Utility of In Vitro Bioactivity as a Lower Bound Estimate of In Vivo Adverse Effect Levels and in Risk-Based Prioritization

Use of high-throughput, in vitro bioactivity data in setting a point-of-departure (POD) has the potential to accelerate the pace of human health safety evaluation by informing screening-level assessments. The primary objective of this work was to compare PODs based on high-throughput predictions of bioactivity, exposure predictions, and traditional hazard information for 448 chemicals. PODs derived from new approach methodologies (NAMs) were obtained for this comparison using the 50th (PODNAM, 50) and the 95th (PODNAM, 95) percentile credible interval estimates for the steady-state plasma concentration used in in vitro to in vivo extrapolation of administered equivalent doses. Of the 448 substances, 89% had a PODNAM, 95 that was less than the traditional POD (PODtraditional) value. For the 48 substances for which PODtraditional < PODNAM, 95, the PODNAM and PODtraditional were typically within a factor of 10 of each other, and there was an enrichment of chemical structural features associated with organophosphate and carbamate insecticides. When PODtraditional < PODNAM, 95, it did not appear to result from an enrichment of PODtraditional based on a particular study type (eg, developmental, reproductive, and chronic studies). Bioactivity:exposure ratios, useful for identification of substances with potential priority, demonstrated that high-throughput exposure predictions were greater than the PODNAM, 95 for 11 substances. When compared with threshold of toxicological concern (TTC) values, the PODNAM, 95 was greater than the corresponding TTC value 90% of the time. This work demonstrates the feasibility, and continuing challenges, of using in vitro bioactivity as a protective estimate of POD in screening-level assessments via a case study.

The using of the Index of Ideality of Correlation (IIC) to improve predictive potential of models of water solubility for pesticides

Models for water solubility of pesticides suggested in this manuscript are important data from point of view of ecologic engineering. The Index of Ideality of Correlation (IIC) of groups of quantitative structure-property relationships (QSPRs) for water solubility of pesticides related to the calibration sets was used to identify good in silico models. This comparison confirmed the high IIC set provides better statistical quality of the model for the validation set. Though there are large databases on solubility, the reliable prediction of the endpoint for new substances which are potential pesticides is an important ecologic task. Unfortunately, predictive models for various endpoints suffer overtraining, and the IIC serves to avoid or at least reduce this. Thus, the approach suggested has both theoretical and economic effects for ecology.

Predicting acute contact toxicity of organic binary mixtures in honey bees (A. mellifera) through innovative QSAR models

Pollinators such as honey bees are of considerable importance, because of the crucial pollination services they provide for food crops and wild plants. Since bees are exposed to a wide range of multiple chemicals "mixtures" both of anthropogenic (e.g. plant protection products) and natural origin (e.g. plant toxins), understanding their combined toxicity is critical. Although honey bees are employed worldwide as surrogate species for Apis and non-Apis bees in toxicity tests, it is practically unfeasible to perform in vivo tests for all mixtures of chemicals. Therefore, Quantitative Structure-Activity Relationships (QSAR) models can be developed using available data and can provide useful tools to predict such combined toxicity. Here, three different QSAR models within the CORAL software have been calibrated and validated for honey bees (A. mellifera) to predict the acute contact mixtures potency (LD_50-mix), in two regression based-models, and the nature of combined toxicity (synergism / non-synergism) in a classification-based model. Experimental data on binary mixtures (n = 123) (LD_50-mix) including dose response data (n = 97) and corresponding Toxic Unit values were retrieved from EFSA databases. The models were built using the principle of extraction of attributes from SMILES (or quasi-SMILES) while calculating so-called correlation weights for these attributes using Monte Carlo techniques. The two regression models were validated for their reliability and robustness (R² = 0.89, CCC = 0.92, Q² = 0.81; R² = 0.87, CCC = 0.89, Q² = 0.75). The classification model was validated using sensitivity (=0.86), specificity (=1), accuracy (=0.96), and Matthews correlation coefficient (MCC = 0.90) as qualitative statistical validation parameters. Results indicate that these QSAR models successfully predict acute contact toxicity of binary mixtures in honey bees and can support prioritisation of multiple chemicals of concerns. Data gaps and further development of QSAR models for honey bees are highlighted particularly for chronic and sub-lethal effects.

Integrating in silico models and read-across methods for predicting toxicity of chemicals: A step-wise strategy

In silico methods and models are increasingly used for predicting properties of chemicals for hazard identification and hazard characterisation in the absence of experimental toxicity data. Many in silico models are available and can be used individually or in an integrated fashion. Whilst such models offer major benefits to toxicologists, risk assessors and the global scientific community, the lack of a consistent framework for the integration of in silico results can lead to uncertainty and even contradictions across models and users, even for the same chemicals. In this context, a range of methods for integrating in silico results have been proposed on a statistical or case-specific basis. Read-across constitutes another strategy for deriving reference points or points of departure for hazard characterisation of untested chemicals, from the available experimental data for structurally-similar compounds, mostly using expert judgment. Recently a number of software systems have been developed to support experts in this task providing a formalised and structured procedure. Such a procedure could also facilitate further integration of the results generated from in silico models and read-across. This article discusses a framework on weight of evidence published by EFSA to identify the stepwise approach for systematic integration of results or values obtained from these "non-testing methods". Key criteria and best practices for selecting and evaluating individual in silico models are also described, together with the means to combining the results, taking into account any limitations, and identifying strategies that are likely to provide consistent results.

Pre- and Postharvest Strategies to Minimize Mycotoxin Contamination in the Rice Food Chain

Rice is part of many people's diet around the world, being the main energy source in some regions. Although fewer reports exist on the occurrence of mycotoxins in rice compared to other cereals, fungal contamination and the associated production of toxic metabolites, even at lower occurrence levels compared to other crops, are of concern because of the high consumption of rice in many countries. Due to the diversity of fungi that may contaminate the rice food chain, the co-occurrence of mycotoxins is frequent. Specific strategies to overcome these problems may be applied at the preharvest part of the crop chain, while assuring good practices at harvest and postharvest stages, since different fungi may find suitable conditions to grow at the various stages of the production chain. Therefore, the aim of this review is to present the state-of-the-art knowledge on such strategies in an integrated way, from the field to the final products, to reduce mycotoxin contamination in rice.

SAR for gastro-intestinal absorption and blood-brain barrier permeation of pesticides

The CORAL software has been applied to the development of classification models for pesticides relative to two pharmacokinetic properties in humans: (i) gastro-intestinal absorption; and (ii) blood-brain barrier permeation. These models were built up using categorical data on pesticide absorption and brain permeation split into training, invisible training, calibration and external validation sets using Monte Carlo simulations. The models were assessed using several random splits into the training and validation sets. Optimal SMILES-based descriptors sensitive to the presence of different chemical elements and types of covalent bonds have been used to build up models. The range of Matthews correlation coefficient for suggested models is 0.64-0.75. The perspectives of studied approach as a tool to build up models for pharmacokinetic properties of chemicals is discussed. The models are built up according to OECD principles.

QSAR models for predicting acute toxicity of pesticides in rainbow trout using the CORAL software and EFSA's OpenFoodTox database

Optimal (flexible) descriptors were used to establish quantitative structure - activity relationships (QSAR) for toxicity of pesticides (n=116) towards rainbow trout. A heterogeneous set of hundreds of pesticides has been used, taken from the EFSA's chemical Hazards Database: OpenFoodTox. Optimal descriptors are preparing from simplified molecular input-line entry system (SMILES). So-called, correlation weights of different fragments of SMILES are calculating by the Monte Carlo optimization procedure where correlation coefficient between endpoint and optimal descriptor plays role of the target function. Having maximum of the correlation coefficient for the training set, one can suggest that the optimal descriptor calculated with these correlation weights can correlate with endpoint for external validation set. This approach was checked up with three different distributions into the training (≈85%) set and external validation (≈15%) set. The statistical characteristics of these models are (i) for training set correlation coefficient (r²) ranges 0.72-0.81, and root mean squared error (RMSE) ranges 0.54-1.25; (ii) for external (validation) set r² ranges 0.74-0.84; and RMSE ranges 0.64-0.75. Computational experiments have shown that presence of chlorine, fluorine, sulfur, and aromatic fragments is promoter of increase for the toxicity.

Comparing bee species responses to chemical mixtures: Common response patterns?

Pollinators in agricultural landscapes can be exposed to mixtures of pesticides and environmental pollutants. Existing mixture toxicity modelling approaches, such as the models of concentration addition and independent action and the mechanistic DEBtox framework have been previously shown as valuable tools for understanding and ultimately predicting joint toxicity. Here we apply these mixture models to investigate the potential to interpret the effects of semi-chronic binary mixture exposure for three bee species: Apis mellifera, Bombus terrestris and Osmia bicornis within potentiation and mixture toxicity experiments. In the potentiation studies, the effect of the insecticide dimethoate with added propiconazole fungicide and neonicotinoid insecticide clothianidin with added tau-fluvalinate pyrethroid acaricide showed no difference in toxicity compared to the single chemical alone. Clothianidin toxicity showed a small scale, but temporally conserved increase in exposure conducted in the presence of propiconazole, particularly for B. terrestris and O. bicornis, the latter showing a near three-fold increase in clothianidin toxicity in the presence of propiconazole. In the mixture toxicity studies, the dominant response patterns were of additivity, however, binary mixtures of clothianidin and dimethoate in A. mellifera, B. terrestris and male O. bicornis there was evidence of a predominant antagonistic interaction. Given the ubiquitous nature of exposures to multiple chemicals, there is an urgent need to consider mixture effects in pollinator risk assessments. Our analyses suggest that current models, particularly those that utilise time-series data, such as DEBtox, can be used to identify additivity as the dominant response pattern and also those examples of interactions, even when small-scale, that may need to be taken into account during risk assessment.

The application of new HARD-descriptor available from the CORAL software to building up NOAEL models

Continuous QSAR models have been developed and validated for the prediction of no-observed-adverse-effect (NOAEL) in rats, using training and test sets from the Fraunhofer RepDose® database and EFSA's Chemical Hazards Database: OpenFoodTox. This paper demonstrates that the HARD index, as an integrated attribute of SMILES, improves the prediction power of NOAEL values using the continuous QSAR models and Monte Carlo simulations. The HARD-index is a line of eleven symbols, which represents the presence, or absence of eight chemical elements (nitrogen, oxygen, sulfur, phosphorus, fluorine, chlorine, bromine, and iodine) and different kinds of chemical bonds (double bond, triple bond, and stereo chemical bond). Optimal molecular descriptors calculated with the Monte Carlo technique (maximization of correlation coefficient between the descriptor and endpoint) give satisfactory predictive models for NOAEL. Optimal molecular descriptors calculated in this way with the Monte Carlo technique (maximization of correlation coefficient between the descriptor and endpoint) give amongst the best results available in the literature. The models are built up in accordance with OECD principles.

Predicting acute contact toxicity of pesticides in honeybees (Apis mellifera) through a k-nearest neighbor model

Ecological risk assessment of plant protection products (PPPs) requires an understanding of both the toxicity and the extent of exposure to assess risks for a range of taxa of ecological importance including target and non-target species. Non-target species such as honey bees (Apis mellifera), solitary bees and bumble bees are of utmost importance because of their vital ecological services as pollinators of wild plants and crops. To improve risk assessment of PPPs in bee species, computational models predicting the acute and chronic toxicity of a range of PPPs and contaminants can play a major role in providing structural and physico-chemical properties for the prioritisation of compounds of concern and future risk assessments. Over the last three decades, scientific advisory bodies and the research community have developed toxicological databases and quantitative structure-activity relationship (QSAR) models that are proving invaluable to predict toxicity using historical data and reduce animal testing. This paper describes the development and validation of a k-Nearest Neighbor (k-NN) model using in-house software for the prediction of acute contact toxicity of pesticides on honey bees. Acute contact toxicity data were collected from different sources for 256 pesticides, which were divided into training and test sets. The k-NN models were validated with good prediction, with an accuracy of 70% for all compounds and of 65% for highly toxic compounds, suggesting that they might reliably predict the toxicity of structurally diverse pesticides and could be used to screen and prioritise new pesticides.

Population effects and variability

Chemical risk assessment for human health requires a multidisciplinary approach through four steps: hazard identification and characterization, exposure assessment, and risk characterization. Hazard identification and characterization aim to identify the metabolism and elimination of the chemical (toxicokinetics) and the toxicological dose-response (toxicodynamics) and to derive a health-based guidance value for safe levels of exposure. Exposure assessment estimates human exposure as the product of the amount of the chemical in the matrix consumed and the consumption itself. Finally, risk characterization evaluates the risk of the exposure to human health by comparing the latter to with the health-based guidance value. Recently, many research efforts in computational toxicology have been put together to characterize population variability and uncertainty in each of the steps of risk assessment to move towards more quantitative and transparent risk assessment. This chapter focuses specifically on modeling population variability and effects for each step of risk assessment in order to provide an overview of the statistical and computational tools available to toxicologists and risk assessors. Three examples are given to illustrate the applicability of those tools: derivation of pathway-related uncertainty factors based on population variability, exposure to dioxins, dose-response modeling of cadmium.

Human variability in glucuronidation in relation to uncertainty factors for risk assessment

The appropriateness of the default uncertainty factor for human variability in kinetics has been investigated for glucuronidation using an extensive database of substrates metabolised primarily by this pathway. Inter-individual variability was quantified for 15 compounds from published pharmacokinetic studies (after oral and intravenous dosing) in healthy adults and other subgroups using parameters relating to chronic exposure (metabolic and total clearances, area under the plasma concentration time-curve (AUC)) and acute exposure (C(max)). Low inter-individual variability (about 30-35%) was found for all parameters (clearance corrected or not corrected for body weight, metabolic clearance, oral AUC and C(max)) after either iv or oral administration to healthy adults. The overall variability of 31% for glucuronidation in healthy adults supported the validity of the default kinetic uncertainty factor of 3.16 for this group, because it would cover more than 99% of individuals. Comparisons between potentially sensitive subgroups and healthy adults using differences in means and variability indicated that neonates showed the greatest impairment of glucuronidation, and that the 3.16 kinetic default factor applied to the mean data for adults would be inadequate for this subpopulation. The in vivo data have been used to derive pathway-related default factors for compounds eliminated largely via glucuronidation.

Uncertainty factors for chemical risk assessment: interspecies differences in glucuronidation

For the risk assessment of effects other than cancer, a safe daily intake in humans is generally derived from a surrogate threshold dose (e.g. NOAEL) in an animal species to which an uncertainty factor of 100 is usually applied. This 100-fold is to allow for possible interspecies (10-fold) and interindividual (10-fold) differences in response to a toxicant, and incorporates toxicodynamic and toxicokinetic aspects of variability. The current study determined the magnitude of the interspecies differences in the internal dose of compounds for which glucuronidation is the major pathway of metabolism in either humans or in the test species. The results showed that there are major interspecies differences in the nature of the biological processes which influence the internal dose, including the route of metabolism, the extent of presystemic metabolism and enterohepatic recirculation. The work presented does not support the refinement of the interspecies toxicokinetic default to species- and pathway-specific values, but demonstrates the necessity for risk assessments to be carried out using quantitative chemical-specific data which define the fundamental processes which will influence the internal dose of a chemical (toxicokinetics), or the interaction of toxicant with its target site (toxicodynamics).

Uncertainty factors for chemical risk assessment. human variability in the pharmacokinetics of CYP1A2 probe substrates

A 100-fold uncertainty factor is used to derive acceptable daily intakes for compounds causing thresholded toxicity. The 10-fold factor for human variability can be further subdivided into two factors of 10(0.5) (3.16) to allow for toxicokinetics and toxicodynamics. The validity of the human kinetic subfactor has been analysed in relation to CYP1A2 metabolism using published in vivo pharmacokinetic parameters selected to reflect chronic exposure (metabolic and total clearances and area under the plasma concentration-time curve: CLm, CL and AUC) and acute exposure (the peak plasma concentration, C(max)). The variability in CYP1A2 activity in healthy adults, based on data after oral and intravenous dosage (CLm, CL and AUC), ranged from 34 to 42%. The variability in C(max) was 21%. The default kinetic factor of 3.16 would cover at least 99% of the healthy adult population, assuming that the data were log-normally distributed, but would give lower protection for some subgroups (pregnant women at term, healthy elderly, patients with liver disease), and was inadequate for neonates. This analysis of in vivo kinetic data for CYP1A2 substrates illustrates the importance of quantifying human variability in specific metabolic pathways, and of identifying potentially susceptible subgroups of the human population, in order to determine the scientific validity of uncertainty factors.

Uncertainty factors for chemical risk assessment: interspecies differences in the in vivo pharmacokinetics and metabolism of human CYP1A2 substrates

The 100-fold default uncertainty factor is used to convert a no-observed-adverse-effect level (NOAEL) from a animal toxicity study, to a "safe" value for human intake. The composite uncertainty factor (100) has to allow for interspecies (10-fold) and interindividual (10-fold) differences in toxicokinetics and toxicodynamics. The aim of the current study was to assess the validity of the interspecies default for toxicokinetics (4.0) for each of the test species (dog, rabbit, rat and mouse), using published data for compounds eliminated by CYP1A2 in humans (caffeine, theobromine, theophylline and paraxanthine). An analysis of the published literature showed that the absorption, bioavailability and route of excretion were generally similar between humans and the test species, for each probe substrate. However, interspecies differences in the route of metabolism, and the enzymes involved in this process, were identified. The magnitude of difference in the internal dose, between species, showed that values for the mouse (10.6) and rat (5.4) exceed the 4.0-fold default, whereas the rabbit (2.6) and dog (1.6) were below this value. This work supports the need to replace the generic default factors by a compound-related value derived from specific, relevant, quantitative data; this would result in more relevant and reliable non-cancer risk assessments.

An analysis of the need for an additional uncertainty factor for infants and children

Uncertainty factors have been used for over 40 years to establish safe intakes of threshold toxicants. Tenfold factors are used to allow for species differences and for human variability, with extra factors for database inadequacies. The proposal to introduce an additional 10-fold factor for pesticides when exposure of infants and children is anticipated implies either age-related differences between species or differences within humans which exceed those present in adults. Alternatively, the extra factor could be related to deficiencies of current testing methods or concerns over irreversibility in developing organ systems. Available data do not provide a scientific rationale for the extra factor due to inadequacy of inter- and intraspecies uncertainty factors. Justification for the factor therefore must relate to the adequacy and sensitivity of current methods or concern about irreversible effects in the developing organism.