A Novel and Fast Online-SPE-LC-MS/MS Method to Quantify Thyroid Hormone Metabolites in Rat Plasma.

Since the focus in regulatory toxicology has drifted toward the identification of endocrine disruptors, the improvement in determination of alterations in the thyroid hormone system has become more important. THs are involved in several molecular processes important for a proper pre- and postnatal development so that disturbances can inter alia lead to incorrect brain maturation and/or disturbed metabolic processes (thermogenesis or lipolysis). In this publication, a new automated online solid-phase extraction (SPE)-liquid chromatography (LC)-tandem mass spectrometry (MS/MS, xLC-MS/MS) is introduced which simultaneously analyzes total T4, T3, rT3, T2, and T1. Method validation parameters are presented, and the method was positively verified by analyzing control and PTU-treated rat plasma samples (time points day 7, 14, and 28) for their total TH content. The obtained results were compared to published results by using a radioimmunoassay method. The automated SPE system ensures a consistent unified sample preparation, and this method overall showed sufficient specificity and accuracy to detect the given analytes in rat plasma. For the preparation of 50 µL of rat plasma, the following LOQs were established: 0.020 nM for T1, 0.029 nM for T2, 0.023 nM for rT3 and T3, and 3.22 nM for T4. This method is suitable to assess the identification of mechanisms leading to adverse effects, such as disturbed TH metabolism and regulation.

Adapting the in vitro micronucleus assay (OECD Test Guideline No. 487) for testing of manufactured nanomaterials: recommendations for best practices.

The current Organisation for Economic Co-Operation and Development test guideline number 487 (OECD TG No. 487) provides instruction on how to conduct the in vitro micronucleus assay. This assay is one of the gold standard approaches for measuring the mutagenicity of test items; however, it is directed at testing low molecular weight molecules and may not be appropriate for particulate materials (e.g. engineered nanoparticles [ENPs]). This study aimed to adapt the in vitro micronucleus assay for ENP testing and underpins the development of an OECD guidance document. A harmonized, nano-specific protocol was generated and evaluated by two independent laboratories. Cell lines utilized were human lymphoblastoid (TK6) cells, human liver hepatocytes (HepG2) cells, Chinese hamster lung fibroblast (V79) cells, whole blood, and buffy coat cells from healthy human volunteers. These cells were exposed to reference ENPs from the Joint Research Council (JRC): SiO2 (RLS-0102), Au5nm and Au30nm (RLS-03, RLS-010), CeO2 (NM212), and BaSO4 (NM220). Tungsten carbide-cobalt (WC/Co) was used as a trial particulate positive control. The chemical controls were positive in all cell cultures, but WC/Co was only positive in TK6 and buffy coat cells. In TK6 cells, mutagenicity was observed for SiO2- and both Au types. In HepG2 cells, Au5nm and SiO2 showed sub-two-fold increases in micronuclei. In V79 cells, whole blood, and buffy coat cells, no genotoxicity was detected with the test materials. The data confirmed that ENPs could be tested with the harmonized protocol, additionally, concordant data were observed across the two laboratories with V79 cells. WC/Co may be a suitable particulate positive control in the in vitro micronucleus assay when using TK6 and buffy coat cells. Detailed recommendations are therefore provided to adapt OECD TG No. 487 for testing ENP.

Toxicological inhalation studies in rats to substantiate grouping of zinc oxide nanoforms.

BACKGROUND: Significant variations exist in the forms of ZnO, making it impossible to test all forms in in vivo inhalation studies. Hence, grouping and read-across is a common approach under REACH to evaluate the toxicological profile of familiar substances. The objective of this paper is to investigate the potential role of dissolution, size, or coating in grouping ZnO (nano)forms for the purpose of hazard assessment. We performed a 90-day inhalation study (OECD test guideline no. (TG) 413) in rats combined with a reproduction/developmental (neuro)toxicity screening test (TG 421/424/426) with coated and uncoated ZnO nanoforms in comparison with microscale ZnO particles and soluble zinc sulfate. In addition, genotoxicity in the nasal cavity, lungs, liver, and bone marrow was examined via comet assay (TG 489) after 14-day inhalation exposure. RESULTS: ZnO nanoparticles caused local toxicity in the respiratory tract. Systemic effects that were not related to the local irritation were not observed. There was no indication of impaired fertility, developmental toxicity, or developmental neurotoxicity. No indication for genotoxicity of any of the test substances was observed. Local effects were similar across the different ZnO test substances and were reversible after the end of the exposure. CONCLUSION: With exception of local toxicity, this study could not confirm the occasional findings in some of the previous studies regarding the above-mentioned toxicological endpoints. The two representative ZnO nanoforms and the microscale particles showed similar local effects. The ZnO nanoforms most likely exhibit their effects by zinc ions as no particles could be detected after the end of the exposure, and exposure to rapidly soluble zinc sulfate had similar effects. Obviously, material differences between the ZnO particles do not substantially alter their toxicokinetics and toxicodynamics. The grouping of ZnO nanoforms into a set of similar nanoforms is justified by these observations.

Avoiding a reproducibility crisis in regulatory toxicology-on the fundamental role of ring trials.

The ongoing transition from chemical hazard and risk assessment based on animal studies to assessment relying mostly on non-animal data, requires a multitude of novel experimental methods, and this means that guidance on the validation and standardisation of test methods intended for international applicability and acceptance, needs to be updated. These so-called new approach methodologies (NAMs) must be applicable to the chemical regulatory domain and provide reliable data which are relevant to hazard and risk assessment. Confidence in and use of NAMs will depend on their reliability and relevance, and both are thoroughly assessed by validation. Validation is, however, a time- and resource-demanding process. As updates on validation guidance are conducted, the valuable components must be kept: Reliable data are and will remain fundamental. In 2016, the scientific community was made aware of the general crisis in scientific reproducibility-validated methods must not fall into this. In this commentary, we emphasize the central importance of ring trials in the validation of experimental methods. Ring trials are sometimes considered to be a major hold-up with little value added to the validation. Here, we clarify that ring trials are indispensable to demonstrate the robustness and reproducibility of a new method. Further, that methods do fail in method transfer and ring trials due to different stumbling blocks, but these provide learnings to ensure the robustness of new methods. At the same time, we identify what it would take to perform ring trials more efficiently, and how ring trials fit into the much-needed update to the guidance on the validation of NAMs.

Molecular signatures of angiogenesis inhibitors: a single-embryo untargeted metabolomics approach in zebrafish.

Angiogenesis is a key process in embryonic development, a disruption of this process can lead to severe developmental defects, such as limb malformations. The identification of molecular perturbations representative of antiangiogenesis in zebrafish embryo (ZFE) may guide the assessment of developmental toxicity from an endpoint- to a mechanism-based approach, thereby improving the extrapolation of findings to humans. Thus, the aim of the study was to discover molecular changes characteristic of antiangiogenesis and developmental toxicity. We exposed ZFEs to two antiangiogenic drugs (SU4312, sorafenib) and two developmental toxicants (methotrexate, rotenone) with putative antiangiogenic action. Molecular changes were measured by performing untargeted metabolomics in single embryos. The metabolome response was accompanied by the occurrence of morphological alterations. Two distinct metabolic effect patterns were observed. The first pattern comprised common effects of two specific angiogenesis inhibitors and the known teratogen methotrexate, strongly suggesting a shared mode of action of antiangiogenesis and developmental toxicity. The second pattern involved joint effects of methotrexate and rotenone, likely related to disturbances in energy metabolism. The metabolites of the first pattern, such as phosphatidylserines, pterines, retinol, or coenzyme Q precursors, represented potential links to antiangiogenesis and related developmental toxicity. The metabolic effect pattern can contribute to biomarker identification for a mechanism-based toxicological testing.

Minimum information for reporting on the TEER (trans-epithelial/endothelial electrical resistance) assay (MIRTA).

Standard information reporting helps to ensure that assay conditions and data are consistently reported and to facilitate inter-laboratory comparisons. Here, we present recommendations on minimum information for reporting on the TEER (trans-epithelial/endothelial electrical resistance) assay (MIRTA). The TEER assay is extensively used to evaluate the health of an epithelial/endothelial cell culture model and as an indicator of the potential toxicity of a test substance. This publication is the result of an international collaboration─called the RespTox (Respiratory Toxicity) Collaborative─through which twelve laboratories shared their protocols for assessing the barrier function of respiratory epithelial cells using the TEER assay following exposure to substances. The protocols from each laboratory were reviewed to identify general steps for performing the TEER assay, interlaboratory differences between steps, the rationale for differences, whether these differences impact results or cross-laboratory comparisons between TEER measurements. While the MIRTA recommendations are focused on respiratory epithelial cell systems, these recommendations can be adapted for other cell systems that form barriers. The use of these recommendations will support data transparency and reproducibility, reduce challenges in data interpretation, enable cross-laboratory comparisons, help assess study quality, and facilitate the incorporation of the TEER assay into national and international testing guidance.

Rapid equilibrium dialysis, ultrafiltration or ultracentrifugation? Evaluation of methods to quantify the unbound fraction of substances in plasma.

In pharmacokinetics plasma protein binding (PPB) is a well-established parameter impacting drug disposition. The unbound fraction (fu) is arguably regarded the effective concentration at the target site. Pharmacology and toxicology, increasingly use in vitro models. The translation of in vitro concentrations to in vivo doses can be supported by toxicokinetic modelling, e.g. physiologically based toxicokinetic models (PBTK). PPB of a test substance is an input parameter for PBTK. We compared three methods to quantify fu: rapid equilibrium dialysis (RED), ultrafiltration (UF) and ultracentrifugation (UC) using twelve substances covering a wide range of Log Pow (-0.1 to 6.8) and molecular weights (151 and 531 g/mol): Acetaminophen, Bisphenol A, Caffeine, Colchicine, Fenarimol, Flutamide, Genistein, Ketoconazole, α-Methyltestosterone, Tamoxifen, Trenbolone and Warfarin. After RED and UF separation, three polar substances (Log Pow < 2) were largely unbound (fu > 70%), while more lipophilic substances were largely bound (fu < 33%). Compared to RED or UF, UC resulted in a generally higher fu of lipophilic substances. fu obtained after RED and UF were more consistent with published data. For half of the substances, UC resulted in fu higher than the reference data. UF, RED and both UF and UC, resulted in lower fu of Flutamide, Ketoconazole and Colchicine, respectively. For fu quantifications, the separation method should be selected according to the test substance's properties. Based on our data, RED is suitable for a broader range of substances while UC and UF are suitable for polar substances.

A high-throughput metabolomics in vitro platform for the characterization of hepatotoxicity.

Cell-based metabolomics provides multiparametric physiologically relevant readouts that can be highly advantageous for improved, biologically based decision making in early stages of compound development. Here, we present the development of a 96-well plate LC-MS/MS-based targeted metabolomics screening platform for the classification of liver toxicity modes of action (MoAs) in HepG2 cells. Different parameters of the workflow (cell seeding density, passage number, cytotoxicity testing, sample preparation, metabolite extraction, analytical method, and data processing) were optimized and standardized to increase the efficiency of the testing platform. The applicability of the system was tested with seven substances known to be representative of three different liver toxicity MoAs (peroxisome proliferation, liver enzyme induction, and liver enzyme inhibition). Five concentrations per substance, aimed at covering the complete dose-response curve, were analyzed and 221 uniquely identified metabolites were measured, annotated, and allocated in 12 different metabolite classes such as amino acids, carbohydrates, energy metabolism, nucleobases, vitamins and cofactors, and diverse lipid classes. Multivariate and univariate analyses showed a dose response of the metabolic effects, a clear differentiation between liver toxicity MoAs and resulted in the identification of metabolite patterns specific for each MoA. Key metabolites indicative of both general and mechanistic specific hepatotoxicity were identified. The method presented here offers a multiparametric, mechanistic-based, and cost-effective hepatotoxicity screening that provides MoA classification and sheds light into the pathways involved in the toxicological mechanism. This assay can be implemented as a reliable compound screening platform for improved safety assessment in early compound development pipelines.

Application of high throughput in vitro metabolomics for hepatotoxicity mode of action characterization and mechanistic-anchored point of departure derivation: a case study with nitrofurantoin.

Omics techniques have been increasingly recognized as promising tools for Next Generation Risk Assessment. Targeted metabolomics offer the advantage of providing readily interpretable mechanistic information about perturbed biological pathways. In this study, a high-throughput LC-MS/MS-based broad targeted metabolomics system was applied to study nitrofurantoin metabolic dynamics over time and concentration and to provide a mechanistic-anchored approach for point of departure (PoD) derivation. Upon nitrofurantoin exposure at five concentrations (7.5 µM, 15 µM, 20 µM, 30 µM and 120 µM) and four time points (3, 6, 24 and 48 h), the intracellular metabolome of HepG2 cells was evaluated. In total, 256 uniquely identified metabolites were measured, annotated, and allocated in 13 different metabolite classes. Principal component analysis (PCA) and univariate statistical analysis showed clear metabolome-based time and concentration effects. Mechanistic information evidenced the differential activation of cellular pathways indicative of early adaptive and hepatotoxic response. At low concentrations, effects were seen mainly in the energy and lipid metabolism, in the mid concentration range, the activation of the antioxidant cellular response was evidenced by increased levels of glutathione (GSH) and metabolites from the de novo GSH synthesis pathway. At the highest concentrations, the depletion of GSH, together with alternations reflective of mitochondrial impairments, were indicative of a hepatotoxic response. Finally, a metabolomics-based PoD was derived by multivariate PCA using the whole set of measured metabolites. This approach allows using the entire dataset and derive PoD that can be mechanistically anchored to established key events. Our results show the suitability of high throughput targeted metabolomics to investigate mechanisms of hepatoxicity and derive point of departures that can be linked to existing adverse outcome pathways and contribute to the development of new ones.

Basic concepts of mixture toxicity and relevance for risk evaluation and regulation.

Exposure to multiple substances is a challenge for risk evaluation. Currently, there is an ongoing debate if generic "mixture assessment/allocation factors" (MAF) should be introduced to increase public health protection. Here, we explore concepts of mixture toxicity and the potential influence of mixture regulation concepts for human health protection. Based on this analysis, we provide recommendations for research and risk assessment. One of the concepts of mixture toxicity is additivity. Substances may act additively by affecting the same molecular mechanism within a common target cell, for example, dioxin-like substances. In a second concept, an "enhancer substance" may act by increasing the target site concentration and aggravating the adverse effect of a "driver substance". For both concepts, adequate risk management of individual substances can reliably prevent adverse effects to humans. Furthermore, we discuss the hypothesis that the large number of substances to which humans are exposed at very low and individually safe doses may interact to cause adverse effects. This commentary identifies knowledge gaps, such as the lack of a comprehensive overview of substances regulated under different silos, including food, environmentally and occupationally relevant substances, the absence of reliable human exposure data and the missing accessibility of ratios of current human exposure to threshold values, which are considered safe for individual substances. Moreover, a comprehensive overview of the molecular mechanisms and most susceptible target cells is required. We conclude that, currently, there is no scientific evidence supporting the need for a generic MAF. Rather, we recommend taking more specific measures, which focus on compounds with relatively small ratios between human exposure and doses, at which adverse effects can be expected.

Three-tiered approach for standard information requirements for polymers requiring registration under REACH.

Polymers are a very large class of chemicals comprising often complex molecules with multiple functions used in everyday products. The EU Commission is seeking to develop environmental and human health standard information requirements (SIRs) for man-made polymers requiring registration (PRR) under a revised Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) Regulation. Conventional risk assessment approaches currently used for small molecules may not apply to most polymers. Therefore, we propose a conceptual three-tiered regulatory approach for data generation to assess individual and groups of polymers requiring registration (PRR). A key element is the grouping of polymers according to chemistry, physico-chemical properties and hazard similarity. The limited bioavailability of many polymers is a prominent difference to many small molecules and is a key consideration of the proposed approach. Methods assessing potential for systemic bioavailability are integral to Tier 1. Decisions for further studies are based on considerations of properties and effects, combined with systemic bioavailability and use and exposure considerations. For many PRRs, Tier 1 data on hazard, use and exposure will likely be sufficient for achieving the protection goals of REACH. Vertebrate animal studies in Tiers 2 and 3 can be limited to targeted testing. The outlined approach aims to make use of current best scientific evidence and to reduce animal testing whilst providing data for an adequate level of protection.

Plant extracts, polymers and new approach methods: Practical experience with skin sensitization assessment.

Over the last decade, research into methodologies to identify skin sensitization hazards has led to the adoption of several non-animal methods as OECD test guidelines. However, predictive accuracy beyond the chemical domains of the individual validation studies remains largely untested. In the present study, skin sensitization test results from in vitro and in chemico methods for 12 plant extracts and 15 polymeric materials are reported and compared to available in vivo skin sensitization data. Eight plant extracts were tested in the DPRA and h-CLAT, with the 2 out of 3 approach resulting in a balanced accuracy of 50%. The balanced accuracy for the 11 plant extracts assessed in the SENS-IS was 88%. Excluding 5 polymers inconclusive in vitro, the remainder, assessed using the 2 out of 3 approach, resulted in 63% balanced accuracy. The SENS-IS method, excluding one polymeric material due to technical inapplicability, showed 68% balanced accuracy. Although based on limited numbers, the results presented here indicate that some substance subgroups may not be in the applicability domains of the method used and careful analysis is required before positive or negative results can be accepted.

Lessons Learned from the Grouping of Chemicals to Assess Risks to Human Health.

In analogy to the periodic system that groups elements by their similarity in structure and chemical properties, the hazard of chemicals can be assessed in groups having similar structures and similar toxicological properties. Here we review case studies of chemical grouping strategies that supported the assessment of hazard, exposure, and risk to human health. By the EU-REACH and the US-TSCA New Chemicals Program, structural similarity is commonly used as the basis for grouping, but that criterion is not always adequate and sufficient. Based on the lessons learned, we derive ten principles for grouping, including: transparency of the purpose, criteria, and boundaries of the group; adequacy of methods used to justify the group; and inclusion or exclusion of substances in the group by toxicological properties. These principles apply to initial grouping to prioritize further actions as well as to definitive grouping to generate data for risk assessment. Both can expedite effective risk management.

Toward Realistic Dosimetry In Vitro: Determining Effective Concentrations of Test Substances in Cell Culture and Their Prediction by an In Silico Mass Balance Model.

Nominal concentrations (CNom) in cell culture media are routinely used to define concentration-effect relationships in the in vitro toxicology. The actual concentration in the medium (CMedium) can be affected by adsorption processes, evaporation, or degradation of chemicals. Therefore, we measured the total and free concentration of 12 chemicals, covering a wide range of lipophilicity (log KOW -0.07-6.84), in the culture medium (CMedium) and cells (CCell) after incubation with Balb/c 3T3 cells for up to 48 h. Measured values were compared to predictions using an as yet unpublished in silico mass balance model that combined relevant equations from similar models published by others. The total CMedium for all chemicals except tamoxifen (TAM) were similar to the CNom. This was attributed to the cellular uptake of TAM and accumulation into lysosomes. The free (i.e., unbound) CMedium for the low/no protein binding chemicals were similar to the CNom, whereas values of all moderately to highly protein-bound chemicals were less than 30% of the CNom. Of the 12 chemicals, the two most hydrophilic chemicals, acetaminophen (APAP) and caffeine (CAF), were the only ones for which the CCell was the same as the CNom. The CCell for all other chemicals tended to increase over time and were all 2- to 274-fold higher than CNom. Measurements of CCytosol, using a digitonin method to release cytosol, compared well with CCell (using a freeze-thaw method) for four chemicals (CAF, APAP, FLU, and KET), indicating that both methods could be used. The mass balance model predicted the total CMedium within 30% of the measured values for 11 chemicals. The free CMedium of all 12 chemicals were predicted within 3-fold of the measured values. There was a poorer prediction of CCell values, with a median overprediction of 3- to 4-fold. In conclusion, while the number of chemicals in the study is limited, it demonstrates the large differences between CNom and total and free CMedium and CCell, which were also relatively well predicted by the mass balance model.

Gut microbiome and plasma metabolome changes in rats after oral gavage of nanoparticles: sensitive indicators of possible adverse health effects.

BACKGROUND: The oral uptake of nanoparticles is an important route of human exposure and requires solid models for hazard assessment. While the systemic availability is generally low, ingestion may not only affect gastrointestinal tissues but also intestinal microbes. The gut microbiota contributes essentially to human health, whereas gut microbial dysbiosis is known to promote several intestinal and extra-intestinal diseases. Gut microbiota-derived metabolites, which are found in the blood stream, serve as key molecular mediators of host metabolism and immunity. RESULTS: Gut microbiota and the plasma metabolome were analyzed in male Wistar rats receiving either SiO2 (1000 mg/kg body weight/day) or Ag nanoparticles (100 mg/kg body weight/day) during a 28-day oral gavage study. Comprehensive clinical, histopathological and hematological examinations showed no signs of nanoparticle-induced toxicity. In contrast, the gut microbiota was affected by both nanoparticles, with significant alterations at all analyzed taxonomical levels. Treatments with each of the nanoparticles led to an increased abundance of Prevotellaceae, a family with gut species known to be correlated with intestinal inflammation. Only in Ag nanoparticle-exposed animals, Akkermansia, a genus known for its protective impact on the intestinal barrier was depleted to hardly detectable levels. In SiO2 nanoparticles-treated animals, several genera were significantly reduced, including probiotics such as Enterococcus. From the analysis of 231 plasma metabolites, we found 18 metabolites to be significantly altered in Ag-or SiO2 nanoparticles-treated rats. For most of these metabolites, an association with gut microbiota has been reported previously. Strikingly, both nanoparticle-treatments led to a significant reduction of gut microbiota-derived indole-3-acetic acid in plasma. This ligand of the arylhydrocarbon receptor is critical for regulating immunity, stem cell maintenance, cellular differentiation and xenobiotic-metabolizing enzymes. CONCLUSIONS: The combined profiling of intestinal microbiome and plasma metabolome may serve as an early and sensitive indicator of gut microbiome changes induced by orally administered nanoparticles; this will help to recognize potential adverse effects of these changes to the host.

Accounting for Precision Uncertainty of Toxicity Testing: Methods to Define Borderline Ranges and Implications for Hazard Assessment of Chemicals.

For hazard classifications of chemicals, continuous data from animal- or nonanimal testing methods are often dichotomized into binary positive/negative outcomes by defining classification thresholds (CT). Experimental data are, however, subject to biological and technical variability. Each test method's precision is limited resulting in uncertainty of the positive/negative outcome if the experimental result is close to the CT. Borderline ranges (BR) around the CT were suggested, which represent ranges in which the study result is ambiguous, that is, positive or negative results are equally likely. The BR reflects a method's precision uncertainty. This article explores and compares different approaches to quantify the BR. Besides using the pooled standard deviation, we determine the BR by means of the median absolute deviation (MAD), with a sequential combination of both methods, and by using nonparametric bootstrapping. Furthermore, we quantify the BR for different hazardous effects, including nonanimal tests for skin corrosion, eye irritation, skin irritation, and skin sensitization as well as for an animal test on skin sensitization (the local lymph node assay, LLNA). Additionally, for one method (direct peptide reactivity assay) the BR was determined experimentally and compared to calculated BRs. Our results demonstrate that (i) the precision of the methods is determining the size of their BRs, (ii) there is no "perfect" method to derive a BR, alas, (iii) a consensus on BR is needed to account for the limited precision of testing methods.

The Evolution of Regulatory Toxicology: Where is the Gardener?

There is a need for paradigm change in the methodology employed for toxicological testing and assessment. It could be said that this change is well on its way, through an evolutionary progress analogous to that of natural selection. Darwin's Theory of Evolution has defined the idea of evolution and descendancy since the last third of the 19th century. Increasingly, this concept of 'evolution' is being applied beyond the field of biology. This Comment article discusses the progress of toxicological testing in the context of 'evolutionary pressure' and deliberates how this process can help foster the development, implementation and acceptance of mechanistic and human-relevant methods in this field. By comparing the current regulatory landscape in toxicity testing and assessment to specific elements in Charles Darwin's evolutionary theory, we aim to better understand the needs and requirements for the future.

The Use of Nanomaterial In Vivo Organ Burden Data for In Vitro Dose Setting.

Effects of nanomaterials are usually observed at higher concentrations in vitro compared to animal studies. This is pointing to differences between in vivo situations and generally less complex in vitro models. These differences concern toxicodynamics and the internal exposure (at the target cells of the in vitro and in vivo test system). The latter can be minimized by appropriate in vivo to in vitro dose extrapolations (IVIVE). An IVIVE six-step procedure is proposed here: 1) determine in vivo exposure; 2) identify in vivo organ burden at lowest observed adverse effect concentration; 3) extrapolate in vivo organ burden to in vitro effective dose; 4) extrapolate in vitro effective dose to nominal concentration; 5) set dose ranges to establish dose-response relationships; and 6) consider uncertainties and specificities of in vitro test system. Assessing the results of in vitro studies needs careful consideration of discrepancies between in vitro and in vivo models: apart from different endpoints (usually cellular responses in vitro and adverse effects on organs or organisms in vivo), nanomaterials can also have a different potency in relatively simple in vitro models and the more complex corresponding organ in vivo. IVIVE can, nonetheless, reduce the differences in exposures.

In Silico Models to Predict the Perturbation of Molecular Initiating Events Related to Thyroid Hormone Homeostasis.

Disturbance of the thyroid hormone homeostasis has been associated with adverse health effects such as goiters and impaired mental development in humans and thyroid tumors in rats. In vitro and in silico methods for predicting the effects of small molecules on thyroid hormone homeostasis are currently being explored as alternatives to animal experiments, but are still in an early stage of development. The aim of this work was the development of a battery of in silico models for a set of targets involved in molecular initiating events of thyroid hormone homeostasis: deiodinases 1, 2, and 3, thyroid peroxidase (TPO), thyroid hormone receptor (TR), sodium/iodide symporter, thyrotropin-releasing hormone receptor, and thyroid-stimulating hormone receptor. The training data sets were compiled from the ToxCast database and related scientific literature. Classical statistical approaches as well as several machine learning methods (including random forest, support vector machine, and neural networks) were explored in combination with three data balancing techniques. The models were trained on molecular descriptors and fingerprints and evaluated on holdout data. Furthermore, multi-task neural networks combining several end points were investigated as a possible way to improve the performance of models for which the experimental data available for model training are limited. Classifiers for TPO and TR performed particularly well, with F1 scores of 0.83 and 0.81 on the holdout data set, respectively. Models for the other studied targets yielded F1 scores of up to 0.77. An in-depth analysis of the reliability of predictions was performed for the most relevant models. All data sets used in this work for model development and validation are available in the Supporting Information.

ChemBioSim: Enhancing Conformal Prediction of In Vivo Toxicity by Use of Predicted Bioactivities.

Computational methods such as machine learning approaches have a strong track record of success in predicting the outcomes of in vitro assays. In contrast, their ability to predict in vivo endpoints is more limited due to the high number of parameters and processes that may influence the outcome. Recent studies have shown that the combination of chemical and biological data can yield better models for in vivo endpoints. The ChemBioSim approach presented in this work aims to enhance the performance of conformal prediction models for in vivo endpoints by combining chemical information with (predicted) bioactivity assay outcomes. Three in vivo toxicological endpoints, capturing genotoxic (MNT), hepatic (DILI), and cardiological (DICC) issues, were selected for this study due to their high relevance for the registration and authorization of new compounds. Since the sparsity of available biological assay data is challenging for predictive modeling, predicted bioactivity descriptors were introduced instead. Thus, a machine learning model for each of the 373 collected biological assays was trained and applied on the compounds of the in vivo toxicity data sets. Besides the chemical descriptors (molecular fingerprints and physicochemical properties), these predicted bioactivities served as descriptors for the models of the three in vivo endpoints. For this study, a workflow based on a conformal prediction framework (a method for confidence estimation) built on random forest models was developed. Furthermore, the most relevant chemical and bioactivity descriptors for each in vivo endpoint were preselected with lasso models. The incorporation of bioactivity descriptors increased the mean F1 scores of the MNT model from 0.61 to 0.70 and for the DICC model from 0.72 to 0.82 while the mean efficiencies increased by roughly 0.10 for both endpoints. In contrast, for the DILI endpoint, no significant improvement in model performance was observed. Besides pure performance improvements, an analysis of the most important bioactivity features allowed detection of novel and less intuitive relationships between the predicted biological assay outcomes used as descriptors and the in vivo endpoints. This study presents how the prediction of in vivo toxicity endpoints can be improved by the incorporation of biological information-which is not necessarily captured by chemical descriptors-in an automated workflow without the need for adding experimental workload for the generation of bioactivity descriptors as predicted outcomes of bioactivity assays were utilized. All bioactivity CP models for deriving the predicted bioactivities, as well as the in vivo toxicity CP models, can be freely downloaded from https://doi.org/10.5281/zenodo.4761225.