SCIENTIFIC QUESTIONS TO THE CSTEE ON ARSENIC

Note: Following the terms of reference, this opinion focuses on specific questions related to human health issues and does not cover ecosystem protection.

Question 1. What is the committee's opinion on the genotoxicity of arsenic inhaled by humans?

The CSTEE considers that, on the basis of the currently available evidence, arsenic should be classified as genotoxic (i.e. it interacts with DNA and probably also with the spindle apparatus) both in vitro and in vivo. It does not produce point mutations in vitro but appears to cause multilocus deletions. There is limited in vivo information that indicates that arsenic is genotoxic in man

Arsenic is a human carcinogen; it has proved difficult to replicate this effect in experimental animal models. There is insufficient information on the dose response relationships for the human cancers to conclude that there is a threshold level below which arsenic will not induce cancer in man. Studies on the mechanism of arsenic carcinogenicity do not provide a clear indication that a threshold exists.

Specific investigations of arsenic genotoxicity have not been conducted using the inhalation route. However inhaled arsenic trioxide particles appear to be well absorbed and distributed throughout the body. It must, therefore, be assumed that arsenic compounds are genotoxic by the inhalation route.

Question 2. If the committee considers that Arsenic is non-genotoxic how would they recommend deriving a limit value?

Since the current understanding is that arsenic is genotoxic this question does not require a response. However, if it were considered appropriate to set a limit value based on non-cancer endpoints the CSTEE is in general agreement with the approach adopted in the report.

Scientific basis for the CSTEE opinion

General remarks.

Based on a limited number studies, arsenic has been claimed by some authors to be beneficial to health in laboratory experiments. However these studies involve the use of unusual diets and relatively high levels of arsenic therefore their relevance is insufficiently substantiated. Arsenic has never been shown to be an essential mineral in man and no deficiency diseases have been reported.

In rodents an inducible tolerance to arsenic occurs. There are claims that this tolerance does not occur in humans. Tolerance has been replicated in vitro in animal cells but not so far in human cells. There is some evidence too that human cells are more sensitive to arsenic than cells of animal origin (Liu et al 2001, Romach et al 2000).

1. Exposure

Arsenic is a metalloid, which occurs in both the trivalent and pentavalent states. Both inorganic and organic compounds of arsenic are widely distributed in the environment.

Food and drinking water are the principal routes of exposure of the public to arsenic with the exception of some industrial workers who are exposed regularly to high airborne levels. The calculated values for absorbed daily dose of inorganic arsenic are summarised (from table 2.1.7 of the position paper) as:

Air <1%
Cigarette smoke 0-16%
Drinking water 0-33%
Food 50-98%

Although air is not an important route of exposure, in quantitative terms, it may be significant toxicologically since a principal site for carcinogenicity is the lung.

Furthermore airborne levels of arsenic are likely, ultimately, to affect the levels of arsenic in food. Consequently, regulating ambient air levels of arsenic may affect concentrations of arsenic in food.

Arsenic in air consists principally of a mixture of particulate arsenite (trivalent) and arsenate (pentavalent). It is derived from both natural sources such as volcanic activity, wind blown dust etc and from anthropogenic sources including: smelting of metals such as copper, combustion of fuels and use of arsenic containing pesticides. Current concentrations of arsenic in ambient air range from 0-1ng/m3 in rural areas, 0.5-3 ng/m3 in urban areas and up to 50ng/m3 in the vicinity of some industrial plants. In addition arsenic is found in tobacco smoke. There appears to be no reliable data on the levels of arsine or organic arsenicals in either ambient or indoor air. It is assumed that such levels are insignificant.

Both, the chemical form of inorganic arsenic in air and the particle size may have a substantial influence on the extent of uptake from the lung. In workers in the smelting industry exposed to arsenic trioxide, the amount of arsenic excreted in the urine was between 40 and 60 % of the estimated inhaled dose (WHO 1997). This indicates good absorption from arsenic containing particles. However autopsy data from workers, who had retired some years previously from the smelting industry, showed levels of arsenic in the lung which were some eight times higher than typical of a control group. Thus under some circumstances inhaled arsenic can accumulate in lung tissue (WHO 1997).

Once absorbed, arsenic is rapidly distributed throughout the body with skin, nails and hair having the highest concentrations. Arsenic appears to be able to transfer from maternal to foetal blood. Metabolic conversion of Arsenite to arsenate is considered to be an important intoxication reaction. Metabolism of Arsenite to arsenate can also occur. Mono- and di- methylation of arsenite occurs in several tissues. These methylated arsenic acids are generally regarded as being less toxic and less well retained by the body than the inorganic forms of arsenic.

2. Genotoxicity

Arsenic induces sister chromatid exchanges and is clastogenic in a range of in vitro systems using both human and animal derived cells (see for example-Position paper 2001, Bernstam and Nriagu 2000). Arsenite is considered to be more potent (perhaps up to ten fold) than arsenate. However the in vivo relevance of this is uncertain, as Arsenite and arsenate are interchangeable in the body. Sodium arsenate inhibits DNA repair in human skin biopsy cells and in human lymphocytes. A single study in human alveolar type 2 cells using arsenite indicated no damage to DNA however.

From the various studies, conducted in bacterial and mammalian cell systems, arsenic is probably not a point mutagen. However arsenite has been shown to cause multiple mutations in some mammalian cell systems.

Limited in vitro studies indicate that very high levels of the metabolites, monomethyl arsenic acid and dimethylarsenic acid (DMA) may be able to cause both strand breaks and mutations.

Arsenic has also been shown to enhance the clastogenicity and mutagenicity of other DNA damaging agents. It should be noted that arsenic produces a number of other changes in vitro, which are related directly, or indirectly to cell replication. For example arsenic activates transcription factors such as AP-1 and induces certain genes (e.g. c-fos and c-jun) which produce products, which stimulate cell proliferation (Simeonova and Luster 2000). Arsenic has been shown also to change the spindle morphology of cultured human keratinocytes (Bernstam and Nriagu 2000).

There have been far fewer in vivo studies. Sodium arsenite has been shown in one study to cause a small increase in chromosomal aberrations in bone marrow cells of mice treated in vivo (IARC 1980). In a more recent study DMA has been demonstrated to cause aneuploidy and mitotic arrest in bone marrow cells after in vivo application to mice.

No experimental studies have been conducted using the inhalation route. Arsenic has not been found to be carcinogenic in several animal studies, and human cells seem to be more sensitive to arsenic than animal cells. Consequently, the use of animal experiments and animal derived cells to assess the genotoxicity of arsenic may be questioned in terms of their relevance to man.

Studies in winegrowers using arsenic pesticides, workers in the copper smelting industry and psoriasis patients treated with arsenicals have indicated an increase in chromosomal aberrations in peripheral blood (e.g. IPCS 2001). In the case of the industrial workers these findings would imply that arsenic is genotoxic following inhalation.

Investigations of dose response relationships are confined to in vitro studies using animal cells. The findings appear to vary according to the endpoint studied (Rudel et al 1996). According to Rudel et al for many endpoints (e.g. Chromosomal aberrations, aberrant metaphases, chromosomal breaks, and potentiation of clastogenicity by various mutagens) a sub-linear dose response occurs. However there is insufficient data to confirm that these findings can be extrapolated to human cells and /or to the in vivo situation.

3. Carcinogenicity

Arsenic has been determined by various expert groups to be a human carcinogen (IARC, IPCS, EC, and USEPA). The principal target organs in man are the lung, skin and bladder (see for example Choiu et al 2001) There is evidence that the colon, liver and kidney may also be targets. Lung cancer appears however to be a critical effect following chronic inhalation exposure of arsenic. Both trivalent and pentavalent arsenic have been implicated. It has proved very difficult to replicate the carcinogenicity of arsenic in animals. There is a debate as to whether this is connected to the phenomenon of arsenic tolerance found in a number of experimental animals but not in man. In one lifetime mouse study, using arsenic in drinking water, an increased tumour incidence in both the lung and the GI tract was observed. Total arsenic exposure rather than airborne exposure alone would appear, from this study, to determine the levels of lung cancer.

Dose response findings in human epidemiological studies do not provide a strong indication as to whether or not there is a threshold dose for the carcinogenicity of arsenic. Recent evaluation of the data from the workforce from two copper smelters led to the conclusion of a probable linear exposure risk relationship (Viren and Silvers 1999, Lubin et al 2000). However these studies involved relatively high arsenic exposures, well above those currently occurring in ambient air.

4. Non-cancer effects

A wide variety of effects can occur after arsenic exposure:

I) Arsenic dust is an irritant to mucous membranes of the nose, throat and upper respiratory tract. The threshold for this irritant effect in workers is around 0.1 to 1mg arsenic /m3. It is uncertain whether some population sub-groups would show an irritant response at lower exposure levels.

II) The skin is a target organ following oral exposure but skin lesions appear to be uncommon following inhalation exposure. Effects include hyperkeratosis, hyperpigmentation, warts and melanomas. No reliable dose response data for these effects has been found.

III) Arsenic has been shown to cause peripheral neuritis. Peripheral neuritis was found in workers inhaling around 50ug arsenic per m3. At this concentration decreased nerve conduction has also been observed.

IV) There is also evidence that arsenic affects the cardiovascular system. Claimed effects range from increased blood pressure and ischaemic heart disease to Raynauds' disease and inhibition of haematopoesis. At a chronic exposure level of 50ug arsenic per m3 both an increased incidence of Raynauds' disease and an enhanced susceptibility to cold initiated blood vessel constriction has been shown (Langerkvist et al 1998).

V) Arsenic has been associated with non-insulin diabetes mellitus (Tseng et al 2000)

VI) Arsenic is known to cause immunotoxic effects in vivo and in vitro (Burns et al 1994). In common with several other metals it appears to produce immunostimulation at low exposure levels but immuno-suppression at higher exposures.

5. Mechanisms of toxicity and carcinogenicity

In principle, if there is good understanding of its mechanism(s) of toxicity, a sound judgement can be made on whether or not there is a threshold for the various toxic effects of arsenic. Mechanisms are not considered in the Position Paper.

Arsenic has the potential to interact with many cellular components. Its direct toxicity is considered to be through an interaction with sulphydryl groups. Arsenic is also thought to act indirectly via generation of active oxygen species and /or by disrupting cellular methylation reactions (Bernstam and Nriagu 2000, Romach et al 2000, Liu et al 2001). Arsenic has been demonstrated to cause oxidative stress analogous to the heat shock response in a number of cell models. This phenomenon has been observed with a number of other inorganic and organic chemicals. It is believed that this mechanism may explain some of the toxic properties of arsenic (Bernstam and Nriagu 2000).

With respect to the mechanism of arsenic induced cancer there remains considerable debate on whether genotoxic or non-genotoxic mechanisms are more important. Arsenic shares some properties with promoters (NAS1999) for example it enhances the clastogenicity of some mutagens. (Simeonova and Luster 2000). Some authors consider that the carcinogenicity of arsenic is due to the alterations it produces in immune-competence, thereby allowing transformed or damaged cells to escape natural immune-surveillance. However, in view of its direct genotoxic effects, summarised above, a genetic mechanism cannot be ruled out on the basis of the presently available evidence.

6. Limit value based on Arsenic being a genotoxic carcinogen

The critical consideration in determining a limit value, based on the carcinogenicity of arsenic, is whether or not a threshold exists for this effect and if so at what exposure level. As discussed above it must be assumed (although an epigenetic mechanism of action is feasible) that the carcinogenicity of arsenic arises from its genotoxic properties. The question is therefore whether a threshold can be identified for arsenic's genotoxicity. There is insufficient data to resolve this directly.

Information, which may be used to support the concept of a threshold, includes:

*It is not a typical mutagen.
*Cellular protective mechanisms might be expected to prevent or reverse changes at low arsenic exposure levels
*In various in vitro studies a sub-linear dose response relationship has been found.

However this data is insufficient to justify the assumption of a threshold.

From epidemiology studies in workers unit risks of between 0.89 x10-3 and 3.77 x10-3 has been identified. WHO (1999) estimate a lifetime risk of cancer of 1:100,000 at an airborne concentration of 6.6ng/m3. It is difficult to translate such data into a limit value because the EU has not established criteria for an 'acceptable risk'.

7. Limit value based on non cancer effects

Studies in workers have indicated both CNS and vascular effects at chronic exposure levels of 50 µg arsenic per m3. There is no reliable animal data to aid in setting a no effect level. The approach used in the position paper is to take the 50ug/m3 figure and divide it by a factor of 5 (to allow for the fact that the public may be exposed 24hrs per day for 365 days rather than 40hrs per week for 225 days). This is a reasonable approach although the workers in the studies reported above were also exposed to ambient air levels of arsenic as well as arsenic from other sources. Two additional uncertainty factors are also introduced in the position paper to allow for the fact that 50ug/m3:

- is a LOAEL rather than a NOAEL (factor of 10 used although it could be argued that a factor of 5 would be more consistent with other applications of a LOAEL to NOAEL safety factor)

- was identified in a limited range of the population rather than the population as a whole (factor of 10)

Bearing in mind the available data this use of uncertainty factors is a reasonable one

It would give an ambient air limit value of 100ng of arsenic per m3

It may be noted that in setting this value the following have not been taken into account:

*arsenic speciation
*the ability of arsenic to promote the genotoxicity of other compounds.

8. Summary

Arsenic is both genotoxic and a well-known human carcinogen. Although there are some reasons for considering that there may be a threshold for carcinogenicity the direct evidence to support this is poor. In the absence of such data it may be considered appropriate to assume that no threshold exists. However this would not allow a basis for setting a threshold limit because the EU has not established criteria for an 'acceptable' risk.

REFERENCES.

- Bernstam L., Nriagu J. (2000) 'Molecular aspects of arsenic stress' in Journal of Toxicology and Environmental Health, Part B, 3:293-322.

- Burns LA, LeVier DG and Munson AE (1994) p213-224 Immunotoxicology of Arsenic. In Immunotoxicology and Immunopharmacology 2nd Edn. Ed JH Dean et al Raven Press NY.

- Chiou HY et al (2001) Amer. J. Epidemiol. 155 422-23

- EC DG Environment (2000) Ambient Air Pollution by AS, CD and NI Compounds. Position Paper. Final Version.

- IARC (1980)

- Langerkvist et al 1998

- Liu S., Athar M., Lippai I., Waldren C. and Hei T. (2001) 'Induction of oxyradicals by arsenic: implications for mechanism of genotoxicity' in Proceedings of National Academy of Sciences USA, Feb 13; 98(4): 1643-1648

- Lubin et al (2000)

- National Academy of Sciences (1999). 'Arsenic in drinking water'. National Academy of Sciences Press, Washington, USA.

- Romach EH. Zhao CQ., Del Razo LM., Cebrian ME., Waalkes MP.(2000) 'Studies of the mechanisms of arsenic-induced self tolerance developed in liver epithelial cells through continuous low-level arsenite exposure' in Toxicological Sciences, 54(2): 500-508.

- Rudel R., Slayton T. and Beck B. (1996) 'Implications of arsenic genotoxicity for dose response of carcinogenic effects' in Regulatory Toxicology and Pharmacology, 23:87-105.

- Simeonova P and Luster M (2000) 'Mechanisms of arsenic Carcinogenicity: Genetic or epigenetic mechanisms?' in Journal of Environmental Pathology, Toxicology and Oncology, 19(3): 281-286

- Tseng CH et al (2000) Environmental Health Perspectives, 108, 847-851.

- Viren and Silvers (1999)

- WHO (1997).