Predictive modeling of biological responses in the rat liver using in vitro Tox21 bioactivity: Benefits from high-throughput toxicokinetics

Abstract

Computational methods are needed to more efficiently leverage data from in vitro cell-based models to predict what occurs within whole body systems after chemical insults. This study set out to test the hypothesis that in vitro high-throughput screening (HTS) data can more effectively predict in vivo biological responses when chemical disposition and toxicokinetic (TK) modeling are employed. In vitro HTS data from the Tox21 consortium were analyzed in concert with chemical disposition modeling to derive nominal, aqueous, and intracellular estimates of concentrations eliciting 50% maximal activity. In vivo biological responses were captured using rat liver transcriptomic data from the DrugMatrix and TG-Gates databases and evaluated for pathway enrichment. In vivo dosing data were translated to equivalent body concentrations using HTTK modeling. Random forest models were then trained and tested to predict in vivo pathway-level activity across 221 chemicals using in vitro bioactivity data and physicochemical properties as predictor variables, incorporating methods to address imbalanced training data resulting from high instances of inactivity. Model performance was quantified using the area under the receiver operator characteristic curve (AUC-ROC) and compared across pathways for different combinations of predictor variables. All models that included toxicokinetics were found to outperform those that excluded toxicokinetics. Biological interpretation of the model features revealed that rather than a direct mapping of in vitro assays to in vivo pathways, unexpected combinations of multiple in vitro assays predicted in vivo pathway-level activities. To demonstrate the utility of these findings, the highest-performing model was leveraged to make new predictions of in vivo biological responses across all biological pathways for remaining chemicals tested in Tox21 with adequate data coverage (n = 6617). These results demonstrate that, when chemical disposition and toxicokinetics are carefully considered, in vitro HT screening data can be used to effectively predict in vivo biological responses to chemicals.

Prediction Models for Agonists and Antagonists of Molecular Initiation Events for Toxicity Pathways Using an Improved Deep-Learning-Based Quantitative Structure–Activity Relationship System

“…Recently, methods for predicting toxicity risk have focused on deep learning (DL)-based models, due to their high accuracy and the use of large datasets, as alternatives to animal testing. These methods follow the principle of the 3Rs (replacement, reduction, and refinement) for the discovery of molecular initiating events (MIEs) in the adverse outcome pathway (AOP), where MIEs are the first point of chemical–biological interaction within the human body [ 1 , 2 , 3 ]. The interaction between endocrine-disrupting chemicals and nuclear–receptor family proteins can affect the endocrine system in the AOP [ 4 ].…”

Mapping Mechanistic Pathways of Acute Oral Systemic Toxicity Using Chemical Structure and Bioactivity Measurements

“…If chemicals that have not been tested in ToxCast are included, the distribution across the categories is broader but still shows a similar pattern. The use of in vitro-in vivo extrapolation to account for toxicokinetics would be expected to improve the correspondence between the bioactivity and in vivo-based toxicity classifications as discussed below ( Honda et al, 2019 ; Ring et al, 2021 ).…”

“…While consideration of the structural features within each cluster coupled with computer-based metabolism predictions can provide some insights regarding the potential for this confounding effect, our work does not specifically address this outstanding issue. There has been considerable progress in recent years both with better in silico predictions ( Pinto et al, 2016 ; Leonard et al, 2018 ; Ring et al, 2021 ) and in vitro approaches ( DeGroot et al, 2018 ; Deisenroth et al, 2020 ; Franzosa et al, 2021 ). As these efforts continue to progress in parallel with efforts such as ours, the robustness of an in vitro testing paradigm should rapidly increase.…”

Use of computational toxicology tools to predict in vivo endpoints associated with Mode of Action and the endocannabinoid system: A case study with chlorpyrifos, chlorpyrifos-oxon and Δ9Tetrahydrocannabinol

“…The HTTK-Pop method simulates human physiological variability by Monte Carlo sampling of the model parameters (e.g., liver flow, glomerular filtration rate, liver volume, concentration at steady state [C ss ], hepatic clearance, plasma protein binding) and reverse dosimetry to predict a human EAD (mg/kg/day) ( Ring et al, 2017 ). Refinements incorporated by use of TK can decrease EAD Human variability by approximately 12% when compared to simpler in vitro to in vivo extrapolation models ( Ring et al, 2017 , Ring et al, 2021 ). The methods of deriving age-specific physiological parameters (e.g., hepato-cellularity and liver mass) in the TK model “httk-pop” package have been detailed in Ring et al (2017) .…”