ADVANCEMENT OF THE PRACTICE
TABLE 1
Summary of the Use, Strengths, and Limitations of Simulation Science Tools Used at the Agency for Toxic Substances and Disease Registry
Method Use
Strengths
Limitations
Physiologically-based pharmacokinetic (PBPK) modeling estimates the absorption, distribution, metabolism, and excretion of a chemical by simulating the body as a series of compartments
• Allows for both cross-route and cross-duration extrapolation • Helps extrapolate data gaps in physicochemical characteristics • Can be useful in emergency response
• Requires appropriate modeling expertise, experience, and training • Lack of confidence in PBPK models for which no tissue or plasma concentration data exist for model evaluation • Lack of transferability across modeling platforms • Chemical-specific data quality and availability • Lack of data for model validation • Requires appropriate modeling expertise, experience, and training • Data quality and availability • The applicability domain of the model • Variable selection and overfeeding • Interpretability • External validation and assumptions of the model • Requires appropriate modeling expertise, experience, and training • Complexity of biological systems • Data quality and availability • Computational power • Integration of multiscale data • Requires appropriate modeling expertise, experience, and training • Complexity of statistical methods • Model selection and benchmark response selection • Data quality and quantity of experimental data • Requires appropriate modeling expertise, experience, and training • Complexity of environmental systems
Quantitative structure-activity relationships (QSAR) modeling predicts the biological activity or toxicity of chemicals based on chemical structure information
• Serves as a starting point to discuss toxicity when no experimental data are available • Assists in prioritization of chemicals for further testing • Can be useful in emergency response
Computational systems biology modeling enables the identification of potential linkages between environmental toxicants, biological pathways, and health outcomes
• Justifies mechanistic hypotheses that can then be investigated via experimentation
Benchmark dose (BMD) modeling considers all available data from an experiment for determining the point of departure for risk assessment
• Uses data from multiple studies to derive the point of departure • Increases consistency in comparing results across studies
Fate and transport modeling estimates cumulative exposure to hazardous substances for persons expected to have come into contact with hazardous waste sites
• Allows for estimation of both past and current exposures
• Data quality and availability • Spatial and temporal scales • Intermedia transfers • Chemical-specific properties challenges
• Risk Assessment: Computational models assist in quantifying the risks associated with specific exposures by integrating information on toxicity, exposure lev- els, and population characteristics. This modeling helps inform risk management decisions and public health interventions. PBPK modeling extrapolates toxicological information across species and sensitive populations and further allows cross- route and cross-duration extrapolation. For example, PBPK modeling can derive an oral chronic HGV based on intermedi-
ate inhalation data. PBPK modeling can also eliminate the need for repeat animal or human studies that require extensive time and resource allocations for HGV estimation. Consequently, PBPK model- ing greatly facilitates chemical risk assess- ment and benefits public health. ATSDR developed a PBPK tool kit for high-pri- ority chemicals by recoding advanced and ecient PBPK models (Mumtaz et al., 2012; Ruiz et al., 2010). The recoded models are more accessible to public health assessors and are more applicable
to the general population (Emond et al., 2017; Satarug et al., 2013). •Prioritizing Research Needs: Computa- tional models can screen large numbers of chemicals for potential hazards, helping to prioritize which substances require fur- ther investigation or monitoring. Scientists from the Simulation Science Section have been involved in eorts to further compu- tational approaches in various collabora- tive projects (Mansouri et al., 2020; Myatt et al., 2018; Satarug et al., 2013; Uwimana et al., 2019). Eorts have been made to
34
Volume 87 • Number 4
Powered by FlippingBook