An In Vitro Bioaccessibility Test Method for the Estimation of Bioavailability of Arsenic From Soil.

Accurate assessment of human health risks from incidental ingestion of soil containing arsenic (As) requires knowledge of the relative oral bioavailability (RBA) of As in those soils.  RBA can be measured in vivo using animal models, but the cost of in vivo bioavailability tests limits widespread application of this approach for most site investigations.  A faster, more economical, yet dependable in vitro method for predicting RBA is highly desirable.  One in vitro approach involves measuring the fraction of As that is released from soil into solution under specified extraction conditions.  This solubilization fraction is referred to as in vitro bioaccessibility (IVBA).  If a strong correlation between RBA and IVBA can be established, then the IVBA method offers a more rapid and less costly alternative to in vivo studies.  This study sought to optimize an IVBA method to estimate As RBA in soil for use in human health risk assessments.  In Phase 1 we reviewed and summarized the available in vivo RBA data from juvenile swine/ monkey studies, and identified 39 candidate test materials for IVBA testing.  Phase II investigated the effect of a wide range of experimental variables in the IVBA protocol, and identified three key variables (pH, phosphate, and hydroxylamine hydrochloride concentration).  Phase III evaluated various combinations of these key variables in a Latin square design using a selected set of text substrates to identify up to three alternative combinations providing the best IVBA-RBA predictive relationship.  Phase IV tested the three most promising extraction conditions on the entire list of test materials.  For juvenile swine the best fit model (R2=0.72) is obtained using IVBA measurements at pH 1.5.  For monkeys the best fit model (R2=0.75) is obtained using IVBA measurements at pH 7 with phosphate and hydroxylamine.  The predictive relationship improves when As mineralogy is included with the IVBA data (R2=0.90 and 0.82 for swine and monkey, respectively).  The precision of the extraction method was verified in a round robin analysis performed by three independent labs.