In Vivo and in Vitro Percutaneous Absorption and Skin Decontamination of Arsenic from Water and Soil

The objective was to determine the percutaneous absorption of arsenic-73 as H3AsO4 from water and soil. Soil (Yolo County 65-California-57-8) was passed through 10-, 20-, and 48-mesh sieves. Soil retained by 80 mesh was mixed with radioactive arsenic-73 at a low (trace) leve of 0.0004 μg/cm2 (micrograms arsenic per square centimeter skin surface area) and a higher dose of 0.6 μg/cm2. Water solutions of arsenic-73 at a low (trace) level of 0.000024 μg/cm2 and a higher dose of 2.1 μg/cm2 were prepared for comparative analysis. In vivo in Rhesus monkey a total of 80.1±6.7% (SD) intravenous arsenic-73 dose was recovered in urine over 7 days; the majority of the dose was excreted in the first day. With topical administration for 24 hr, absorption of the low dose from water was 6.4±3.9% and 2.0±1.2% from the high dose. In vitro percutaneous absorption of the low dose from water with human skin resulted in 24-hr receptor fluid (phosphate-buffered saline) accumulation of 0.93±1.1% dose and skin concentration (after washing) of 0.98±0.96%. Combining receptor fluid accumulation and skin concentration gave a combined amount of 1.9%, a value less than that in vivo (6.4%) in the Rhesus monkey. From soil, receptor fluid accumulation was 0.43±0.54% and skin concentration was 0.33±0.25%. Combining receptor fluid plus skin concentrations gave an absorption value of 0.8%, an amount less than that with in vivo absorption (4.5%) in the Rhesus. These absorption values did not match current EPA default assumptions. Washing with soap and water readily removed residual skin surface arsenic, both in vitro and in vivo. The partition coefficient of arsenic in water to powdered human stratum corneum was 1.1×104 and from water to soil it was 2.5×104. This relative similarity in arsenic binding to powdered human stratum corneum and soil may indicate why arsenic absorption was similar from water and soil. This powdered human stratum corneum partition coefficient model may provide a facile method for such predictions.