Registration Dossier
Registration Dossier
Data platform availability banner - registered substances factsheets
Please be aware that this old REACH registration data factsheet is no longer maintained; it remains frozen as of 19th May 2023.
The new ECHA CHEM database has been released by ECHA, and it now contains all REACH registration data. There are more details on the transition of ECHA's published data to ECHA CHEM here.
Diss Factsheets
Use of this information is subject to copyright laws and may require the permission of the owner of the information, as described in the ECHA Legal Notice.
EC number: 231-158-0 | CAS number: 7440-48-4
- Life Cycle description
- Uses advised against
- Endpoint summary
- Appearance / physical state / colour
- Melting point / freezing point
- Boiling point
- Density
- Particle size distribution (Granulometry)
- Vapour pressure
- Partition coefficient
- Water solubility
- Solubility in organic solvents / fat solubility
- Surface tension
- Flash point
- Auto flammability
- Flammability
- Explosiveness
- Oxidising properties
- Oxidation reduction potential
- Stability in organic solvents and identity of relevant degradation products
- Storage stability and reactivity towards container material
- Stability: thermal, sunlight, metals
- pH
- Dissociation constant
- Viscosity
- Additional physico-chemical information
- Additional physico-chemical properties of nanomaterials
- Nanomaterial agglomeration / aggregation
- Nanomaterial crystalline phase
- Nanomaterial crystallite and grain size
- Nanomaterial aspect ratio / shape
- Nanomaterial specific surface area
- Nanomaterial Zeta potential
- Nanomaterial surface chemistry
- Nanomaterial dustiness
- Nanomaterial porosity
- Nanomaterial pour density
- Nanomaterial photocatalytic activity
- Nanomaterial radical formation potential
- Nanomaterial catalytic activity
- Endpoint summary
- Stability
- Biodegradation
- Bioaccumulation
- Transport and distribution
- Environmental data
- Additional information on environmental fate and behaviour
- Ecotoxicological Summary
- Aquatic toxicity
- Endpoint summary
- Short-term toxicity to fish
- Long-term toxicity to fish
- Short-term toxicity to aquatic invertebrates
- Long-term toxicity to aquatic invertebrates
- Toxicity to aquatic algae and cyanobacteria
- Toxicity to aquatic plants other than algae
- Toxicity to microorganisms
- Endocrine disrupter testing in aquatic vertebrates – in vivo
- Toxicity to other aquatic organisms
- Sediment toxicity
- Terrestrial toxicity
- Biological effects monitoring
- Biotransformation and kinetics
- Additional ecotoxological information
- Toxicological Summary
- Toxicokinetics, metabolism and distribution
- Acute Toxicity
- Irritation / corrosion
- Sensitisation
- Repeated dose toxicity
- Genetic toxicity
- Carcinogenicity
- Toxicity to reproduction
- Specific investigations
- Exposure related observations in humans
- Toxic effects on livestock and pets
- Additional toxicological data
Endpoint summary
Administrative data
Key value for chemical safety assessment
Additional information
In this dossier, the endpoint genetic toxicity is not addressed by substance-specific information, but instead by a weight of evidence approach based on collected information for all cobalt substances of the CoRC cobalt category. Further details on the read-across approach are given in the reports attached in IUCLID section 13.2. The assessment of the genotoxic properties of cobalt and its substances is related to the assumption that once inorganic cobalt compounds or cobalt metal become bioavailable, this will be in the form of the divalent cobalt cation. Further assuming that the anion of such inorganic cobalt compounds can be regarded as “inert” with regard to genetic toxicity, the subsequent discussion focuses on the cobalt cation. This assessment is restricted to those studies which are relevant from a regulatory point of view and fulfil the relevance reliability and adequacy criteria as e.g. laid down in the ECHA Guidance on information requirements and chemical safety assessment, Chapter R.4: Evaluation of available information. In contrast, studies reporting DNA damage in bacteria, induction of SCE in mammalian cells or tests in inappropriate test systems such as yeasts are considered to contribute far less to the overall assessment and are therefore not further discussed.
in vitro results
bacterial test systems:
Some papers report negative results for bacterial mutation tests with cobalt salts. However, many of them have defects either in terms of the design, or the amount of detail reported. Four reports showing some evidence for mutagenic activity. However, follow-up GLP studies have not been able to confirm these indications of mutagenic activity:
Evidence that cobalt sulfate heptahydrate was a weak inducer of mutations in Salmonella typhimurium strain TA100 when using the pre-incubation method was published by Zeiger et al (1992), when reporting on NTP studies. Approximately 2-fold increases in TA100 revertants were seen in 3 separate experiments in the absence of metabolic activation (rat or hamster liver S9), with a peak response at 333 µg/plate in each experiment. Two-fold increases in TA100 revertants are usually considered to be biologically meaningful, and the peak responses were either just below or just greater than 2-fold, hence the conclusion of “weak”. However, the reproducibility of response across 3 independent experiments suggests the responses were not due to chance.
In the NTP study (NTP, 2013) cobalt metal was evaluated for bacterial mutagenicity in 2 strains of Salmonella typhimurium (TA98 and TA100) and 1 strain of E. coli (WP2 uvrA/pKM101) using the pre-incubation method. Five concentrations were tested in triplicate in the absence and presence of S9 in 2 independent experiments. Cobalt metal (100 to 5,000 μg/plate) gave an equivocal response in TA100 in the absence of S9 activation mix. In the presence of 10% rat liver S9, doses up to 7,500 μg/plate did not induce an increase in mutant colonies in TA100. In strain TA98 without S9, cobalt metal (100 to 3,500 μg/plate) was mutagenic, although the responses observed were weak and not well correlated with dose level. In the presence of S9, no mutagenic activity was observed in TA98. In E. coli WP2uvrA/pKM101, doses of cobalt metal up to 450 μg/plate were not associated with mutagenic activity, with or without S9.
Pagano & Zeiger, 1992: Cobalt chloride induced mutations in TA97 when using the pre-incubation method with different aqueous (sterile water or buffers, including phosphate buffer) solvents. The increases in TA97 revertants were dose-related and reached 5-6-fold control levels at 800 µM, which is clearly biologically and statistically significant. The top concentration = 104 µg/ml in the pre-incubation mixture. As the final volume of the pre-incubation mixture was 700 µl, this would equate to 72.8 µg/plate, which is well below the required upper limit for testing. However, it needs to be noted that the Salmonella strain TA97 was commonly replaced by the TA97a, since the former showed genetic instability in reverse mutation testing leading to a not substance induced increased mutation rate (“false positive”).
Wong, 1988: Reported evidence of mutagenic activity with cobalt chloride, using the plate incorporation method. No data from individual concentrations was presented, only the slope (revertants/µg) at the linear part of the dose-response curve. The control revertant numbers for 2 of the strains (TA1535 and TA1537) are unusually high, and therefore the reliability of the study is questionable. The range of concentrations used induced 50-90% toxicity. As there are no details, it is not clear how the possible outcome of non-mutant microcolonies resembling mutant colonies at high levels of toxicity was controlled for. The authors reported mutagenic responses in the absence of S9 in TA98 and TA1537.
In light of the above publications, a series of GLP studies were performed on cobalt metal, cobalt chloride and cobalt sulfate. In order to investigate the bacterial strains that had shown evidence of potential mutagenic effects, metallic cobalt was tested in strain TA98, cobalt chloride was tested in strain TA97a and cobalt sulfate was tested in strain TA100. The studies were performed in two different laboratories. The studies in both laboratories with all 3 test chemicals were therefore very similar and equally robust. In both laboratories, there was no evidence of any increases in revertant numbers with any of the test chemicals under any of the treatment conditions, and all 3 were appropriately concluded as negative. Thus, the indications of mutagenic potential for cobalt metal and its salts found in the published literature and by NTP have not been confirmed in recent, robust, GLP studies where similar or higher concentrations were evaluated.
Thus, the indications of induction of bacterial mutation found in published papers and NTP reports has not been substantiated in recent, robust GLP studies. Overall, there is no convincing evidence that either metallic cobalt or soluble cobalt salts (chloride and sulfate) are mutagenic in the Ames test.
Several studies on bacterial reverse mutation testing (Ames test) were identified which do not fulfil the relevance, reliability and adequacy criteria as foreseen by the ECHA Guidance on information requirements. These studies are discussed below in brief for information purposes only and were included in the SIDS (IUCLID) as summary entry.
Schultz et al. (1982): A method description was not given. Unclear information on dose levels. Cobalt coordination complex substances were used with large organic ligands which are deemed unsuitable, since not covered by the current assessment and read-across concept.
Arlauskas et al. (1985): Reasonably well described methods description. However, although the authors state cobalt dichloride as test item, no results were reported.
Tso & Fung (1981): Single strain experiment with no individual results reported. No control cultures used, confirmatory experiment not performed. Hence, the test system was not in line with guideline recommendations.
Turoczi et al. (1987): Only a conference/poste abstract was available, which does not provide sufficient information for an assessment. Crucial information was missing such as: control cultures apparently not used, evaluation criteria, toxicity, information on revertants per plate.
Mochizuki & Kada, (1982): Detailed method description missing, e.g. exposure duration, metabolic activation, plate preparation.
Chromosomal aberrations (CA) in mammalian cells, published references:
Surprisingly there are few studies on induction of CA in mammalian cells. There are only 2 relevant papers.
Olivero et al (1995) studied both CA and micronuclei (MN) in human lymphocytes treated with cobalt chloride, sulfate and nitrate (the MN results are discussed below). Treatment (only in the absence of metabolic activation) was begun 24 hr after PHA stimulation and continued for 48 hrs. The control frequencies of CA were very high (7-16%) compared to normal levels (around 1.5%), and the classes of aberrations scored are unclear – it is possible that some of these abnormalities were gaps, which are normally not considered in the assessment of a positive response. Although the percentage of aberrant cells increased (to 2-fold) following cobalt chloride treatment, the results were not statistically significant. However, the data are sufficiently questionable that a negative result is not convincing.
Paton & Allison (1972) studied the induction of CA by cobalt nitrate in 2 human cell lines, WI38 and MRC5, and human leukocytes. The leukocytes were cultured for 72 hr before treatment (presumably with mitogen stimulation), which means they were in exponential growth at the start of treatment. Treatments were only in the absence of metabolic activation, and were for times ranging from 2-24 hr in the cell lines, and for 48 hr in the leukocytes. In the cell lines, no quantitative measure of cytotoxicity was employed – only visual observation of rounding up of cells, although apparently sub-toxic concentrations were tested for CA induction. Despite these extended treatments, no CA were induced either in the cell lines or in human leukocytes. However, the methods are far from standard and the negative result must be viewed with caution.
Several studies on in vitro clastogenicity tests were identified which do not fulfil the relevance, reliability and adequacy criteria as foreseen by the ECHA Guidance on information requirements. These studies are discussed below in brief for information purposes only and were included in the SIDS (IUCLID) as summary entry.
Miller et al., 2001: The experimental procedure is incompletely described. The suitability of the cellular system was not demonstrated and cobalt metal particle-cell interaction was not monitored. Clastogenic events were not correlated with cytotoxicity, hence it remains unclear whether clastogenic events were due to direct interaction or secondary to cytotoxicity. The exposure duration of 1hr is too short, the relevant OECD TG recommends 3-4 hrs.
Daley et al., 2004: The chemical identity of the test item was not conclusively determined/reported. It is unclear at which doses the cells were exposed. The centromere staining was not described. CytoB was added directly with the test item.
Suzuki et al., 1993: The cell system was not further described/characterised, the suitability for detecting clastogenic events was not demonstrated. A positive control substance was not used and therefore data on the specificity and selectivity not reported. No correlation of mutagenic effects with cytotoxicity, hence it remains unclear whether clastogenic events were due to direct interaction or secondary to cytotoxicity.
Resende de Souza Nazareth, 1976: The reference exhibits serious reporting and experimental deficiencies, such as an incomplete description of the experimental procedure. It is unclear how many replicates were used and in which density the cells were plated. The sex, health status and age of the blood donors was not reported. A single dose regime was used which renders this reference unsuitable for risk assessment purposes. Authors postulate to analyse numerical aberrations in human blood cells after treatment with a cobalt salt. However, the induction of numerical aberrations is a rare event so that it appears highly implausible that a positive finding can be detected after analysing an average of <20 cells per culture.
Voroshilin et al., 1978: Authors conclude that cobalt diacetate induces hyperdiploidy in cultured peripheral blood lymphocytes. However, the results of this reference appear questionable due to serious experimental and reporting deficiencies as follows: the health status of donors was not reported, the culture conditions were not reported in sufficient detail, the information on cytotoxicity cannot be reproduced and are therefore unclear, the cell analysis in G0 phase is implausible for a chromosome aberration analysis (no cytokinesis blocker was added), the analysis of hyperdiploid cells appears implausible given the unsuitable study design and the low number of analysed cells.
Speer et al., 2017: Cobalt (II) oxide and cobalt chloride hexahydrate are considered to induce chromosomal damage. Authors hypothesise that the effect is caused by the Co2+ ion, since proportion of metaphases with chromosomal damage and total number of aberrations in 100 metaphases showed similar responses at similar intracellular Co2+ concentrations. The publication shows major reporting and experimental deficiencies, such as purity and source of test material not specified, pH value, osmolality, and precipitation of test item-treated cell culture not reported. Authors did not account for any particle effect caused by poorly soluble cobalt (II) oxide particles. These can elicit secondary effects confounding the finding of the experiment, thus it is not discriminable whether positive findings are due to a direct effect or only due to a secondary particle effect. Positive controls were either not reported or are missing. The authors used the cell line hTU1-38, which is a subclone with a p-arm deficient chromosome 9. Potentially confounding effects, due to this deficiency, could not be excluded based on the data available. Basic parameters on the cell line, such as doubling time and culture conditions are missing. HCD is missing and thus labs proficiency in CA analysis is not proven.The experimental procedure is described in sufficient detail.
Chromosomal aberrations (CA) in mammalian cells, unpublished study reports:
A recent GLP CA study has been conducted by Haddouk (2007). Peconal H (cobalt acetyl acetonate) was tested for CA induction in cultures of human lymphocytes that had been stimulated to divide by treatment with phytohaemagglutinin for 48 hr before start of treatment. Two independent experiments were performed, and in each case cells were treated with cobalt acetyl acetonate for 3 hr in the absence or presence of S9, and harvested 20 hr after the start of treatment.
A summary of the key results is given in the table below. Although some statistically significant increases in % cells with CA were seen at low and middle concentrations, those below 5% would probably not be considered biologically significant, although they do exceed the historical control ranges of maximum 2.5 and 3% in the absence and presence of S9 respectively. At first sight it appears that biologically significant induction of CA is only seen at concentrations of cobalt acetyl acetonate inducing >50% toxicity. The second experiment in the absence of S9 gave a most unusual toxicity response (not dose-related) and therefore the value of 18% mitotic inhibition at 150 µg/ml is probably not reliable. However, biologically significant increases in CA were clearly seen at modest levels of mitotic inhibition in the second experiment in the presence of S9. Therefore it does not seem likely that the increased levels of CA are an indirect result of excessive toxicity.
Given that:
- There is a consistent pattern of increased CA frequencies in both the presence and absence of S9
- The effects are reproducible across independent experiments
- The increased CA are not exclusively associated with high (>50%) levels of cytotoxicity
it has to be concluded that this study indicates a biologically meaningful clastogenic response for cobalt acetyl acetonate.
Table. Chromosomal aberration test with cobalt acetyl acetonate, key results
Expt. |
S9 |
Treat + Recovery (hr) |
Conc. of |
% cells with CA (excl. gaps) |
% toxicity (reduction in mitotic index) |
||||||
1 |
- |
3 + 17 |
0 |
0 |
0 |
||||||
1 |
- |
3 + 17 |
34 |
3.5 |
31 |
||||||
1 |
- |
3 + 17 |
68 |
4.7 |
47 |
||||||
1 |
- |
3 + 17 |
136 |
18.0 |
56 |
||||||
2 |
- |
3 + 17 |
0 |
0 |
0 |
||||||
2 |
- |
3 + 17 |
75 |
3.5 |
0 |
||||||
2 |
- |
3 + 17 |
100 |
3.5 |
43 |
||||||
2 |
- |
3 + 17 |
150 |
12.0 |
18 |
||||||
1 |
+ |
3 + 17 |
0 |
0.5 |
0 |
||||||
1 |
+ |
3 + 17 |
17 |
4.5 |
16 |
||||||
1 |
+ |
3 + 17 |
34 |
4.5 |
31 |
||||||
1 |
+ |
3 + 17 |
68 |
32.0 |
68 |
||||||
2 |
+ |
3 + 17 |
0 |
0.5 |
0 |
||||||
2 |
+ |
3 + 17 |
50 |
8.0 |
35 |
||||||
2 |
+ |
3 + 17 |
75 |
12.0 |
46 |
||||||
2 |
+ |
3 + 17 |
100 |
26.0 |
64 |
Bold figures indicate statistical significance. Shaded boxes indicate clear biological significance.
Another GLP CA study has recently been conducted (Sire, 2007) on Produit Y (cobalt resinate). As before, cultures of human lymphocytes were stimulated to divide by treatment with phytohaemagglutinin for 48 hr before start of treatment. Three independent experiments were performed. In the first experiment cells were treated with cobalt resinate for 3 hr in the absence or presence of S9, and harvested 20 hr after the start of treatment. In the second experiment, cells were again treated for 3 hr in the presence of S9 and harvested at both 20 and 44 hr, but in the absence of S9 treatment was continuous for 20 or 44 hr until harvest. In the third experiments the 3 hr treatments in the presence of S9 with 20 hr harvest were again repeated.
A summary of the key results is given in the table below. The data from this study are not easy to interpret, possibly because of the impact of treating suspension cultures with precipitating concentrations. The following inconsistencies arise:
- In experiment 1 in the absence of S9 it is not clear why cytotoxicity increases through the precipitating range.
- In experiment 2 with 20 hr treatment in the absence of S9 there is less (in fact no) toxicity at the same concentrations that induced increasing toxicity after only 3 hr treatment. On the other hand, 44 hr treatment in the absence of S9 is much more toxic.
- In the 3 hr treatment in the presence of S9 in experiment 1 the toxicity plateaus across the precipitating range (which would be expected), yet in experiment 2 it neither plateaus nor shows a dose response, and in experiment 3 it shows some evidence of a dose response. On the other hand, after 41 hr recovery, significant toxicity was induced.
Some of the inconsistencies in toxicity probably reflect the difficulties in removing precipitate from suspension cultures following 3 hr treatments. However, the inconsistencies following continuous treatments are not readily explained.
Table. Chromosomal aberration test with cobalt resinate, key results
Expt. |
S9 |
Treat + Recovery (hr) |
Conc. of cobalt resinate (µg/ml) |
% cells with CA (excl. gaps) |
% toxicity (reduction in mitotic index) |
1 |
- |
3 + 17 |
0 |
1.0 |
0 |
1 |
- |
3 + 17 |
75P |
4.5 |
39 |
1 |
- |
3 + 17 |
150P |
2.0 |
49 |
1 |
- |
3 + 17 |
300P |
2.0 |
61 |
2 |
- |
20 + 0 |
0 |
0.5 |
0 |
2 |
- |
20 + 0 |
75P |
0.5 |
0 |
2 |
- |
20 + 0 |
150P |
2.0 |
0 |
2 |
- |
20 + 0 |
300P |
1.5 |
0 |
2 |
- |
44 + 0 |
0 |
1.5 |
0 |
2 |
- |
44 + 0 |
18.8 |
2.0 |
61 |
1 |
+ |
3 + 17 |
0 |
1.5 |
0 |
1 |
+ |
3 + 17 |
75P |
6.5 |
44 |
1 |
+ |
3 + 17 |
150P |
18.0 |
44 |
1 |
+ |
3 + 17 |
300P |
16.0 |
49 |
2 |
+ |
3 + 17 |
0 |
0 |
0 |
2 |
+ |
3 + 17 |
75P |
0.5 |
15 |
2 |
+ |
3 + 17 |
150P |
1.5 |
21 |
2 |
+ |
3 + 17 |
300P |
2.5 |
0 |
2 |
+ |
3 + 41 |
0 |
0.5 |
0 |
2 |
+ |
3 + 41 |
75P |
3.0 |
59 |
3 |
+ |
3 + 17 |
0 |
0.5 |
0 |
3 |
+ |
3 + 17 |
37.5 |
4.0 |
47 |
3 |
+ |
3 + 17 |
75P |
7.3 |
34 |
3 |
+ |
3 + 17 |
150P |
8.7 |
69 |
Bold figures indicate statistical significance. Shaded boxes indicate clear biological significance.
P= precipitate
The small increase in CA frequency (4.5%) at the lowest concentration in experiment 1 in the absence of S9 is probably due to chance and not biologically significant. The increases in CA in experiment 1 in the presence of S9 were not reproduced in experiment 2 (but then neither was the toxicity), however they were reproduced in experiment 3 where toxicity similar to that seen in experiment 1 was induced. Overall it can be concluded that cobalt resinate does induce CA in the presence of S9, although the impact of cytotoxicity on these responses is unclear because of the confounding factor of precipitation.
In vitro micronucleus (MN) tests:
Van Goethem et al (1997) tested ultrafine cobalt metal for induction of MN in human leukocytes using the cytokinesis block method. Cells were treated for 15 minutes in the absence of metabolic activation (which would not in any case be relevant for testing of pure metal) starting 24 hr after mitogen stimulation, at which point leukocytes would be mainly in the G1 phase of the cell cycle. Cells were harvested for MN after a total of 72 hr in culture. 200 binucleate cells per concentration were scored for MN. Although the treatment time was short, and cells were not in exponential growth at the time of treatment, this is a reasonably robust study by current standards. There was significant induction (>2-fold) in MN frequency at concentrations from 0.6 µg/ml and higher. Only 19% toxicity was induced at the lowest concentration.
De Boeck et al (2003) also tested cobalt metal, suspended in sterile water, for induction of MN in peripheral blood mononuclear cells from 2 different donors. Three different concentrations of cobalt, namely 0.6, 3.0 and 6.0 µg/ml, were tested. The treatment and sampling protocol was the same as that used by Van Goethem et al (1997). However, 1000 binucleate cells per culture were examined for presence of MN. Statistically significant and dose-related increases in binucleate cells with MN were seen in several different experiments in cultures from both blood donors. Significant increases in MN frequency were seen even at the lowest concentration of cobalt (0.6 µg/ml) which generally induced <40% reduction in relative division index. Thus the findings of Van Goethem et al (1997) were confirmed in this study.
Olivero et al (1995) studied induction of MN in human lymphocytes treated with cobalt chloride, sulfate and nitrate. Treatment (only in the absence of metabolic activation) was begun 24 hr after PHA stimulation and continued for 48 hrs. Cytochalasin B (to produce binucleate cells for scoring) was included for the final 28 hrs. Curiously, toxicity was measured by mitotic index (presumably from the CA treatments with the same concentrations) and not by replication index (RI) or cytokinesis-block proliferation index (CBPI), which are more usual. 1000 binucleate cells per culture were scored. As with the CA study by these authors, the control MN frequencies (5-10%) are quite high. There was a >2-fold increase in MN frequency at all concentrations of cobalt chloride, but seemingly no effect with sulfate or nitrate. The MN response with cobalt chloride was flat (a plateau), which suggests saturation of effect, and yet toxicity (mitotic inhibition) increased from zero at 0.0045 µg/ml to 72% at 0.45 µg/ml, so this would suggest the biological effects were not saturated across this range. This study is difficult to evaluate, and is therefore of limited relevance for risk assessment purposes.
Gibson et al (1997) studied MN induction in Syrian hamster embryo (SHE) cells. Exponentially growing cells were treated with cobalt sulfate for 24 hr in the presence of cytochalasin B (in the absence of metabolic activation, although early-passage SHE cells have considerable metabolic activity). Toxicity was determined both by trypan blue exclusion and percentage of binucleate cells. At the highest concentration (4 µg/ml) toxicity was <50%. Where possible, 1000 binucleate cells were scored for MN. The frequencies of MN were increased in a dose related manner, were significantly different from controls at all 3 concentrations (1, 2 and 4 µg/l) and reached 2.5x control levels at the top concentration. This frequency of MN was similar to that induced by the positive control chemicals (5 µg/ml cyclophosphamide and 0.25 µg/ml colchicines). This seems a robust positive response, and is consistent with findings in the studies discussed above.
A paper by Miller et al (2001) studied cell transformation induced by heavy metal alloys in human osteosarcoma cells. However, ultrafine metallic cobalt was also tested alone. Supposedly measures of chromosome damage (MN) were included in this study, however the Materials and Methods are confused, partly describing the technique for measuring SCE, but also stating that cytochalasin B was included, which would be involved in measurement of MN. Treatment was for 1 hr. No concurrent measure of cytotoxicity was included in the MN assay, but 6 µg/ml cobalt only induced 40% reduction in cell survival after a 24 hr treatment, so presumably the 1 hr treatments were even less toxic. Only 100 cells/concentration were scored for MN, which is low. However, cobalt induced significant (3-6-fold or greater) increases in MN frequency at all 4 concentrations scored (0.75, 1, 3 and 6 µg/ml). Considering the questionable methods used, this reference is considered of limited relevance for risk assessment purposes.
There are a number of papers indicating induction of MN by soluble cobalt salts and ultrafine metal cobalt. These effects do not appear to be artefacts of excessive toxicity. The consistency of effects across several publications indicates a potential to induce genotoxic effects in mammalian cells. Although none of these studies investigated whether the MN were due to a chromosome breakage (clastogenic) or chromosome loss/gain (aneuploidy) mechanism, the consistency with the unpublished chromosome aberration studies would suggest a clastogenic mode of action. This could be due to a reactive oxygen mechanism.
Uboldi et al. (2016) investigated effects of tricobalt tetraoxide and cobalt chloride on chromosomal damage using transformed human bronchial epithelial cell line (BEAS-2B): The authors concluded that tricobalt tetraoxide and cobalt chloride induce MN formation. However, validity of the experiments is not given, since the true negative control (LB-3 polystyrene beads) showed a significant induction of MN, when compared to untreated results. Moreover, information on pH value, osmolality, and precipitation are not provided. These factors are known variables that may lead to false positive findings. Especially for poorly soluble tricobalt tetraoxide information on precipitation are necessary for demonstration of reliability. Details on cell culture, MN scoring and evaluation are missing.
Gene mutations in mammalian cells, published references:
Amacher and Paillet (1980) tested cobalt chloride for induction of tk mutations in mouse lymphoma cells. Treatment was only for 3 hrs in the absence of metabolic activation. Concentrations ranged from 5.69-57.11 µg/ml, but there is no indication what levels of toxicity (if any) these treatments induced. There were no increases in mutant frequency as a result of these treatments. However, at the time this study was conducted there was no optimisation of the assay to detect small colony mutants (which would be expected to be representative of clastogenic or aneugenic modes of action), and no continuous 24 hr treatment in the absence of metabolic activation, as is now routine.
In 1990, Hartwig et al reported that cobalt chloride induced hprt mutations (4.2-fold increase) in V79 cells after 24 hr treatment at a concentration of 100 µM (23.8 µg/ml). This concentration induced 56% reduction in colony forming ability, and so is not excessively toxic for this assay. In 1991, the same group (Hartwig et al, 1991) made further studies (in the presence and absence of UV light) and appeared to confirm these findings.
Miyaki et al (1979) also studied induction of hprt mutations in V79 cells treated with cobalt chloride. Treatment was for 20 hr in the absence of metabolic activation. Only one concentration of cobalt chloride was tested (0.2 mM) and the nature of the salt was not given. This concentration induced a significant amount of toxicity (almost 90%) and induced 2.3-fold increase in hprt mutant frequency. Since only one concentration was tested which also induced excessive cytotoxicity, this result is of little relevance for human health risk assessment purposes.
Several studies on in vitro DNA damage studies in mammalian cells (involving comet assays) were identified which do not fulfil the relevance, reliability and adequacy criteria as foreseen by the ECHA Guidance on information requirements. The comet assay is a powerful tool to detect even low levels of DNA damage. However, this assay is also prone to positive findings not caused by the test item but by e. g. inappropriate test design. Minimum quality criteria were therefore used to rate these comet studies for their relevance and adequacy. The criteria as published by Tice et al. (2000) were used for this screening. In case a reference did not fulfil the criteria stated therein, it was rated as “not rateable” and not considered further.
Caicedo, M.C. et al. (2008): Limitations in experimental performance: identical handling of samples not sufficiently described, tail moment not given, positive control not included, information on the correlation of apoptosis and necrosis towards DNA damage detected in COMET not given.
Duerksen-Hughes et al. (1999): Study reports increased p53 levels upon cobalt exposure. Increase of p53 levels is not a specific indicator for a mutagenic mode of action, hence is not relevant for the assessment of substance induced mutagenicity.
Painter, R.B. and Howard, R. (1982): Study investigates altered DNA repair activity in mammalian cells upon cobalt exposure. The DNA repair assay does not detect chemical induced DNA mutation, hence is not relevant for the assessment of substance induced mutagenicity.
Robison, S.H. et al. (1984): Study investigates substance induced DNA lesions via gradient sedimentation. The reference exhibits several reporting deficiencies: a correlation between DNA damage and cytotoxicity was not performed, hence it remains unclear whether any positive findings were secondary to toxicity or due to direct substance interaction. Further, the amount of cells counted for DNA repair was not stated, only one concentration used which does not allow a dose-response analysis.
Wedrychowski, A. et al. (1986): Study investigates on protein-DNA complex formation for various metal substances and compared their molecular weights. Whereas such effect might impair cell survival, it does not report on permanent potentially heritable DNA mutation. Further, effects were not correlated with cytotoxicity, only one concentration used which does not allow a dose-response analysis, no information on positive or negative control cultures given.
Uboldi, C. et al. (2016): Study evaluated DNA damage after treatment with an insoluble and a soluble cobalt compound using the alkaline comet assay. Reporting and methodological deficiencies: Number of analysed cells is not specified, cytotoxicity and apoptosis rate are partially missing, evaluation criteria are missing, details on staining procedure are missing.
Uboldi, C. et al (2016): Thegamma-H2Ax assay was performed in order to investigate DNA lesions induced by treatment with tricobalt tetraoxide and cobalt chloride. In a second experiment, the cells were pre-treated with a ROS scavenger in order to assess the proportion of DNA damage induced by oxidative stress. There is no validated guideline for the gamma-H2Ax assay. Moreover, the publication shows several reporting deficiencies.
Liu, Deng, Yang (2016): Primary DNA damage was assessed after cobalt chloride hexahydrate treatment using the alkaline comet assay. The publication shows major reporting was well as methodological deficiencies. Purity of the test item is not specified. Only qualitative description of the comets found, quantitative results are not specified. Osmolality and pH of test item-treated cell culture medium are not specified. Type of negative control and metabolic activation are not specified. Scoring and evaluation criteria as well as HCD are not specified. Only two test item concentrations were used.
Kopp, B. et al. (2018): The effect of tricobalt tetraoxide and cobalt chloride on DNA lesions was investigated performing the gamma-H2Ax assay. The authors concluded both cobalt substances to induce no DNA damage in human HepG2 and LS-174T cells. However, the negative outcome is due to major reporting deficiencies questionable. Changes in gamma-H2Ax levels are only shown as fold induction, data for negative controls, however, are not shown. Positive control was run concurrently but the results are not specified. HCD are completely missing. Information on pH value, osmolality, and precipitation are missing. Justification for top dose used is not specified. Especially, in the case of tricobalt tetraoxide, which showed no cytotoxicity under the conditions tested and for which statements on precipitation are missing. Number of treated and scored cells is not stated. Method description lacks details. Details on the cell lines used are missing. Moreover,there is no validated guideline for the gamma-H2Ax assay.
Several studies were identified which do not fulfil the relevance, reliability and adequacy criteria as foreseen by the ECHA Guidance on information requirements. DNA damage in bacteria, induction of SCE in mammalian cells, or tests in yeasts or drosophila are no longer recommended as part of regulatory testing by many agencies worldwide and there are no up-to-date OECD guidelines for their conduct. All references in the following summary entries are more than 30 years of age and do not comply with today’s standards in genetic toxicity testing. Interpretation of the relevance of both positive and negative results from such tests is therefore unclear and was not used for the current assessment.
The studies are discussed below in brief for information purposes only and were included in the SIDS (IUCLID) as summary entry.
Summary entry on the assessment of DNA damage in bacteria: Nishioka, H. (1975), Kanematsu, N. et al., (1980), Kada, T. et al., (1980), Kumar, V., et al. (2017)
Summary entry on tests in yeast: Fukunaga, M. et al. (1982), Singh, I. (1983), Kharab, P. and Singh, I. (1985 and 1987), Prazmo, W. et al. (1975), Egilsson, V. et al. (1979), Lindegren, C.C. et al. (1958), Putrament, A. et al. (1977)
Summary entry on other unsuitable test systems, such as SOS chromotest, direct interaction of cobalt with isolated DNA, cytotoxicity assays with no direct correlation of a mutagenic mode of action, cytotoxic/mutagenic activity of cobalt towards pre-damaged cells: Olivier, Ph. and Marzin, D. (1987), Ogawa, H. I. et al.(1999), Nackerdien, Z. et al. (1991), Kada, T. and Kanematsu, N. (1978), Rossman, T.G. et al.(1984), Snyder, R.D, et al. (1989)
Gene mutations in mammalian cells, unpublished study reports:
Cobalt resinate has been tested for induction of tk mutations in mouse lymphoma L5178Y cells. Cobalt resinate dissolved in tetrahydrofuran, was tested using the microwell method (Sarlang, 2010). Treatments were for 3 hr and 24 hrs in the absence of S9, and for 3 hrs (2 experiments) in the presence of S9. The salt precipitated in culture medium and persisted to the end of treatment at concentrations of 25 µg/ml and higher for the 3 hr treatments and at 50 µg/ml and higher for the 24 hr treatment. However, 100 µg/ml was chosen as the top concentration in all of the main mutation experiments in order to try to achieve acceptable levels of toxicity. In the absence of S9 the top concentration induced approximately 50% toxicity (reduction in relative total growth, RTG) in both 3 and 24 hr treatments. In the presence of S9 the top concentration was less toxic and induced only 15-25% reduction in RTG. Although target levels of toxicity (>80%) were not reached, the inclusion of several precipitating concentrations constitutes a valid assay. Mutant frequencies in all treated cultures were very similar to concurrent controls, and there were no increases that approached a biologically significant level, i. e. did not approach the Global Evaluation Factor (GEF) of +126 mutants per 10^6 cells as recommended by Moore et al (2006). Therefore cobalt resinate did not induce tk mutations in mouse lymphoma cells in a robust test up to precipitating concentrations.
Eleven different cobalt substances have been tested for induction of hprt mutations in mouse lymphoma L5178Y cells. In each case the salts were tested for 3 hr in the absence and presence of S9. Usually 2 independent experiments were performed, but occasionally additional experiments were performed for clarification of results. A continuous 24h treatment in the absence of S9 was performed with cobalt sulfate, cobalt sulfide, cobalt monoxide and cobalt borate neodecanoate.
The results for the 7 cobalt substances tested in the absence and presence of S9 with a 3h exposure duration are summarised as follows:
Cobalt dihydroxide, suspended in 0.5% methyl cellulose, precipitated in culture medium and persisted to the end of treatment at concentrations of 116.1 µg/ml and higher (Stone, 2011). However, it was toxic at much lower concentrations, inducing >80% reduction in relative survival in the range 19-35 µg/ml, and therefore all concentrations evaluated for mutations were soluble. It did not induce any statistically or biologically significant increases in hprt mutant frequency when tested up to toxic concentrations in the absence of S-9. Statistically significant increases in mutant frequency were seen at 1 or 2 concentrations in the presence of S-9 in two out of three experiments, and did exceed the historical control range (calculated as mean ± 2 standard deviations), although the mutant frequencies were close to those observed in control cultures elsewhere in this series of studies. However, the increases were not dose-related and not reproduced at similar concentrations and similar levels of toxicity in all experiments. They are therefore of questionable biological relevance. Overall it is concluded that cobalt dihydroxide did not induce biologically relevant hprt mutations when tested up to toxic concentrations in the absence or presence of S9.
Tricobalt tetraoxide, suspended in DMSO, precipitated in culture medium and persisted to the end of treatment at various different concentrations in the absence and presence of S9 in the different experiments (Lloyd, 2010). However, the top concentration in each experiment was observed to be the lowest producing persistent precipitate through the treatment period. The top concentrations were 2000 and 2408 µg/ml in the absence of S9, and 600 and 750 µg/ml in the presence of S9. None of these induced significant toxicity (maximum 34% reduction in relative survival). Although hprt mutant frequencies exceeded the historical control range in some of the treated cultures, they did not exceed control mutant frequencies seen elsewhere in this series of studies, and none were statistically significant when compared to concurrent controls. It is therefore concluded that tricobalt tetraoxide did not induce hprt mutations when tested to precipitating concentrations in the absence or presence of S9.
Lithium cobalt dioxide, suspended in 0.5% methyl cellulose, precipitated in culture medium and persisted to the end of treatment at concentrations of 50-60 µg/ml and higher (Lloyd, 2010). Curiously, concentrations up to 600 µg/ml were tested in the presence of S9 in the first experiment, but although no precipitate was visible at lower concentrations the cell pellets obtained when treatment medium was removed after 3 hrs were discoloured. Therefore the solubility limit was almost certainly exceeded in this experiment. None of the treatments induced significant toxicity (maximum 22% reduction in relative survival). Although hprt mutant frequencies exceeded the historical control range in some of the treated cultures, none were statistically significant when compared to concurrent controls. It is therefore concluded that lithium cobalt dioxide did not induce hprt mutations when tested to toxic concentrations in the absence or presence of S9.
Cobalt 2-ethyl hexanoate, dissolved in DMSO, precipitated in culture medium and persisted to the end of treatment at concentrations of 1000 µg/ml and higher (Lloyd, 2010). However, it was toxic at much lower concentrations, inducing >80% reduction in relative survival in the range 80-120 µg/ml, and therefore all concentrations evaluated for mutations were soluble. Although hprt mutant frequencies exceeded the historical control range in some of the treated cultures, they did not exceed control mutant frequencies seen elsewhere in this series of studies, and none were statistically significant when compared to concurrent controls. It is therefore concluded that cobalt 2-ethyl hexanoate did not induce hprt mutations when tested to toxic concentrations in the absence or presence of S9.
Cobalt oxide hydroxide, suspended in 0.5% methyl cellulose, precipitated in culture medium and persisted to the end of treatment at various different concentrations in the absence and presence of S9 in the different experiments (Lloyd, 2010). However, the top concentration in each experiment was observed to be the lowest producing persistent precipitate through the treatment period. The top concentrations were 15 and 12.5 µg/ml in both the absence and presence of S9 in the 2 main experiments respectively. Neither of these induced significant toxicity (maximum 18% reduction in relative survival). Mutant frequencies in all treated cultures were similar to those in controls, not statistically different, and did not exceed historical control ranges. It is therefore concluded that cobalt oxide hydroxide did not induce hprt mutations when tested to precipitating concentrations in the absence or presence of S9.
Cobalt oxalate, suspended in 0.5% methyl cellulose, precipitated in culture medium and persisted to the end of treatment at concentrations of 70 µg/ml and higher (Stone, 2011). However, it was toxic at this and lower concentrations, inducing >80% reduction in relative survival at 55 µg/ml in the absence of S9, and at 60 or 70 µg/ml in the presence of S9. Therefore most concentrations evaluated for mutations were soluble. It did not induce any statistically or biologically significant increases in hprt mutant frequency when tested up to toxic concentrations in the absence of S-9. A statistically significant increase in mutant frequency was seen at 1 intermediate concentration in the second experiment in the presence of S-9, and did exceed the historical control range (calculated as mean ± 2 standard deviations), although it was close to control mutant frequencies observed elsewhere in this series of studies. Moreover, the isolated increase in mutant frequency was not seen at a similar concentration and similar level of toxicity in the first experiment, and there was no dose-response. The increase in mutant frequency is therefore of questionable biological relevance. Overall it is concluded that cobalt oxalate did not induce biologically relevant hprt mutations when tested up to toxic concentrations in the absence or presence of S9.
Cobalt metal powder, suspended in 0.5% methyl cellulose, clearly precipitated in culture medium and persisted to the end of treatment at concentrations of 589.3 µg/ml and higher (Stone, 2011). However, undissolved cobalt metal powder was present at many lower concentrations. It was toxic at lower concentrations, inducing >80% reduction in relative survival at 37.5-40 µg/ml in the absence of S9, and at 40-200 µg/ml in the presence of S9. The marked variability in the toxic responses in the presence of S9 between the different experiments may be a reflection of the fact that undissolved cobalt metal powder was present, and may have exerted toxic effects on the cells due to its physical presence. These factors contribute to the problematic interpretation of the data. It did not induce any statistically or biologically significant increases in hprt mutant frequency when tested up to toxic concentrations in the absence of S-9, although a weak dose-response was seen in the second experiment. Statistically significant increases in mutant frequency were seen at single intermediate concentrations in each of the experiments in the presence of S-9, and a dose-response was seen in the second experiment. Although these isolated increases in mutant frequency did exceed the historical control range (calculated as mean ± 2 standard deviations), they did not exceed control mutant frequencies observed elsewhere in this series of studies. However, a third experiment in the presence of S9 was performed in order to investigate these findings further. In this experiment, significant increases in mutant frequency were seen at 30 and 40 µg/ml, accompanied by a significant dose-response. The mutant frequencies at these concentrations exceeded both historical control ranges and concurrent control mutant frequencies seen elsewhere in this series of experiments. Overall it is concluded that cobalt metal powder did not induce biologically relevant hprt mutations when tested up to toxic concentrations in the absence of S9, but did induce weak, reproducible mutagenic effects in the presence of S9. However, due to the formulation tested, the biological relevance of these results is questionable.
The results for the 4 substances tested both for 3 hrs in the absence and presence of S9 and for 24 hrs in the absence of S9 are summarised as follows:
Cobalt sulfate, dissolved in water, precipitated in culture medium and persisted to the end of treatment at concentrations of 2809 µg/ml and higher (Stone, 2011). However, it was toxic at lower concentrations, inducing >80% reduction in relative survival at 50-100 µg/ml for the 3 hr treatments, and at 35 µg/ml for the 24 hr treatment. Therefore all concentrations evaluated for mutations were soluble. It did not induce any statistically or biologically significant increases in hprt mutant frequency when tested up to toxic concentrations in the absence of S-9 using either 3 or 24 hr treatments. Statistically significant increases in mutant frequency was seen at the 2 highest concentrations in the second experiment in the presence of S-9 leading to a dose-response, but the concurrent control mutant frequency was unusually low, and the mutant frequencies in the treated cultures did not exceed the historical control range (calculated as mean ± 2 standard deviations). Moreover, these increases in mutant frequency were not seen at even higher concentrations and similar levels of toxicity in the first experiment. The increases in mutant frequency are therefore of questionable biological relevance. Overall it is concluded that cobalt sulfate did not induce biologically relevant hprt mutations when tested up to toxic concentrations in the absence or presence of S9, even when extended treatment periods were used in the absence of S9.
Cobalt sulfide, suspended in 1.0% carboxymethyl cellulose, precipitated at several concentrations in culture medium at the beginning of treatment but had dissolved by the end of treatment (Stone, 2011). It was therefore possible for the 3 hr treatments to test it up to 10 mM (922 µg/ml), which was not toxic, and is the maximum required for testing of non–toxic chemicals. Cobalt sulfide exhibited some toxicity following 24 hr treatments, and so the maximum concentration (that reduced relative survival to 11% of controls) was 800 µg/ml (Lloyd, 2012). Although hprt mutant frequencies exceeded the historical control range in some of the treated cultures, they did not exceed control mutant frequencies seen elsewhere in this series of studies, and none of the mutant frequencies were statistically significant when compared to concurrent controls. There were also no dose-responses. It is therefore concluded that cobalt sulfide did not induce hprt mutations when tested either to 10 mM or toxic concentrations in the absence or presence of S9.
Cobalt monoxide, suspended in 0.5% methyl cellulose, precipitated in culture medium and persisted to the end of treatment at concentrations of 60-80 µg/ml and higher (Stone, 2010). Following 3 hr treatments, significant (86%) toxicity was induced at 80 µg/ml in the absence of S9 in the first experiment, but at 60 µg/ml (the top concentration for the second experiment in the absence of S9 and both experiments in the presence of S9), only slight to moderate (15-41%) toxicity was induced. It was also toxic following 24 hr treatments in the absence of S9, and the top concentration tested for mutation induction was 60 µg/ml, which induced 90% reduction in relative survival (Lloyd, 2012). Although hprt mutant frequencies exceeded the historical control range in some of the cultures treated in the presence of S9, they did not exceed control mutant frequencies seen elsewhere in this series of studies. A 3-fold increase in mutant frequency was seen at one of the intermediate concentrations in the 24 hr treatment protocol. However, it was not significantly different from the concurrent control and there was no dose response. In fact none of the mutant frequencies in any part of the study, either in the absence of presence of S9, were statistically significant when compared to concurrent controls. It is therefore concluded that cobalt monoxide did not induce hprt mutations when tested to precipitating concentrations in the absence or presence of S9.
Cobalt borate neodecanoate, dissolved in tetrahydrofuran, precipitated in culture medium and persisted to the end of treatment at concentrations of 100 µg/ml and higher (Stone, 2011). However, it was toxic at lower concentrations, inducing >80% reduction in relative survival at 2.75-3.0 µg/ml in the absence of S9, and at 50-60 µg/ml in the presence of S9 following short (3 hr) treatments. It was also toxic following 24 hr treatments, and the maximum concentration used was 2.0 µg/ml, which induced 90% reduction in relative survival (Lloyd, 2012). Therefore all concentrations evaluated for mutations were soluble. A statistically significant increase in mutant frequency was seen at the highest concentration in the second experiment with 3 hr treatment in the absence of S-9, and did exceed the historical control range (calculated as mean ± 2 standard deviations), although it did not exceed control mutant frequencies observed elsewhere in this series of studies, and there was no dose-response. Moreover, the isolated increase in mutant frequency was not seen at similar or higher concentrations and similar levels of toxicity in the first experiment with 3 hrs treatment. Dose-responses were seen in the first experiment with 3 hr treatment in the absence of S9 and in both experiments with 3 hr treatment in the presence of S9, but these were strongly influenced by low negative control mutant frequencies, and, apart from the exception described earlier, none of the mutant frequencies in treated cultures were statistically significant when compared to concurrent controls. Following 24 hr treatment in the absence of S9 all mutant frequencies were similar to control and there were no significant differences or dose response. Overall it is concluded that cobalt borate neodecanoate did not induce biologically relevant hprt mutations when tested up to toxic concentrations in the absence or presence of S9.
The overwhelming conclusion is that cobalt salts do not induce biologically relevant mutagenic responses in mammalian cells.
in vivo somatic cell results
Micronucleus (MN) test, published reference:
In an abstract, Turocziet al (1987) reported that cobalt acetate did not induce MN in mouse bone marrow at an intraperitoneal dose of 33 mg/kg. No details of numbers of animals, sampling times or cells scored are given, and therefore this negative result will not be used in the cobalt hazard assessment.
At the end of a 13 week (5 days per week) inhalation study with 5 different doses of cobalt metal powder, NTP (NTP, 2013) measured MN in normochromatic erythrocytes (NCE) of mice. The top dose (10 mg/m³) was >3x higher than the top dose used in the inhalation carcinogenicity study with cobalt sulfate (see later). Two thousand NCE per animal (10,000 cells per sex per dose group) were examined for MN. There were no increases in MN frequency in any of the dose groups, males or females, and no evidence of bone marrow toxicity.
By contrast, a recent paper by Rasgele et al (2013) reported significant induction of MN in bone marrow of mice treated with cobalt chloride. Three different doses (11.2, 22.5 and 45 mg/kg) of cobalt chloride dissolved in distilled water were injected by the intraperitoneal route to groups of 5 mice (sex is not specified). The top dose is considered to be 50% of the LD50. Mice were sacrificed 24 and 48 hrs after injection. Bone marrow was sampled, slides made and stained with May-Grünwald-Giemsa, and 1000 polychromatic erythrocytes (PCE) per animal scored for MN. The MN frequencies are given as %. If that is correct and not per thousand, then the frequencies in the control animals (around 0.7%) are very high. However, the positive control chemical did induce clear and significant increases in MN frequency at both sampling times. There were statistically significant increases in MN frequency at the top 2 doses (2 to 2.5-fold increases) at 24 hrs, and at all 3 doses (3 to 4-fold increases, but not dose-related) at 48 hrs. There was no evidence of bone marrow toxicity (based on PCE:NCE ratio) at any dose level, although the ratio in control animals was higher than usually experienced. Thus, there appears to be a weak induction of MN by cobalt chloride in this study, but there are concerns that suggest the results should be viewed with caution. Firstly the intraperitoneal route in not considered a physiologically relevant route of exposure by many regulatory agencies. Secondly, doses as high as 50% of the LD50 may be considered unreasonably high (even though no animals died). Thirdly, the control MN frequencies appear to be unusually high. Consequently this study is considered of limited relevance for human health risk assessment purposes.
A peripheral blood micronucleus (MN) study has been performed with cobalt metal powder following inhalation exposure to B6C3F1 mice (NTP, 2013). Groups of 10 male and 10 female mice received the following doses: 0.625, 1.25, 2.5, 5.0 and 10.0 mg/m³ (nominal concentration) over a period of 13 weeks, 5 days/week. Blood samples were taken from mice at the study termination. 1,000 to 10,000 mature erythrocytes were scored per animal. A summary of the key results is given in the table below, detailed results can be found in the endpoint study record in section 7.6.2 of the IUCLID file.
Table. Micronucleus test in mice: Summary Data
Dose |
Animals |
Sex |
Percent NCE |
MN cells per 1000 |
Vehicle control |
5 |
Male |
97.44 ± 0.17 |
2.40 ± 0.33 |
Vehicle control |
5 |
Female |
97.38 ± 0.18 |
2.50 ± 0.35 |
0.625 mg/m³ |
5 |
Male |
96.96 ± 0.29 |
2.40 ± 0.33 |
0.625 mg/m³ |
5 |
Female |
97.36 ± 0.29 |
2.60 ± 0.29 |
1.25 mg/m³ |
5 |
Male |
97.36 ± 0.14 |
2.30 ± 0.37 |
1.25 mg/m³ |
5 |
Female |
97.46 ± 0.16 |
2.00 ± 0.22 |
2.5 mg/m³ |
5 |
Male |
97.30 ± 0.15 |
3.10 ± 0.19 |
2.5 mg/m³ |
5 |
Female |
97.68 ± 0.15 |
2.80 ± 0.30 |
5.0 mg/m³ |
5 |
Male |
97.66 ± 0.02 |
2.80 ± 0.34 |
5.0 mg/m³ |
5 |
Female |
97.70 ± 0.11 |
2.00 ± 0.32 |
10 mg/m³ |
5 |
Male |
97.36 ± 0.07 |
2.80 ± 0.37 |
10 mg/m³ |
5 |
Female |
97.78 ± 0.20 |
2.30 ± 0.34 |
Sample collection time: 24 hours
Route and dose regimen: inhalation, 13 weeks, 5 days/week
No significant increase of the micronucleus frequency was observed at any of the dose groups, males or females following 13-week inhalation exposure. The study is concluded as negative.
Micronucleus (MN) test, unpublished study reports:
In a GLP study, CD-1 mice were dosed orally on 2 consecutive days (24 hrs apart) with a series of chemicals that included cobalt 2-ethyl hexanoate (Richold et al, 1981). The chemical was mixed with corn oil to form a dosing preparation. The top dose was 2500 mg/kg/day which, based on preliminary toxicity studies, was expected to induce some lethality. Deaths did occur at this dose in the main study, indicating it exceeded the maximum tolerated dose (MTD), and signs of systemic exposure (hypopnoea, lethargy and piloerection) were seen at this and lower doses. Animals were sacrificed 24 hr after the second dose, bone marrow smears were made and stained. The numbers of polychromatic erythrocytes (PCE) scored for presence of MN was not given, but from the tables it appears that 1000 PCE/animal were scored, which is less than the currently recommended 2000 PCE/animal (OECD, 1997).
From the ratio of PCE to total erythrocytes, there was no clear evidence of bone marrow toxicity. MN frequencies in treated groups were similar to those in control groups, and there were no significant differences. No blood samples were taken for determination of test chemical in plasma, and as there was no bone marrow toxicity, there was no direct evidence of bone marrow exposure. However, the clinical signs indicate systemic exposure, and therefore it is anticipated that the bone marrow would have been exposed. Thus, cobalt 2-ethyl hexanoate did not induce MN in mouse bone marrow at lethal doses.
A GLP study (Sire, 2009) with cobalt acetyl acetonate has been conducted to measure MN induction in mouse bone marrow. Groups of at least 5 male and 5 female mice were dosed orally (by gavage) with 125, 250 or 500 mg/kg cobalt acetyl acetonate on 2 consecutive days, and bone marrow was sampled 24 hr after the second dose. The top dose of 500 mg/kg was chosen as a maximum tolerated dose as clinical signs of toxicity were observed at this dose, and higher doses caused mortality, although deaths occurred even at this dose in the main study. For MN analysis 2000 polychromatic erythrocytes (PCE) per animal were scored. Bone marrow toxicity was assessed by determining the ratio of PCE to normochromatic erythrocytes (NCE). This study design complies with current recommendations. In addition satellite animals were dosed at 500 mg/kg on 2 consecutive days and blood samples were taken 2 hr after the second dose for determination of cobalt acetyl acetonate in plasma. However, this was not possible due to “interferences with the biological matrix”.
There was no reduction in PCE: NCE ratio in cobalt acetyl acetonate treated animals and therefore no evidence of bone marrow toxicity. The frequencies of micronucleated PCE were similar to controls in all cobalt acetyl acetonate-treated groups, and therefore there was no evidence of a genotoxic effect.
A GLP study (Sire, 2007) with cobalt resinate has also been conducted recently to measure MN induction in mouse bone marrow. Groups of at least 5 male mice were dosed orally (by gavage) with 375, 750 or 1500 mg/kg cobalt resinate on 2 consecutive days, and groups of at least 5 female mice were dosed orally (by gavage) with 187.5, 375 or 750 mg/kg cobalt resinate on 2 consecutive days. Bone marrow was sampled 24 hr after the second dose. The top doses of 1500 and 750 mg/kg for males and females respectively were chosen as maximum tolerated doses as clinical signs of toxicity were observed at this dose, and higher doses caused mortality. One male dosed with 1500 mg/kg died in the main study. For MN analysis 2000 polychromatic erythrocytes (PCE) per animal were scored. Bone marrow toxicity was assessed by determining the ratio of PCE to normochromatic erythrocytes (NCE). This study design complies with current recommendations. In addition satellite animals were dosed at 1500 (males) or 750 mg/kg (females) on 2 consecutive days and blood samples were taken 2 hr after the second dose for determination of cobalt resinate in plasma. However, this analysis was not preformed due to evidence of bone marrow toxicity (see below).
There was a significant (at least 50%) reduction in PCE: NCE ratio in cobalt resinate-treated males and females in the top dose groups, and therefore clear evidence of bone marrow toxicity. The frequencies of micronucleated PCE were slightly higher than controls in all cobalt resinate-treated groups, but did not exceed historical control ranges and did not achieve statistical significance, and therefore there was no evidence of a genotoxic effect.
When an in vivo study is performed to follow up on positive results in vitro (as is the case here), a negative result in that in vivo study usually requires proof of exposure of the target tissue. The fact that there was clear evidence of bone marrow toxicity indicates that there was systemic, and therefore bone marrow, exposure. Thus the absence of plasma analysis of cobalt resinate is not critical.
Chromosome aberration (CA) test, published reference:
Palit et al (1991) dosed groups of mice orally with cobalt chloride at 20, 40 and 80 mg/kg, the top dose being 10% of a lethal dose. Bone marrow preparations were made 6, 12, 18 and 24 hr later. Fifty metaphases were scored from each animal (i.e. 250 metaphases per dose group per sampling time). The frequencies of CA in control animals were normal. Dose- and time-related increases in CA frequency were seen in all groups. This is most unusual, and raises suspicions. Different sampling times are included in this assay to allow for the different kinetics and metabolism of different chemicals, and well-studied in vivo genotoxins do not usually produce both dose- and time-related responses at all doses tested.
Farah (1983) studied the effects of cobalt chloride for induction of CA in bone marrow of male Syrian hamsters. Animals were dosed intraperitoneally on 5 consecutive days, although it appears that animals also received additional daily doses after 2 days of dosing. The total dose was given as 400 mg/kg, although it is not clear whether this was a total daily dose, or the total dose summed over 5 days. It is also not clear how this dose was chosen. Fifty cells per animal were scored for CA (i. e. a total of 510 cells from the treated group and 650 cells from the control group). Cells were not scored for structural aberrations but only for numerical aberrations. The frequency of hypodiploid cells was high in both groups, probably due to chromosome loss during swelling of the cells during preparation of metaphases. Hyperdiploidy was seen to increase from 0.5 to 3.1%. The frequency of hyperdiploid cells in control preparations seems high – normally a frequency of 0.1% or less would be expected, and thousands of cells per group would need to be scored in order to detect any increase. The frequency of pseudodiploid cells also increased, although there is not description as to what “pseudodiploid” means in this case. It is therefore not clear how reliable are these results. When cells are swollen to provide acceptable metaphase spreads, adjacent metaphases may be mis-scored, with one cell appearing to be hypodiploid while the other appears hyperdiploid. It may be important that no measure of bone marrow toxicity was included in this study.
Several studies on in vivo clastogenicity tests were identified which do not fulfil the relevance, reliability and adequacy criteria as foreseen by the ECHA Guidance on information requirements. These studies are discussed below in brief for information purposes only and were included in the SIDS (IUCLID) as summary entry.
Suzuki, Y. et al. (1993): The maximum dose used in the study is equal to the LD50 which is considered excessively high. The intraperitoneal injection is an unphysiological route of application, thus is of no relevance for the human health hazard assessment for industrial chemicals. The standard deviation for the mid dose and high dose group indicates a high variability in MPCE and P/N ratios by up to 16-fold, indicating a high intra-animal variability, which is usually only seen at incidences of excessive toxicity. Further the MPCS frequency of the control animals varied significantly between the various experiments by up to 6-fold, which raises questions whether the selected mouse strain is suitable for clastogenic experiments.
Nehéz, M. et al. (1982): Only a conference/poste abstract was available, which does not provide sufficient information for an assessment. Crucial information was missing such as: dose regime, route of exposure, frequency of treatment, description of cell preparation, evaluation criteria, toxicity.
Awoyemi et al. (2017): The publication shows major reporting and experimental deficiencies. The number of scored polychromatic erythrocytes is not stated. The doses actually received by the test animals is not specified, treatment was only specified as different CoCl2 solutions used. Proportion of immature vs total erythrocytes is not specified. The only measure for cytotoxicity was immunochemistry of liver BAX expression, used as marker for apoptosis. Positive controls and historical control data is missing. Details on test animals and animal housing are missing. Statements on scoring and evaluation criteria are missing.
The references contained in the summary entry for the in vivo test systems drosophila (Ogawa, H.I. et al, 1994; Yesilda, E., 2001) are of limited value for risk assessment purposes, since this test system is no longer recommended as part of regulatory testing by many agencies worldwide and there is no up-to-date OECD guideline for its conduct. Interpretation of the relevance of both positive and negative results from such tests is therefore unclear and was not used for the current assessment. The information contained therein was included for information purposes only.
Chromosome aberration (CA) test, unpublished study reports:
A bone marrow chromosomal aberration (CA) study has been performed with cobalt sulfate, cobalt monoxide and tricobalt tetraoxide following oral administration to rats (Legault, 2009). Both single and multiple (5x daily) dose schedules were used in this non-GLP experiment.
For the single dose study groups of 2 male and 2 female rats received the following doses:
Cobalt sulfate – 80, 160 and 320 mg/kg
Cobalt monoxide – 100, 300 and 1000 mg/kg
Tricobalt tetraoxide – 500, 1000 and 2000 mg/kg.
The top doses were estimated to be maximum tolerated doses based on preliminary toxicity studies. Animals from the low dose oxide and tetraoxide groups were not processed for CA. The group sizes for this phase of the study are much smaller than required by OECD guidelines.
Groups of 5 males + 5 females were used in the multi-dose phase, which complies with OECD recommendations, and should provide for a sensitive study. However, both cobalt sulfate and cobalt monoxide were toxic in the multi-dose study, and therefore animals from some of the dose groups with these compounds were sacrificed early and did not complete the 5-day schedule. Only the low dose sulfate group (100 mg/kg/day) survived until the end of the 5-day dosing, and no CA data could be obtained from the mid or high dose group animals. Also, no CA data were obtained from females in the high dose monoxide group. Thus, although the initial design of the study complied with OECD guidelines, deaths and early sacrifices compromised the compliance. Tricobalt tetraoxide was less toxic and animals survived dosing with 200, 600 or 2000 mg/kg/day for 5 days.
In order to accumulate bone marrow cells in metaphase, animals were given 4 µg/kg colchicine 1 hour before sacrifice, which was set at 16 hours after the last (or only) dose. Bone marrow smears were made and stained, and 100 cells/animal scored for CA.
By measurement of mitotic index, cobalt monoxide induced some toxicity in the bone marrow in the multi-dose phase. In one part of the report the author indicated some slight mitotic inhibition by cobalt sulfate in the multi-dose phase, and some slight mitotic stimulation by tricobalt tetraoxide in the single-dose phase, but these could not be substantiated from the tables of data. In fact the most significant toxicity finding was for the high dose of monoxide in the multi-dose phase where there was clear mitotic inhibition.
In most groups, CA frequencies were within normal ranges. However in the low dose sulfate (males and females) and low dose oxide (females) groups, CA frequencies in the 5-7% range were seen. These would likely be outside of historical control range, but were not different from the vehicle control values without colchicine. Increased CA frequencies at a low dose, not seen at mid and high doses, can be meaningful if there is severe bone marrow toxicity, such that cells are prevented from dividing because of the severe toxic effects and therefore damaged metaphases are not seen at the time of sampling. This cannot be the explanation for either the sulfate of oxide groups because there was no noticeable mitotic inhibition with these treatments in the single-dose phase. Thus, the increases in CA are not dose-related, and not explained by toxicity, and must therefore be deemed to be due to chance.
In the multi-dose phase, the CA frequencies in vehicle control animals were very low (0.2%). Thus, the finding of 1.8% cells with CA in the high dose sulfate and monoxide groups of males could be indicative of a clastogenic response. However, such levels are likely within the normal range (0-2% is normal), and similar frequencies were seen in control rats in the single-dose phase. Thus, it is considered that none of the cobalt salts induced CA in the multi-dose phase.
There do not appear to be any clastogenic responses from sulfate, monoxide or tetraoxide treatments.
Cobalt chloride hexahydrate was dissolved in water and administered once by oral gavage to groups of male and female Sprague-Dawley rats at doses up to 600 mg/kg (Gudi and Ritter, 1998). Some animals died at this dose, and also at the next lower dose (200 mg/kg), and so both were higher than the MTD. Clinical signs, including lethargy and piloerection, indicated systemic exposure. Animals were administered colchicine 2-4 hrs before sacrifice in order to arrest dividing cells of the bone marrow in metaphase. Rats from 7 dose groups were sacrificed 18 hrs after dosing, and animals from the top 3 dose groups were sacrificed 42 hrs after dosing. Bone marrow was aspirated from 1 femur per animal (the other femur was used for micronucleus analysis), cells swollen for 10 minutes in 0.075M KCl, slides made and stained with Giemsa. Where possible, 100 cells per animal were analysed microscopically for presence of CA. This study design complies with current recommendations (OECD, 1997).
There were some reductions in mitotic index in the bone marrow preparations of treated animals (up to 34% reduction compared to control), which may be taken as indicative of bone marrow toxicity. However, as described earlier, in the micronucleus part of this same study, severe reductions in the percentage of PCE gave clear indications of bone marrow toxicity. Frequencies of CA in treated groups were low and similar to control, and there were no significant increases. Thus, cobalt chloride hexahydrate did not induce CA in bone marrow of rats at lethal doses that induced bone marrow toxicity.
In vivo studies in germ cells
Farah (1983) studied the effects of cobalt chloride for induction of CA in metaphase I and II meiotic testicular cells of male Syrian hamsters. Animals were dosed intraperitoneally on 5 consecutive days, although it appears that animals also received additional daily doses after 2 days of dosing. The total dose was given as 400 mg/kg, although it is not clear whether this was a total daily dose, or the total dose summed over 5 days. It is also not clear how this dose was chosen. An increase in cells with at least 23 bivalents (instead of the normal 22) was seen in metaphase I preparations from the treated group. As there are no guidelines for this type of study it is difficult to assess whether the numbers of cells scored were appropriate or not.
This study exhibits serious shortcomings which renders it unsuitable for the human health risk assessment :
- the dosing of the animals is not properly described, so that it is unclear which doses the animals received
- only a single dose was administered, thus a dose response relationship cannot be established
- the intraperitoneal injection is an unphysiological route of application, thus is of no relevance for the human health hazard assessment for industrial chemicals
- the study design was non-GLP and not according to any technical guideline
- there is no known genotoxic agent which acts solely against germ cells without affecting somatic cells mutagens (Guidance on information requirements and Chemical Safety Assessment, Chapter R.7a: Endpoint specific guidance, Section R.7.7.1.2). In the light of conclusively negative findings in in vivo animal micronucleus and chromosome aberration assays as well as negative micronuclei findings in humans the findings described in this paper are not conclusive
In consequence, this study is considered unreliable and is not considered further in the chemical safety assessment of cobalt and cobalt substances.
In an in vivo spermatogonial chromosomal aberration (CA) test, no changes in behaviour, external appearance of the animals or the faeces were noted for control animals or those treated with cobalt chloride at any of the 3 dose levels (Leuschner, 2015). Although no obvious clinical signs were observed at the doses used, some small but statistically insignificant reductions in body weight were seen in animals treated at 30 mg/kg/day (by 5.0% at day 8 and by 14.1% at day 29). Reductions in absolute (25.0%) and relative liver weights (12.3%) were seen in the 30 mg/kg/day group, but these were also not statistically significant. Sinceit would not have been possible to use higher doses due to deaths at 100 and 300 mg/kg, the top dose used in this study (30 mg/kg/day) is considered an acceptable maximum tolerated dose.
Table: Chromosomal aberration frequencies in spermatogonia of rats treated with cobalt dichloride hexahydrate
Dose (mg/kg/day) |
MI |
No. of cells examined (M+F) |
Group mean % cells with CA (range) |
Group mean % cells with CA (range) |
% polyploidy cells |
|
|
|
Including gaps |
Excluding gapsa |
|
0 |
1.00 |
1000 |
2.8 (1.5-4.0) |
1.1 (1.0-1.5) |
0.0 |
3 |
1.48 |
1000 |
1.3 (0.5-3.0) |
0.7 (0.5-1.5) |
0.0 |
10 |
1.26 |
1000 |
2.2 (1.5-2.5) |
0.7 (0.5-1.0) |
0.0 |
30 |
1.11 |
1000 |
2.2 (1.5-2.5) |
0.9 (0.5-1.5) |
0.0 |
ahistorical control range = 0.7-1.5% for group mean CA, excluding gaps
The results of the CA analysis in spermatogonia are summarized in the Table above. It can be seen that there was no bone marrow toxicity as measured by mitotic index, and there were also no increases in the frequency of CA. In the cobalt chloride groups all group mean structural CA frequencies fell within the historical control range. Also, there were no polyploid cells found from 1000 metaphases scored in each of the groups. Although no concurrent positive control was employed, the testing laboratory has consistently observed significantly increased CA frequencies (in the range 9-13%) in animals treated with mitomycin C. Although plasma was not analysed for presence of cobalt chloride, Nationet al(1983) demonstrated significant exposure of multiple tissues after oral dosing of cobalt chloride to rats.
In vivo studies in humans
There are no available studies on genotoxic effects in humans exposed to cobalt by the oral and dermal routes of exposure. A cohort of 26 male workers who were occupationally exposed to cobalt, chromium, nickel, and iron exhibited increased sister chromatid exchange rank values (by analysis of variance) that were related to metal exposures and smoking habits (Gennart et al., 1993). However, due to the mixed exposure towards various metals, the effects could not be attributed to a single metal – study not regarded further.
De Boeck et al. (2000) performed a comet assay and micronuclei detection on lymphocytes from a group of 35 workers occupationally exposed to cobalt, inorganic cobalt substances or another group of 29 workers exposure to hard metal dusts. The exposure level was approx. 20 µg/m³, measured over a period of 5 weeks (exposure levels were estimated by urinary concentrations of 20µ Co/g creatinine). Blood samples were drawn at least 48 hours after the last exposure. The frequency of micronucleated binucleates (MNCB) was not statistically different between control (3.9‰ ± 1.7‰) and exposed workers (5.3‰ ± 1.9‰). The frequency of micronucleated mononucleates (MNMC) did not vary among the different worker groups (5.3‰ ±- 1.6‰ exposed group; 6.1‰ ± 1.5‰ control group). In all groups, the mean frequency of MNMC was always higher or equal to that of MNCB. When both control and exposed workers were considered together, smoking status significantly influenced the frequency of MNCB. The relative influence of smoking status categorised as smoking or non-smoking on the level of MNCB was tested by stepwise multivariate linear regression analysis. This survey did not detect significant increases of genotoxic effects in workers exposed to cobalt-containing dust at a mean level of 20 µg Co per gram of creatinine in urine equivalent to a TWA exposure of 20 µg/m³ Co.
Discussion
The positive Ames result of Pagano and Zeiger (1992) has not been reproduced. Moreover, although earlier publications (Hartwig et al, 1990; Miyaki et al, 1979) suggested soluble cobalt salts could induce gene mutation in mammalian cells, the recent GLP studies on hprt mutations have not confirmed this. Several of these studies have included what might be considered the most critical treatment regime, namely a 24 hr treatment in the absence of S9. Thus, there was no convincing or consistent evidence of induction of gene mutations in either bacteria or mammalian cells in vitro.
Clastogenicity (chromosome breakage) can often be caused by oxidative damage, or by indirect mechanisms such as excessive cytotoxicity, disruption of non-DNA targets etc. Such mechanisms would be expected to have a threshold. The clastogenic potential of cobalt salts in vitro, as seen in chromosomal aberration, micronucleus and tk mutation (small colony mutants) assays, has been satisfactorily addressed by negative in vivo bone marrow micronucleus and chromosomal aberration results with cobalt chloride, cobalt 2-ethyl hexanoate, cobalt acetyl acetonate and cobalt resinate. Further, a survey in workers occupationally exposed to cobalt, inorganic cobalt substances did not detect significant increases of genotoxic effects (micronuclei and DNA damage in peripheral blood) in workers exposed to cobalt-containing dust at a mean level of 20 µg Co/m³.
Based on the entire database of genetic toxicity studies and the review by the OECD and an external peer reviewer it is concluded that in summary, poorly soluble cobalt salts/compounds do not appear to be genotoxic in vitro or in vivo at all, soluble cobalt salts do not elicit any mutagenic activity either in bacterial or mammalian test systems. However they induce some genotoxic effects in vitro, mainly manifest as DNA strand or chromosome breaks, which are consistent with a reactive oxygen mechanism, as has been proposed by various authors. A weight-of-evidence approach was applied, considering positive as well as negative in vivo clastogenicity studies and the absence of such chromosome damage in humans that are occupationally exposed to inorganic cobalt substances. It was concluded that effective protective processes exist in vivo to prevent genetic toxicity with relevance for humans from the soluble cobalt salts category (OECD 2014, Kirkland et al. 2015).
References
OECD SIAP. (2014) SIDS Initial Assessment Profile (SIAP) for Soluble cobalt salts category. Summary of SIDS Submission to the Organisation for Economic Cooperation and Development (OECD) High Production Volume Chemicals Program. October, 2014. http://webnet.oecd.org/HPV/UI/handler.axd?id=b789fd1c-bab3-433c-9f47-3cbd49042976
Kirkland, D. et al. (2015) New investigations into the genotoxicity of cobalt compounds and their impact on overall assessment of genotoxic risk. Accepted for publication in Regulatory Toxicology and Pharmacology, 20 July 2015 (available online 22 July 2015http://dx.doi.org/10.1016/j.yrtph.2015.07.016)
Justification for classification or non-classification
Based on the entire database of genetic toxicity studies and the review by the OECD and an external peer reviewer it is concluded that in summary, poorly soluble cobalt salts/compounds do not appear to be genotoxic in vitro or in vivo at all, soluble cobalt salts do not elicit any mutagenic activity either in bacterial or mammalian test systems. However they induce some genotoxic effects in vitro, mainly manifest as DNA strand or chromosome breaks, which are consistent with a reactive oxygen mechanism, as has been proposed by various authors. A weight-of-evidence approach was applied, considering positive as well as negative in vivo clastogenicity studies and the absence of such chromosome damage in humans that are occupationally exposed to inorganic cobalt substances. It was concluded that effective protective processes exist in vivo to prevent genetic toxicity with relevance for humans from the soluble cobalt salts category (OECD 2014, Kirkland et al. 2015).
Based on the above information, the classification criteria for germ cell mutagenicity according to regulation (EC) 1272/2008 are not met, thus no classification required.
Information on Registered Substances comes from registration dossiers which have been assigned a registration number. The assignment of a registration number does however not guarantee that the information in the dossier is correct or that the dossier is compliant with Regulation (EC) No 1907/2006 (the REACH Regulation). This information has not been reviewed or verified by the Agency or any other authority. The content is subject to change without prior notice.
Reproduction or further distribution of this information may be subject to copyright protection. Use of the information without obtaining the permission from the owner(s) of the respective information might violate the rights of the owner.