Science is based on facts and their discourse. Willingly or unwillingly, facts are mixed with opinion, i.e., views or judg­ments formed, not necessarily based on fact or knowledge. This is often necessary, where we have controversial facts or no definitive evidence yet, because we need to take decisions or have to prioritize. Evidence-based approaches aim at identifying the facts and their quality objectively and transparently; they are now increasingly embraced in toxicology, especially by employing systematic reviews, meta-analyses, quality scoring, risk-of-bias tools, etc. These are core to Evidence-based Toxicology. Such approaches aim at minimizing opinion, the “eminence-based” part of science.

Animal experiments are the basis of a lot of our textbook knowledge in the life sciences, have helped to develop des­perately needed therapies, and have made this world a safer place. However, they represent only one of the many possible approaches to accomplish all these things. Like all approaches, they come with shortcomings, and their true contribution is often overrated. This article aims to summarize their limitations and challenges beside the ethical and economical concerns (i.e., costs and duration as well as costs following wrong decisions in product development): they include reproducibility, inadequate reporting, statistical under-powering, lack of inter-species predictivity, lack of reflection of human diversity and of real-life exposure. Each and every one of these increasingly discussed aspects of animal experiments can be amended, but this would require enormous additional resources. Together, they prompt a need to engineer a new paradigm to ensure the safety of patients and consumers, new products and therapies.

Development of reliable cell-based nanotoxicology assays is important for evaluation of potentially hazardous engineered nanomaterials. Challenges to producing a reliable assay protocol include working with nanoparticle dispersions and living cell lines, and the potential for nano-related interference effects. Here we demonstrate the use of a 96-well plate design with several measurement controls and an interlaboratory comparison study involving five laboratories to characterize the robustness of a nanocytotoxicity MTS cell viability assay based on the A549 cell line. The consensus EC50 values were 22.1 mg/L (95% confidence intervals 16.9 mg/L to 27.2 mg/L) and 52.6 mg/L (44.1 mg/L to 62.6 mg/L) for positively charged polystyrene nanoparticles for the serum-free and serum conditions, respectively, and 49.7 μmol/L (47.5 μmol/L to 51.5 μmol/L) and 77.0 μmol/L (54.3 μmol/L to 99.4 μmol/L) for positive chemical control cadmium sulfate for the serum-free and serum conditions, respectively. Results from the measurement controls can be used to evaluate the sources of variability and their relative magnitudes within and between laboratories. This information revealed steps of the protocol that may need to be modified to improve the overall robustness and precision. The results suggest that protocol details such as cell line ID, media exchange, cell handling, and nanoparticle dispersion are critical to ensure protocol robustness and comparability of nanocytotoxicity assay results. The combi­nation of system control measurements and interlaboratory comparison data yielded insights that would not have been available by either approach by itself.

The present study applies a systems biology approach for the in silico predictive modeling of drug toxicity on the basis of high-quality preclinical drug toxicity data with the aim of increasing the mechanistic understanding of toxic effects of compounds at different levels (pathway, cell, tissue, organ). The model development was carried out using 77 compounds for which gene expression data for treated primary human hepatocytes is available in the LINCS database and for which rodent in vivo hepatotoxicity information is available in the eTOX database. The data from LINCS were used to determine the type and number of pathways disturbed by each compound and to estimate the extent of disturbance (network perturbation elasticity), and were used to analyze the correspondence with the in vivo information from eTOX. Predictive models were developed through this integrative analysis, and their specificity and sensitivity were assessed. The quality of the predictions was determined on the basis of the area under the curve (AUC) of plots of true positive vs. false positive rates (ROC curves). The ROC AUC reached values of up to 0.9 (out of 1.0) for some hepatotoxicity endpoints. Moreover, the most frequently disturbed metabolic pathways were determined across the studied toxicants. They included, e.g., mitochondrial beta-oxidation of fatty acids and amino acid metabolism. The process was exemplified by successful predictions on various statins. In conclusion, an entirely new approach linking gene expression alterations to the prediction of complex organ toxicity was developed and evaluated.

An approach to systematically describe the uncertainties and complexity of the standard animal testing and assessment approach for carcinogenicity is explored by using a OECD Guidance Document that was originally developed for reporting defined in vitro approaches to testing and assessment. The format is suitable for this re-purposing and it appears that the potential multitude of approaches for integrating and interpreting data from standard animal testing may ultimately be conceptually similar to the challenge of integrating relevant in vitro and in silico data. This structured approach shall allow 1) fostering interest in developing improved defined in silico and in vitro approaches; 2) the definition of what type of effects should be predicted by the new approach; 3) selection of the most suitable reference data and assessments; 4) definition of the weight that the standard animal reference data should have compared to human reference data and mechanistic information in the context of assessing the fitness of the new in vitro and in silico approach; 5) definition of a benchmark for the minimum performance of the new approach, based on a conceptual recognition that correlation of alternative assessment results with reference animal results is limited by the uncertainties and complexity of the latter. A longer term perspective is indicated for evolving the definition of adversity for classifi­cation and regulatory purposes. This work will be further discussed and developed within the OECD expert group on non-genotoxic carcinogenicity IATA development.

Surgical implantation of a biomaterial triggers foreign-body-induced fibrous encapsulation. Two major mechanisms of this complex physiological process are (I) chemotaxis of fibroblasts from surrounding tissue to the implant region, followed by (II) tissue remodeling. As an alternative to animal studies, we here propose a process-aligned in vitro test platform to investigate the material dependency of fibroblast chemotaxis and tissue remodeling mediated by material-resident macrophages.

Embedded in a biomimetic three-dimensional collagen hydrogel, chemotaxis of fibroblasts in the direction of macrophage-material-conditioned cell culture supernatant was analyzed by live cell imaging. A combination of statistical analysis with a complementary parameterized random walk model allowed quantitative and qualitative characterization of the cellular walk process. We thereby identified an increasing macrophage-mediated chemotactic potential ranking of biomaterials from glass over polytetrafluorethylene to titanium. To address long-term effects of bio-material-resident macrophages on fibroblasts in a three-dimensional microenvironment, we further studied tissue remod­eling by applying macrophage-material-conditioned medium on fibrous in vitro tissue models. A high correlation of the in vitro tissue model to state of the art in vivo study data was found. Titanium exhibited a significantly lower tissue remodeling capacity compared to polytetrafluorethylene. With this approach, we identified a material dependency of both chemotaxis and tissue remodeling processes, strengthening knowledge on their specific contribution to the foreign body reaction.

There is a great need for valuable ex vivo models that allow for assessment of cartilage repair strategies to reduce the high number of animal experiments. In this paper we present three studies with our novel ex vivo osteochondral culture platform. It consists of two separated media compartments for cartilage and bone, which better represents the in vivo situation and enables supply of factors specific to the different needs of bone and cartilage. We investigated whether separation of the cartilage and bone compartments and/or culture media results in the maintenance of viability, struc­tural and functional properties of cartilage tissue. Next, we evaluated for how long we can preserve cartilage matrix stability of osteochondral explants during long-term culture over 84 days. Finally, we determined the optimal defect size that does not show spontaneous self-healing in this culture system.

It was demonstrated that separated compartments for cartilage and bone in combination with tissue-specific medium allow for long-term culture of osteochondral explants while maintaining cartilage viability, matrix tissue content, structure and mechanical properties for at least 56 days. Furthermore, we could create critical size cartilage defects of different sizes in the model.

The osteochondral model represents a valuable preclinical ex vivo tool for studying clinically relevant cartilage ther­apies, such as cartilage biomaterials, for their regenerative potential, for evaluation of drug and cell therapies, or to study mechanisms of cartilage regeneration. It will undoubtedly reduce the number of animals needed for in vivo testing.

In vitro approaches to address key steps of chemical-induced skin sensitization have been developed, but there is uncertainty how keratinocytes, which play a crucial role not only regarding xenobiotic metabolism but also skin inflammation, impact on a chemical’s potential and potency to activate dendritic cells. We investigated these aspects by coculturing THP-1 cells, as surrogate dendritic cells, with HaCaT keratinocytes. We tested our HaCaT/THP-1 model with a set of 14 sensitizers, containing 7 prohaptens, and 10 non-sensitizers. Compared to exposing THP-1 alone, coculturing resulted in up to 3.1-fold enhanced maximal CD86 and/or CD54 upregulation on THP-1, and improved concentration-dependency. All 14 sensitizers were found positive for CD86 and/or CD54 upregulation based on Δ mean fluorescence intensity (MFI) ≥ 10 for CD86 and ΔMFI ≥ 50 for CD54. Only 1 of 10 non-sensitizers was false-positive. Remarkably, coculture with HaCaT keratinocytes improved the rank correlation of the estimated minimum chemical concentrations inducing a positive response in vitro with in vivo data on sensitization potency, especially for CD54 (Spearman: r = 0.739, p = 0.006; CD86: r = 0.571, p = 0.041). These promising data suggest that the coculture model has the potential to support the prediction of sensitization potency based on in vitro data.

During the past decades, many epidemiological, toxicological and biological studies have been performed to assess the role of environmental chemicals as potential toxicants associated with diverse human disorders. However, the relationships between diseases based on chemical exposure rarely have been studied by computational biology.

We developed a human environmental disease network (EDN) to explore and suggest novel disease-disease and chemical-disease relationships. The presented scored EDN model is built upon the integration of systems biology and chemical toxicology using information on chemical contaminants and their disease relationships reported in the TDDB database. The resulting human EDN takes into consideration the level of evidence of the toxicant-disease relationships, allowing inclusion of some degrees of significance in the disease-disease associations. Such a network can be used to identify uncharacterized connections between diseases. Examples are discussed for type 2 diabetes (T2D). Addi­tionally, this computational model allows confirmation of already known links between chemicals and diseases (e.g., between bisphenol A and behavioral disorders) and also reveals unexpected associations between chemicals and diseases (e.g., between chlordane and olfactory alteration), thus predicting which chemicals may be risk factors to human health.

The proposed human EDN model allows exploration of common biological mechanisms of diseases associated with chemical exposure, helping us to gain insight into disease etiology and comorbidity. This computational approach is an alternative to animal testing supporting the 3R concept.

Translating in vitro biological data into actionable information related to human health holds the potential to improve disease treatment and risk assessment of chemical exposures. While genomics has identified regulatory pathways at the cellular level, translation to the organism level requires a multiscale approach accounting for intra-cellular regu­lation, inter-cellular interaction, and tissue/organ-level effects. Tissue-level effects can now be probed in vitro thanks to recently developed systems of three-dimensional (3D), multicellular, “organotypic” cell cultures, which mimic functional responses of living tissue. However, there remains a knowledge gap regarding interactions across different biological scales, complicating accurate prediction of health outcomes from molecular/genomic data and tissue responses. Systems biology aims at mathematical modeling of complex, non-linear biological systems. We propose to apply a systems biology approach to achieve a computational representation of tissue-level physiological responses by inte­grating empirical data derived from organotypic culture systems with computational models of intracellular pathways to better predict human responses. Successful implementation of this integrated approach will provide a powerful tool for faster, more accurate and cost-effective screening of potential toxicants and therapeutics.

On September 11, 2015, an interdisciplinary group of scientists, engineers, and clinicians gathered for a workshop in Research Triangle Park, North Carolina, to discuss this ambitious goal. Participants represented laboratory-based and computational modeling approaches to pharmacology and toxicology, as well as the pharmaceutical industry, government, non-profits, and academia. Discussions focused on identifying critical system perturbations to model, the computational tools required, and the experimental approaches best suited to generating key data.