Probabilistic Analysis of Human Health Risks Associated with Background Concentrations of Inorganic Arsenic: Use of a Margin of Exposure Approach

ABSTRACT

Substantial evidence exists from epidemiological and mechanistic studies supporting a sublinear or threshold dose–response relationship for the carcinogenicity of ingested arsenic; nonetheless, current regulatory agency evaluations have quantified arsenic risks using default, generic risk assessment procedures that assume a linear, no-threshold dose–response relationship. The resulting slope factors predict risks from U.S. background arsenic exposures that exceed certain regulatory levels of concern, an outcome that presents challenges for risk communication and risk management decisions. To better reflect the available scientific evidence, this article presents the results of a Margin of Exposure (MOE) analysis to characterize risks associated with typical and high-end background exposures of the U.S. population to arsenic from food, water, and soil. MOE values were calculated by comparing a no-observable-adverse-effect-level (NOAEL) derived from the epidemiological literature with exposure estimates generated using a probabilistic (Monte Carlo) model. The plausibility and conservative nature of the exposure and risk estimates evaluated in this analysis are supported by sensitivity and uncertainty analyses and by comparing predicted urinary arsenic concentrations with empirical data. Using the more scientifically supported MOE approach, the analysis presented in this article indicates that typical and high-end background exposures to inorganic arsenic in U.S. populations do not present elevated risks of carcinogenicity.

Key Words:

• dose–response relationship
• carcinogenicity
• margin of exposure
• arsenic
• background exposures

ACKNOWLEDGMENTS

Preparation of this article was supported in part by the MAA Research Task Force. We gratefully acknowledge the valuable assistance of Jasmine Lai, Jennifer O'Brien, and Ruth Buchman in the preparation of this article.

Notes

¹ A threshold dose is a dose below which health effects, including carcinogenic health effects, are not induced.

² This value is calculated using the following equation: ([arsenic concentration in water × water intake] + dietary intake)/ (body weight × conversion factor), or (150 μ g/L × 4.5 L/day) + 40 μ g/day)/(55 kg × 1,000 μ g/mg).

³ Note that these articles did not provide raw data or specific information regarding the shapes of the distributions they generated for dietary intake. Based on the available information, it was found that use of a normal distribution required truncation of the lower portion of the distribution to avoid incorporation of negative dietary intake estimates in the Monte Carlo analysis. Because a lognormal distribution also provided a reasonable fit to the available information and avoided the issue of truncation, such a distribution was used in the Monte Carlo analysis. To assess the implications of the alternative distribution assumptions, the exposure calculations were performed using both a lognormal and a truncated normal distribution for the dietary intake estimates. Use of the truncated normal distribution did not yield substantially different results from those presented in this article.

⁴ Note that a recent publication indicates that regional variations may exist in arsenic concentrations in U.S. rice, an important dietary source of background arsenic intake (Williams et al. 2007). Review of the recent data indicates, however, that consideration of these data would not significantly alter the exposure and risk assessment conclusions derived in the current MOE analysis.

⁵ USEPA data indicate that 11,403 community water systems rely on surface water sources and serve 178.1 million people, whereas 42,661 systems rely on groundwater sources and serve 85.9 million people (USEPA 2006b).

As discussed in the Toxicity Assessment section.

^a Calculated from exposure estimates assuming a 70-kg body weight.

^b This analysis incorporated consideration of lifetime spatial and temporal differences in exposures.

^c This analysis used an integrated, biologically based model (MENTOR) to assess exposures and also included results for Hunterton County, NJ and Pima County, AZ. Although specific exposure estimates weren't provided for the 5th and 95th percentiles, these authors indicated that their exposure results were generally in agreement with those of Meacher et al. (2002).

^a MOE calculation based on Point of Departure value of 0.013 mg/kg-day.

^b Calculation based on CSF value presented in USEPA's IRIS database: 1.5 (mg/kg-day)⁻¹.

^c Calculation based on alternative CSF value used in recent USEPA risk assessments: 3.67 (mg/kg-day)⁻¹.

^aThe “worst-case” results were calculated using 95th percentile value estimates for all input parameters with the exception of the following parameters: the relative bioavailability adjustment (RBA) factor assumed for arsenic ingested in soil, and arsenic concentrations in surface water, groundwater, and soil. The assumed RBA value was set at 50% (the maximum value included in the distribution used in this analysis) and the media concentrations were set at the maximum values reported in underlying documentation (i.e., soil data from Dragun and Chekiri 2005 and water data from Chappell et al. 1994).

⁷ No actual U.S. populations experiencing such combined high-end exposures are known to exist. For example, an evaluation of regional differences in background arsenic exposures found that typical exposures to inorganic arsenic via dietary sources were relatively similar across the United States, with the highest intakes estimated for residents of the northeastern United States (Meacher et al. 2002). These differences were attributed to differences in the types of food consumed rather than regional differences in arsenic concentrations in food. By contrast, inorganic arsenic exposures via drinking water were more variable, with the highest exposures estimated for residents of the western United States and the second highest exposures found in residents of the midwestern United States.

⁷ For epidemiological data, the ED₀₁ or LED₀₁ is typically used as a point of departure; it can be used as the “NOAEL” in an MOE analysis or the starting point for linear extrapolation.

⁸ In the absence of specific pharmacokinetic information, the USEPA recommends deriving a human equivalent dose using a default scaling factor. This step stems from the general consideration that distribution, clearance, and metabolism of toxins are more rapid in smaller animals than in humans.

⁹ Angiogenesis is the growth of new blood vessels, particularly those that supply blood to cancerous tissues.