Antimony in drinking-water
Background document for development of WHO Guidelines for Drinking-water Quality
Effects on laboratory animals and in vitro test systems
1 Acute exposure
Oral LD50 values reported for APT in experimental animals range from about 115 mg/kg of body weight in rabbits and rats to 600 mg/kg of body weight in mice. ATO is practically non-toxic (LD50 > 20 000 mg/kg of body weight) due to its extremely low solubility in water (Gebel, 1999a).
2 Short-term exposure
In a 14-day NTP drinking-water study, APT was tolerated by rats and mice in doses up to 168 and 273 mg of antimony per kg of body weight per day, respectively. Lesions of the forestomach and liver were observed at a dose of 407 mg/kg of body weight per day (NTP, 1992). In contrast, APT administered intraperitoneally to rats and mice in 16-day toxicity studies provoked clear signs of toxicity. In rats, increased mortality and histopathological lesions in liver and kidney were produced generally at the highest dose levels (11 and 22 mg of antimony per kg of body weight per day). In mice, increased mortality and minimal to mild hepatocellular necrosis were observed at 50 and 100 mg of antimony per kg of body weight per day (NTP, 1992). The large differences in toxicity between the intraperitoneal and the oral exposure routes were due to significant differences in systemic availability and absorption of APT (Lynch et al., 1999).
In a 90-day NTP study in which groups of rats and mice were intraperitoneally injected with APT at doses equal to 0, 1.5, 3.0, 6.0, 12 or 24 mg of antimony per kg of body weight per day, rats were about 4 times more sensitive than mice to APT treatment. They exhibited adverse effects (increased mortality, decreased body weight) at 12 and 24 mg/kg of body weight per day. At 6.0 mg/kg of body weight per day (males) and 12 mg/kg of body weight per day (females), hepatocellular degeneration and/or necrosis occurred in association with dose-related increases in the activities of sorbitol dehydrogenase and alanine aminotransferase. The NOAEL of antimony (APT) resulting from intraperitoneal injection was 3.0 mg of antimony per kg of body weight per day. This would be equivalent to an oral NOAEL of about 15 mg of antimony per kg of body weight per day, assuming 20% absorption (NTP, 1992).
Poon et al. (1998) administered APT in drinking-water for 90 days to Sprague-Dawley rats (15 per sex per dose) at concentrations equivalent to 0, 0.5, 5.0, 50 and 500 mg of antimony per litre. This corresponded to daily antimony intakes of 0.06–45.39 mg/kg of body weight in females and 0.06–42.17 mg/kg of body weight in males. Additional groups of 10 rats per sex were exposed to either 0 or 500 mg of antimony per litre in drinking-water for 90 days and then observed over 4 weeks for reversibility of any antimony-mediated adverse or non-adverse effects. No signs of overt clinical toxicity were observed in any animal. In the high-dose males, a marked but reversible loss of body weight gain occurred, probably in conjunction with distinctly reduced food and water intake at this dose. Based on subtle histopathological changes in the thyroids of males (increased epithelial height, decreased follicular size), the authors identified a NOAEC of 0.5 mg of antimony per litre in drinking-water, which corresponded to a NOAEL of 0.06 mg of antimony per kg of body weight per day.
However, Lynch et al. (1999) questioned the authors’ evaluation of the otherwise “generally well designed study,” pointing to the reversible/adaptive nature of the “critical” thyroidal and other biochemical and histological observations in this study, the absence of any quantitative dose–response relationship, although a more than 1000-fold dose range was applied, and the high physiological variability and/or treatment-related occurrence of the observed “critical” effects. Moreover, “none of the subtle histological changes recorded by Poon et al. (1998) were detected in the NTP (1992) study.” Instead of 0.06 mg of antimony per kg of body weight per day, Lynch et al. (1999) proposed a subchronic NOAEL of 6.0 mg of antimony per kg of body weight per day, which corresponds to 50 mg of antimony per litre in the Poon et al. (1998) drinking-water study. This NOAEL was based on the decreased body weight gain and reduced food and water intake observed in that same study at 500 mg of antimony per litre, which is the LOAEC corresponding to a LOAEL of 60 mg of antimony per kg of body weight per day (Lynch et al., 1999).
In a dose range-finding GLP study, ATO incorporated in the diet of Wistar-derived rats (Alpk:ApfSD) (eight per sex per group) at concentrations of 0, 1000, 5000 or 20 000 mg/kg for 28 days caused significant changes only in two top-dose females. The lesions in kidney and liver were not of marked severity, but the lesion seen in the adrenal capsule was unusual and could be treatment-related. The LOAEC of 20 000 mg of ATO per kg in the diet was equivalent to a daily dose of 1000 mg of antimony per kg of body weight per day (LOAEL), the exact dose depending on the weekly feed consumption/body weight ratio (Central Toxicology Laboratory, 1996).
In a subchronic GLP study, ATO given to Wistar-derived rats (Alpk:ApfSD) (12 per sex per dose) in the diet at concentrations of 0, 1000, 5000 or 20 000 mg/kg for 90 days provoked no toxicologically significant findings in any of the dose groups (Hext et al., 1999). There were no adverse effects on body weight, food consumption or haematological parameters. A number of statistically significant, but inconsistent, biochemical changes were observed, including a decrease in alkaline phosphatase in the plasma at the two highest doses in males and the highest dose in females; an increase in triglycerides in the plasma at the highest dose in males, but with no trend; an increase in cholesterol at the highest dose in females; and an increase in aspartate aminotransferase in the plasma at the highest dose in males. In addition, liver weights were slightly increased at 20 000 mg/kg in animals of both sexes, and urine volume and specific gravity were reduced in the 20 000 mg/kg females. The authors considered none of these changes to be of toxicological significance. At necropsy, no treatment-related findings were made. The NOAEC of 20 000 mg of ATO per kg in the diet was equivalent to a NOAEL of 1685.9 mg of ATO per kg of body weight per day (or 1407.7 mg of antimony per kg of body weight per day).
2.3 Comparison between APT and ATO
The large difference in toxicity between ATO antimony and APT antimony is due to the significant difference between bioavailabilities and corresponding systemic exposures. This is not unexpected, in view of the high water solubility of APT and the insoluble nature of ATO.
3 Long-term exposure
Early studies on the chronic and subchronic oral toxicity of antimony were carried out using APT. Lynch et al. (1999) extensively reviewed these studies and raised several points of criticism. One of these studies, carried out by Schröder et al. (1970), was the experimental basis on which the provisional guideline value in the second edition of the Guidelines was derived.
4 Reproductive and developmental toxicity
Exposure to 250 mg of antimony per m3 in the air for 4 h per day during a 2-month period was reported to cause some adverse effects on the reproductive outcome of rats (Belyeava, 1967). Teratogenicity of antimony(V) dextraneglycoside or of antimony trichloride could not be demonstrated in rats and sheep (James et al., 1966; Casals, 1972; Rossi et al., 1987). However, 125Sb was shown to cross the placenta and was also found in the milk of lactating rats (Gerber et al., 1982).
5 Mutagenicity and related end-points
ATO was genotoxic in a number of older bacterial mutation assays but not in more recent ones (Lantzsch & Gebel, 1997; Elliott et al., 1998). Conflicting results were also obtained with respect to the genotoxicity of antimony in cultured mammalian cells. Positive results were observed with ATO in the in vitro cytogenetic assay with human lymphocytes (Elliott et al., 1998) and the sister chromatid exchange assay with V79 cells (Kuroda et al., 1991), but not in the L5178Y mutation assay (Elliott et al., 1998).
The in vivo genotoxicity of ATO was extensively investigated by Elliott et al. (1998) using single- and repeat-dose mouse bone marrow micronucleus tests and the rat liver unscheduled DNA synthesis assay. All three studies were negative. In contrast, Gurnani et al. (1992a) reported chromosomal damage by ATO in mouse bone marrow cells after repeat dosing but not after single dosing. This discrepancy between Elliott et al. (1998) and Gurnani et al. (1992a) with respect to repeat dosing may be explained (Elliott et al., 1998) by the “not specified purity” and much higher systemic toxicity of the ATO sample used by Gurnani et al. (1992a). For this reason, and because of the poor water solubility of ATO (17 µg/litre), Elliott et al. (1998) concluded that ATO was not genotoxic in vivo.
Different results were obtained with more water soluble antimony compounds. The compounds antimony trichloride and antimony pentachloride were reported to be genotoxic in the rec-assay with Bacillus subtilis (Kanematsu et al., 1980; Kuroda et al., 1991). Antimony(III) acetate enhanced the simian adenovirus-7-mediated transformation of Syrian hamster embryo cells (Casto et al., 1979), and enhanced rates of chromosomal breaks in human leukocytes were reported after treatment with APT (Paton & Allison, 1972).
The potency of antimony(III) to induce micronuclei in vitro in V79 cells and human lymphocytes was about 1 order of magnitude lower than that of arsenic(III) (Gebel, 1998; Schaumlöffel & Gebel, 1998). The comparable dose levels in human lymphocytes were 0.5 µmol/litre for arsenic(III) and 5 µmol/litre for antimony(III). In contrast to sodium arsenite, antimony trichloride did not induce DNA–protein cross-links in V79 cells and peripheral human lymphocytes (Gebel et al., 1998; Schaumlöffel & Gebel, 1998). The genotoxicity of antimony(III) was also lower than that of arsenic(III) in the test for sister chromatid exchange and in a single-cell gel test reviewed by Gebel (1999a).
In vivo, tartar emetic (APT) and bilharcid (piperazine antimony tartrate), two important antischistosomal drugs, were reported to be genotoxic after acute and subacute application to rats (El Nahas et al., 1982). Seven days after being given orally to Swiss mice, antimony trichloride was reported to cause increased chromosomal aberration rates in bone marrow cells (Gurnani et al., 1992b).
The greatest concern with regard to the carcinogenicity of antimony compounds relates to the inhalation route. ATO has been found to be carcinogenic to experimental animals in inhalation studies (IARC, 1989) and to cause direct lung damage following chronic inhalation as a consequence of overload with insoluble particulates (Newton et al., 1994).
In contrast, oral lifetime studies with LE rats (Kanisawa & Schröder, 1969; Schröder et al., 1970) or Charles River CD mice (Schröder et al., 1968; Kanisawa & Schröder, 1969), in which animals were exposed to 5 and/or 50 mg of antimony (as APT) per litre in drinking-water, did not give any indication that antimony(III) showed carcinogenic potential by the oral route. However, Lynch et al. (1999) critically reviewed all three studies and concluded that they were not suitable for making a definitive assessment of the carcinogenicity of antimony because they contained many flaws in design and experimental methodology. In addition, the use of these studies as a quantitative starting point to assess the cancer risks associated with oral antimony exposure was deemed to be inappropriate.