Toxicokinetics

Per- and polyfluoroalkyl substances (PFASs) are a group of fluorinated substances of interest to researchers, regulators, and the public due to their widespread presence in the environment. A few PFASs have comparatively extensive amounts of human epidemiological, exposure, and experimental animal toxicity data (e.g., perfluorooctanoic acid), whereas little toxicity and exposure information exists for much of the broader set of PFASs. Given that traditional approaches to generate toxicity information are resource intensive, new approach methods, including in vitro high-throughput toxicity (HTT) testing, are being employed to inform PFAS hazard characterization and further (in vivo) testing. The U.S. Environmental Protection Agency (EPA) and the National Toxicology Program (NTP) are collaborating to develop a risk-based approach for conducting PFAS toxicity testing to facilitate PFAS human health assessments. This article describes the construction of a PFAS screening library and the process by which a targeted subset of 75 PFASs were selected. Multiple factors were considered, including interest to the U.S. EPA, compounds within targeted categories, structural diversity, exposure considerations, procurability and testability, and availability of existing toxicity data. Generating targeted HTT data for PFASs represents a new frontier for informing priority setting. https://doi.org/10.1289/EHP4555.

Scientifically sound, risk-informed evaluation of chemicals is essential to protecting public health. Systematically leveraging information from exposure, toxicology, and epidemiology studies can provide a holistic understanding of how real-world exposure to chemicals may impact the health of populations, including sensitive and vulnerable individuals and life-stages. Increasingly, public health policy makers are employing toxicokinetic (TK) modeling tools to integrate these data streams and predict potential human health impact. Development of a suite of tools for predicting internal exposure, including physiologically-based toxicokinetic (PBTK) models, is being driven by needs to address large numbers of data-poor chemicals efficiently, translate bioactivity, and mechanistic information from new in vitro test systems, and integrate multiple lines of evidence to enable scientifically sound, risk-informed decisions. New modeling approaches are being designed "fit for purpose" to inform specific decision contexts, with applications ranging from rapid screening of hundreds of chemicals, to improved prediction of risks during sensitive stages of development. New data are being generated experimentally and computationally to support these models. Progress to meet the demand for internal exposure and PBTK modeling tools will require transparent publication of models and data to build credibility in results, as well as opportunities to partner with decision makers to evaluate and build confidence in use of these for improved decisions that promote safe use of chemicals.

Regulatory toxicology seeks to ensure that exposures to chemicals encountered in the environment, in the workplace, or in products pose no significant hazards and produce no harm to humans or other organisms, i.e., that chemicals are used safely. The most practical and direct means of ensuring that hazards and harms are avoided is to identify the doses and conditions under which chemical toxicity does not occur so that chemical concentrations and exposures can be appropriately limited. Modern advancements in pharmacology and toxicology have revealed that the rates and mechanisms by which organisms absorb, distribute, metabolize and eliminate chemicals-i.e., the field of kinetics-often determine the doses and conditions under which hazard, and harm, are absent, i.e., the safe dose range. Since kinetics, like chemical hazard and toxicity, are extensive properties that depend on the amount of the chemical encountered, it is possible to identify the maximum dose under which organisms can efficiently metabolize and eliminate the chemicals to which they are exposed, a dose that has been referred to as the kinetic maximum dose, or KMD. This review explains the rationale that compels regulatory toxicology to embrace the advancements made possible by kinetics, why understanding the kinetic relationship between the blood level produced and the administered dose of a chemical is essential for identifying the safe dose range, and why dose-setting in regulatory toxicology studies should be informed by estimates of the KMD rather than rely on the flawed concept of maximum-tolerated toxic dose, or MTD.

The development of artificial intelligence (AI) tools and technology has made AI-driven drug discovery a more prominent field. We are firmly in the AI era, with hybrid designs that eventually comprise deep learning (DL) and conventional machine learning (ML). Although traditional models can predict ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties, they remain relatively unsuccessful, and improving the accuracy of predictions remains challenging. Recently, several researchers have developed a hybrid learning model that successfully addresses these problems and improves prediction accuracy. The systematic tendencies facing AI-powered transformation from conventional DL and ML to hybrid learning AI models are examined in this review. Compared with traditional ML and DL, hybrid AI models have increased efficiency by reducing drug development time and costs, and improved success rates. In this context, the ongoing development of new ADMET software based on hybrid AI and multimodeling techniques can enhance the accuracy of pharmacokinetic-pharmacodynamic predictions, improve ADMET endpoint predictions, and expedite the drug discovery of new chemical entities. Moreover, this review covers the future of AI in pharmaceutical sciences and ADMET predictions, including AI-driven prediction models that range from basic ML/DL to newly developed hybrid models, evaluation parameters, and their applications in ADMET property prediction. SIGNIFICANCE STATEMENT: The article covers the compilation of ongoing research in the development of ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) software based on hybrid artificial intelligence and multimodeling techniques, which may increase the accuracy of pharmacokinetic-pharmacodynamic predictions, improve ADMET endpoint predictions, and accelerate drug discovery.