Registration Dossier

Administrative data

Description of key information

There is sufficient evidence in humans for the carcinogenicity of inorganic arsenic compounds, including arsenic trioxide and arsenites. For the risk assessment incl. the derivation of a DNEL for diarsenic trioxide, carcinogenicity is considered the most relevant endpoint; other effects of repeated exposure to arsenic including reproduction toxicity are considered to occur at higher exposures and therefore secondary to cancer. For a further discussion please refer to the overall endpoint summary on toxicological information.

Key value for chemical safety assessment

Additional information

Animal data:

In their most recent monograph , IARC (2012) state that many animal studies have been reported with essentially negative results, but which are not referred to because of inadequacies in experimental design (too few animals, too short a duration, poor survival, too low a level of exposure etc.).

Studies on inhalation exposure in animals with inorganic arsenicals are not available. Whereas investigations involving intra-tracheal administration of arsenite in hamsters elicited formation of lung tumours (adenoma and/or carcinoma) (ATSDR, 2007; IARC, 2004), this route of administration is not of particular relevance for human health risk assessment.

There is merely one 2-year dose–response study of adequate design with sodium arsenite administered via drinking water to Sprague-Dawley rats which showed some evidence of renal tumour formation in female animals but not in males; since the tumour incidence however did not reach significance, the overall outcome with respect to carcinogenicity is negative (Soffrittiet al., 2006).

With respect to animal to human extrapolation however, rodents are generally recognised as a poor model system.ATSDR (2007) for example summarises some of the major drawbacks of animals as a model of arsenic toxicity: (i) there is currently no animal model for the health effect of greatest concern for human exposure, i.e. carcinogenicity in skin and other organs after oral exposure; (ii) the metabolism of arsenic in humans (significant formation of methylated forms) is unlike that of most other mammalian species, and the ratio of inorganic to organic arsenic excreted also varies between species; (iii) rats sequester arsenic in their erythrocytes and thus are not a suitable model for human toxicity.



Human data:

With relevance to human health risk assessment for cancer from exposure to diarsenic trioxide, the available studies consistently point to causal relationships between (I) skin cancer and high exposure to inorganic arsenic in drinking-water with high arsenic content, or (II) lung cancer in the occupational environment. The risk of lung cancer is clearly increased in certain smelter workers who inhale high levels of arsenic trioxide. However, the causative role of arsenic is uncertain, since the influence of other constituents of the working atmosphere cannot be determined. Cases of lung cancer have also been associated with the medicinal use of inorganic arsenic compounds, and of liver haemangioendothelioma following various kinds of arsenic exposure, but these may be chance associations. No evidence exists that other forms of cancer occur excessively with heavy arsenic exposure (IARC, 1998).The pivotal studies for human cancer risk assessment can be summarised briefly as follows:


Oral route, humans:

The latency period for the onset of cancer in humans is approx. 30-50 years. For this reason, most epidemiological studies on arsenic-induced carcinogenicity originate from countries like Taiwan, USA, Chile and Argentina, where environmentally mediated exposure to arsenic has been prevalent for more than 50 years.Previous risk assessments for cancer via the oral route have focused on drinking water studies (UBA, 2007), namely:


(I) Tseng et al. (1968, 1977), ecological study on the incidence of skin cancer and morbidity in southwest Taiwan;

(II) Chen et al. (1985, 1992) and Wu et al. (1989), ecological studies of cancer of bladder and lungs in southwest Taiwan;

(III) Chiou et al. (2001), cohort study on the incidence cancer of the bladder in in northeast Taiwan

(IV) Ferrecio et al. (2000), case-control study on the incidence of lung cancer in Chile.

Previous studies on the relationship between ingestion of arsenic via drinking water and cancer focused on populations exposed to rather high arsenic concentrations, which suggest a linear dose-response. However, the shape of this dose-response curve and where a threshold can be reasonably established are subjects of controversy, particularly since more recent studies (Chen et al. 2004; Ferrecio et al. 2000; both as cited in UBA, 2007) appear to suggest a risk of cancer at concentrations below 100 ug/L (UBA, 2007). In contrast, Lamm et al. (2003, 2006) extrapolate a threshold for carcinogenicity of 160 ug/L based on the epidemiological data from Taiwan. Overall, there is no consistent clear picture regarding the dose-response.

The most recent study by Ahsan et al. (2006) designated the Health Effects of Arsenic Longitudinal Study (“HEALS”) conducted in Araihazar (Bangladesh) may be considered a step forward in that it (i) addresses a large number of subjects (n=11,746), (ii) covers a wide range of arsenic concentrations in drinking water (0.1-864 ug/L), and (iii) exposure to arsenic was well characterised: not only the arsenic in well water was analysed, but also drinking water consumption monitored and most relevant, creatinine-adjusted urinary arsenic excretion as monitored for each participant enrolling in the study.



Inhalation route, humans:

Several studies have been reported in the past that document the association between inhalation exposure to arsenic and lung cancer, mores specifically to arsenic trioxide dust in air at copper smelters. These have previously been described in detail (ATSDR, 2007) and are therefore addressed here only briefly:

(i) a cohort study of works in the Asarco copper refinery in Washington, USA (Enterline et al., 1987;

(ii) a cohort study in the Anaconda copper refinery in Montana, USA (Lee-Feldstein et al., 1983 & 1986);

(iii) a study in a collective of Swedish copper refinery workers (Järup et al., 1989).

The most recent and reliable study however is considered to be the one by Lubin et al. (2008) which is an updated evaluation of the cohort of workers in the Anaconda (USA) refinery. This latter investigation relates to the most recent data with the highest number of man-years and also an attempt at a more thorough exposure assessment. However, some inherent imprecision of such retrospective exposure estimates cannot be excluded.


Justification for classification or non-classification

In their most recent monograph, IARC (2012) conclude that there is sufficient evidence in humans for the carcinogenicity of mixed exposure to inorganic arsenic compounds, including arsenic trioxide and arsenite. These compounds cause cancer of the lung, urinary bladder, and skin. Also, a positive association has been observed between exposure to inorganic arsenic compounds and cancer of the kidney, liver, and prostate. In contrast, IARC also concludes that there is only limited evidence in experimental animals for the carcinogenicity of arsenic trioxide.

In the EU, diarsenic trioxide is subject to a harmonise classification (Regulation (EC) No 1272/2007, Annex I, Index No. 033 -003 -00 -0), including the classification as a Carcinogen Cat 1A.