Society For Risk Analysis Annual Meeting 2017

Session Schedule & Abstracts


* Disclaimer: All presentations represent the views of the authors, and not the organizations that support their research. Please apply the standard disclaimer that any opinions, findings, and conclusions or recommendations in abstracts, posters, and presentations at the meeting are those of the authors and do not necessarily reflect the views of any other organization or agency. Meeting attendees and authors should be aware that this disclaimer is intended to apply to all abstracts contained in this document. Authors who wish to emphasize this disclaimer should do so in their presentation or poster. In an effort to make the abstracts as concise as possible and easy for meeting participants to read, the abstracts have been formatted such that they exclude references to papers, affiliations, and/or funding sources. Authors who wish to provide attendees with this information should do so in their presentation or poster.

Common abbreviations

T2-I
New Models for Dose-Response

Room: Salon K   10:30 am–12:00 pm

Chair(s): Allen Davis   davis.allen@epa.gov

Sponsored by Dose Response Specialty Group



T2-I.1  10:30 am  Rat and human PBPK model for malathion: application for risk assessment. Reiss R*, Loccisano A, Whatling P, Wang W; Exponent (1,2), FMC Corporation (3,4)   rreiss@exponent.com

Abstract: Malathion is an effective organophosphorus insecticides and has been used on the market for decades. one of the most widely used insecticides in the United States and-It is currently under reregistration review at the U.S. Environmental Protection Agency (EPA). In recent years, physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling has been established as a scientifically sound tool for predicting internal exposure and the new approach has been well-received by EPA for estimating the uncertainty in human health risk assessment. The PBPK/PD models have been used to develop chemical-specific uncertainty factors for interspecies extrapolation and intraspecies variability. This paper describes the development of rat and human PBPK/PD models for malathion for acetylcholinesterase inhibition, the most sensitive toxicological endpoint for malathion. Data of the chemical, physiological and biochemical properties are available for developing the models. In addition, anAn in vitro testing program was has been undertaken to develop kinetic constants for malathion activation to its active toxicity moiety malaoxon, detoxication of malathion and malaoxon, and inhibition of acetylcholinesterase. The detailed architecture of the PBPK/PD model will be described, including the malathion/malaoxon metabolism scheme, principally through carboxylesterase, and inhibition, reactivation, and aging of acetylcholinesterase and butrylcholinesterase. The model can estimate acetylcholinesterase inhibition for oral and dermal exposures for malathion and oral exposures for malaoxon. The presentation will describe the use of the PBPK/PD models to estimate chemical specific uncertainty factors for pharmacokinetics and pharmacodynamics between rats and humans and within the human population.

T2-I.2  10:50 am  Assessing Uncertainty and Variability in Biochemical Parameters in a PBTK Model for Perchlorate. Kapraun DF*, Schlosser PM; US Environmental Protection Agency   kapraun.dustin@epa.gov

Abstract: Physiologically based toxicokinetic (PBTK) models can serve an important role in chemical risk assessments because they allow for extrapolations across different species and dosing regimens and between in vitro and in vivo conditions. In order to make predictions, however, such models require various physiological and chemical-specific parameters that can only be known to finite precision and whose values tend to differ between individuals in a population. We sought to quantify intra-individual uncertainty and inter-individual variability for four biochemical parameters in a PBTK model relating perchlorate exposures and thyroid hormone levels in women of child-bearing age. Using data from a previously published controlled human perchlorate exposure study, we applied Markov chain Monte Carlo (MCMC) sampling to obtain Bayesian distributional estimates of the following: urinary clearance rates for (1) perchlorate and (2) iodide, (3) the Michaelis-Menten constant for perchlorate’s competitive inhibition of thyroid iodide uptake, and (4) the maximum thyroid iodide uptake rate. The resulting probability distributions for these four parameters for each of 12 adult female subjects illustrate intra-individual uncertainty and inter-individual variability. Unlike point (single value) estimates, distributional estimates allow for quantification of uncertainty and variability in the parameters. Furthermore, propagating parameter uncertainty through a PBTK model allows one to quantify uncertainty and variability in model outputs, such as serum concentrations of perchlorate or thyroid hormones. Using credible intervals for quantities of interest that are based on distributional estimates provides an alternative to the standard practice of applying uncertainty factors, and may improve confidence in risk analyses and assessment conclusions. The views expressed herein do not necessarily reflect the views or policies of the US EPA.

T2-I.3  11:10 am  Dose-Response Assessment of Arsenic in Drinking Water: A Bayesian Network Model of Diabetes Risks. MacDonald-Gibson J*, Zabinski J; University of North Carolina at Chapel Hill   jackie.macdonald@unc.edu

Abstract: Regulatory risk assessment requires dose-response models that accurately link exposure to toxicants to the probability of adverse health outcomes. In current U.S. practice for regulating contaminants in drinking water, dose-response models are binary: they assume a threshold (the reference dose) above which risk is presumed present and below which it is presumed absent. This approach does not allow for the computation of a quantitative risk measure that can be used in comparing health benefits of programs to reduce toxicant exposure. In this research, we demonstrate the use of an alternative dose-response assessment approach based on Bayesian belief networks for quantifying noncancer risks from arsenic in drinking water. Using a data set of 1,050 individuals from an arsenic-endemic region of Chihuahua, Mexico, we fit a Bayesian belief network model to estimate the risks of pre-diabetes and diabetes from arsenic exposure, biomarkers of arsenic metabolism, and demographic characteristics. Using a training-testing approach, we compare the predictive ability of the Bayesian network model to that of a reference dose model and to a logistic regression model. We find that the Bayesian network model achieves a higher sum of sensitivity and specificity than the reference dose and logistic regression models. The results could inform the development of improved approaches for dose-response assessment of chemicals in drinking water.

T2-I.4  11:30 am  Impact of Generalized Informative Prior on BMD Estimation using Dichotomous Data. Shao K*; Indiana University   kshao@indiana.edu

Abstract: Historical toxicological data can provide risk assessors invaluable information regarding the shape of dose-response curves, which is especially useful when the data being analyzed have limited dose-response information. The Bayesian dose-response modeling framework is an important tool for combining prior information from historical studies with the information from the specific dataset being considered. In this study, we employ the recently developed Bayesian benchmark dose (BBMD) estimation system to examine how generalized informative prior distributions can impact BMD estimates. We first use 518 toxicological data sets (mainly from the IRIS database, including cancer and noncancer) to identify an appropriate distribution for each parameter in commonly used dichotomous dose-response models. Then, to mitigate the estimation error and sensitivity to potential data errors in specific data sets included in the analysis, a Bayesian hierarchical model is applied to estimate the distribution of the parameters in these dose-response models. These resulting distributions of model parameters are used as generalized informative prior for further analyses. Both real data analyses and simulation studies are employed to examine how significantly the informative priors can impact BMD estimates individually. Preliminary results show that the background and potency parameters can usually be adequately informed by the data being analyzed (i.e., informative priors generally have limited impact on these two parameters). On the other hand, an informative prior for the shape parameter can be greatly beneficial for the precision of the estimated BMD.



[back to schedule]