Computational toxicology of chloroform: reverse dosimetry using Bayesian inference, Markov chain Monte Carlo simulation, and human biomonitoring data

BACKGROUND: One problem of interpreting population-based biomonitoring data is the reconstruction of corresponding external exposure in cases where no such data are available. OBJECTIVES: We demonstrate the use of a computational framework that integrates physiologically based pharmacokinetic (PBPK) modeling, Bayesian inference, and Markov chain Monte Carlo simulation to obtain a population estimate of environmental chloroform source concentrations consistent with human biomonitoring data. The biomonitoring data consist of chloroform blood concentrations measured as part of the Third National Health and Nutrition Examination Survey (NHANES III), and for which no corresponding exposure data were collected. METHODS: We used a combined PBPK and shower exposure model to consider several routes and sources of exposure: ingestion of tap water, inhalation of ambient household air, and inhalation and dermal absorption while showering. We determined posterior distributions for chloroform concentration in tap water and ambient household air using U.S. Environmental Protection Agency Total Exposure Assessment Methodology (TEAM) data as prior distributions for the Bayesian analysis. RESULTS: Posterior distributions for exposure indicate that 95% of the population represented by the NHANES III data had likely chloroform exposures < or = 67 microg/L [corrected] in tap water and < or = 0.02 microg/L in ambient household air. CONCLUSIONS: Our results demonstrate the application of computer simulation to aid in the interpretation of human biomonitoring data in the context of the exposure-health evaluation-risk assessment continuum. These results should be considered as a demonstration of the method and can be improved with the addition of more detailed data.     

Reconstructing population exposures from dose biomarkers: inhalation of trichloroethylene (TCE) as a case study

Physiologically based pharmacokinetic (PBPK) modeling is a well-established toxicological tool designed to relate exposure to a target tissue dose. The emergence of federal and state programs for environmental health tracking and the availability of exposure monitoring through biomarkers creates the opportunity to apply PBPK models to estimate exposures to environmental contaminants from urine, blood, and tissue samples. However, reconstructing exposures for large populations is complicated by often having too few biomarker samples, large uncertainties about exposures, and large interindividual variability. In this paper, we use an illustrative case study to identify some of these difficulties, and for a process for confronting them by reconstructing population-scale exposures using Bayesian inference. The application consists of interpreting biomarker data from eight adult males with controlled exposures to trichloroethylene (TCE) as if the biomarkers were random samples from a large population with unknown exposure conditions. The TCE concentrations in blood from the individuals fell into two distinctly different groups even though the individuals were simultaneously in a single exposure chamber. We successfully reconstructed the exposure scenarios for both subgroups - although the reconstruction of one subgroup is different than what is believed to be the true experimental conditions. We were however unable to predict with high certainty the concentration of TCE in air.     

Monte Carlo modeling in CT-based geometries: dosimetry for biological modeling experiments with particle beam radiation

The space radiation environment imposes increased dangers of exposure to ionizing radiation, particularly during a solar particle event (SPE). These events consist primarily of low energy protons that produce a highly inhomogeneous dose distribution. Due to this inherent dose heterogeneity, experiments designed to investigate the radiobiological effects of SPE radiation present difficulties in evaluating and interpreting dose to sensitive organs. To address this challenge, we used the Geant4 Monte Carlo simulation framework to develop dosimetry software that uses computed tomography (CT) images and provides radiation transport simulations incorporating all relevant physical interaction processes. We found that this simulation accurately predicts measured data in phantoms and can be applied to model dose in radiobiological experiments with animal models exposed to charged particle (electron and proton) beams. This study clearly demonstrates the value of Monte Carlo radiation transport methods for two critically interrelated uses: (i) determining the overall dose distribution and dose levels to specific organ systems for animal experiments with SPE-like radiation, and (ii) interpreting the effect of random and systematic variations in experimental variables (e.g. animal movement during long exposures) on the dose distributions and consequent biological effects from SPE-like radiation exposure. The software developed and validated in this study represents a critically important new tool that allows integration of computational and biological modeling for evaluating the biological outcomes of exposures to inhomogeneous SPE-like radiation dose distributions, and has potential applications for other environmental and therapeutic exposure simulations.     

Interpreting variability in population biomonitoring data: role of elimination kinetics

Biomarker concentrations in spot samples of blood and urine are implicitly interpreted as direct surrogates for long-term exposure magnitude in a variety of contexts including (1) epidemiological studies of potential health outcomes associated with general population chemical exposure, and (2) cross-sectional population biomonitoring studies. However, numerous factors in addition to exposure magnitude influence biomarker concentrations in spot samples, including temporal variation in spot samples because of elimination kinetics. The influence of half-life of elimination relative to exposure interval is examined here using simple first-order pharmacokinetic simulations of urinary concentrations in spot samples collected at random times relative to exposure events. Repeated exposures were modeled for each individual in the simulation with exposure amounts drawn from lognormal distributions with varying geometric standard deviations. Relative variation in predicted spot sample concentrations was greater than the variation in underlying dose distributions when the half-life of elimination was shorter than the interval between exposures, with the degree of relative variation increasing as the ratio of half-life to exposure interval decreased. Results of the modeling agreed well with data from a serial urine collection data set from the Centers for Disease Control. Data from previous studies examining intra-class correlation coefficients for a range of chemicals relying upon repeated sampling support the importance of considering the half-life relative to exposure frequency in design and interpretation of studies using spot samples for exposure classification and exposure estimation. The modeling and data sets presented here provide tools that can assist in interpretation of variability in cross-sectional biomonitoring studies and in design of studies utilizing biomonitoring data as markers for exposure.     

Human blood concentrations of cotinine, a biomonitoring marker for tobacco smoke, extrapolated from nicotine metabolism in rats and humans and physiologically based pharmacokinetic modeling

The present study defined a simplified physiologically based pharmacokinetic (PBPK) model for nicotine and its primary metabolite cotinine in humans, based on metabolic parameters determined in vitro using relevant liver microsomes, coefficients derived in silico, physiological parameters derived from the literature, and an established rat PBPK model. The model consists of an absorption compartment, a metabolizing compartment, and a central compartment for nicotine and three equivalent compartments for cotinine. Evaluation of a rat model was performed by making comparisons with predicted concentrations in blood and in vivo experimental pharmacokinetic values obtained from rats after oral treatment with nicotine (1.0 mg/kg, a no-observed-adverseeffect level) for 14 days. Elimination rates of nicotine in vitro were established from data from rat liver microsomes and from human pooled liver microsomes. Human biomonitoring data (17 ng nicotine and 150 ng cotinine per mL plasma 1 h after smoking) from pooled five male Japanese smokers (daily intake of 43 mg nicotine by smoking) revealed that these blood concentrations could be calculated using a human PBPK model. These results indicate that a simplified PBPK model for nicotine/cotinine is useful for a forward dosimetry approach in humans and for estimating blood concentrations of other related compounds resulting from exposure to low chemical doses.     

Implications for epidemiologic research on variation by pesticide in studies of farmers and their families

Epidemiologic studies rely on participants' recall to classify their exposure to specific pesticides. As exposure classification evolves, an important issue is whether the recall of pesticide application details can be used to derive measures of exposure intensity or cumulative exposure. A preliminary analysis of biomonitoring data for farmers before, during, and after a pesticide application suggests variation for different pesticides in the proportion with detectable urinary concentrations, urinary levels, and patterns of uptake and elimination. These findings, and the limited predictive modeling done to date, suggest that chemical-specific differences need to be considered in exposure classification schemes. An analysis of biomonitoring data for farm spouses and children found few with appreciable changes in the urinary concentration after a pesticide application. These findings point to the need to validate assumptions about exposures in studies of people who are not directly involved in pesticide application.     

The Implications of Using a Physiologically Based Pharmacokinetic (PBPK) Model for Pesticide Risk Assessment

Background

A physiologically based pharmacokinetic (PBPK) model would make it possible to simulate the dynamics of chemical absorption, distribution, metabolism, and elimination (ADME) from different routes of exposures and, in theory, could be used to evaluate associations between exposures and biomarker measurements in blood or urine.

Objective

We used a PBPK model to predict urinary excretion of 3,5,6-trichloro-2-pyridinol (TCPY), the specific metabolite of chlorpyrifos (CPF), in young children.

Methods

We developed a child-specific PBPK model for CPF using PBPK models previously developed for rats and adult humans. Data used in the model simulation were collected from 13 children 3–6 years of age who participated in a cross-sectional pesticide exposure assessment study with repeated environmental and biological sampling.

Results

The model-predicted urinary TCPY excretion estimates were consistent with measured levels for 2 children with two 24-hr duplicate food samples that contained 350 and 12 ng/g of CPF, respectively. However, we found that the majority of model outputs underpredicted the measured urinary TCPY excretion.

Conclusions

We concluded that the potential measurement errors associated with the aggregate exposure measurements will probably limit the applicability of PBPK model estimates for interpreting urinary TCPY excretion and absorbed CPF dose from multiple sources of exposure. However, recent changes in organophosphorus (OP) use have shifted exposures from multipathways to dietary ingestion only. Thus, we concluded that the PBPK model is still a valuable tool for converting dietary pesticide exposures to absorbed dose estimates when the model input data are accurate estimates of dietary pesticide exposures.     

Interpreting human biomonitoring data in a public health risk context using Biomonitoring Equivalents

Human biomonitoring (HBM) has proven an extremely valuable tool for determining which chemicals are getting into people, detecting trends in population exposures over time, and identifying populations with exposures above background. The potential significance of the HBM data in the context of existing toxicology data and risk assessments can be assessed if chemical-specific quantitative screening criteria are available. Such screening criteria would ideally be based on robust datasets relating potential adverse effects to biomarker concentrations in human populations. However, such assessments are data intensive and exist for only a few chemicals. As an interim approach, the concept of Biomonitoring Equivalents (BEs) has been developed. A Biomonitoring Equivalent (BE) is defined as the concentration or range of concentrations of a chemical or its metabolites in a biological medium (blood, urine, or other medium) that is consistent with an existing health-based exposure guidance value such as a Reference Dose (RfD) or Tolerable or Acceptable Daily Intake (TDI or ADI). This paper provides an overview of the derivation of BEs and how BEs can be used to interpret human biomonitoring data in a public health risk context.     

The U.S. National Health and Nutrition Examination Survey and human exposure to environmental chemicals

Researchers are increasingly interested in using human biomonitoring - the measurement of chemicals, their metabolites or specific reaction products in biological specimens/body fluids - for investigating exposure to environmental chemicals. General population human biomonitoring programs are useful for investigating human exposure to environmental chemicals and an important tool for integrating environment and health. One of these programs, the National Health and Nutrition Examination Survey (NHANES), conducted in the United States is designed to collect data on the health and nutritional status of the noninstitutionalized, civilian U.S. population. NHANES includes a physical examination, collecting a detailed medical history, and collecting biological specimens (i.e., blood and urine). These biological specimens can be used to assess exposure to environmental chemicals. NHANES human biomonitoring data can be used to establish reference ranges for selected chemicals, provide exposure data for risk assessment, and monitor exposure trends.     

Assessment of population exposure to air pollution by benzene

Biomonitoring is one of the methods that allow to identify population groups that have significantly higher exposures to a particular chemical than the general population. However, use of biomonitoring is particularly useful when applied in combination with other methods of pollution exposure assessment. The current study is focused on the developing of the modelling approach to estimate population exposure to benzene through inhalation. The model is based on a microenvironment approach and is adapted to be applied in urban areas where the pattern of exposure is complex. The results provided by the model may be used in combination with human biomonitoring in order to select who and where should monitoring be done, as well as for interpretation and extrapolation of biomonitoring results.     

Dose to the cell nucleus from exposure to tritiated pump oil or formaldehyde

A Monte Carlo simulation of tritium decays in a cell composed of two parts, a nucleus and surrounding cytoplasm, was developed to evaluate the beta-radiation dose to the nucleus. A dose modifying factor (DMF), which is a ratio of the average nuclear dose to the whole-tissue dose, after skin-contact exposure of rats to tritiated pump oil or tritiated formaldehyde was estimated. Biokinetic data characterizing the retention of tritium in liver were available in the form of tritium-specific activities and biological half-times for tritiated water and five macromolecular species (DNA, RNA, acid-soluble fraction, acid-insoluble protein, and lipids). The spatial distribution of tissue-free water and macromolecular species in the nucleus and cytoplasm of rat liver cells was based on published data. In the case of exposure to tritiated pump oil, tritium incorporated into lipids provides the largest percentage (60%) of the absorbed dose to the nucleus. For the tritiated-formaldehyde exposure, the tritium dose to the nucleus is overwhelmingly contributed by tritiated water (58%) and in acid-insoluble proteins (40%). For both these tritiated organic exposures, the tritium-labeled DNA has a negligible effect on the DMF. The DMF for the tritiated pump oil and formaldehyde exposures was estimated as 0.81 and 1.05, respectively: the DMF of both exposures was close to unity. Given the other uncertainties in tritium dosimetry, our results suggest that for these skin-contact exposures a uniform distribution of tritium in tissue is an adequate assumption for dosimetry.     

User subjectivity in Monte Carlo modeling of pesticide exposure

Monte Carlo techniques are increasingly used in pesticide exposure modeling to evaluate the uncertainty in predictions arising from uncertainty in input parameters and to estimate the confidence that should be assigned to the modeling results. The approach typically involves running a deterministic model repeatedly for a large number of input values sampled from statistical distributions. In the present study, six modelers made choices regarding the type and parameterization of distributions assigned to degradation and sorption data for an example pesticide, the correlation between the parameters, the tool and method used for sampling, and the number of samples generated. A leaching assessment was carried out using a single model and scenario and all data for sorption and degradation generated by the six modelers. The distributions of sampled parameters differed between the modelers, and the agreement with the measured data was variable. Large differences were found between the upper percentiles of simulated concentrations in leachate. The probability of exceeding 0.1 microg/L ranged from 0 to 35.7%. The present study demonstrated that subjective choices made in Monte Carlo modeling introduce variability into probabilistic modeling and that the results need to be interpreted with care.     

Evaluation of the U.S. EPA/OSWER preliminary remediation goal for perchlorate in groundwater: focus on exposure to nursing infants

BACKGROUND: Perchlorate is a common contaminant of drinking water and food. It competes with iodide for uptake into the thyroid, thus interfering with thyroid hormone production. The U.S. Environmental Protection Agency's Office of Solid Waste and Emergency Response (OSWER) set a groundwater preliminary remediation goal (PRG) of 24.5 microg/L to prevent exposure of pregnant women that would affect the fetus. This does not account for the greater exposure that is possible in nursing infants or for the relative source contribution (RSC), a factor normally used to lower the PRG due to nonwater exposures. OBJECTIVES: Our goal was to assess whether the OSWER PRG protects infants against exposures from breast-feeding, and to evaluate the perchlorate RSC. METHODS: We used Monte Carlo analysis to simulate nursing infant exposures associated with the OSWER PRG when combined with background perchlorate. RESULTS: The PRG can lead to a 7-fold increase in breast milk concentration, causing 90% of nursing infants to exceed the reference dose (RfD) (average exceedance, 2.8-fold). Drinking-water perchlorate must be < 6.9 microg/L to keep the median, and < 1.3 microg/L to keep the 90th-percentile nursing infant exposure below the RfD. This is 3.6- to 19-fold below the PRG. Analysis of biomonitoring data suggests an RSC of 0.7 for pregnant women and of 0.2 for nursing infants. Recent data from the Centers for Disease Control and Prevention (CDC) suggest that the RfD itself needs to be reevaluated because of hormonal effects in the general population. CONCLUSIONS: The OSWER PRG for perchlorate can be improved by considering infant exposures, by incorporating an RSC, and by being responsive to any changes in the RfD resulting from the new CDC data.     

Biological monitoring: the role of toxicokinetics and physiologically based pharmacokinetic modeling

This short review outlines the contribution of modeling techniques, particularly physiologically based pharmacokinetic (PBPK) modeling, in promulgating biological monitoring as a practical tool for the occupational health professional. The impact of modeling techniques is discussed in helping to establish the relevant biomarkers to measure, the appropriate time of sampling, and the relationship between atmospheric exposure limits and concentration of biological analyte. Of particular interest is the use of "population" PBPK techniques. These can explore the influence of physiological differences between workers or of particular susceptible subgroups (e.g., pregnant women, breast-feeding mothers, and infants) on the relationship between atmospheric exposure levels and biomarker concentration. Such techniques will become more widely used as biological monitoring guidance values (e.g., biological exposure indices, biological tolerance values) are increasingly established by various international professional and regulatory bodies.     

Use of biomonitoring data to evaluate methyl eugenol exposure

Methyl eugenol is a naturally occurring material found in a variety of food sources, including spices, oils, and nutritionally important foods such as bananas and oranges. Given its natural occurrence, a broad cross-section of the population is likely exposed. The availability of biomonitoring and toxicology data offers an opportunity to examine how biomonitoring data can be integrated into risk assessment. Methyl eugenol has been used as a biomarker of exposure. An analytical method to detect methyl eugenol in human blood samples is well characterized but not readily available. Human studies indicate that methyl eugenol is short-lived in the body, and despite the high potential for exposure through the diet and environment, human blood levels are relatively low. The toxicology studies in animals demonstrate that relatively high-bolus doses administered orally result in hepatic neoplasms. However, an understanding is lacking regarding how this effect relates to the exposures that result when food containing methyl eugenol is consumed. Overall, the level of methyl eugenol detected in biomonitoring studies indicates that human exposure is several orders of magnitude lower than the lowest dose used in the bioassay. Furthermore, there are no known health effects in humans that result from typical dietary exposure to methyl eugenol.     

Evaluation of potential toxicity from co-exposure to three CNS depressants (toluene, ethylbenzene, and xylene) under resting and working conditions using PBPK modeling

Under OSHA and American Conference of Governmental Industrial Hygienists (ACGIH) guidelines, the mixture formula (unity calculation) provides a method for evaluating exposures to mixtures of chemicals that cause similar toxicities. According to the formula, if exposures are reduced in proportion to the number of chemicals and their respective exposure limits, the overall exposure is acceptable. This approach assumes that responses are additive, which is not the case when pharmacokinetic interactions occur. To determine the validity of the additivity assumption, we performed unity calculations for a variety of exposures to toluene, ethylbenzene, and/or xylene using the concentration of each chemical in blood in the calculation instead of the inhaled concentration. The blood concentrations were predicted using a validated physiologically based pharmacokinetic (PBPK) model to allow exploration of a variety of exposure scenarios. In addition, the Occupational Safety and Health Administration and ACGIH occupational exposure limits were largely based on studies of humans or animals that were resting during exposure. The PBPK model was also used to determine the increased concentration of chemicals in the blood when employees were exercising or performing manual work. At rest, a modest overexposure occurs due to pharmacokinetic interactions when exposure is equal to levels where a unity calculation is 1.0 based on threshold limit values (TLVs). Under work load, however, internal exposure was 87%higher than provided by the TLVs. When exposures were controlled by a unity calculation based on permissible exposure limits (PELs), internal exposure was 2.9 and 4.6 times the exposures at the TLVs at rest and workload, respectively. If exposure was equal to PELs outright, internal exposure was 12.5 and 16 times the exposure at the TLVs at rest and workload, respectively. These analyses indicate the importance of (1) selecting appropriate exposure limits, (2) performing unity calculations, and (3) considering the effect of work load on internal doses, and they illustrate the utility of PBPK modeling in occupational health risk assessment.     

Population Physiologically Based Pharmacokinetic Modeling for the Human Lactational Transfer of PCB-153 with Consideration of Worldwide Human Biomonitoring Results

Background

One of the most serious human health concerns related to environmental contamination with polychlorinated biphenyls (PCBs) is the presence of these chemicals in breast milk.

Objectives

We developed a physiologically based pharmacokinetic model of PCB-153 in women, and predict its transfer via lactation to infants. The model is the first human, population-scale lactational model for PCB-153. Data in the literature provided estimates for model development and for performance assessment.

Methods

We used physiologic parameters from a cohort in Taiwan and reference values given in the literature to estimate partition coefficients based on chemical structure and the lipid content in various body tissues. Using exposure data from Japan, we predicted acquired body burden of PCB-153 at an average childbearing age of 25 years and compared predictions to measurements from studies in multiple countries. We attempted one example of reverse dosimetry modeling using our PBPK model for possible exposure scenarios in Canadian Inuits, the population with the highest breast milk PCB-153 level in the world.

Results

Forward-model predictions agree well with human biomonitoring measurements, as represented by summary statistics and uncertainty estimates.

Conclusion

The model successfully describes the range of possible PCB-153 dispositions in maternal milk, suggesting a promising option for back-estimating doses for various populations.     

Occupational exposures to influenza among healthcare workers in the United States

The objective of this study is to estimate the annual number of occupational exposures to influenza among healthcare workers that result from providing direct and supportive care to influenza patients in acute care, home care and long-term care settings. Literature review was used to identify healthcare utilization for influenza, and worker activity patterns. This information was used, with Monte Carlo simulation, to tabulate the mean annual number of occupational exposures. Given a medium-sized epidemic with a 6% annual symptomatic influenza incidence proportion, the mean number of occupational exposures was estimated to be 81.8 million annually. Among the approximately 14 million healthcare workers, this corresponds to 5.8 exposures per worker annually, on average. Exposures, however, are likely concentrated among subsets of healthcare workers. Occupational exposures were most numerous in ambulatory care settings (38%), followed by long-term care facilities (30%) and home care settings (21%). The annual number of occupational exposures to influenza is high, but not every occupational exposure will result in infection. Some infection control activities, like patient isolation, can reduce the number of occupational exposures.     

Quantifying children's aggregate (dietary and residential) exposure and dose to permethrin: application and evaluation of EPA's probabilistic SHEDS-Multimedia model

Reliable, evaluated human exposure and dose models are important for understanding the health risks from chemicals. A case study focusing on permethrin was conducted because of this insecticide's widespread use and potential health effects. SHEDS-Multimedia was applied to estimate US population permethrin exposures for 3- to 5-year-old children from residential, dietary, and combined exposure routes, using available dietary consumption data, food residue data, residential concentrations, and exposure factors. Sensitivity and uncertainty analyses were conducted to identify key factors, pathways, and research needs. Model evaluation was conducted using duplicate diet data and biomonitoring data from multiple field studies, and comparison to other models. Key exposure variables were consumption of spinach, lettuce, and cabbage; surface-to-skin transfer efficiency; hand mouthing frequency; fraction of hand mouthed; saliva removal efficiency; fraction of house treated; and usage frequency. For children in households using residential permethrin, the non-dietary exposure route was most important, and when all households were included, dietary exposure dominated. SHEDS-Multimedia model estimates compared well to real-world measurements data; this exposure assessment tool can enhance human health risk assessments and inform children's health research. The case study provides insights into children's aggregate exposures to permethrin and lays the foundation for a future cumulative pyrethroid pesticides risk assessment.     

Accommodating the ecological fallacy in disease mapping in the absence of individual exposures

In health exposure modeling, in particular, disease mapping, the ecological fallacy arises because the relationship between aggregated disease incidence on areal units and average exposure on those units differs from the relationship between the event of individual incidence and the associated individual exposure. This article presents a novel modeling approach to address the ecological fallacy in the least informative data setting. We assume the known population at risk with an observed incidence for a collection of areal units and, separately, environmental exposure recorded during the period of incidence at a collection of monitoring stations. We do not assume any partial individual level information or random allocation of individuals to observed exposures. We specify a conceptual incidence surface over the study region as a function of an exposure surface resulting in a stochastic integral of the block average disease incidence. The true block level incidence is an unavailable Monte Carlo integration for this stochastic integral. We propose an alternative manageable Monte Carlo integration for the integral. Modeling in this setting is immediately hierarchical, and we fit our model within a Bayesian framework. To alleviate the resulting computational burden, we offer 2 strategies for efficient model fitting: one is through modularization, the other is through sparse or dimension‐reduced Gaussian processes. We illustrate the performance of our model with simulations based on a heat‐related mortality dataset in Ohio and then analyze associated real data.