Biofuel blends of 10% ethanol (EtOH) and gasoline are common in the USA, and higher EtOH concentrations are being considered (15-85%). Currently, no physiologically-based pharmacokinetic (PBPK) models are available to describe the kinetics of EtOH-based biofuels. PBPK models were developed to describe life-stage differences in the kinetics of EtOH alone in adult, pregnant, and neonatal rats for inhalation, oral, and intravenous routes of exposure, using data available in the open literature. Whereas ample data exist from gavage and intravenous routes of exposure, kinetic data from inhalation exposures are limited, particularly at concentrations producing blood and target tissue concentrations associated with developmental neurotoxicity. Compared to available data, the three models reported in this paper accurately predicted the kinetics of EtOH, including the absorption, peak concentration, and clearance across multiple datasets. In general, model predictions for adult and pregnant animals matched inhalation and intravenous datasets better than gavage data. The adult model was initially better able to predict the time-course of blood concentrations than was the neonatal model. However, after accounting for age-related changes in gastric uptake using the calibrated neonate model, simulations consistently reproduced the early kinetic behavior in blood. This work provides comprehensive multi-route life-stage models of EtOH pharmacokinetics and represents a first step in development of models for use with gasoline-EtOH blends, with additional potential applicability in investigation of the pharmacokinetics of EtOH abuse, addiction, and toxicity.

To provide useful alternatives to in vivo animal studies, in vitro assays for dose-response assessments of xenobiotic chemicals must use concentrations in media and target tissues that are within biologically-plausible limits. Determining these concentrations is a complex matter, which can be facilitated by applying physiologically-based pharmacokinetic (PBPK) models in an in vitro to in vivo extrapolation (IVIVE) paradigm. We used ethanol (EtOH), a ubiquitous chemical with defined metrics for in vivo and in vitro embryotoxicity, as a model chemical to evaluate this paradigm. A published series of life-stage PBPK models for rats was extended to mice, yielding simulations that adequately predicted in vivo blood EtOH concentrations (BECs) from oral, intraperitoneal, and intravenous routes in nonpregnant and pregnant adult mice. The models were then extrapolated to nonpregnant and pregnant humans, replicating BEC data within a factor of two. The rodent models were then used to conduct IVIVEs for rodent and whole-embryo culture embryotoxicity data (neural tube closure defects, morphological changes). A second IVIVE was conducted for exposure scenarios in pregnant women during critical windows of susceptibility for developmental toxicity, such as the first 6-to-8 weeks (prerecognition period) or mid-to-late pregnancy period, when EtOH consumption is associated with fetal alcohol spectrum disorders. Incorporation of data from human embryonic stem cell studies led to a model-supported linkage of in vitro concentrations with plausible exposure ranges for pregnant women. This effort demonstrates benefits and challenges associated with use of multispecies PBPK models to estimate in vivo tissue concentrations associated with in vitro embryotoxicity studies.

People are often exposed to complex mixtures of environmental chemicals such as gasoline, tobacco smoke, water contaminants, or food additives. We developed an approach that applies chemical lumping methods to complex mixtures, in this case gasoline, based on biologically relevant parameters used in physiologically based pharmacokinetic (PBPK) modeling. Inhalation exposures were performed with rats to evaluate the performance of our PBPK model and chemical lumping method. There were 109 chemicals identified and quantified in the vapor in the chamber. The time-course toxicokinetic profiles of 10 target chemicals were also determined from blood samples collected during and following the in vivo experiments. A general PBPK model was used to compare the experimental data to the simulated values of blood concentration for 10 target chemicals with various numbers of lumps, iteratively increasing from 0 to 99. Large reductions in simulation error were gained by incorporating enzymatic chemical interactions, in comparison to simulating the individual chemicals separately. The error was further reduced by lumping the 99 nontarget chemicals. The same biologically based lumping approach can be used to simplify any complex mixture with tens, hundreds, or thousands of constituents.