Using publicly available data, a physiologically-based pharmacokinetic model and Bayesian simulation to improve arsenic non-cancer dose-response
Graphical abstract
Introduction
Chronic exposure to elevated levels of arsenic (As) has resulted in many adverse effects appearing in humans (Maull et al., 2012, Naujokas et al., 2013). Epidemiological evidence provides opportunities to undertake a dose-response study, and furthermore to assist in assessment and management. For example, one study over a mean follow-up period of 9.7 years for 52,931 eligible participants suggested that the adjusted incidence rate ratios per 1 μg/L increment in arsenic levels in drinking water were 1.03 (95% confidence interval (CI): 1.01, 1.06) for all diabetes cases (Bräuner et al., 2014). Such epidemiological studies have convincingly linked the As exposure level and risk (Bräuner et al., 2014, U.S. EPA, 1988).
Excepting exposure level, previous research has also demonstrated the incidence of diseases increases with exposure duration (Liao et al., 2008, Mazumder et al., 1998, U.S. EPA, 1988). To quantify the exposure duration effects, mathematical functions (such as Weibull and Hill functions) have usually been employed, by parameterizing age factor to represent exposure duration effect (Liao et al., 2008, U.S. EPA, 1988). For long-term chronic exposure, since the dose metric emerging from exposure duration is not a linear or explicit variable, it is difficult to address these effects simply based on mathematical parameterization (Hodgson and Darnton, 2000, Philippe and Mansi, 1998). The case study on dioxin has successfully illustrated how to use toxicokinetic model to convert external exposure level and exposure duration into a cumulative dose metric, which was further applied in dose-response study (Becher et al., 1998, Crump et al., 2003). To understand the influence of exposure duration to public health requires a toxicokinetic model to appropriately quantify the impact of exposure duration on delivered dose and ultimately risk in a quantitative dose-response framework.
Several toxicokinetic models have been previously developed (El-Masri and Kenyon, 2008, Liao et al., 2008, Yu, 1999). Based on short-term oral exposures, Yu (1999) developed a seven-compartment physiologically-based pharmacokinetic (PBPK) model for inorganic As (iAs). More recently, El-Masri and Kenyon (2008) published an individual PBPK model that traced the relationships among iAs, monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA) for oral exposure. While these models offered an overview of the absorption, metabolism, distribution and excretion mechanisms in human systems, all such models were developed based on normal people at an individual level. To relate exposure to public health, a PBPK model needs to account for intrinsic heterogeneity at a population and lifetime scale.
Publicly available data have the potential to support the optimization of population PBPK models for use in quantitative risk assessment (Bernillon and Bois, 2000, Lyons et al., 2008), particularly in dose-response study. Specifically, the U.S. FDA has conducted a total diet study (TDS) program to monitor the levels of multiple elements, as well as As, in the country's food supply (Tao and Bolger, 1999). Also, the National Health and Nutrition Examination Survey (NHANES) program was initiated to assess the health and nutritional status of adults and children in the United States (Aylward et al., 2014). Fitting of PBPK models to available data using Bayesian methods such as Markov Chain Monte Carlo (MCMC), these publicly available data can be utilized to bridge As exposure and public health. To the best of our knowledge, this type of research has not previously been attempted and represents a novel interpretation of human health from existing data sets.
In this study, the aim is to illustrate how to integrate publicly available data, PBPK model and Bayesian simulation to refine human health risk assessment, using arsenic as a case study. In particular, the objectives include: 1) assessment of As exposure from U.S. TDS; 2) reporting As biomonitoring information based on the latest U.S. NHANES data (2011 − 2012); 3) optimizing an As population lifetime PBPK model; and 4) improving As non-cancer dose-response study. The newly proposed dose-response study has the potential to protect human health from arsenic exposure.
Section snippets
Procedure for establishing arsenic dose response
As shown in Fig. 1, the procedure for establishing As dose response consisted of three steps. In step 1, a national As exposure assessment was conducted based on TDS data. Then, the urinary As data was retrieved from NHANES database. The As exposure information and urinary As concentration were set as PBPK model input and output, respectively. Therefore a population, lifetime PBPK model was optimized by using Bayesian simulation (step 2). Finally, the optimized PBPK model assisted in As
Exposure estimation
Of the 272 types of food, only the median value of 24 types was above the detection limit (U.S. FDA, 2014). Together with consumption data (U.S. FDA, 2009), the median of daily dietary tAs exposure was estimated to be 0.15 μg/kg/day (body weight was used as 70 kg). Specifically, the values for age groups 0–0.5, 2, 6, 10, 14–16, 25–30, 40–45, 60–65, and 70 + were 0.24, 0.39, 0.19, 0.18, 0.15, 0.16, 0.15, 0.20 and 0.16 μg/kg/day, respectively (Fig. 2, age-specific body weights as presented in Table 1
Limitations and conclusions
Some limitations have been acknowledged in this study. The total exposures considered only diet and drinking water, since it was difficult to trace other pathways. This treatment may bring the bias since this value was used as input to optimize the PBPK model parameters. However, previous studies have demonstrated that diet and drinking water were the major exposures, and such estimations agree well with the biomonitoring in our analysis. Also, only As(III) was used for fitting the model
Acknowledgements
We would like to thank the Cooperative Research Centre for Contamination Assessment and Remediation of the Environment (CRC CARE) for funding support, and the Global Centre for Environmental Remediation, University of Newcastle for use of its facilities.
References (47)
- et al.
Plasmid DNA damage caused by methylated arsenicals, ascorbic acid and human liver ferritin
Toxicol. Lett.
(2002) - et al.
Evaluation of urinary speciated arsenic in NHANES: issues in interpretation in the context of potential inorganic arsenic exposure
Regul. Toxicol. Pharmacol.
(2014) - et al.
Introduction to benchmark dose methods and US EPA's benchmark dose software (BMDS) version 2.1.1
Toxicol. Appl. Pharmacol.
(2011) - et al.
The quantitative risks of mesothelioma and lung cancer in relation to asbestos exposure
Ann. Occup. Hyg.
(2000) - et al.
A Weibull-PBPK model for assessing risk of arsenic-induced skin lesions in children
Sci. Total Environ.
(2008) - et al.
A comprehensive evaluation of inorganic arsenic in food and considerations for dietary intake analyses
Sci. Total Environ.
(2014) - et al.
Human retention studies with 74 As
Toxicol. Appl. Pharmacol.
(1980) - et al.
A market basket survey of inorganic arsenic in food
Food Chem. Toxicol.
(1999) A physiologically based pharmacokinetic model of inorganic arsenic
Regul. Toxicol. Pharmacol.
(1999)- et al.
Dimethylarsine and trimethylarsine are potent genotoxins in vitro
Chem. Res. Toxicol.
(2003)
India's spreading health crisis draws global arsenic experts
Science
Quantitative cancer risk assessment for dioxins using an occupational cohort
Environ. Health Perspect.
Arsenic speciation in human organs following fatal arsenic trioxide poisoning - a case report
Clin. Chem.
Statistical issues in toxicokinetic modeling: a Bayesian perspective
Environ. Health Perspect.
Long-term exposure to low-level arsenic in drinking water and diabetes incidence: a prospective study of the diet, cancer and health cohort
Environ. Health Perspect.
Physiological parameter values for physiologically based pharmacokinetic models
Toxicol. Ind. Health
Methods of Dealing With Values Below the Limit of Detection Using SAS
Meta-analysis of dioxin cancer dose response for three occupational cohorts
Environ. Health Perspect.
Development of lead source-specific exposure standards based on aggregate exposure assessment: Bayesian inversion from biomonitoring information to multipathway exposure
Environ. Sci. Technol.
Dietary Exposure to Inorganic Arsenic in the European Population
Development of a human physiologically based pharmacokinetic (PBPK) model for inorganic arsenic and its mono-and di-methylated metabolites
J. Pharmacokinet. Pharmacodyn.
Computational Toxicology: Methods and Applications for Risk Assessment
Elements in rice from the swedish market: 1. Cadmium, lead and arsenic (total and inorganic)
Food Addit. Contam.
Cited by (17)
Physiologically based kinetic model for assessing intermittent chronic internal exposure to chemicals: Application for disinfection by-products in swimming pool water
2022, Computational ToxicologyCitation Excerpt :The USEPA’s external exposure assessment model is highly effective in conducting screening level risk assessment and regulatory management of THMs in swimming pool waters owing to its simplicity and efficiency. Given that toxicity in humans is determined by the concentration of chemicals in human organs and tissues, it is important to evaluate the internal exposure doses of chemicals, for which physiologically based kinetic (PBK) models have been developed and applied to predict the distribution of chemicals in human bodies [14–22]. Modeling efforts have been made to evaluate time-independent internal exposure doses of THMs during swimming [21,23].
Accumulation, transformation and subcellular distribution of arsenite associated with five carbon nanomaterials in freshwater zebrafish specific-tissues
2021, Journal of Hazardous MaterialsCitation Excerpt :Thus, the interactions between CNMs and other pollutants in aquatic environments have attracted increasing research interest, such as the toxic effects of co-exposure to inorganic arsenic (As) and CNMs (Freixa et al., 2018). As pollution is a serious concern due to its carcinogenicity and widespread occurrence worldwide, especially in aquatic environments (Podgorski and Berg, 2020; Zheng, 2020; Yan et al., 2018; Dong et al., 2016). It has been reported that As levels vary from < 0.001 mg/L to 5.1 mg/L in groundwater resources worldwide (Sharma and Sohn, 2009).
Arsenic toxicokinetic modeling and risk analysis: Progress, needs and applications
2021, ToxicologyCitation Excerpt :A common strength of all these studies is the use of Monte Carlo simulation to explicitly quantify uncertainty/variability in the input data which allows characterization of uncertainty in resulting drinking water limits and risk estimates. Dong et al. (2016) presented a framework in which publicly available exposure (U.S. Total Diet Study) and biomonitoring data (NHANES) were used with Bayesian simulation to optimize a lifetime population PBTK model for ingested arsenic (diet and drinking water). The model was a simplification of earlier PBTK models with most parameters fixed and sensitive parameters (liver:blood partition coefficient AsIII, Vmax AsIII → MMA, urinary elimination constant for AsIII) optimized using Bayesian techniques.
A Hybrid Bayesian Network Framework for Risk Assessment of Arsenic Exposure and Adverse Reproductive Outcomes
2020, Ecotoxicology and Environmental SafetyCitation Excerpt :However, several previous case studies have implemented BN to predict health outcomes. Dong et al. (2016) estimated tAs and iAs based on the U.S. total diet study and National Health and Nutrition Examination Survey data. They used the optimized model to derive an oral reference dose for iAs exposure.
Influence of humic acid on arsenic bioaccumulation and biotransformation to zebrafish: A comparative study between As(III) and As(V) exposure
2020, Environmental PollutionCitation Excerpt :Arsenic (As) is a well-known toxic element and inorganic As has been shown to cause adverse human health effects such as neurological, gastrointestinal and renal toxicity, skin cancer, and others (Dong et al., 2016; Yan et al., 2018).