The Department of Defense (DoD) faces a potentially daunting task of remediating thousands of metal-contaminated sites within the United States and its territories that contain unacceptable levels of the toxic metal(loid)s arsenic (As), cadmium (Cd), chromium (Cr), and lead (Pb). With the exception of Pb-contaminated soils, human health and ecological risk drivers have prompted EPA to assume that the total soil metal concentration is 100% bioavailable. Previous SERDP-funded research (ER-1166) has shown that the ubiquitous metal-sequestering properties of soil can significantly lower the bioavailability and risk of heavy metals to human and ecological receptors.

The technical objectives of this project were to: (1) provide validation that the relationships between soil properties and in vitro bioaccessibility methods can serve as a screening tool for estimating in vivo toxic metal bioavailability in DoD soils; (2) provide DoD with a scientifically and technically sound method for estimating human and ecological risk associated with metal contaminated soils in place of or as justification for more detailed, site-specific bioavailability (e.g., animal dosing), and (3) promote the use of in vitro methods in human health and ecological risk assessments through the upfront involvement of end-users and regulators.

Soil properties, total metal content, speciation, and bioaccessibility and bioavailability (as measured by various in vitro and in vivo methods, respectively) were determined for metal contaminated soils collected from three DoD sites for the human health models. A similar approach was taken for the in vitro ecological model, which was made more robust by considering an additional eight DoD soils (total of eleven contaminated and eleven control soils). 

Human Health. Metal bioaccessibility and metal bioavailability for three study soils was calculated using soil property-driven models developed from earlier SERDP studies. Calculated bioaccessibility values were compared with measured bioaccessibility values using in vitro gastrointestinal methods for study soils. The physiologically based extraction test (PBET) developed by Ruby et al. (1999), was utilized at a variety of pH conditions to estimate metal bioaccessibility for a variety of stomach environments indicative of food intake, or lack thereof. Additional soil property-driven models were constructed using the PBET method at these pH values. This is particularly important for Pb-contaminated soils since Pb bioaccessibility decreases with an increase in pH. In addition to PBET, the Ohio State University In vitro Gastrointestinal Method (OSU-IVG) method was used to measure bioaccessible As. The ability of the OSU-IVG method to predict contaminant bioavailability was determined.

Ecological. For ecological risk estimates, metal bioavailability was estimated from multiple regression models developed using bioaccumulation data from 26 soils. Also, the ability of soil extraction methods to predict phytoavailable metals was investigated. Additionally, eight selected DoD sites were tested in addition to the three soils used in the swine study. This was necessary to enhance the robustness of the ecological model as had already been done for the human-based model in SERDP project ER-1166. In the ecological investigations, metal concentrations from in vitro DoD soil metal extractions or DoD soil chemical and physical properties were used to predict metal bioavailability to plants and soil invertebrates. Initially, statistical relationships developed for metal availability from a set of 26 soils were used to estimate the chemical availability of metals in DoD soils, based upon total metal levels and soil physical/chemical characteristics. This was followed by extraction of the DoD soils using several soil extraction methods using pore water, dilute calcium nitrate solution, and Mehlich 3 solution. The ability of soil chemical extractants to predict metal bioavailability to plants was determined. Plant and soil invertebrate bioassays were conducted with DoD soils to determine actual toxicity and bioaccumulation, and these results were compared to the model predictions of toxicity and bioaccumulation.

Ruby, M.V., R. Schoof, W. Brattin, M. Goldade, G. Post, M. Harnois, D.E. Mosby, S.W. Casteel, W. Berti, M. Carpenter, D. Edwards, D. Cragin, and W. Chappell, Advances in evaluating the oral bioavailability of inorganics in soil for use in human health risk assessment. Environmental Science & Technology, 1999. 33(21): p. 3697-3705.

Summary of Soil Properties to Predict Metal Bioavailability. At a minimum, soil property information needed from a site investigation for all contaminants studied were soil pH, clay content, organic carbon (C), inorganic C, reactive iron (Fe) and aluminum (Al) (FEAL, Feox and/or CBD Fe). Other properties not studied that will affect ecological endpoints include soil salinity and the presence of other toxicants.

These properties will  predict metal bioavailability for all soils. A major finding of this study is that the contaminant source and likely speciation greatly affects the ability of soil property to predict metal bioavailability. Metal bioavailability was not able to be predicted for several soils where the contaminant source was unweathered mining waste or discrete inorganic mineral forms such as coal ash. Soil properties should  be used to predict contaminant bioavailability in these soils. More research on contaminant source and speciation is needed to determine when soil properties can provide an accurate assessment of metal bioavailability. Currently research is in progress, including research funded by SERDP (i.e., ER-1742), to determine the relationship between As speciation and ability to predict As bioavailability to humans.

Summary of Soil Extraction Methods to Predict Metal Bioavailability. Both PBET and OSU-IVG were able to very accurately predict relative bioavailability (RBA) As and Pb, but only for one soil each. The number of soils evaluated was very limited because of cost constraints associated with in vivo dosing trials required to measure contaminant RBA. More research is needed to evaluate the ability of these methods to predict RBA Pb and RBA As on other contaminated soils.

Soil pore water was able to predict plant tissue concentration of Pb, As, and Cd. Soil extraction with 0.1 M Ca(NO₃)₂ was able to predict cationic metal contaminants (i.e., Pb, Cd) but was not evaluated for anionic As contamination. The ability of simply water or dilute calcium nitrate to predict phytoavailable contaminant suggests high solubility of these contaminants in soils. Thus, it is likely that 0.1 M Ca(NO₃)₂ would also have been a good predictor of plant As. However, two cautions should be heeded. The accuracy of these extraction methods to predict plant tissue contamination was limited to ±35%. Similar to metal bioaccessibility results, metal bioavailability was not able to be predicted for several soils where the contaminant source was unweathered mining waste or discrete inorganic mineral forms such as coal ash. Soil extraction methods listed in the summary table in the Final Report should  be used to predict contaminant bioavailability in these soils. More research on contaminant source and speciation is needed to determine which soil extraction methods can provide an accurate assessment of metal bioavailability.

This investigation brought together regulators, EPA, end-users, and scientists to demonstrate the applicability of these concepts by showing that simple, readily available soil properties can often be used to predict the bioavailability of As, Cd, Cr, and Pb with a reasonable level of confidence. This project has shown that in vitro methods can often be used for risk assessment of toxic metals in soil by comparing in vitro and in vivo metal bioavailability studies.