Bioavailability of inorganic arsenic from bog ore-containing soil in the dog.

In some parts of The Netherlands, bog ore-containing soils predominate, which have natural arsenic levels that exceed, by a factor of 10, existing standards for maximum allowable levels of inorganic arsenic in soil. These standards are based on the assumption that in humans the bioavailability of arsenic from ingested soil is equal to that from an aqueous solution. In view of the regulatory problem that the arsenic levels of these soils present, we questioned the validity of this assumption. To obtain a more realistic estimate, the bioavailability of inorganic arsenic from soil in a suitable animal model was studied. In this report, a study performed in six dogs in a two-way cross-over design is presented. The dogs received orally, in random order, arsenic both as an intravenous solution and as arsenic-containing soil. During a 120-hr period after administration urine was collected in 24-hr fractions. Levels of arsenic were determined using a method of wet digestion, isolation and complexation of arsine, followed by molecule absorption spectrometry. Within 120 hr after intravenous administration, 88 +/- 16% of the dose was excreted renally. After oral administration of arsenic-containing soil, only 7.0 +/- 1.5% was excreted renally. From the urinary excretion data for these two routes of administration, the calculated bioavailability of inorganic arsenic from soil was 8.3 +/- 2.0%. The results from this study demonstrate the need to reconsider the present risk assessment for arsenic in soil.

In general, in toxicological risk assessment of environmental contaminants in humans and mammals, there is little information available on the bioavailability of toxic compounds. Applying inadequate corrections for incomplete absorption can lead to considerable overestimation of the internal dose. Recent investigations in cows on the bioavailability of dioxins bound to fly ash have clearly shown that the bioavailability of the different dioxin congeners strongly depends on the matrix to which the dioxins are bound (1). Mean bioavailability of dioxins from fly ash in cows is about 1%, which is considerably less than the 30% bioavailability of dioxin congeners from oil (2).
A situation analogous to that for the dioxins may hold true for metals. In some parts of the Netherlands, bog ore-containing soils predominate, which have natural arsenic levels that exceed, by a factor of 10, existing Dutch standards for maximum allowable levels for inorganic arsenic in soil, currently set at 50 mg/kg dry weight. In setting the tolerable intake of arsenic in humans, bioavailability from soil was assumed to be equal to that from a solution as used in toxicity studies. The main problems in setting the standards were that the limit setters took into account neither the form of arsenic nor route of exposure or the toxicological effects of its different forms (3). In a recent paper (4), adjusted, toxicologically based soil clean-up criteria have been proposed, taking into account differences in species, bioavailability, etc.
Bioavailability of inorganic arsenic from solutions was found to be more than 90% (5)(6)(7)(8)(9). However, as shown in Figure   1, arsenate (the most common form of arsenic in soil) is specifically bound to bog ore (10). This makes the assumption that the bioavailability of arsenic from bog orecontaining soil is equal to that from solutions questionable.
The aim of the present study was to examine whether there is a scientific basis for reconsidering the present practice of risk assessment for arsenic in soil from residential areas. For this purpose, dogs were used as an animal model because they are relatively easy to administer large amounts of soil. Furthermore, the rat is not a suitable animal model for studies on arsenic because in this species, 50% of the arsenic accumulates in the erythrocytes (11), which is not the case in dog and man.
We collected soil from an area in Doetinchem in the province of Gelderland, the Netherlands. Samples were taken at 60 and 80 cm depth. Detailed information about this area is available from earlier investigations (12). Soil samples were stored at 40C until use.
For intravenous injection, we prepared a solution with a final concentration of 2 mg As/ml in saline using As2O5 (Tritisol lot no. 09014068, Merck, Darmstadt, Germany).
The experiments were performed using three male and three female beagles (DoBe) from the cohort of the Unit Biotransformation, Pharmacokinetics and Toxicokinetics, National Institute of Public Health and Environmental Protection, the Netherlands. These animals were kept under conventional conditions. Their weight was between 10 and 15 kg and their age between 2 and 7 years. Individual data are given in Table 1. The dogs received 200 g of commercially available food twice daily. The experiment was performed according to a two-way crossover design, allowing each dog to be its own reference, and excluding the time factor as a possible cause of differences in results.
The bioavailability of arsenic from soil was related to an intravenous (IV) administered dose of arsenic. The IV dose was 2 mg As, administered as As2O in saline via the vena cephalica. For oral administration, soil from 80 cm depth was used because this sample was more homogeneous than the sample from 60 cm. Each animal received 20 g of soil mixed with 200 g of food. Dose levels for arsenic in soil were chosen based on data obtained in a pilot experiment. The objective was to obtain comparable plasma levels of arsenic after both routes of administration.
Before each administration, the animals were fasted for 16 hr. Administration always took place between 0900 and 1000 hr. At 1600 hr the day of administration and on every nonadministration day at 0900 am and 1600 hr, the animals received 200 g of food. During the experiment, tap water was available ad libitum. The dogs were individually housed in metabolism cages, and during a 120-hr period after administration, urine was  caught in 24-hr fractions in glass bottles native step, and a blank determination to containing 5 ml of 4 M nitric acid. Urine check for the reagents' purity. The detecsamples were stored at -18 C until analysis. tion limit of the procedure was 1 pg/g. We determined arsenic in soil using Arsenic measured in soil-5 was 97 pg/g, atomic emission spectrometry with induc-being well within the 95% confidence tively coupled plasma (ICP-AES). The interval of 93.9 ± 7.5 pg/g. Recovery of the apparatus used was a Perkin Elmer ICP standard addition experiments was 102 ± 6 6000 and instrumental settings were as rec-%, indicating that no matrix-induced sensiommended by the manufacturer. The spec-tivity changes were present. Arsenic was not tral line was the 189.042 line employing detected in the reagent blank. background correction and nitrogen flush We determined levels of total arsenic in for the spectrometer to reduce light absorpurine, food, and the solution for IV injection tion at this wavelength.
using a procedure of wet digestion of arsenic The soil samples (1-3 kg each) were to As(V), reduction to As(III), isolation and hand-homogenized using a plastic spatula, complexation of arsine (AsH3) with silver after which four subsamples of about 5 g diethyldithiocarbamate (AgDDTC), foleach were taken from each sample. The lowed by molecular absorption spectrometry samples were processed as received, i.e., in (MAS). Absorption was measured at 526 wet condition. The subsamples were dinm. Sample sizes were taken depending on gested with aqua regia under reflux (13). In the expected arsenic content, resulting in this procedure, 4 ml of nitric acid, 65%, 12 amounts of arsenic within the linear range of ml of hydrochloric acid, 37%, and 50 ml of the method (i.e. 0.5-20 pg per test soludistilled water were added to the sample, tion). The validity of this method has been and the mixture was boiled for 2.5 hr. After demonstrated for several products and biocooling down, we diluted the digests with logical materials, such as fruit (14). Furtherwater to 100 ml. After decantation, the more, the arsenic content in several standard digests were ready for analysis. reference materials has been determined Quality control included the digestion using this method (15). In the present study, of a standard reference material (soil-5 of the levels of arsenic in eight urine samples the IAEA Vienna), application of standard were determined in duplicate from separate addition to the digests to check for matrix-test portions and in two samples by dupliinduced sensitivity changes in the determicate measurement in the same digest. Total content of arsenic in food was determined in samples from the two batches used during the experiment. Results were 0.063 and 0.059 mg/kg, respectively. The arsenic content in the solution for IV injection was 2.0 mg/ml.
Comparison of the excretion half-lives after IV and oral administration of arsenic was performed using the Student's t-test on log-transformed values.
From the results of the determination of arsenic in soil, it was obvious that at 60 cm depth, arsenic was less homogeneously distributed (concentration 323 ± 74 pg/g wet weight) than at 80 cm depth (339 ± 19 pg/g wet weight). Therefore, the latter was used in the animal experiments. From previous reports, we knew that arsenate was the predominant species in soil from this area (12,13).
In Table 2, the data for the excretion of arsenic in urine after IV and oral administration are presented. After IV administration, most of the dose appeared to be excreted within 24 hr. Within 120 hr after administration, almost the entire dose was recovered in urine of five out of six dogs. Only one dog, Anonymus, had a low urinary recovery of 57% within 120 hr.
After administration of arsenic containing bog-ore mixed with food, no more than 7.0 ± 1.5% of the dose was recovered in urine within 120 hr.
Typical excretion curves after IV and oral administration are shown in Figure 2. After IV administration, two phases in the excretion curve are observed. The first part of the curve shows a half-life of 8.1 ± 3.7 hr, whereas the second part of the curve shows a half-life of 28.4 ± 8.7 hr. After oral administration of arsenic-containing soil, only an excretion phase with a halflife of 18.7 ± 5.2 hr was observed. The terminal half-lives of excretion were not significantly different between the routes of administration (Table 3).  A(0-120)po x DiV x 100% = F A(O-120)iv Dpo where A(0 120)p, and A(0o120)iv are the amounts of arsenic excreted in urine within 120 hr after oral and IV administration, respectively. DPO and DA. are the oral and IV dose, respectively. Bioavailability of arsenic from bog orecontaining soil was 8.3 ± 2.0%. Individual data are given in Table 3.
The study described here was performed to determine the bioavailability of arsenic from bog ore-containing soil. The dog was used as an animal model for this objective. This type of study was triggered by the discussion about whether to reconsider the present regulation for arsenic in soil from domestic areas, since in some areas, natural arsenic levels significantly exceed existing standards (4). In general, regulation is, among other things, based on the assumption that the bioavailability of contaminants from soil is the same as the bioavailability from a solution or from food. This assumption is questionable because the matrix can significantly influence the bioavailability.
Based on literature data (5,16,1), the study was designed assuming that excretion of arsenic only takes place renally. This was confirmed in our study after IV administration of arsenic. The values for the elimination half-life were comparable to those reported in literature (11). Our data show that only about 8% of arsenic from bog ore-containing soil is absorbed from the gastrointestinal tract and excreted into urine. Thus, bioavailability of arsenic from soil is much lower than from a solution, either containing As(III) or As(V) (5,8,9). Seventy-three percent of orally administered As203 in water was recovered in urine of female monkeys (5). After ingestion of dissolved inorganic As(V) by two volunteers, 64-69% of the dose was excreted into urine within 166 hr (9). However, several factors, such as species tested and dietary status, could modify the extent of absorption.
Comparing data from the present study with those from a study using a gastric juice simulation test reveals that the simulation test underestimated the release of arsenic from bog ore-containing soil. In the gastric simulation test, a release of less than 1% was found (12). The results obtained from the present investigation on the bioavailability of inorganic arsenic from bog ore-containing soil in dogs stress the need to reconsider the present practice of risk assessment for arsenic in soil in residential areas.