexibility in ligand-receptor interactions. The approach provides flexibility in screening chemical data sets in that it helps establish criteria for identifying false positives and false negatives, and is not dependent upon a predetermined and specified toxicophore or an alignment of conformers to a lead compound. The algorithm was recently used to screen chemical data sets for rat androgen receptor binding affinity. To further explore the potential application of the algorithm in establishing reactivity patterns for human estrogen receptor alpha (hERalpha) binding affinity, the stereoelectronic requirements associated with the binding affinity of 45 steroidal and nonsteroidal ligands to the receptor were defined. Reactivity patterns for relative hERalpha binding affinity (RBA; 17ss-estradiol = 100%) were established based on global nucleophilicity, interatomic distances between electronegative heteroatoms, and electron donor capability of heteroatoms. These reactivity patterns were used to establish descriptor profiles for identifying and ranking compounds with RBA of > 150%, 100-10%, 10-1%, and 1-0.1%. Increasing specificity of reactivity patterns was detected for ligand data sets with RBAs above 10%. Using the results of this analysis, an exploratory expert system was developed for use in ranking relative ER binding affinity potential for large chemical data sets.

A computationally based identification algorithm for estrogen receptor ligands: part 2. Evaluation of a hERalpha binding affinity model.

The objective of this study was to evaluate the capability of an expert system described in the previous paper (S. Bradbury et al., Toxicol. Sci. 58, 253-269) to identify the potential for chemicals to act as ligands of mammalian estrogen receptors (ERs). The basis of the expert system was a structure activity relationship (SAR) model, based on relative binding affinity (RBA) values for steroidal and nonsteroidal chemicals derived from human ERalpha (hERalpha) competitive binding assays. The expert system enables categorization of chemicals into (RBA ranges of < 0.1, 0.1 to 1, 1 to 10, 10 to 100, and >150% relative to 17ss-estradiol. In the current analysis, the algorithm was evaluated with respect to predicting RBAs of chemicals assayed with ERs from MCF7 cells, and mouse and rat uterine preparations. The best correspondence between predicted and observed RBA ranges was obtained with MCF7 cells. The agreement between predictions from the expert system and data from binding assays with mouse and rat ER(s) were less reliable, especially for chemicals with RBAs less than 10%. Prediction errors often were false positives, i.e., predictions of greater than observed RBA values. While discrepancies were likely due, in part, to species-specific variations in ER structure and ligand binding affinity, a systematic bias in structural characteristics of chemicals in the hERalpha training set, compared to the rodent evaluation data sets, also contributed to prediction errors. False-positive predictions were typically associated with ligands that had shielded electronegative sites. Ligands with these structural characteristics were not well represented in the training set used to derive the expert system. Inclusion of a shielding criterion into the original expert system significantly increased the accuracy of RBA predictions. With this additional structural requirement, 38 of 46 compounds with measured RBA values greater than 10% in hERalpha, MCF7, and rodent uterine preparations were correctly categorized. Of the remaining 129 compounds in the combined data sets, RBA values for 65 compounds were correctly predicted, with 47 of the incorrect predictions being false positives. Based upon this exploratory analysis, the modeling approach, combined with a high-quality training set of RBA values derived from a diverse set of chemical structures, could provide a credible tool for prioritizing chemicals with moderate to high ER binding affinity for subsequent in vitro or in vivo assessments.

Relationships between descriptors for hydrophobicity and soft electrophilicity in predicting toxicity.

The toxicity of chemicals is orthogonal with individual molecular descriptors used to quantify hydrophobicity and soft electrophilicity when considering large data sets. Estimating the toxicity of reactive chemicals requires descriptors of both passive transport and the stereoelectronic interaction, which are largely independent processes. QSARs using either log P or an electronic parameter alone are only significant for sets of chemicals that represent special, albeit some important, cases in QSAR. Chemicals were clustered according to their reactivity as soft electrophiles by defining isoelectrophilic windows along the toxicity response surface. Within these narrow windows of reactivity, the variation of toxicity was explained by the variation of log P. We observed that the dependence of toxicity on log P in different isoelectrophilic windows decreased as reactivity increased. The data are consistent with toxicity models where competing nucleophilic interaction sites are distributed along the transport route of the chemicals.

A reactivity pattern for discrimination of ER agonism and antagonism based on 3-D molecular attributes.

Various models have been developed to predict the relative binding affinity (RBA) of chemicals to estrogen receptors (ER). These models can be used to prioritize chemicals for further tiered biological testing to assess the potential for endocrine disruption. One shortcoming of models predicting RBA has been the inability to distinguish potential receptor antagonism from agonism, and hence in vivo response. It has been suggested that steroid receptor antagonists are less compact than agonists; thus, ER binding of antagonists may prohibit proper alignment of receptor protein helices preventing subsequent transactivation. The current study tests the theory of chemical bulk as a defining parameter of antagonism by employing a 3-D structural approach for development of reactivity patterns for ER antagonists and agonists. Using a dataset of 23 potent ER ligands (16 agonists, 7 antagonists), molecular parameters previously found to be associated with ER binding affinity, namely global (E(HOMO)) and local (donor delocalizabilities and charges) electron donating ability of electronegative sites and steric distances between those sites, were found insufficient to discriminate ER antagonists from agonists. However, parameters related to molecular bulk, including solvent accessible surface and negatively charged Van der Waal's surface, provided reactivity patterns that were 100% successful in discriminating antagonists from agonists in the limited data set tested. The model also shows potential to discriminate pure antagonists from partial agonist/antagonist structures. Using this exploratory model it is possible to predict additional chemicals for their ability to bind but inactivate the ER, providing a further tool for hypothesis testing to elucidate chemical structural characteristics associated with estrogenicity and anti-estrogenicity.

Androgen receptor binding affinity of pesticide "active" formulation ingredients. QSAR evaluation by COREPA method.

The COREPA approach for identifying the COmmon REactivity PAttern of biologically similar chemicals was employed to upgrade the recently derived affinity pattern for high androgen receptor (AR) binding affinity. The training set consisted of 28 steroidal and nonsteroidal ligands whose AR binding affinity was determined in competitive binding assays (in terms of pKi). The interatomic distances between nucleophilic sites and their charges providing distinct and non-overlapping integral patterns for active and inactive chemicals were assumed that it was related with the endpoint, which was under study. These stereoelectronic characteristics were used to predict pKi values of pesticide "active" formulation ingredients in an attempt to identify chemicals with potential AR binding affinity.

The electronic factor in QSAR: MO-parameters, competing interactions, reactivity and toxicity.

Reactive chemicals pose unique problems in the development of SAR and QSAR in environmental chemistry and toxicology. Models of the stereoelectronic interactions of reactive toxicants with biological systems require formulation of parameters that quantify the electronic structure of the chemicals. A review of early approaches to modeling reactivity is presented in this work, with emphasis on the generalized polyelectronic perturbation theory. Applications of GPPT are demonstrated with QSARs for predicting toxicity of soft electrophiles and proelectrophiles using superdelocalizability and the charges on frontier orbitals. Prediction of toxicity for hard electrophiles such as organophosphates require atomic charges and bond orders in the QSAR. Special considerations for the orthogonality of factors and for the classification of reactive chemicals are reviewed.

Probabilistic assessment of biodegradability based on metabolic pathways: catabol system.

A novel mechanistic modeling approach has been developed that assesses chemical biodegradability in a quantitative manner. It is an expert system predicting biotransformation pathway working together with a probabilistic model that calculates probabilities of the individual transformations. The expert system contains a library of hierarchically ordered individual transformations and matching substructure engine. The hierarchy in the expert system was set according to the descending order of the individual transformation probabilities. The integrated principal catabolic steps are derived from set of metabolic pathways predicted for each chemical from the training set and encompass more than one real biodegradation step to improve the speed of predictions. In the current work, we modeled O2 yield during OECD 302 C (MITI I) test. MITI-I database of 532 chemicals was used as a training set. To make biodegradability predictions, the model only needs structure of a chemical. The output is given as percentage of theoretical biological oxygen demand (BOD). The model allows for identifying potentially persistent catabolic intermediates and their molar amounts. The data in the training set agreed well with the calculated BODs (r2 = 0.90) in the entire range i.e. a good fit was observed for readily, intermediate and difficult to degrade chemicals. After introducing 60% ThOD as a cut off value the model predicted correctly 98% ready biodegradable structures and 96% not ready biodegradable structures. Crossvalidation by four times leaving 25% of data resulted in Q2 = 0.88 between observed and predicted values. Presented approach and obtained results were used to develop computer software for biodegradability prediction CATABOL.