Antimony

Carcinogenicity

Description of key information

Cf. Scientific opinion on lung toxicity and carcinogenicity in Section 13 for complete weight of evidence and read-across assessments.

Several long-term historical occupational inhalation exposures to Sb compounds have been associated with impairments of lung function resulting from chronic inflammation and fibrosis. Two studies of relatively good quality, and other studies of lower quality document such effects in Sb-exposed workers. In general, however, the pneumoconiosis observed in workers tends to be benign with only little evidence of pathological changes (e.g. fibrosis) that would indicate progressive reactivity of the Sb burden in the lungs of workers.

Studies confirm the historical incidence of pneumoconiosis in workers employed at Sb processing facilities, with an exposure exceeding the current Occupational Exposure Limit of 0.5 mg/m3 by a factor of 10 or more. This is consistent with medical surveillance data reported to the International Antimony Association by its membership and with the decrease of the incidence of pneumoconiosis at an Sb trioxide production facility after the implementation of OELs.

Although epidemiology studies in smelter environments (Jones et al., 1994; Schnorr et al., 1995 and Jones et al., 2007) have reported small increases in lung cancer in workers occupationally exposed to Sb trioxide, attribution to Sb compounds has not been possible due to significant levels of co-exposure to known lung carcinogens such as arsenic and cadmium. An older study of glassblowers suggested an association with colon and stomach cancer (Wingren and Axelson, 1987) but measurements of Sb exposure were not available and the association with the cancer data not presented. An ecological study of breast cancer incidence in the United States (White et al., 2019) failed to find an association with airborne Sb levels (although exposure levels were low). Finally, Deng et al (2019) observed no impacts upon early biomarkers of potential health effects (e.g. the Comet Assay) except for altered expression of microRNAs associated with the metabolism of polycyclic aromatic hydrocarbons. Urinary Sb levels served as the index of exposure and, curiously enough, was higher in the non-exposed controls. No consistent association between cancer incidence and Sb exposure is evident in these studies.

Animal studies using exposures of sufficient duration and respirable particle aerosols with a small size facilitating penetration to the deep lung (< 4 µm) confirmed that Sb trioxide can have toxic impacts upon the lung. This affirmation needs to be properly evaluated regarding to the workers exposed in industrial facilities as the occupational aerosols possess a larger particle size distribution (Hughson, 2005) with larger particles (the inhalable fraction) preferentially depositing in the nose, throat and upper airways. The respirable fraction is, on average, only about one third the size of inhalable fraction.

Animal inhalation studies are thus designed to maximize the likelihood of damage to tissues of the deep lung. Although three initial experimental inhalation studies deviate from standard protocols (one year of exposure opposed to the two years specified by most cancer bioassay guidelines), it has been demonstrated that Sb trioxide could impair particle lung clearance. More recently, a two-year cancer bioassay (NTP, 2017) reported evidence of a relationship between exposure to respirable Sb trioxide and lung tumors in the mouse and, to a lesser extent, the rat.

Evidence is only available for Sb metal, Sb trioxide, and Sb trisulfide. There is no information on the potential to cause lung toxicity or cancer for any other Sb substance. While all studies have reported some degree of lung toxicity, only one reports clear evidence of cancer (in mice). Evidence in rats is either less clear or can be attributed an overload response.

Key value for chemical safety assessment

Carcinogenicity: via oral route

Endpoint conclusion

Endpoint conclusion:

no study available

Carcinogenicity: via inhalation route

Endpoint conclusion

Endpoint conclusion:

adverse effect observed

Dose descriptor:

LOAEC

3 mg/m³

Study duration:

chronic

Species:

rat

Quality of whole database:

Test was conducted with diantimony trioxide. A weight of evidence and read-across approach is applied to assess the toxicological endpoint. Cf. scientific opinion in Section 13.

Organ:

lungs

Carcinogenicity: via dermal route

Endpoint conclusion

Endpoint conclusion:

no study available

Mode of Action Analysis / Human Relevance Framework

Cf. Scientific opinion on lung toxicity and carcinogenicity in Section 13 for complete weight of evidence and read-across assessments.

In order to assess the potential lung toxicity and carcinogenicity of Sb substances, it is important to nature of the effects which have been reported and try to understand the mechanism(s) by which Sb compounds may damage the lung, and the chemical species involved in such a response.

Lung toxicity

The historical medical surveillance literature, documenting the impacts of inhalation exposure to Sb trioxide in occupational setting, is concordant with the animal studies in that impairments of pulmonary function (e.g. spirometry deficits and radiographic indications of mild pulmonary fibrosis) were associated with occupational exposures experienced prior to the adoption of modern OELs (McCallum, 1967; Potkonjak and Pavlovich, 1983). Although the pulmonary changes associated with occupational exposures were much less severe than those evident in rats and mice, they confirm that the human lung can be adversely impacted by inhalation exposure to Sb trioxide and ore materials containing Sb trisulfide (stibnite). The combined animal and human exposure data support a STOT RE classification for impacts upon the lung after repeated inhalation exposure to Sb trioxide and Sb trisulfide.

STOT RE classifications are further assigned either category 1 or category 2 (ECHA, 2017). Category 1 classifications are indicative of high potency for the product of significant to severe health effects whereas a category 2 classification indicates moderate potency to induce significant health effects. Although the ECHA Classification and Labelling guidance indicates that Category 1 designations are often indicated when there is “good quality evidence from human case or epidemiology studies”, it is further noted that “In exceptional cases human evidence can also be used to place a substance in Category 2” (ECHA, 2017). Category assignment of a STOT RE substance thus entails a weight of evidence analysis that results in a classification which accurately conveys both the potency of the substance and the severity of the health effects observed.

STOT RE Category 1 designations are triggered by the observation of significant or severe impacts in rats at aerosol concentrations less than 0.02 mg/liter/6h/day (20 mg/m³) in a 90-day exposure study, whereas category 2 is indicated for effects induced between 0.02 and 0.2 mg/liter/6r/day (20 – 200 mg/m³). These values are not intended as strict demarcation points but as general guidance to be used in conjunction with expert judgement. Adjustment of these values in accordance with Haber’s rule is also suggested – thus the demarcation value of 20 mg/m³ would be reduced by a factor of at least 4 (to < 5 mg/m³) in comparing results of a 90-day study and to those from a 1 – 2-year inhalation study.

The comprehensive two-year inhalation studies of NTP (2017) observed a LOAEL for pulmonary impacts of 3 mg/m³, just below the Haber’s law adjusted Category 1 demarcation value. However, the NTP (2017) studies utilized experimentally generated respirable Sb trioxide aerosols capable of deep lung penetration and deposition. Studies of real-world occupational aerosols indicate that their particle size distribution has a relatively low content of respirable particles. On average, for exposed humans, inhalable aerosols capable of yielding pulmonary deposition fractions comparable to those produced by experimentally generated Sb trioxide aerosols used in rodent inhalation studies would require a 5-fold higher concentration of Sb trioxide in air (Hughson, 2005; Vetter, 2018).

Using the Multiple-Path Particle Dosimetry Model (v. 3.01) described by Ashgarian and Price (2009), one can further compare the pulmonary deposition of the experimental aerosols used by NTP (2017), with those of the real-world occupational Sb trioxide aerosols measured by Hughson (2005). Whereas the NTP aerosols (MMAD 1.2 µm +/- 1.9 GSD) would yield a pulmonary deposition rate in rats of 7.6%, the average particle size distribution of the aerosols sampled by Hughson (2005), as calculated by Vetter (2018) had an MMAD of 17.2 µm with a GSD of 2.7. This would yield a pulmonary deposition rate in the rat of 0.3%. In terms of potency for the rat, a real-world Sb trioxide occupational aerosol of 75 mg/m³ would be required to produce the pulmonary impacts observed by NTP (2017) at their airborne LOAEL of 3 mg/m³. This total aerosol Sb trioxide concentration is significantly above the 5 mg/m³ demarcation point for chronic exposure potency in establishing category 1 vs. category 2 in STOT RE inhalation classifications.

Granulometry studies are described in the CSRs for Sb metal powder and Sb trisulfide, and predict the characteristics of the aerosols each would produce. Sb metal powder would be expected to generate an occupational aerosol with an MMAD of 19.05 µm +/- 2.75 GSD. A bimodal distribution is predicted for Sb trisulfide aerosols with 11% of the particle mass having an MMAD of 2.69 µm +/- 2.38 GSD, and 88.7% of the aerosol mass with a MMAD of 28.48 µm +/- 1.56 GSD). MPPD modelling predicts that pulmonary deposition rates of 0.05% and 0.22% would result from aerosols of Sb metal powder and Sb trisulfide, respectively. Pulmonary deposition rates equivalent to those for rats exposed to 3 mg/m³ of Sb trioxide in the NTP studies would thus require Sb metal powder aerosols of approximately 100 mg/m³ and Sb trisulfide aerosols of 450 mg/m³. Real-world aerosols of Sb trioxide, Sb trisulfide and Sb metal powder would be judged to have a moderate to low potency as pulmonary toxicants when viewed from the perspective of deposition rates in the lung regions that are the targets for pulmonary toxicity.

Historical exposures capable of producing human pulmonary impacts after years of chronic exposure, although not precisely defined, were most likely in significant excess of 10 mg/m³ (ECHA, 2008). The historical exposure levels associated with changes to lung pathology and function confirm that Sb trioxide has only moderate potency for inducing pulmonary impacts in humans. The nature of the pulmonary alterations associated with exposure of humans to Sb trioxide provides further indications that Sb trioxide has only moderate potency as a pulmonary toxicant.

Inhalation of Sb trioxide by rats and mice produced severe impairment of both pulmonary structure and function (NTP, 2017). The severity of the impacts in rodents, contrasts with the observed impacts in humans. Although impacts upon human lung function are judged as clinically significant, the pulmonary function impacts observed are generally mild. The underlying alterations to human lung tissue that mediate these modest functional changes are in turn associated with comparatively modest inflammatory responses and rather benign and generally non-progressive fibrotic changes.

The concentrations of Sb trioxide associated with pulmonary toxicity in both humans and rodents indicate moderate potency that is consistent with a STOT RE category 2 classification. The relatively benign and non-progressive nature of the structural alterations documented in workers with high-level historical occupational exposures similarly indicates relatively mild potency consistent with STOT RE category 2 classification for lung toxicity from inhalation exposure. Modelling of the alveolar deposition fractions predicted for rats exposed to aerosols of Sb metal powder and Sb trisulfide further indicates that the potency of these substances could be lower than Sb trioxide and thus also consistent with a category 2 STOT RE classification.

(Lung) carcinogenicity

As just noted, chronic inhalation of Sb trioxide by rats and mice can produce damage to the lungs characterized by the progressive development of pulmonary inflammation, tissue damage and fibrotic changes. These dose-dependent changes, at sufficiently high exposures, can produce significant impairment of pulmonary function and severe systemic hypoxia that induces adaptive physiological changes (e.g. erythroid hyperplasia).

Activation and alteration of oncogenes

NTPs (2017) analysis of mouse and rat lung tumors extended to an evaluation of oncogene alterations associated with tumor formation. The presence of activated oncogenes in tumors is informative but can be the result of a myriad of direct and indirect processes. Focusing on the mouse lung tumors, which were observed with far higher frequency, permits more robust analysis of the “molecular pathology” responsible for activated oncogenes in spontaneous and induced neoplasms. Spontaneous lung tumors were found to contain altered Kras genes with the activating mutations generally mapping to established “hot spots” (i.e. G to A transitions in codon 12). Altered Kras oncogenes were detected in 43% of the tumors observed in Sb trioxide treated animals. NTP notes that tumors in Sb trioxide treated animals possessed base sequence changes in hot spots similar to those observed in spontaneous tumors, and suggests that the Kras altered genes observed in the tumors of Sb trioxide treated animals were the result of spontaneous lesions, permitted to undergo clonal expansion by the pulmonary toxicity of Sb trioxide. This suggestion is consistent with the observation that spontaneous activated oncogenes are now known to be present in the normal tissues of animals used in cancer bioassays (Parsons et al., 2009), exhibiting both tissue and animal strain specificity with respect to the prevalence of different activated oncogenes.

In addition to Kras alterations, 46% of lung tumors in Sb trioxide treated mice were observed to contain altered Egfr oncogenes. The high prevalence of tumors with Egfr alterations in exposed animals could be interpreted as evidence of mutagenic oncogene alterations induced by Sb trioxide. However, the origin of Egfr alterations is potentially more complex than is described. In humans, lung cancer tumors are increased in subjects with disease syndromes (e.g. chronic obstructive pulmonary disease) that impair lung function and lead to hypoxic conditions. Signaling pathways involving EGFR appear to play a role in the growth of such tumors under hypoxic condition (Karoor et al., 2012). Egfr alterations are further linked to the ability of cancer cells to survive in hypoxic microenvironments (Murakami et al., 2014). The prevalence of Egfr alterations in Sb trioxide treated animals may thus be a result of selection for tumors capable of undergoing rapid clonal expansion, under the hypoxic conditions associated with the pulmonary toxicity produced by Sb trioxide. The activated Egfr oncogenes may thus be spontaneous in origin or produced by a variety of indirect processes during tumor progression (e.g. ROS generation, error prone DNA repair) with an increased prevalence in tumors that is more indicative of the conditions that permitted clonal expansion of neoplastic lesions. The mere observation of an activated oncogene in a tumor, in and of itself, confers little information that permits determination of the mechanism(s) that may have produced it.

These oncogene structural alterations do not represent the only potential means by which Sb trioxide might influence the expression of oncogenes or other cellular constituents. The electrophilic nature of Sb is such that binding to a variety of cellular macromolecules occurs (Verdugo et al., 2017) and could facilitate neoplastic development. For example, although Sb has not been linked to prostate cancer, Sb 3+ ions will activate signaling pathways that stabilizes the c-myc oncogene that stimulates cell proliferation (Zhang et al., 2018).

Whereas NTP suggests that many of the lung tumors in mice originate from cells with spontaneous Kras oncogene activation that are permitted to undergo clonal expansion in response to the pulmonary toxicity induced by Sb trioxide, mouse lung tumors with Egfr lesions may similarly reflect selection for, and clonal expansion of, cells with enhanced proliferative capacity under hypoxic conditions. It is not possible to ascertain whether Egfr alterations are spontaneous or induced. The etiology of Kras and Egfr oncogene alterations observed in lung tumors merits investigation to determine if they are pre-existing spontaneous lesions, lesions induced by Sb via indirect mechanisms of genotoxicity and/or lesions selected for clonal expansion as a consequence of pulmonary toxicity and hypoxia.

Body weight suppression

The impacts of Sb trioxide exposure upon the overall health status of rats and mice should not be neglected, and may explain other adverse effects observed in the NTP studies. Exposure of rats to 3, 10 or 30 mg/m³ Sb trioxide was associated with end of study body weight suppression of 7, 8 and 20% in male rats; and 10, 20 and 28% in female rats, respectively. Corresponding body weight suppression in male mice was 8, 11 and 25%; and 3, 8 and 21% in female mice. Much of the data generated by the NTP bioassays reflects effects near, or in excess of, the maximum tolerated dose for Sb trioxide. This conclusion is bolstered by the observations of labored breathing, hypoxia and premature mortality due to pulmonary inflammation in exposed animals. These observations do not negate the induction of pulmonary lesions, but indicate that care must be exercised in the interpretation of systemic effects that might be associated with inhalation exposure to Sb trioxide.

Pheochromocytomas

Adrenal gland neoplasms (pheochromocytomas) are cited by the US NTP Monograph (p. 65 – 66) as supporting classification of Sb trioxide as a carcinogen. However, adrenal gland neoplasms (pheochromocytomas) lesions are expected to develop under conditions of pulmonary inflammation and hypoxia. As reviewed by Greim et al. (2009), the association of this adrenal lesion with pulmonary impairment is sufficiently robust that, within the context of the EU REACH process, pheochromocytomas secondary to pulmonary impairment are not considered as relevant for cancer classification or risk assessment. The adrenal lesions are a response to pulmonary damage induced by Sb trioxide and not a direct substance-specific effect of Sb trioxide. Indeed, they can be interpreted as confirmation that maximum tolerated doses have been exceeded in the rat.

Lymphomas

Sb trioxide exposures in mice were also associated with an increase in lymphomas. Interpretation of increased lymphoma incidence in female mice poses diagnostic challenges that were not addressed by NTP’s histopathological analysis. Whereas lymphomas induced by chemicals are usually T cell in origin (Ward, 2005), those associated with Sb trioxide exposure were predominantly B-cell or mixed B- and T-cell in origin, and many appeared to be reactive lesions responding to Sb trioxide lung toxicity. Mouse B-cell lymphomas are further difficult to interpret due to their high spontaneous incidence and complex etiology that likely includes endogenous retrovirus activity. In NTP inhalation studies, the average historical control incidence of lymphomas in B6C3F1 female mice is 25.2% (range 14 – 36%). Thus, lymphoma incidence at 10 and 30 mg/m³ Sb trioxide, but not 3 mg/m³, was significantly elevated over historical controls. The complex and diverse mechanisms for B-cell lymphoma induction have prompted the development of histopathological diagnosis and classification strategies to distinguish between spontaneous and induced lesions (Ward, 2005). Unfortunately, none of these diagnostic criteria were applied in the NTP study. Based upon the limited data provided, the excess lymphomas associated with Sb trioxide exposure appear to be similar to the naturally occurring lesions in the B6C3F1 mouse; it can be plausibly postulated that the chronic inflammation and hypoxic conditions in the Sb trioxide exposed lung produced adaptive responses in the lung and spleen that promoted the development of what is already a high incidence spontaneous neoplasm in the female mouse. As such, the increased incidence of lymphomas would not provide clear evidence of carcinogenicity.

Skin lesions

Neoplastic skin lesions were also observed in mice exposed to Sb trioxide and different types of skin lesions were pooled to yield statistical significance. Given the high-level whole-body inhalation exposures employed by NTP, the appearance of histiocytomas (a benign skin lesion) is mostly likely an immunological response, as opposed to neoplastic response, and not a precursor lesion to fibrosarcoma (malignant tumors of fibrous tissues). Histiocytomas are not generally known to be precursor lesions to fibrosarcoma and there appears to be no legitimate scientific rationale to support data pooling. The observation of two squamous cell carcinomas in Sb trioxide treated female mice is unusual but is similarly difficult to interpret in the absence of preneoplastic precursor lesions. Moreover, no other study has suggested skin as a target organ for Sb trioxide carcinogenesis. There is no legitimate scientific rationale to support that skin tumors are induced by Sb trioxide.

Lung cancer

According to the experimental studies, Sb compounds might pose a carcinogenic risk to the lungs of rats through particle overload (Newton et al., 1994; Schroeder, 2003). Rat’s lungs do not have the capacity to remove excessive quantity of respirable particle and this triggers a cascade of inflammatory responses leading to a tumor formation, by accumulation of inert particles. This response to inflammation from particle overload is not observed in mice or humans. There is no statistically significant increase in rat lung tumors at Sb trioxide concentrations (3 mg/m3) that do not produce pulmonary overload. Rat lung tumor incidence at higher exposure levels is low, lacking in dose-response and most likely the result of pulmonary overload. As such, the rat pulmonary lesions are not reflective of human risk. Therefore, rat lung tumors, if induced by particle overload, would be of questionable significance for hazard classification or risk assessment.

NTP (2017) concluded that overload did not occur in rats at an Sb trioxide exposure of 3 mg/m³ and therefore that pulmonary overload is not required for the induction of lung neoplasms in the rat lung. The rationale for this conclusion is tenuous in that 3 mg/m³ is indeed associated with impaired clearance in the rat in the NTP studies – the departure from modeled clearance rates is just not sufficient to attain the lung burden levels that meet an arbitrary criterion for overload. Moreover, significant impairment of clearance has been reported at levels much lower than those used in the NTP studies (e.g. Newton et al., 1994). Finally, the incidence of lung neoplasms in both male and female rats is not statistically elevated over that in controls at 3 mg/m³ Sb trioxide exposures. The lack of both overload and a carcinogenic response in the rat at 3 mg/m³ Sb trioxide cannot be taken as an indication that tumors produced in the rat lung at higher levels of exposure were not the result of the pulmonary overload. Particle overload and the subsequent cascade of inflammatory responses leading to a tumor formation can be retained as a possible mode of action for lung cancer in rats, but would be of questionable relevance for hazard classification or risk assessment.

As discussed in the scientific opinion on genotoxicity, any Sb genotoxicity that might facilitate neoplastic development is likely mediated by indirect mechanisms, such as induction of oxidative stress or interference with DNA repair processes. The available data do not permit discrimination between alternative mechanisms, nor do the mechanisms need to be mutually exclusive, but there is relatively high confidence that the lung carcinogenicity is not a result of direct genotoxicity of Sb. Excess tumors observed may reflect the clonal expansion of pre-existing preneoplastic cells with activated oncogenes in the absence of genotoxicity (direct or indirect). If lesions are induced, it is most likely via a local indirect genotoxic mode of action. The most probable indirect modes of action (e.g. overload in the rat, inflammation and ROS generation in the mouse) would be expected to exhibit effect thresholds that produce neoplastic response only above a given level of inhalation exposure.

Sb trioxide appears to induce cancer at tissue sites (adrenal, lymphoma and skin) that are likely side-effects of pulmonary toxicity or the irritant properties of Sb trioxide. These lesions are not relevant to an evaluation of the carcinogenic properties of Sb trioxide. The primary target organ of inhaled Sb substances appears to be the lung, and mode of action considerations should look at local effects in the lung rather than systemic effects. The inhalation exposure route is the only route of exposure relevant for the assessment of carcinogenicity properties. Exposure route specificity (the lung by inhalation exposure) is further evidenced by lack of pulmonary changes after sub-chronic oral exposures to high doses of Sb trioxide (Hext et al., 1999) and high sub-chronic i.p. dosing with the Sb (III) potassium tartrate (Dieter, 1992).

Data from experimental animal studies do not yield compelling evidence of cancer risk at exposure levels, or via mechanisms, that are likely to be relevant to present occupational or consumer exposure scenarios. Epidemiological studies have failed to demonstrate elevated cancer risk that can be attributed to Sb trioxide exposure. According to the ECHA Guidance on the Application of CLP criteria, the present evidence satisfies, and likely exceeds, that required for a Category 2 cancer via inhalation classification. Indeed, according to the ECHA Guidance on the Application of the CLP Criteria (July 2017), suspected human carcinogens are those for which the evidence obtained from human and/or animal studies is not sufficiently convincing to place the substance in Category 1A or 1B, in particular, when e.g. the data suggest a carcinogenic effect but are limited for making a definitive evaluation because there are unresolved questions regarding the adequacy of the interpretation of the results of the studies. In light of the discussion presented above, there are clear interpretation issues which do not permit to conclude on a category 1 carcinogenicity, and rather suggest maintaining the current Category 2 cancer classification:

• Tumour background incidence - comparison of the tumour incidence with historical control tumour data. This can be particularly relevant for animal strains which have a propensity to develop a particular type of tumour spontaneously with variable and potentially high incidence. In such a case, the tumour incidence may not be providing reliable evidence of treatment related carcinogenicity;
• The possibility of a confounding effect of excessive toxicity at test doses. In lifetime bioassays, compounds are routinely tested using at least three dose levels, of which the highest dose needs to induce minimal toxicity, such as characterized by an approximately 10% reduction in body weight gain (maximal tolerated dose, MTD dose). The MTD is the highest dose of the test agent during the bioassay that can be predicted not to alter the animal’s normal longevity from effects other than carcinogenicity. If a test compound is only found to be carcinogenic at the highest dose(s) used in a lifetime bioassay, and the characteristics associated with doses exceeding the MTD are present, this could be an indication of a confounding effect of excessive toxicity/excessive loading. This may support a classification of the test compound in Category 2 or no classification; and
• Mode of action and its relevance for humans, such as cytotoxicity with growth stimulation, mitogenesis, immunosuppression, mutagenicity. The various international documents on carcinogen assessment all note that mode of action in and of itself, or consideration of comparative metabolism, should be evaluated on a case-by-case basis and are part of an analytic evaluative approach. One must look closely at any mode of action in animal experiments taking into consideration comparative toxicokinetics/toxicodynamics between the animal test species and humans to determine the relevance of the results to humans. The criteria for a Carcinogenicity Category 1B classification are not met due to the uncertain relevance of both rat and mouse lung tumors for humans.

Justification for classification or non-classification

Cf. Scientific opinion on lung toxicity and carcinogenicity in Section 13 for complete weight of evidence and read-across assessments.

The combined animal and human exposure data support a Carcinogenicity category 2 via inhalation classification for Sb trioxide. Based on physical form/particle size, water solubility, and Sb speciation/valency, the same classification can be applied to Sb metal and Sb trisulfide. Sb trichloride and Sb tris (ethylene glycolate) do not satisfy the criteria to be grouped with Sb metal, Sb trioxude and Sb trisulfide for purpose of lung carcinogenicity classification, and are not classified for carcinogenicity.

The main uncertainty underpinning this pertains to the mode of action related to the carcinogenicity, but the conclusion can be established (based on human exposure evidence), with a relatively high level of confidence, but the following research options could reinforce the classification justification:

• Determine whether the Kras and Egfr oncogene alterations observed in lung tumors are pre-existing spontaneous lesions, lesions induced by Sb via indirect mechanisms of genotoxicity, and/or lesions selected for clonal expansion as a consequence of pulmonary toxicity and hypoxia;
• Clarify the (local) genotoxicity of Sb substances (cf. scientific opinion on genotoxicity);
• Run one or more in vivo inhalation study to confirm the observations gathered so far.

The research strategy developed by the International Antimony Association, which supports REACH registrants with their Registration and Evaluation obligations, already foresees the above research options.