This report contains the collective views of an international group of experts and does not necessarily represent the decisions or the stated policy of the United Nations Environment Programme, the International Labour Organization, or the World Health Organization.

Concise International Chemical Assessment Document 69

First draft prepared by Dr James H. Kim and Dr Herman J. Gibb, Sciences International Inc., Alexandria, Virginia, USA; and Mr Paul D. Howe, Centre for Ecology and Hydrology, Monks Wood, Huntingdon, Cambridgeshire, United Kingdom

Published under the joint sponsorship of the United Nations Environment Programme, the International Labour Organization, and the World Health Organization, and produced within the framework of the Inter-Organization Programme for the Sound Management of Chemicals.

The International Programme on Chemical Safety (IPCS), established in 1980, is a joint venture of the United Nations Environment Programme (UNEP), the International Labour Organization (ILO), and the World Health Organization (WHO). The overall objectives of the IPCS are to establish the scientific basis for assessment of the risk to human health and the environment from exposure to chemicals, through international peer review processes, as a prerequisite for the promotion of chemical safety, and to provide technical assistance in strengthening national capacities for the sound management of chemicals.

The Inter-Organization Programme for the Sound Management of Chemicals (IOMC) was established in 1995 by UNEP, ILO, the Food and Agriculture Organization of the United Nations, WHO, the United Nations Industrial Development Organization, the United Nations Institute for Training and Research, and the Organisation for Economic Co-operation and Development (Participating Organizations), following recommendations made by the 1992 UN Conference on Environment and Development to strengthen cooperation and increase coordination in the field of chemical safety. The purpose of the IOMC is to promote coordination of the policies and activities pursued by the Participating Organizations, jointly or separately, to achieve the sound management of chemicals in relation to human health and the environment.

WHO Library Cataloguing-in-Publication Data

Kim, James H.

Cobalt and inorganic cobalt compounds/prepared by James H. Kim, Herman J. Gibb, Paul D. Howe

(Concise international chemical assessment document ; 69)

1. Cobalt - adverse effects. 2.Cobalt - toxicity. 3. Environmental exposure. 4 Risk
assessment. I.Gibb, Herman J. II. Howe, Paul D. III. World Health Organization
IV. International Programme on Chemical Safety. V.Title. VI. Series

ISBN 92 4 153069 3          (NLM Classification: QV 290)
ISSN 978 92 4 153069 9

©World Health Organization 2006

All rights reserved. Publications of the World Health Organization can be obtained from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel: +41 22 791 3264; fax: +41 22 791 4857; email: bookorders@who.int). Requests for permission to reproduce or translate WHO publications — whether for sale or for noncommercial distribution — should be addressed to WHO Press, at the above address (fax: +41 22 791 4806; email: permissions@who.int).

The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement.

The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters.

All reasonable precautions have been taken by WHO to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either express or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use.

Risk assessment activities of the International Programme on Chemical Safety, including the production of Concise International Chemical Assessment Documents, are supported financially by the Department of Health and Department for Environment, Food & Rural Affairs, UK, Environmental Protection Agency, Food and Drug Administration, and National Institute of Environmental Health Sciences, USA, European Commission, German Federal Ministry of Environment, Nature Conservation and Nuclear Safety, Health Canada, Japanese Ministry of Health, Labour and Welfare, and Swiss Agency for Environment, Forests and Landscape.

FOREWORD

Concise International Chemical Assessment Documents (CICADs) are published by the International Programme on Chemical Safety (IPCS) — a cooperative programme of the World Health Organization (WHO), the International Labour Organization (ILO), and the United Nations Environment Programme (UNEP). CICADs have been developed from the Environmental Health Criteria documents (EHCs), more than 200 of which have been published since 1976 as authoritative documents on the risk assessment of chemicals.

International Chemical Safety Cards on the relevant chemical(s) are attached at the end of the CICAD, to provide the reader with concise information on the protection of human health and on emergency action. They are produced in a separate peer-reviewed procedure at IPCS. They may be complemented by information from IPCS Poison Information Monographs (PIM), similarly produced separately from the CICAD process.

CICADs are concise documents that provide summaries of the relevant scientific information concerning the potential effects of chemicals upon human health and/or the environment. They are usually based on selected national or regional evaluation documents or on existing EHCs. Before acceptance for publication as CICADs by IPCS, these documents undergo extensive peer review by internationally selected experts to ensure their completeness, accuracy in the way in which the original data are represented, and the validity of the conclusions drawn.

The primary objective of CICADs is characterization of hazard and dose–response from exposure to a chemical. CICADs are not a summary of all available data on a particular chemical; rather, they include only that information considered critical for characterization of the risk posed by the chemical. The critical studies are, however, presented in sufficient detail to support the conclusions drawn. For additional information, the reader should consult the identified source documents upon which the CICAD has been based.

Risks to human health and the environment will vary considerably depending upon the type and extent of exposure. Responsible authorities are strongly encouraged to characterize risk on the basis of locally measured or predicted exposure scenarios. To assist the reader, examples of exposure estimation and risk characterization are provided in CICADs, whenever possible. These examples cannot be considered as representing all possible exposure situations, but are provided as guidance only. The reader is referred to EHC 170.¹

While every effort is made to ensure that CICADs represent the current status of knowledge, new information is being developed constantly. Unless otherwise stated, CICADs are based on a search of the scientific literature to the date shown in the executive summary. In the event that a reader becomes aware of new information that would change the conclusions drawn in a CICAD, the reader is requested to contact IPCS to inform it of the new information.

Procedures

The flow chart shows the procedures followed to produce a CICAD. These procedures are designed to take advantage of the expertise that exists around the world — expertise that is required to produce the high-quality evaluations of toxicological, exposure, and other data that are necessary for assessing risks to human health and/or the environment. The IPCS Risk Assessment Steering Group advises the Coordinator, IPCS, on the selection of chemicals for an IPCS risk assessment based on the following criteria:

• there is the probability of exposure; and/or

• there is significant toxicity/ecotoxicity.

Thus, it is typical of a priority chemical that:

• it is of transboundary concern;

• it is of concern to a range of countries (developed, developing, and those with economies in transition) for possible risk management;

• there is significant international trade;

• it has high production volume;

• it has dispersive use.

Advice from Risk Assessment Steering Group Criteria of priority: there is the probability of exposure; and/or there is significant toxicity/ ecotoxicity. Thus, it is typical of a priority chemical that it is of transboundary concern; it is of concern to a range of countries (developed, developing, and those with economies in transition) for possible risk management; there is significant international trade; the production volume is high; the use is dispersive. Special emphasis is placed on avoiding duplication of effort by WHO and other international organizations. A prerequisite of the production of a CICAD is the availability of a recent high-quality national/regional risk assessment document = source document. The source document and the CICAD may be produced in parallel. If the source document does not contain an environmental section, this may be produced de novo, provided it is not controversial. If no source document is available, IPCS may produce a de novo risk assessment document if the cost is justified. Depending on the complexity and extent of controversy of the issues involved, the steering group may advise on different levels of peer review: standard IPCS Contact Points above + specialized experts above + consultative group

The Steering Group will also advise IPCS on the appropriate form of the document (i.e. a standard CICAD or a de novo CICAD) and which institution bears the responsibility of the document production, as well as on the type and extent of the international peer review.

The first draft is usually based on an existing national, regional, or international review. When no appropriate source document is available, a CICAD may be produced de novo. Authors of the first draft are usually, but not necessarily, from the institution that developed the original review. A standard outline has been developed to encourage consistency in form. The first draft undergoes primary review by IPCS to ensure that it meets the specified criteria for CICADs.

The second stage involves international peer review by scientists known for their particular expertise and by scientists selected from an international roster compiled by IPCS through recommendations from IPCS national Contact Points and from IPCS Participating Institutions. Adequate time is allowed for the selected experts to undertake a thorough review. Authors are required to take reviewers’ comments into account and revise their draft, if necessary. The resulting second draft is submitted to a Final Review Board together with the reviewers’ comments. At any stage in the international review process, a consultative group may be necessary to address specific areas of the science. When a CICAD is prepared de novo, a consultative group is normally convened.

The CICAD Final Review Board has several important functions:

• to ensure that each CICAD has been subjected to an appropriate and thorough peer review;
• to verify that the peer reviewers’ comments have been addressed appropriately;
• to provide guidance to those responsible for the preparation of CICADs on how to resolve any remaining issues if, in the opinion of the Board, the author has not adequately addressed all comments of the reviewers; and
• to approve CICADs as international assessments.

Board members serve in their personal capacity, not as representatives of any organization, government, or industry. They are selected because of their expertise in human and environmental toxicology or because of their experience in the regulation of chemicals. Boards are chosen according to the range of expertise required for a meeting and the need for balanced geographic representation.

Board members, authors, reviewers, consultants, and advisers who participate in the preparation of a CICAD are required to declare any real or potential conflict of interest in relation to the subjects under discussion at any stage of the process. Representatives of nongovernmental organizations may be invited to observe the proceedings of the Final Review Board. Observers may participate in Board discussions only at the invitation of the Chairperson, and they may not participate in the final decision-making process.

This CICAD² on cobalt and inorganic cobalt compounds was prepared by Sciences International, Inc. in the United States and the Centre for Ecology and Hydrology in the United Kingdom and was based on reviews prepared by the Agency for Toxic Substances and Disease Registry (ATSDR, 2004) and the International Agency for Research on Cancer (IARC, 2005). To address literature citations not included in either of these reviews, a comprehensive literature search of several online databases was conducted in April 2005. Information on the source documents and their peer reviews is presented in Appendix 2. Information on the peer review of this CICAD is presented in Appendix 3. This CICAD was considered and approved as an international assessment at a meeting of the Final Review Board held in Nagpur, India, on 31 October – 3 November 2005. Participants at the Final Review Board meeting are presented in Appendix 4. International Chemical Safety Cards for cobalt, cobalt(II) oxide, cobalt(III) oxide, cobalt(II) sulfide, cobalt(II) chloride, cobalt(II) sulfate, cobalt(II) sulfate heptahydrate, cobalt(II) nitrate, cobalt(II) nitrate hexahydrate, cobalt(II) acetate tetrahydrate, cobalt naphthenate, and cobalt carbonyl, produced by the International Programme on Chemical Safety (IPCS, 2000, 2001a–e, 2004a–f) in a separate, peer-reviewed process, have also been presented in this CICAD.

Cobalt (atomic number 27) is a naturally occurring element with one stable isotope (⁵⁹Co) and 26 known radioactive isotopes. There are three valence states of cobalt (0, +2, and +3). Because cobalt may occur as a radioactive isotope, it can produce ionizing radiation. This document focuses primarily on stable cobalt. The reader should consult other sources, such as ATSDR (2004), for information on the effects of ionizing radiation from radioactive cobalt isotopes.

Cobalt (CAS No. 7440-48-4) is a silvery grey solid at room temperature. It is the 33rd most abundant element and has been found in a variety of media, including air, surface water, leachate from hazardous waste sites, groundwater, soil, and sediment. Sources of exposure to cobalt and inorganic cobalt compounds are both natural and anthropogenic. Natural sources include wind-blown dust, seawater spray, volcanoes, forest fires, and continental and marine biogenic emissions. Anthropogenic sources include the burning of fossil fuels, sewage sludge, phosphate fertilizers, mining and smelting of cobalt ores, processing of cobalt alloys, and industries that use or process cobalt compounds.

Cobalt and inorganic cobalt compounds are non-volatile and released into the atmosphere in particulate form. Anthropogenic cobalt from combustion sources is assumed to be primarily in the form of oxides. Sulfide and arsenide forms are also released into the atmosphere during ore extraction and refining processes.

Cobalt released into the atmosphere is deposited on soil, and cobalt released to water may sorb to particles and settle into sediment or sorb directly to sediment. The distribution coefficient of cobalt (e.g. from water to sediment) varies due to pH, redox conditions, ionic strength, and dissolved organic matter concentrations. Factors affecting the speciation and fate of cobalt in water, sediments, and soil include organic ligands such as humic acids, anions, pH, and redox potential. The soil mobility of cobalt is inversely related to the strength of adsorption by soil constituents. Although plants may take up cobalt from the soil, the translocation of cobalt from the roots to other parts of the plant is not significant.

Measured atmospheric concentrations of cobalt are about 1 ng/m³ or less in non-source areas and generally less than 10 ng/m³ in source areas, although higher concentrations in source areas have been reported. Surface water and groundwater concentrations of cobalt are low, below 1 µg/l in pristine areas and 1–10 µg/l in populated areas. Surface water and groundwater concentrations can be much higher in mining and agricultural areas — as much as several hundred milligrams per litre. Mean cobalt concentrations in seawater have been reported to be less than 1 µg/l. Cobalt concentrations in drinking-water are generally <1–2 µg/l. In rainwater, mean concentrations are 0.3–1.7 µg/l. The earth’s crust contains an average cobalt concentration of 20–25 mg/kg. Near some anthropogenic sources, the concentration of cobalt in soil may be several hundred milligrams per kilogram.

The largest source of exposure to cobalt for the general population is the food supply. The estimated intake from food is 5–40 µg/day, most of which is inorganic cobalt. Occupational exposure to cobalt occurs in several industries. Levels of cobalt in tobacco range from <0.3 to 2.3 µg/g dry weight, and approximately 0.5% of this cobalt is present in mainstream smoke. Cobalt concentrations in coal, crude oil, fuel oil, and gasoline in the United States were found to be 5 mg/kg, 0.001–10 mg/kg, 0.03–0.3 mg/kg, and <0.1 mg/kg, respectively.

Inhalation of cobalt particles results in deposition in the upper and lower respiratory tract, where they can be retained or absorbed into the blood after dissolution or mechanically transferred to the gastrointestinal tract by mucociliary action and swallowing. Approximately 50% of the cobalt that enters the gastrointestinal tract will be absorbed. Cobalt absorption is increased among individuals who are iron deficient. Water-soluble forms are better absorbed than insoluble forms. Cobalt is essential as a component of vitamin B₁₂; therefore, it is found in most tissues. Total body burden is estimated as 1.1–1.5 mg, with 0.11 mg in the liver. After inhalation exposure, higher levels of cobalt have been found in the lung. No studies describe the distribution of cobalt in humans following ingestion, but animal studies indicate that cobalt is retained primarily in the liver. In a controlled human aerosol exposure study, 40% of the initial lung burden of cobalt oxide was retained at 6 months after exposure. Urinary excretion increases with time following inhalation exposure. Particle size affects elimination of inhaled cobalt, since more cobalt is mechanically cleared to the gastrointestinal tract when particles are larger. Faecal elimination is the primary route of excretion following oral exposure in humans.

The inhalation LC₅₀ for cobalt hydrocarbonyl in rats was found to be 165 mg/m³ for a 30-min exposure. Oral LD₅₀s for soluble cobalt compounds have been reported to range from 42.4 to 317 mg/kg body weight, depending on the compound and species tested. Tricobalt tetraoxide, an insoluble cobalt compound, is reported to have an LD₅₀ of 3672 mg of cobalt per kilogram body weight in rats.

Rats and mice exposed short term (16 days) to cobalt sulfate by inhalation at cobalt concentrations of 19 mg/m³ and 1.9 mg/m³, respectively, exhibited necrosis and inflammation of the respiratory tract epithelium. Rats also developed thymus necrosis and testicular atrophy. Male rats exposed orally to cobalt chloride at a cobalt concentration of 12.4 mg/kg body weight per day for 3 weeks exhibited cardiac damage. Rats, rabbits, and mice exposed by inhalation to cobalt compounds at concentrations of >0.3 mg/m³ (cobalt concentrations of >0.11 mg/m³) for 3–4 months exhibited lesions of the respiratory tract. Rats exposed for 2–3 months to cobalt sulfate in the diet or to cobalt chloride in the drinking-water at cobalt doses of 26–30.2 mg/kg body weight per day exhibited increased heart weight and degenerative heart lesions. Rats exposed to cobalt sulfate at a cobalt dose of 8.4 mg/kg body weight per day in the diet for 24 weeks had significant reductions in heart enzyme activity levels. Rats exposed to cobalt chloride for 4–5 months at cobalt doses of 10–18 mg/kg body weight per day exhibited kidney damage.

Hamsters exposed by inhalation to cobalt oxide for a lifetime developed emphysema. Mice and rats exposed to cobalt sulfate by inhalation for 105 weeks developed lung tumours in a dose-related manner. Cobalt (as cobalt metal powder) produces tumours such as sarcomas in rats when injected intramuscularly.

Many cobalt compounds are genotoxic in mammals and in mammalian and bacterial test systems. Cobalt(III) compounds are positive in bacterial test systems. Cobalt(II) compounds were positive for genetic conversions in Saccharomyces cerevisiae but otherwise demonstrated little genotoxic activity.

Cobalt has been found to have reproductive and developmental effects in animals. Rats exposed to cobalt (as cobalt chloride) at 13.3–58.9 mg/kg body weight per day for 2–3 months and mice exposed to cobalt (as cobalt chloride) at 43.4 mg/kg body weight per day for 13 weeks exhibited testicular degeneration and atrophy. Male mice exposed to cobalt chloride at doses of 46.9 or 93.0 mg/kg body weight per day and mated with unexposed female mice displayed decreased epididymal weight, sperm count, testes weight, and fertility, as measured by the number of successful matings. In developmental studies, pregnant rats exposed to maternally toxic doses of cobalt chloride (5.4 or 21.8 mg of cobalt per kilogram body weight per day) produced newborn pups with stunted growth and decreased survival, but no teratogenic effects were observed. Rabbits exposed to cobalt (as cobalt sulfate) at 7.6 mg/kg body weight per day had increased fetal resorption and an increased number of fetuses with retarded body weight.

Inhalation and dermal exposure to cobalt in humans can result in sensitization. Bronchial asthma has been described in workers exposed to various forms of cobalt.

Humans ingesting cobalt chloride at 150 mg/day for 22 days experienced polycythaemia and an increase in haemoglobin. Studies have also reported cardiomyopathy in humans who had consumed large quantities of beer that contained cobalt sulfate.

Interstitial lung disease caused by metallic cobalt-containing particles is an occupational lung disease generally referred to as hard metal lung disease.

Mortality studies of the hard metal industry suggest an increase in lung cancer mortality. Cobalt is used as a binder in this industry, and exposures to other compounds, including tungsten carbide and other metallic compounds, such as titanium carbide, tantalium carbide, and niobium carbide, also occur.

A cross-sectional study of diamond polishers exposed to cobalt was used to derive an inhalation tolerable concentration of 1 × 10⁻⁴ mg/m³ based on lung function decrement. The difference between the tolerable concentration and the cobalt concentrations found in the ambient air near anthropogenic sources is generally about 10-fold.

A 96-h EC₅₀ for cobalt based on growth of the freshwater green alga Chlorella vulgaris was reported as 0.6 mg/l, whereas EC₅₀s for aquatic vascular plants were 0.1 and 0.2 mg/l. The 5-day EC₅₀ for cobalt based on growth of the marine diatom Ditylum brightwellii was 0.3 mg/l. For freshwater invertebrates, acute LC₅₀s (24–96 h) range from 1.1 to 239 mg/l. Several studies on Daphnia magna reproduction were reported, with a 21-day EC₅₀ at 0.01 mg/l and a 28-day NOEC of 0.003 mg/l; however, later studies found 21-day NOECs ranging from 0.03 to 0.05 mg/l for varying levels of calcium carbonate. The lowest reported NOEC for aquatic organisms was for the water flea Ceriodaphnia dubia in a 7-day test, at <0.003 mg/l. The most sensitive marine invertebrates were lobster larvae, with 96-h LC₅₀s ranging from 4.5 to 22.7 mg/l. Ninety-six-hour LC₅₀s for freshwater fish range from 1.4 to 333 mg/l. A 16-day NOEC based on survival was reported at 0.06 mg/l. Test results for marine fish suggest that at least the species tested are relatively insensitive to cobalt, with 96-h LC₅₀s ranging from 52.5 to >1000 mg/l. Ca²⁺ competition and dissolved organic matter complexation were the most important factors preventing Co²⁺ from binding at the gills in natural water tests. However, the effect of Ca²⁺ ions on the uptake and potential toxicity of cobalt occurs at very low Ca²⁺ concentrations, probably lower than those used in any of the reported toxicity tests.

Moderate-reliability guidance values were determined for the marine environment at 20 µg/l (for the protection of 99% of marine species with 50% confidence) and for the freshwater environment at 8 µg/l (for the protection of 95% of freshwater species with 50% confidence). A comparison of the guidance values with environmental concentrations would suggest that effects are likely only in the vicinity of major anthropogenic releases. There is some evidence that under freshwater conditions of extremely low Ca²⁺ there is less competition for cobalt at fish gill binding sites and therefore greater uptake of cobalt. Therefore, the greatest risk to aquatic organisms might be in very soft water areas (where the Ca²⁺ ion concentration is extremely low) close to sources of anthropogenic release.

Data regarding the toxicity of cobalt to soil microorganisms are limited. There is little evidence of cobalt toxicity to plants due to elevated concentrations in soil. Cobalt tolerance, along with tolerance to other metals, has been found in plant populations growing on soils high in particular metals. Exclusion of the metal has been demonstrated in the cobalt tolerance of some species, whereas others growing on cobalt-rich copper clearings are hyperaccumulators of cobalt. Adverse effects on earthworm growth and springtail reproduction have been reported at 300–400 mg/kg dry weight. In the terrestrial environment, adverse effects of cobalt on birds and wild mammals would appear unlikely, with cobalt deficiency in ruminants more likely than cobalt toxicosis.

Cobalt (CAS No. 7440-48-4) is a naturally occurring element (atomic number 27) in the first transition series of Group 9 of the periodic chart of elements. ⁵⁹Co is the only stable isotope. There are 26 known radioactive isotopes, of which only ⁵⁷Co and ⁶⁰Co are commercially important.

Cobalt occurs in the 0, +2, and +3 valence states. Cobalt(II) is more stable than cobalt(III), which is a powerful oxidizing agent that can oxidize water and liberate oxygen. Metallic cobalt(0) occurs in two allotropic forms, hexagonal and cubic, which are stable at room temperature. Cobalt has a relative molecular mass of 58.93 and is a silvery grey solid at room temperature. Its melting point is 1493 °C. At room temperature (20 °C), the density of cobalt is 8.9 g/cm³. Cobalt is soluble in dilute acids, and ultrafine metal cobalt powder is soluble in water at 1.1 mg/l.

Selected chemical and physical properties of cobalt and several inorganic cobalt compounds are presented in Table 1, with further details contained in the International Chemical Safety Cards reproduced at the end of this document.

Table 1: Physical and chemical properties of selected cobalt compounds.

Cobalt can be analysed in human biological samples, such as urine, blood, serum, and tissues. Analysis of cobalt in urine usually involves sample chelation and/or acid digestion, followed by GF-AAS (Heinrick & Angerer, 1984; Ichikawa et al., 1985; Bouman et al., 1986; Kimberly et al., 1987; Alexandersson, 1988; Sunderman et al., 1989; Templeton, 1996). Detection limits range from 0.1 to 2.4 μg/l. Analysis in whole blood can be done by GF-AAS, by acid digestion, chelation, preconcentration, and extraction followed by differential pulse cathodic stripping voltammetry, or by a colorimetric method (Heinrick & Angerer, 1984; Afeworki & Chandravanshi, 1987). GF-AAS and differential pulse cathodic stripping voltammetry have detection limits of 2 µg/l and 0.8 μg/l, respectively. The colorimetric method has a detection limit of 150 μg/l. Analysis of cobalt in serum also uses the GF-AAS method, with a detection limit of 0.02 μg/l (Sunderman et al., 1989). NIOSH method 8005 utilizes ICP-AES, with detection limits of 10 µg/l for blood and 0.2 μg/g for tissue (NIOSH, 1994b). ICP-MS is more widely available since the 1990s and is used for multi-elemental analysis of human blood, serum, and urine.

Environmental samples are analysed by atomic absorption spectrometry, instrumental neutron activation analysis, and mass spectrometry (USEPA, 1982, 1986; Haddad & Zikovsky, 1985; Nojiri et al., 1985; Fishman et al., 1986; Hansson et al., 1988; Nakashima et al., 1988; NIOSH, 1994a). Using these methods, the detection limits for cobalt in air range from 0.17 to 0.5 μg/m³. A more recent NIOSH method for the analysis of cobalt in workplace air utilizes sample collection on cellulose or PVC membrane and analysis by ICP-AES; the limit of detection for a 2-m³ sample is 6 ng/m³ (NIOSH, 2003). Detection limits for cobalt in water range from 0.004 μg/l (from lake water using ICP-AES) to 0.05 mg/l (using flame atomic absorption spectrometry).

Cobalt comprises 0.0025% of the weight of the earth’s crust and is the 33rd most abundant element (Smith & Carson, 1981; Merian, 1985; Abbasi et al., 1989). Cobalt does not occur naturally as a base metal, but is a component of more than 70 naturally occurring minerals, including various sulfides, arsenides, sulfoarsenides, hydrates, and oxides. The most common cobalt minerals are the arsenide CoAs_2–3 (smeltite), the arsenosulfide CoAsS (cobaltine), and the sulfide Co₃S₄ (linneite) (IARC, 1991). Identified world cobalt resources are about 14 million tonnes. The vast majority of these resources are in nickel-bearing laterite deposits, with most of the rest occurring in nickel–copper sulfide deposits hosted in mafic and ultramafic rocks in Australia, Canada, and the Russian Federation and in the sedimentary copper deposits in Kinshasha, Democratic Republic of Congo, and Zambia (USGS, 2005). Significant resources of cobalt are also present in the deep-sea nodules and crusts that occur in the mid-Pacific Ocean and are estimated to contain anywhere from 2.5 to 10 million tonnes of cobalt (Cobalt Development Institute, undated a). Encrustation deposits ("cobalt-rich crusts") in shallow waters close to the Hawaiian Islands are believed to contain up to 2.5% cobalt and constitute an important potential source of cobalt (Cobalt Development Institute, 2004).

Sources of environmental cobalt are both natural and anthropogenic (Barceloux, 1999). Natural sources include erosion (wind-blown continental dusts), weathering of rocks and soil, seawater spray, volcanoes, forest fires, extraction by plants, and continental and marine biogenic emissions. The worldwide estimate for atmospheric cobalt emissions is 5350–6170 tonnes per year (Lantzy & Mackenzie, 1979; Nriagu, 1989). Cobalt compounds have been found to occur naturally in seawater, surface water, spring water, and groundwater (Smith & Carson, 1981).

Cobalt is normally associated with copper or nickel; mined ore often contains only 0.1% elemental cobalt. About 44% of world production of cobalt comes from nickel ores. Cobalt is extracted from the metals in the ore by both flotation (sulfide ores) and gravity (arsenide ores); roasting or acid leaching is necessary to concentrate the cobalt (Barceloux, 1999). Cobalt is also extracted from the ore and concentrated by pyrometallurgical, hydrometallurgical, and electrolytic processes alone or in combination (Donaldson et al., 1986). Cobalt is currently mined in 12 countries and refined in 23 countries. The global mine production of cobalt in 2003 totalled 46 900 tonnes, with the principal nine producing countries as follows (production in tonnes): Democratic Republic of Congo, 11 000; Zambia, 9000; Australia, 7000; Canada, 5200; Russian Federation, 4800; Cuba, 3400; New Caledonia, 1500; Brazil, 1300; Morocco, 1300; and other counties, 2400 (USGS, 2005). The approximate refined quantity of cobalt in 2004 was 43 000 tonnes, with the largest amounts (in tonnes) produced in Finland (8000), Zambia (6500), Canada, China, Russian Federation, and Norway (4500 each), Australia (3900), Belgium, Morocco, New Caledonia, and Democratic Republic of Congo (1200 each) (Cobalt Development Institute, 2004). A significant source of cobalt is the recycling of scrap metal. In 1998, an estimated 32% of cobalt supply in the United States was derived from scrap, and the ratio of cobalt derived from new scrap to that derived from old scrap was estimated to be 50:50. Of all the cobalt in old scrap available for recycling, an estimated 68% was either consumed in the United States or exported to be recycled (Shedd, 2004). In 2003, 2200 tonnes of cobalt were recycled in the United States (USGS, 2004).

In 2002, consumption of cobalt metal, organic and inorganic cobalt compounds, and cobalt scrap in the United States was 3870, 1270, and 2800 tonnes, respectively (Shedd, 2002). The use pattern (end use: tonnes) was as follows: superalloys: 3700; steel alloys: 555; other alloys including magnetic alloys: 1050; cemented carbides: 617; chemical and ceramic use: 1950; and miscellaneous: 63 (Shedd, 2002). Cobalt metal is used in alloys with iron, nickel, and other metals to make Alnico, an alloy of unusual magnetic strength; and in Stellite alloys, which contain cobalt, chromium, and tungsten and are used for high-speed, heavy-duty, high-temperature cutting tools (Cobalt Development Institute, 2004). Cobalt metal has three major uses in the petrochemical and plastic industries as both heterogeneous and homogeneous catalysts: (1) hydro-treating and desulfurization catalysts for oil and gas; these catalysts are typically 3–5% cobalt oxide (Co₃O₄), 14% manganese trioxide (MnO₃), and the balance aluminium oxide (Al₂O₃); (2) mixed cobalt acetate/manganese–sodium bromide homogeneous catalyst for the production of terephthalic acid and dimethyl terephthalate; and (3) cobalt catalyst in the oxo synthesis (hydroformylation) for the production of alcohols and aldehydes for plastic and detergent production, employing freshly reduced cobalt metal, carbonyls, or cobalt salts (transformed in situ to carbonyl) (Cobalt Development Institute, undated b; USGS, 2005).

The major anthropogenic sources of environmental cobalt include mining and processing (smelting) of cobalt-bearing ores, the use of cobalt-containing sludge or phosphate fertilizers on soil, the disposal of cobalt-containing waste, and atmospheric deposition from activities such as the burning of fossil fuels and smelting and refining of metals (Smith & Carson, 1981). Cobalt-containing sewage sludge, phosphate fertilizers, processing of cobalt alloys, and industries that use or process cobalt compounds are estimated to emit an estimated 4000 tonnes per year of atmospheric cobalt (Lantzy & Mackenzie, 1979). More than 2000 tonnes of cobalt are released annually from mining and mineral processing in the United States, including 480 tonnes of cobalt in coal produced; losses generated during cobalt chemical and powder processing were estimated at 50–80 tonnes annually, whereas losses from alloy processing and manufacture of parts and products were estimated to be 360 tonnes and 120 tonnes, respectively (Donaldson, 1986; Donaldson et al., 1986; Shedd, 1993). The total environmental release of cobalt by industrial sources in the United States that was reportable to the Toxics Release Inventory for 2000 was approximately 228 400 kg, which included air release (16 150 kg), water release (1633 kg), and land release (210 600 kg). Additionally, the total off-site waste transfer of cobalt was 2 967 000 kg (USEPA, 2002).

Cobalt and inorganic cobalt compounds are non-volatile. Therefore, they are released into the atmosphere in particulate form. Atmospheric transport depends on particle size and density and meteorological conditions. Coarse particles with diameters >2 μm may deposit within 10 km from the point of emission, while smaller particles may travel longer distances. The mass median diameter of atmospheric cobalt was found to be 2.6 μm in one study (Milford & Davidson, 1985). Data on the transformations of cobalt in the atmosphere are limited. Anthropogenic cobalt from combustion processes is assumed to be primarily oxides (Schroeder et al., 1987). Arsenide and sulfide forms are also released into the atmosphere during ore extraction processes. It is unclear whether these forms of cobalt are transformed in the atmosphere. If oxides are transformed into more soluble species such as sulfates, then these may be washed out of the atmosphere in rain.

Ultimately, the final repository for cobalt is soil and sediment. Released into water, cobalt may sorb to particles and settle into sediment or sorb directly to sediment. Complexation of cobalt to dissolved organic substances can reduce sediment sorption (Albrecht, 2003). Interparticle migration of cobalt can affect the transport of metal ions in sediments (Jackman et al., 2001). In addition, cobalt can be transported in dissolved form or as suspended sediment by rivers and by sea and ocean currents. Concentration profiles of cobalt in deep water suggest that dissolved amounts decrease with increasing depth and that dissolved cobalt is precipitated in the adsorbed state with oxides of iron and manganese and with crystalline sediments such as aluminosilicate and goethite. In the deep sea, formation of manganese nodules removes cobalt by interaction with manganese oxide (MnO) (Barceloux, 1999). Polluted water with higher concentrations of organic pollutants may result in higher concentrations of soluble organic cobalt complexes (Nriagu & Coker, 1980; Glooschenko et al., 1981; Smith & Carson, 1981; Knauer et al., 1982; Brügmann, 1988; Finney & Huh, 1989; Windom et al., 1989; Shine et al., 1995; Szefer et al., 1996; Bargagli, 2000). Humic substances/humic acids are naturally present in aquatic environments and bind strongly to cobalt (Burba et al., 1994). Over time, these complexes may transform into stronger complexes where cobalt is less readily disassociated (Zhang et al., 1990).

The distribution coefficient of cobalt in water varies due to pH, redox conditions, ionic strength, and dissolved organic matter concentrations (Mahara & Kudo, 1981). For example, as pH is increased from 5 to 7.5, the uptake of ⁶⁰Co from the water to sediment increased rapidly (Benes et al., 1989a, 1989b). Liquid-to-solids ratio and ionic strength did not affect ⁶⁰Co uptake by sediment. ⁶⁰Co has also been found to be more mobile in anaerobic aquatic environments than in aerobic freshwater environments (Mahara & Kudo, 1981). For example, in anaerobic seawater–sediment systems, ⁶⁰Co was 250 times more mobile than in aerobic freshwater–sediment systems. In anaerobic conditions, 30% of ⁶⁰Co added to a freshwater–sediment system was mobile, whereas in aerobic conditions, 98% was permanently fixed. In anaerobic seawater systems, mobile ⁶⁰Co consisted of non-ionic forms associated with low molecular weight organic substances that were stable as pH changed. Mobile ⁶⁰Co was mostly ionic.

Factors that affect the speciation and fate of cobalt in water and sediments include organic ligands such as humic acids and EDTA, anions such as Cl⁻, OH⁻, CO₃²⁻, HCO₃⁻, and SO₄²⁻, pH, and redox potential. The mole percentages of cobalt species in a Welsh lake were 76% free Co²⁺, 9.8% CoCO₃, 9.6% CoHCO₃⁺, 4.0% humate complexes, and 0.5% CoSO₄ based on stability constant data used in conjunction with the HALTAFALL program (Mantoura et al., 1978). Similarly, Smith & Carson (1981) reported the rank concentrations of cobalt species in fresh water as free Co²⁺ > CoCO₃ > CoSO₄. In the Rhone River in France, where organic wastes are present in high levels, cobalt is almost completely complexed. The distribution of ⁶⁰Co in the Rhone River at Arles, France, was 45% particulate phase, 30% dissolved phase, and 25% colloidal phase (Eyrolle & Charmasson, 2001). As pH decreases, adsorption of cobalt by particulate matter also decreases, since increasing H⁺ concentrations compete with metal binding sites. Therefore, levels of dissolved cobalt will be increased at low pH (ATSDR, 2004). In a study of riverine, estuarine, and marine surface water in England, cobalt carbonate complexes (HCO₃⁻ and CO₃²⁻) constituted 70% of dissolved cobalt, whereas free Co²⁺ was a major species at 25% (Tipping et al., 1998). As water alkalinity increases, the proportion of cobalt carbonate complexes increases as free Co²⁺ decreases. In seawater, the proportion of carbonate and free cobalt species is similar. Sulfate complexes are estimated to make up 20% of cobalt in seawater (Tipping et al., 1998). Smith & Carson (1981) estimated the rank concentrations of cobalt species in seawater to be CoCl⁺ > free Co²⁺ > CoCO₃ > CoSO₄, whereas Mantoura et al. (1978) reported the rank concentrations of cobalt species in seawater (35‰) as CoCO₃ > free Co²⁺ > CoSO₄ > CoHCO₃⁺ > CoCl⁺ > CoOH⁺. Redox potential can also affect speciation of cobalt. For example, the concentration of dissolved cobalt has been found to increase by several orders of magnitude with increasing depth in Baltic waters. This is because of the formation of soluble bisulfide and polysulfide complexes in anoxic zones (ATSDR, 2004).

Soil mobility of cobalt is inversely related to the strength of adsorption by soil constituents. The adsorption of cobalt to soil occurs rapidly, within 1–2 h. Mineral oxides such as iron and manganese oxide, crystalline materials such as aluminosilicate and goethite, and organic substances can retain cobalt. Soil oxides adsorb larger levels of cobalt than do other materials. Clay minerals adsorb relatively smaller amounts of cobalt (McLaren et al., 1986). Desorption of cobalt from soil oxides is low, although humic acids and montmorillonite desorb substantial amounts. Adsorption in clay soils is most likely due to ion exchange at cationic sites of clay with simple ionic cobalt or hydrolysed ionic species such as CoOH⁺. Adsorption of cobalt with iron or manganese increases with pH (Brooks et al., 1998). As pH increases, insoluble hydroxides and carbonates may form that also reduce cobalt mobility. In contrast, adsorption to mobile colloids would enhance cobalt mobility. Typically, cobalt is more mobile than other metals, such as lead, chromium(II), zinc, and nickel, in soil, but less mobile than cadmium (Mahara & Kudo, 1981; Smith & Carson, 1981; Baes & Sharp, 1983; King, 1988). The partition coefficient, K_D, of cobalt ranged from 0.2 to 3800 l/kg in a wide variety of soils. In 36 Japanese agricultural soils, the mean K_D was 1840 l/kg (minimum 130 l/kg, maximum 104 000 l/kg, median 1735 l/kg) (Yasuda et al., 1995). Soil properties that exhibited the highest correlation with K_D were exchangeable calcium, pH, water content, and cation exchange capacity. The mean Freundlich adsorption constant, K_F, and isotherm exponent, n, values in 11 soils in the United States were 37 l/kg and 0.754, respectively (Buchter et al., 1989). The K_F values ranged from 2.6 to 363 l/kg and correlated with soil pH and cation exchange capacity. In another study, 13 soils from the south-eastern United States had soil pH values that ranged from 3.9 to 6.5, and cobalt sorption ranged from 15% to 93% (King, 1988). Soil pH accounted for 84–95% of sorption variation.

Decontamination at nuclear facilities involves the use of organic complexing agents such as EDTA, which greatly enhances cobalt mobility in soil (Killey et al., 1984; Toste et al., 1984; McLaren et al., 1986). Cobalt has been found to leach from municipal and low-level radioactive waste sites (Czyscinski et al., 1982; Cyr et al., 1987; Friedman & Kelmers, 1988). In soils from two sites in Nevada, USA, cobalt was sorbed at >90% when the pH was above 7 and the solids concentration was 20 g/l (USDOE, 1996). Only under extreme conditions, such as pH <4 or high ionic strength soil (0.1 mol/l), would cobalt be capable of migrating.

Factors that affect cobalt speciation in soil and sediment include the nature of the soil and sediment, the concentration of chelating and complexing agents, pH, and redox potential. Dissolved cobalt may form complexes with fulvic acid, humic acid, or other organic ligands, or it may be absorbed by ion exchange mechanisms. However, humic and fulvic cobalt complexes are not as stable as those of copper, lead, iron, and nickel. Sediment from nine sites in the Red Sea was assessed for cobalt speciation using a sequential extraction technique: 5.5% exchangeable, 5% carbonate, 24% iron/manganese oxides, 30.4% organic, 13% sulfides, and 22% lithogenous (Hanna, 1992). The Red Sea is unique, since no permanent streams flow into it. Mean cobalt concentrations increased from 3 mg/kg in 1934 to 6 mg/kg in 1984, although the cobalt distribution was not altered. A reduction of soil redox potential may occur when soil is flooded or in deeper oxygen-depleted layers. This may result in the reduction of iron and manganese and the release of adsorbed cobalt from mineral oxides. A decrease in soil pH may also cause a solubilization of precipitated cobalt, desorption of cobalt, and an increase in cobalt mobility (Smith & Carson, 1981).

Although plants may take up cobalt from the soil, the translocation of cobalt from the roots to other parts of the plants is not significant (Smith & Carson, 1981; Mermut et al., 1996). The transfer coefficient, defined as the ratio of the plant concentration to the soil cobalt concentration, for cobalt is 0.01–0.3 (Mascanzoni, 1989). Higher amounts of cobalt translocation are observed in highly acidic soils (pH 3.3) and in some higher plants (Tolle et al., 1983; Kloke et al., 1984; Watabe et al., 1984; Boikat et al., 1985; Francis et al., 1985; Mejstrik & Svacha, 1988; Palko & Yli-Halla, 1988).

⁶⁰Co is taken up by unicellular algae with reported concentration factors (dry weight) of 40 000 for Scenedesmus obliquus and 18 000 for Selenastrum capricornutum (Nucho et al., 1988; Corisco & Carreiro, 1999). Freshwater molluscs have concentration factors of 100–14 000 (~1–300 in soft tissue). Much of the cobalt taken up by molluscs and crustaceans from water or sediment is adsorbed to the shell or exoskeleton; very little cobalt is generally accumulated in the edible parts (Amiard & Amiard-Triquet, 1979; Smith & Carson, 1981). Similarly, in laboratory studies with Daphnia magna, adsorption to the exoskeleton was the major contamination process (Adam et al., 2001). In studies with starfish (Asterias rubens), accumulation of ⁵⁷Co was found to be predominately from seawater rather than from food (Warnau et al., 1999). Bioaccumulation factors for marine fish and freshwater fish are 100–4000 and <10–1000, respectively (Smith & Carson, 1981). However, accumulation is mostly in the viscera and skin of the fish, not the edible parts of the fish (Smith & Carson, 1981). In carp (Cyprinus carpio), accumulation from water accounted for 75% of ⁶⁰Co accumulated from both water and food; accumulation from water and food was additive (Baudin & Fritsch, 1989). Depuration half-lives were 53 and 87 days for fish contaminated from food and water, respectively (Bandin & Fritsch, 1989). Biomagnification of cobalt up the food-chain does not occur (Smith & Carson, 1981).

Atmospheric cobalt is associated with particulate matter principally to the extent to which particles of soil are dispersed by the wind. At unpolluted sites, mean cobalt levels are typically <1–2 ng/m³ (Smith & Carson, 1981; Hamilton, 1994). At the South Pole, the concentration of cobalt was 0.000 49 ± 0.000 15 ng/m³ in 1974–1975 (Maenhaut et al., 1979). In open-ocean environments, mean cobalt concentrations ranged from 0.0004 to 0.08 ng/m³ (Chester et al., 1991). As examples of cobalt concentrations in urban areas, the annual average cobalt concentration at Nahant, Massachusetts (near Boston), USA, in 1992–1993 was 1.7 ng/m³ (Golomb et al., 1997), whereas in Seville, Spain, during 1996 it was 0.5 ng/m³ (Espinosa et al., 2001). In southern Norway, the mean cobalt level was 0.10 ng/m³ in 1985–1986 (Amundsen et al., 1992). In source areas, cobalt concentrations may exceed 10 ng/m³. The highest average atmospheric cobalt concentration was recorded near a nickel refinery in Wales, at 48 ng/m³ (Smith & Carson, 1981).

Surface water and groundwater concentrations of stable cobalt are low: <1 μg/l in pristine areas and 1–10 μg/l in populated areas (Smith & Carson, 1981; Hamilton, 1994). In 1962–1967, cobalt was detected in 2.8% of 1577 raw surface waters in the United States, with a detection limit of 1 μg/l and a maximum level of 48 μg/l (NAS, 1977). United States Geological Survey data for 6805 ambient surface water stations reported mean and median cobalt levels of 2.9 and 2.0 μg/l, respectively (Eckel & Jacob, 1988). Mean dissolved cobalt concentrations ranging from 0.1 to 1.1 µg/l were reported for rivers in the United Kingdom sampled between 1993 and 1998 (Neal et al., 1996, 1998, 2000). Water concentrations can be much higher in mining and agricultural areas. For example, surface water and groundwater samples collected near the Blackbird Mine in Idaho, USA, where lead and silver mining was conducted from the 1880s to 1982, exhibited cobalt concentrations that ranged from <1 to 625 000 μg/l and from not detected to 315 000 μg/l, respectively (ATSDR, 1995). Levels in Mineral Creek, Arizona, USA (which is near a copper mine and smelter), were recorded at 4500 μg/l, and levels in the Little St. Francis River, Missouri, USA (which receives cobalt mining and milling effluent), were 6500 μg/l (Smith & Carson, 1981).

Mean cobalt levels in seawater were reported as 0.078 μg/l in the Caribbean Sea and 0.17–0.39 μg/l in the Indian Ocean (Hamilton, 1994).

Cobalt is rarely detected in drinking-water. The concentration of cobalt in drinking-water is low and ranges from 0.1 to 5 µg/l (Barceloux, 1999). Only 0.5% of 380 finished drinking-waters in the United States were found to contain cobalt at concentrations above 1 μg/l, with a maximum concentration of 29 μg/l (NAS, 1977). In finished drinking-water in Canada, the median and maximum cobalt concentrations were <2.0 μg/l and 6.0 μg/l, respectively (Meranger et al., 1981). Household tap water in the United States from 35 geographical areas had cobalt concentrations ranging from 2.6 to 107 µg/l in 9.8% of 3834 grab samples (Greathouse & Craun, 1978). In the National Community Water Supply Study in the United States, 62% of 2500 samples contained <1 μg/l, whereas the average and maximum cobalt concentrations were 2.2 and 19 μg/l, respectively (Smith & Carson, 1981).

In rainwater, mean cobalt concentrations are 0.3–1.7 μg/l, with ranges from 0.002 μg/l at Enewetak Atoll to 2.9 μg/l at Swansea Valley, Wales (Smith & Carson, 1981; Arimoto et al., 1985; Hansson et al., 1988; Dasch & Wolff, 1989; Heaton et al., 1990; Nimmo & Chester, 1993; Helmers & Schrems, 1995; Nimmo & Fones, 1997). The highest recorded concentration was 68.9 μg/l in the vicinity of a nickel smelter at Monchegorsk in the Russian Arctic (Reimann et al., 1997). Data on rain from the Mediterranean and the United Kingdom demonstrated that 33–44% of the cobalt occurred as stable organic complexes (Nimmo & Chester, 1993; Nimmo & Fones, 1997).

The earth’s crust contains an average cobalt concentration of 20–25 mg/kg (Smith & Carson, 1981; Merian, 1985; Abbasi et al., 1989). The average concentration of cobalt in soil in the United States is 7.2 mg/kg, with a range of 1–40 mg/kg (Smith & Carson, 1981). Soils that contain cobalt at <0.5–3 mg/kg are considered deficient, since vegetation growing on such soils has insufficient cobalt (<0.08–0.1 mg/kg) to meet the dietary requirements of cattle and sheep. Generally, concentrations of up to 800 mg/kg have been reported in soils near ore deposits, phosphate rocks, ore smelting facilities, and soils contaminated by airport traffic, highway traffic, or other industrial pollution (Smith & Carson, 1981; Kloke et al., 1984). However, soils near the aforementioned Blackbird Mine in Idaho, USA, had cobalt concentrations ranging from 26.5 to 7410 mg/kg (ATSDR, 1995). Cobalt levels in surface soils from two active volcano islands of Sicily ranged from 5.1 to 59.0 mg/kg (Bargagli et al., 1991). Soils surrounding large copper–nickel smelters in Sudbury, Ontario, Canada, illustrate the increasing concentrations of cobalt with closer proximity: 42–154 mg/kg between 0.8 and 1.3 km from the smelter, 33 mg/kg at 10 km, 48 mg/kg at 19 km, and 19 mg/kg at 50 km (Smith & Carson, 1981). Soils surrounding a tungsten carbide tool grinding factory had cobalt levels as high as 12 700 mg/kg; however, neighbourhood soils located 30 and 160 m from the factory had 12–18 mg/kg (Abraham & Hunt, 1995).

Unpolluted freshwater sediment contains about the same levels of cobalt as does cobalt-sufficient soil, generally <20 mg/kg. Cobalt concentrations in polluted lake and river sediment ranged from 0.16 to 133 mg/kg (Smith & Carson, 1981). Knutson et al. (1987) reported cobalt concentrations of up to 700 mg/kg in surficial sediment (Hudson River, New York, USA) near a disused nickel–cadmium battery plant (4 years after closure). In the Hudson River estuary, cobalt levels were an order of magnitude higher in suspended sediment than in bottom sediment (Gibbs, 1994). This can be attributed to the finer grain size of suspended sediment or local sources. Cobalt levels in core samples (surface to 42 cm deep) from the Upper St. Lawrence River estuary in Canada were independent of depth, indicating the lack of any recent significant anthropogenic releases (Coakley et al., 1993).

The cobalt content of living plants depends on the species, the cobalt content of the soil, and numerous environmental factors. The mean cobalt concentration reported for terrestrial plants was 0.48 µg/g (Bowen, 1966). Median cobalt concentrations in freshwater vascular plants of 0.32 and 0.37 µg/g dry weight were reported for unpolluted and polluted environments, respectively (Outridge & Noller, 1991). Grasses normally contain cobalt concentrations of 0.2–0.35 µg/g, but grasses from cobalt-deficient regions contain only 0.02–0.06 µg/g (Hamilton, 1994). Cobalt tolerance, along with tolerance to other metals, has been found in plant populations growing on soils high in particular metals. For example, some plants growing on cobalt-rich soils in Zaire are hyperaccumulators of cobalt, with the plant Haumaniastrum robertii containing a mean concentration of 4304 mg/kg dry weight (1368–10 222 mg/kg) (Brooks, 1977).

Cobalt concentrations have been reported in various aquatic animals. Fish from three Dutch polder lakes contained cobalt at 2.5–25 mg/kg wet weight (Badsha & Goldspink, 1988). Muscle tissue of ocean fish and rock crabs caught near dump sites off New York City, New Haven, Connecticut, and Delaware Bay, USA, contained 10–40 µg/kg and 16.0 µg/kg, respectively (Greig & Jones, 1976). Cobalt has also been detected at remote sites; mean cobalt levels in fish and amphipods in Antarctica were 0.11–0.14 µg/g dry weight and 1.01 µg/g dry weight, respectively (Szefer et al., 1993). The concentration of cobalt in the tissue of 14 bluefin tuna (Thunnus thynnus) caught by various commercial fishing vessels off Newfoundland, Canada, was essentially the same, 0.01 ± 0.004 µg/g (Hellou et al., 1992). In a broad survey of contaminant levels in nine species of fish and fiddler crabs from 11 sites in the lower Savannah River, Georgia, and the Savannah National Wildlife Refuge, USA, mean cobalt levels (0.1–2.5 mg/kg wet weight) among different species and sites were statistically indistinguishable (Winger et al., 1990). These studies suggest that cobalt does not biomagnify up the food-chain (Smith & Carson, 1981).

In a study of the levels and distribution of 14 elements in oceanic seabirds, the concentration of cobalt, an essential element, appeared to be highly regulated, with over 80% of the body burden residing in the skeleton. The mean cobalt concentration in the livers of 11 seabird species ranged from 0.048 to 0.078 µg/g dry weight; of the elements studied, cobalt had the lowest coefficient of variation in the different species (Kim et al., 1998). Mean cobalt levels in the tissues of penguin and other Antarctic seabirds ranged from 0.09 to 0.11 µg/g (Szefer et al., 1993). The geometric mean concentrations of cobalt in tern eggs collected from coastal New Jersey, USA, in 1971 and 1982 were 0.48 mg/kg and 0.50 mg/kg, respectively. Unlike the levels of many other metals, the level of cobalt showed no decline over the 11-year period (Burger & Gochfeld, 1988).

Cobalt concentrations in coal, crude oil, fuel oil, and gasoline in the United States are 5 mg/kg, 0.001–10 mg/kg, 0.03–0.3 mg/kg, and <0.1 mg/kg, respectively (Smith & Carson, 1981).

The largest potential source of cobalt exposure for the general population is food. Most of the cobalt that is ingested is inorganic. Vitamin B₁₂ contains cobalt but occurs in foods of animal origin and represents only a small fraction of cobalt intake. Green vegetables and fresh cereals are the richest sources of cobalt (0.2–0.6 µg/g dry mass), whereas dairy products, refined cereals, and sugar contain the least cobalt (0.01–0.03 µg/g dry mass) (IARC, 1991; Cobalt Development Institute, 2003). Plant products have been estimated to contribute up to 88% of the total cobalt in the Japanese diet (Yamagata et al., 1963; IARC, 1991). The Total Diet Study in the United Kingdom in 1994 estimated the population average intake of cobalt to be 0.12 mg/day (MAFF, 1997; EVM, 2002). Cobalt intake in the United States has been estimated to be 5–40 μg/day (Jenkins, 1980), with relatively high concentrations of cobalt occurring in fish and vegetables (Barceloux, 1999). In Canada, the estimated average daily intake is 11 μg/day (Dabeka & McKenzie, 1995). Bakery goods/cereals and vegetables contributed most to this daily intake, at 29.8% and 21.9%, respectively. The cobalt intake of Canadian children (age 1–19 years) has been estimated to range from 7 to 14 μg/day (Dabeka & McKenzie, 1995). In France, the estimated average daily intake is 29 μg/day (Biego et al., 1998). Foodstuffs that contributed most to this intake were milk and dairy products (32%), fish/ crustaceans (20%), and condiments/sugar/oil (16%). A study in Sweden from 1983 to 1990 evaluated cobalt levels in various foodstuffs (Jorhem & Sundström, 1993). Cobalt levels were highest in seeds (alfalfa seeds, 0.86 μg/g fresh weight; linseed, 0.56 μg/g), beef liver (0.043 μg/g), and milk chocolate (0.34 μg/g), whereas fish, fruit, and leafy vegetables contained <0.01 μg/g fresh weight. In Spain, cobalt concentrations in 20 brands of beer ranged from 0.16 to 0.56 μg/l, with a median concentration of 0.39 μg/l (Cameαn et al., 1998). The cobalt content of five brewed teas averaged 0.2 µg/g (range 0.16–0.34 µg/g), and that of seven brewed coffees, 0.75 µg/g (range 0.42–2.0 µg/g) (Horwitz & van der Linden, 1974).

Tobacco contains cobalt at <0.3–2.3 μg/g dry weight, and 0.5% of the cobalt is present in mainstream smoke (Munita & Mazzilli, 1986; Ostapczuk et al., 1987; Stebbins et al., 1992; Barceloux, 1999).

Occupational exposure to cobalt occurs in several industries, including hard metal manufacturing, welding, and grinding. Air concentrations of cobalt in occupational settings generally range from 1.0 × 10⁴ to 1.7 × 10⁶ ng/m³ (IARC, 1991; Barceloux, 1999).

Inhalation of cobalt particles results in deposition in the upper and lower respiratory tract (Casarett & Doull, 1986). Particle size is the primary factor determining deposition patterns. Large particles (diameter >2 μm) deposit in the upper respiratory tract. High airstream velocities promote the inertial impaction of these large particles. Smaller particles tend to escape this inertial impaction and deposit in the lower respiratory tract, where sedimentation and diffusion can occur. Fractional deposition varies due to particle size and the age and breathing patterns of the exposed individual. Fractional deposition of cobalt oxide in humans varied from approximately 50% of the inhaled dose for particles with a geometric mean diameter of 0.8 μm to approximately 75% for particles with a geometric mean diameter of 1.7 μm (Foster et al., 1989). Studies in hamsters suggest that the lungs absorb approximately 30% of an inhaled dose of cobalt oxide (Wehner et al., 1977).

Transfer pathways in humans and laboratory animals have been studied using ⁵⁷Co in the form of cobalt oxide (Bailey et al., 1989). Cobalt particles deposited in the respiratory tract can be absorbed into the blood after dissolution or mechanically transferred to the gastrointestinal tract by mucociliary action and swallowing. Approximately 50% of the cobalt that enters the gastrointestinal tract will be absorbed. Large particles (>2 μm) tend to deposit in the upper respiratory tract, where mechanical clearance processes occur more readily than translocation. Smaller particles that deposit in the lower respiratory tract will usually remain dissolved or be phagocytosed by macrophages and then translocated. The ratio of translocation to mechanical clearance in humans is 5:1 for particle sizes ranging from 0.8 to 1.7 μm (Foster et al., 1989).

Cobalt oxide (using ⁵⁷Co tracer) was found to persist in the respiratory tracts of humans at half the original lung burden after 6 months of exposure (Bailey et al., 1989). In contrast, rats exhibited nearly complete clearance after 6 months. Since cobalt may bind to cellular components in human lung, the elimination half-time in human lung increases with increasing times after exposure (Sedlet et al., 1958; Foster et al., 1989).

Gastrointestinal absorption of cobalt in humans has been found to vary from 18% to 97% of the administered dose, depending on the type and dose of the cobalt compound and the nutritional status of the individual (Harp & Scoular, 1952; Valberg et al., 1969; Sorbie et al., 1971; Smith et al., 1972). Studies of the absorption of cobalt chloride in volunteers indicate that the absorption rate from the gastrointestinal tract ranges from 5% to >20% between doses of <1 μg and 1.2 mg cobalt (Smith et al., 1972). Cobalt absorption was increased among individuals who were iron deficient (31–71% absorption in iron-deficient subjects, 18–44% in controls) (Valberg et al., 1969; Sorbie et al., 1971). The absorption of vitamin B₁₂ occurs by a complex, yet specific pathway that involves the interaction of the molecule with factors in the stomach and intestine that facilitate absorption (Russell-Jones & Alpers, 1999).

Data on absorption via the gastrointestinal tract are available from animal experiments. Several rat studies have found that soluble cobalt chloride was 13–34% absorbed, whereas insoluble cobalt oxides were only 1–3% absorbed (Taylor, 1962; Barnaby et al., 1968; Schade et al., 1970; Hollins & McCullough, 1971; Bailey et al., 1989; Collier et al., 1989; Patrick et al., 1989; Kirchgessner et al., 1994; Ayala-Fierro et al., 1999). Particle size did not affect gastrointestinal absorption in baboons, guinea-pigs, HMT rats, F-344 rats, hamsters, or CBA/H mice (Bailey et al., 1989). In rats, cobalt chloride (with ⁵⁸Co tracer) that was complexed with histidine, lysine, glycylglycine, EDTA, casein, or glycine was absorbed less than free cobalt chloride (Taylor, 1962). Cobalt chloride administered in conjunction with cow’s milk resulted in significantly greater gastrointestinal absorption (~40%) (Taylor, 1962). Water-soluble cobalt compounds have been found to exhibit greater absorption than non-water-soluble forms (Kinoshita & Fujita, 1972; Inaba et al., 1980; Deka et al., 1981; Firriolo et al., 1999). As in humans, iron deficiency in animals increased cobalt absorption, while simultaneous administration of cobalt and iron resulted in less cobalt absorption (Schade et al., 1970; Reuber et al., 1994). As oral cobalt doses increase, fractional absorption decreases (Houk et al., 1946; Taylor, 1962; Kirchgessner et al., 1994). Rats and guinea-pigs aged 1–60 days have 3- to 15-fold greater absorption than adult animals aged 200 days or more (Naylor & Harrison, 1995). Species differences in absorption rates have not been observed; however, absorption of soluble cobalt compounds is greater in rats (13–34%) than in cows (1–2%) or guinea-pigs (4–5%) (Taylor, 1962; Barnaby et al., 1968; Schade et al., 1970; Hollins & McCullough, 1971; van Bruwaene et al., 1984; Bailey et al., 1989; Kirchgessner et al., 1994; Naylor & Harrison, 1995; Ayala-Fierro et al., 1999).

Since cobalt is an essential metal and a component of vitamin B₁₂, it has been found in most tissues, such as muscle, lung, lymph nodes, heart, skin, bone, hair, stomach, brain, pancreatic juice, kidneys, plasma, urinary bladder, and liver (highest levels), of non-occupationally exposed subjects (Forbes et al., 1954; Yamagata et al., 1962; Yukawa et al., 1980; Teraoka, 1981; Collecchi et al., 1986; Ishihara et al., 1987; Hewitt, 1988; Muramatsu & Parr, 1988). These tissue levels reflect exposure from all routes and all sources. Total body burden in humans has been estimated as 1.1–1.5 mg, with 0.11 mg in the liver (Yamagata et al., 1962; ICRP, 1979).

Workers exposed occupationally to airborne cobalt had higher tissue levels of cobalt when examined at death. Lung concentrations of cobalt are significantly higher in copper smelter workers, metal workers, and coal miners who were occupationally exposed compared with non-occupationally exposed workers (Teraoka, 1981; Hillerdal & Hartung, 1983; Gerhardsson et al., 1984; Hewitt, 1988). In copper smelter workers, no increases in liver or kidney cobalt levels were observed compared with controls (Gerhardsson et al., 1984). However, metal workers had increased cobalt levels in lymph nodes, liver, spleen, and kidneys (Teraoka, 1981; Hillerdal & Hartung, 1983).

Tissue distribution of cobalt in laboratory animals is similar to that in humans. After inhalation exposure, marked increases of cobalt have been found in the lung (Barnes et al., 1976; Brune et al., 1980; Kreyling et al., 1986; Patrick et al., 1989; Talbot & Morgan, 1989; Collier et al., 1991; Kyono et al., 1992). Histological analysis revealed that cobalt particles were localized to macrophages within the bronchial wall or in the interstitium close to the terminal bronchioli (Brune et al., 1980). Cobalt has been found in significant amounts in the liver, kidney, trachea, spleen, bones, and heart, with the highest levels in the liver and kidney (Wehner & Craig, 1972; Kerfoot, 1975; Barnes et al., 1976; Brune et al., 1980; Kreyling et al., 1986).

Although no studies describe the distribution of cobalt after oral exposure in humans, laboratory animal studies indicate that cobalt absorbed in the gastrointestinal tract is primarily retained in the liver (Simesen, 1939; Greenberg et al., 1943; Ayala-Fierro et al., 1999). Cobalt was also found in the kidneys, heart, stomach, and intestines (Simesen, 1939; Persson et al., 1992; Ayala-Fierro et al., 1999). In pregnant rats, oral exposure to cobalt caused a dose-dependent increase in fetal blood and amniotic fluid (Szakmary et al., 2001). Long-term oral exposure of rats caused significantly increased cobalt levels in the liver, kidney, muscle, brain, and testes (Barnaby et al., 1968; Thomas et al., 1976; Bourg et al., 1985).

Cobalt (as ⁵⁵CoCl₂ and ⁵⁶CoCl₂) administered to two human volunteers by intravenous injection was found to be distributed primarily to the liver and kidneys (Jansen et al., 1996). In rats, intravenous injection of ⁵⁷CoCl₂ resulted in cobalt accumulation in the liver (22.8%), kidneys (10.2%), and intestines (3.16%) 2 h after exposure (Gregus & Klaassen, 1986). When rats were intracardially injected with cobalt nitrate, similar results were observed: 29% accumulation in the liver, 10% in the kidneys, and 4.6% in the intestines (Patrick et al., 1989). In a rat study in which tissue cobalt levels were determined 100 days after intravenous injection of ⁶⁰CoCl₂, the highest levels were found in spleen, followed by heart and then bone (Thomas et al., 1976). Liver and kidney had the highest initial cobalt concentrations, but concentrations were comparatively low at 100 days. Intramuscular injection of cobalt mesoporphyrin in rats resulted in the highest levels in liver and blood, followed by kidneys, lung, spleen, adrenal glands, and heart, at 7 days after exposure (Feng et al., 1998). Subcutaneous injection of cobalt protoporphyrin resulted in the highest levels in kidney, followed by spleen, liver, lung, thymus, and gonads, at 4 weeks after exposure (Rosenberg, 1993).

In humans, there are no available data on the elimination of soluble cobalt particles after inhalation exposure. Elimination of insoluble cobalt particles after inhalation exposure appears to follow three-phase kinetics. The first phase is the mucociliary clearance of particles deposited in the tracheobronchial region and has a half-time of 2–44 h (Apostoli et al., 1994; Mosconi et al., 1994). The second phase is the macrophage-mediated clearance of lung cobalt particles and has a half-time of 10–78 days (Beleznay & Osvay, 1994; Mosconi et al., 1994). The third phase represents long-term lung clearance and has a half-time on the order of years (Newton & Rundo, 1971; Bailey et al., 1989; Beleznay & Osvay, 1994; Mosconi et al., 1994). In a controlled aerosol exposure study in humans, 40% of the initial lung burden of cobalt oxide (using ⁵⁷Co as a tracer) was retained at 6 months after exposure (Foster et al., 1989). In the first week after exposure, 17% of the initial lung burden was eliminated, of which 90% was mechanically cleared to the gastrointestinal tract and excreted in faeces. By 6 months after exposure, 33% of the initial lung burden was cumulatively eliminated in urine and 28% was cumulatively eliminated in faeces. The ratio of peak absorption rate and mechanical clearance rate is 5:1. Urinary excretion increases with time following exposure. Particle size affects elimination, since more cobalt is mechanically cleared to the gastrointestinal tract when particles are larger (Bailey et al., 1989; Foster et al., 1989). Smokers with no occupational exposure to cobalt had a significantly higher mean concentration in urine (0.6 μg/l; SD 0.6) than non-smokers (0.3 μg/l; SD 0.1). There was no difference between smokers and non-smokers in the cobalt levels in blood (Alexandersson, 1988).

Animal data on cobalt elimination indicate that the solubility of the cobalt compound greatly affects its long-term clearance. For example, the more soluble cobalt(II) oxide is cleared from the lungs at a faster rate than the less soluble cobalt(II,III) oxide (Barnes et al., 1976; Kreyling, 1984). Soluble cobalt compounds are absorbed into the blood at a faster rate than less soluble compounds and excreted in the urine and faeces (Barnes et al., 1976). Urinary excretion rates seem to correlate with the translocation rate of cobalt from the lungs to blood, whereas faecal excretion rates seem to correlate with mechanical clearance rates of cobalt from the lungs to the gastrointestinal tract (Kreyling et al., 1986, 1989; Andre et al., 1989; Bailey et al., 1989; Collier et al., 1989; Patrick et al., 1989; Talbot & Morgan, 1989). Following an initial high rate of faecal clearance, urinary excretion is the primary route of cobalt elimination after a single inhalation exposure or 3 months of exposure (Palmes et al., 1959; Kerfoot, 1975).

Faecal elimination is the primary route of excretion following oral exposure in humans. Faecal elimination has been found to vary (3–99% of the dose) depending on the amount and type of cobalt administered and the nutritional status of the subject (Harp & Scoular, 1952; Paley et al., 1958; Valberg et al., 1969; Sorbie et al., 1971; Smith et al., 1972). Several days after an oral exposure, 10 times more cobalt was excreted in faeces than in urine (Paley et al., 1958). In subjects with an iron deficiency, less cobalt was eliminated in the faeces, and more was absorbed (Valberg et al., 1969; Sorbie et al., 1971).

Faecal elimination is also the primary route of excretion in animals following oral exposure. Faecal clearance has been noted to decrease as cobalt particle solubility increases. In several species, oral exposure to cobalt(II,III) oxide (with ⁵⁷Co tracer) resulted in little gastrointestinal absorption and a rapid elimination in faeces (>96%) (Bailey et al., 1989). No significant differences in cobalt(II,III) oxide elimination were observed among species (Andre et al., 1989; Bailey et al., 1989; Collier et al., 1989; Patrick et al., 1989; Talbot & Morgan, 1989). Cobalt(II) chloride, which is more soluble, was excreted primarily via faeces (70–83% of the administered dose) in rats, with urinary excretion accounting for the remainder of the dose (Barnaby et al., 1968; Hollins & McCullough, 1971; Ayala-Fierro et al., 1999). In lactating dairy cows, 97% of an oral dose of cobalt chloride was recovered in the faeces by 70 days after exposure, whereas urine and milk contained 0.26% and 0.012% of the dose, respectively (van Bruwaene et al., 1984). Single exposures in beagle dogs demonstrated that insoluble cobalt(II,III) oxide was eliminated in the faeces and urine at 90% and 5%, respectively, while the more soluble cobalt nitrate was eliminated at 70% in the faeces and 25% in the urine (Kreyling et al., 1986). Similar to humans, iron deficiency in rats also caused less elimination in the faeces, whereas co-administration of iron caused an increase in faecal elimination (Schade et al., 1970; Reuber et al., 1994).

After an intravenous injection of cobalt chloride in humans, 30% of the dose was excreted in urine within 24 h, 56–73% within 48 h, and 57% within 2 weeks (Kent & McCance, 1941; Paley et al., 1958; Smith et al., 1972). In various animal species, urinary excretion has also been shown to be the primary elimination route following intravenous injection of cobalt nitrate (Andre et al., 1989; Bailey et al., 1989; Collier et al., 1989; Patrick et al., 1989; Talbot & Morgan, 1989). In animals, 80% of the dose was excreted via urine within 21 days. Most of the remaining dose (5–30% of the total dose) was excreted in the faeces, with little long-term retention. Biliary excretion has also been reported in animals, at 2–7% of the injected dose (Sheline et al., 1945; Cikrt & Tich, 1981; Gregus & Klaassen, 1986).

The ICRP has developed two physiologically based pharmacokinetic/pharmacodynamic models that are applicable to cobalt: a human respiratory tract model for radiological protection (ICRP, 1994) and a biokinetic model of ingested cobalt in humans (ICRP, 1979, 1994). The PBPK model for the human respiratory tract was developed for a wide variety of radionuclides and their chemical forms. It models the behaviour of aerosols and vapours in the respiratory tract. It provides inhalation dose coefficients for estimating the committed dose equivalents and the effective doses to organs and tissues based on a unit intake of radioactive material, the distribution and retention of the material, the radioactive decay, and the energy of radiation emitted from the material and absorbed by tissues. This model applies to various particle sizes (0.0005–100 μm in diameter) and can be adjusted for population characteristics such as sex, age, and level of physical activity. Inhaled particles may be redistributed either upwards into the respiratory tract or to the lymph and blood by particle removal mechanisms. Deposition of vapours and gases is modelled as a partitioning process and is based on physiological parameters and the solubility and reactivity of the compound of interest. The solubility and reactivity of compounds are classified into three categories: SR-0, insoluble and non-reactive gases; SR-1, soluble and reactive gases and vapours that are expected to be taken up and deposited on respiratory tract tissues; and SR-2, soluble and reactive gases and vapours that are completely retained in the extrathoracic regions of the respiratory tract. This model also accounts for mechanical particle clearance and is based primarily on human data, although particle retention in airway walls is based on data from experimental animals. This model assumes that blood absorption occurs at equivalent rates in all areas of the respiratory tract, except for the anterior nasal passages, where absorption does not occur. Particles undergo dissociation, after which the dissolved molecules diffuse across capillary walls into the blood. Absorption is classified into four types: Type V (complete and instantaneous absorption), Type F (fast, 100% absorption within 10 min), Type M (medium, 70% absorption within 10 min), and Type S (slow, 0.1% absorption within 10 min). Cobalt compounds have been classified as follows: Type F, cobalt chloride and nitrate; Type M or S, cobalt oxides, cobalt metal, and metal alloys; Type M, cobalt in mineral dusts, such as fly ash and volcano ash, and all cobalt aerosols in the absence of specific information; and Type S, cobalt-infused aluminosilicate or polystyrene.

The ICRP’s cobalt biokinetics model (ICRP, 1994) is a three-compartment model for ingested cobalt that is applicable to infants, children, adolescents, and adults. Absorption of ingested cobalt is assumed to be 60% in infants up to 3 months of age, 30% from 3 months to 15 years of age, and 10% after 15 years of age. The distribution of cobalt is assumed to be 50% excreted in the urine and faeces at a 6:1 ratio, 5% in the liver, and 45% in other tissues. Elimination from tissues is assumed to follow three first-order rate constants that represent slow, medium, and fast, with half-times of 6, 80, and 600 days, respectively. These half-times are assumed to be independent of age. The validation of this model is not described by the ICRP, but the model has been used to establish radiation dose equivalents (Sv/Bq) of ingested ⁵⁷Co, ⁵⁸Co, and ⁶⁰Co for ages 3 months to 70 years (ICRP, 1994). This model is designed for human dosimetry and would need modification for other species. Radiation doses from cobalt radionuclides to all major organs can be estimated using this model and can be used to assess environmental and occupational exposures to radioactive cobalt.

Analysis of urinary cobalt has been recommended for biological monitoring of exposure to cobalt at work (Templeton, 1996), and different institutions have proposed biomonitoring action limits for acceptable exposure (ACGIH, 1999; FIOH, 1999). The concentration of cobalt in a urine specimen collected at the end of the last work shift of the week reflects exposure over the preceding work week, and that collected on Monday morning reflects chronic occupational exposure in comparison with the reference population (Templeton, 1996).

Urine and blood cobalt levels correlate positively with occupational exposure to cobalt. Unexposed humans have blood cobalt levels ranging from 0.05 to 0.19 μg/dl and urine cobalt levels ranging from 0.04 to 2 μg/l (Ichikawa et al., 1985; Alexandersson, 1988). Workers exposed to 0.1 mg/m³ reported blood levels ranging from 0.57 to 0.79 μg/dl (95% CI), compared with 0.19 μg/dl in unexposed subjects (Ichikawa et al., 1985). This study also reported urine cobalt levels of 59–78 μg/l in the workers compared with 2 μg/l in unexposed subjects (Ichikawa et al., 1985).

The LC₅₀ for a 30-min inhalation exposure of rats was 165 mg/m³ as cobalt hydrocarbonyl (Palmes et al., 1959). One of 14 Syrian golden hamsters exposed by inhalation to cobalt oxide at 106 mg/m³ for 3 h died within 24 h; after a 6-h exposure, the 24-h mortality rate was 2/44 (Wehner & Craig, 1972). Oral LD₅₀ values are dependent on the type of cobalt compound tested and the test species. Wistar rats and Sprague-Dawley rats had LD₅₀ values ranging from 42.4 mg of cobalt (as cobalt chloride) per kilogram body weight to 317 mg of cobalt (as cobalt carbonate) per kilogram body weight (FDRL, 1984a, 1984b, 1984c; Singh & Junnarkar, 1991). Tricobalt tetraoxide, an insoluble compound, had an LD₅₀ in Sprague-Dawley rats of 3672 mg of cobalt per kilogram body weight (FDRL, 1984c). Speijers et al. (1982) reported an LD₅₀ of 418 mg/kg body weight for cobalt chloride in Wistar rats. In male Swiss mice, LD₅₀ values ranged from 89.3 mg of cobalt (as cobalt chloride) per kilogram body weight to 123 mg of cobalt (as cobalt sulfate) per kilogram body weight (Singh & Junnarkar, 1991).

Rats and mice exposed by inhalation to cobalt sulfate heptahydrate at cobalt concentrations of 19 and 1.9 mg/m³, respectively, for 16 days exhibited necrosis and inflammation of the respiratory tract epithelium. Rats also developed thymus necrosis and testicular atrophy (Bucher et al., 1990; NTP, 1991). Male CFY rats exposed orally to cobalt chloride at 50 mg/kg body weight per day (equivalent to 12.4 mg of cobalt per kilogram body weight per day) for 3 weeks and co-exposed to drinking-water that contained 10% ethanol and 5% sugar exhibited cardiac damage that presented as incipient, multifocal myocytolysis with degeneration of myofibrils (Morvai et al., 1993). Rats exposed to ultrafine cobalt particles (diameter 20 nm) at concentrations of 2.72 mg/m³ for 5 h or 2.12 mg/m³ for 5 h/day for 4 days displayed focal hypertrophy or proliferation of lower airway epithelium, macrophage damage, intracellular oedema of type I alveolar epithelium, interstitial oedema, and proliferation of type II alveolar epithelium (Kyono et al., 1992).

Rats (strain not specified), guinea-pigs (strain not specified), and beagle dogs exposed to cobalt (as cobalt hydrocarbonyl) at 9 mg/m³ for 6 h/day, 5 days/week, for 3 months displayed foam cell aggregates (Palmes et al., 1959). These foam cell aggregates were composed of nodules of large macrophages with foamy cytoplasm, accompanied by moderate interstitial and peribronchial fibrosis, mild emphysema, and moderate peribronchial lymphoid hyperplasia. These aggregates were not present when animals were sacrificed and evaluated at 3 months or 6 months post-exposure. Rabbits exposed by inhalation for 1–4 months to cobalt chloride (0.4–2 mg/m³) exhibited lesions of the alveolar region of the respiratory tract that were characterized by nodular accumulation of Type II epithelial cells and interstitial inflammation (Johansson et al., 1984, 1987, 1991, 1992).

F344/N rats and B6C3F1 mice exposed by inhalation to cobalt sulfate heptahydrate (0, 0.3, 1, 3, 10, and 30 mg/m³; equivalent to cobalt concentrations of 0, 0.11, 0.38, 1.14, 3.80, and 11.38 mg/m³) for 6 h/day, 5 days/ week, for 13 weeks developed adverse effects throughout the respiratory tract (Bucher et al., 1990; NTP, 1991). At concentrations >0.3 mg/m³ (cobalt concentrations >0.11 mg/m³), both rats and mice developed squamous metaplasia of the larynx (the most sensitive tissue), such that a NOAEC could not be determined. Rats developed chronic inflammation of the larynx at >1 mg/m³ and more severe effects in the nose, larynx, and lung at higher exposures. Mice exhibited acute inflammation of the nose at >3 mg/m³ and more severe effects in the nose, larynx, and lung at higher exposures. At 30 mg/m³, mice exhibited hyperplasia of the mediastinal lymph nodes and testicular atrophy and increased estrous cycle length in females. Both rats and mice exhibited histiocytic infiltrates of the lung at similar exposure levels. Sperm motility was decreased in mice exposed to 3 mg/m³ or higher (lower exposures not assessed), and increased abnormal sperm and decreased testis and epididymal weights were observed in mice exposed to 30 mg/m³.

Rats exposed for 2–3 months to cobalt at 26–30.2 mg/kg body weight per day in the diet (as cobalt sulfate) or in drinking-water (as cobalt chloride) exhibited increased heart weight and degenerative heart lesions (Grice et al., 1969; Domingo et al., 1984). Rats exposed to cobalt (as cobalt sulfate) in the diet at 8.4 mg/kg body weight per day for 24 weeks experienced significant reductions in cardiac enzyme activity levels, such as manganese superoxide dismutase, succinate cytochrome c oxidase, NADH cytochrome c reductase, and cytochrome c oxidase, and a reduction in mitochondrial ATP production (Clyne et al., 2001). Rats exposed to cobalt (as cobalt chloride) at 10–18 mg/kg body weight per day for 4–5 months exhibited renal injury, such as histological alteration of proximal tubules (Holly, 1955; Murdock, 1959).

A study by the NTP examined the carcinogenicity of cobalt by inhalation in mice (NTP, 1998; Bucher et al., 1999). Groups of 50 male and 50 female B6C3F1 mice were exposed to cobalt sulfate heptahydrate at 0, 0.3, 1, or 3 mg/m³ for 6 h/day, 5 days/week, for 105 weeks. Cobalt concentrations in this study were 0, 0.11, 0.38, 1.14, and 3.80 mg/m³. Mean body weights were increased in all treated females and decreased only in the high-dose males. Survival was not adversely affected by treatment. The incidences of benign and malignant alveolar/bronchiolar neoplasms were increased in a concentration-dependent manner: males, 11/50, 14/50, 19/50, and 28/50 for 0, 0.3, 1, and 3 mg/m³, respectively; females, 4/50, 7/50, 13/50, and 18/50 for 0, 0.3, 1, and 3 mg/m³, respectively. There were no increased incidences of neoplasms in other tissues. The NTP concluded that there was clear evidence of carcinogenic activity.

Another study by the NTP examined the carcinogenicity of cobalt by inhalation in rats (NTP, 1998; Bucher et al., 1999). Groups of 50 male and 50 female Fischer 344/N rats were exposed to cobalt sulfate heptahydrate at 0, 0.3, 1, and 3 mg/m³ (equivalent to cobalt concentrations of 0, 0.11, 0.38, 1.14, and 3.80 mg/m³) for 6 h/day, 5 days/week, for 105 weeks. Mean body weights and survival were unaffected by treatment. Rats exhibited a concentration-related increase in the incidence of benign and malignant alveolar/bronchiolar neoplasms in male and female rats and benign and malignant pheochromocytomas in female rats. The incidences of benign and malignant alveolar/bronchiolar neoplasms were 1/50, 4/50, 4/48, and 7/50 for 0, 0.3, 1, and 3 mg/m³, respectively, in males and 0/50, 3/49, 16/50, and 16/50 for 0, 0.3, 1, and 3 mg/m³, respectively, in females. Although many of the alveolar/bronchiolar lesions were morphologically similar to those that arise spontaneously, the lesions in rats, unlike those in mice, were predominantly fibrotic, squamous, or mixtures of alveolar/bronchiolar epithelium and squamous or fibrous components. Squamous metaplasia of alveolar/bronchiolar epithelium, which is a common response to pulmonary injury, was observed in a number of rats. In females, incidences of benign and malignant pheochromocytomas of the adrenal medulla were 2/48, 1/49, 4/50, and 10/48 for 0, 0.3, 1, and 3 mg/m³, respectively. In males, the incidences of benign and malignant pheochromocytomas of the adrenal medulla were 15/50, 19/50, 25/50, and 20/50 for 0, 0.3, 1, and 3 mg/m³, respectively. Pheochromocytomas are common spontaneous neoplasms in male Fischer F344/N rats, but have a lower spontaneous occurrence in females. There were no increased incidences of neoplasms in other tissues. The NTP concluded that there was some evidence of carcinogenic activity in male rats, but clear evidence in female rats.

Syrian golden hamsters (51 per group) exposed by inhalation to cobalt (as cobalt oxide) at 10.0 mg/m³ for 7 h/day, 5 days/week, for a lifetime developed emphysema, but the incidence of pulmonary tumours was not different from controls. While tobacco smoke exposure induced pulmonary tumours in 14/51 animals, the incidence in animals exposed to both tobacco smoke and cobalt oxide was 11/51 (Wehner et al., 1977).

Steinhoff & Mohr (1991) conducted a study of rats exposed to a cobalt–aluminium–chromium spinel, with an empirical formula Co(II) 0.66, Al 0.7, Cr(III) 0.3, and O 3.66 (80% of the particles < 1.5 μm), or to cobalt(II) oxide. Groups of 50 male and 50 female Sprague-Dawley rats were exposed by intratracheal instillations of the spinel in saline at 10 mg/kg body weight every 2 weeks for 18 treatments, followed by every 4 weeks from the 19th to the 30th treatments for a total of 2 years. The rats were allowed to live until a natural death, or they were sacrificed when moribund. Alveolar/bronchiolar proliferation was not observed in 100 untreated controls and 100 saline controls; however, 61/100 rats exhibited this effect in the spinel treatment group. Likewise, no pulmonary tumours were observed in the untreated or saline controls. In the spinel-treated groups, one male and two female rats exhibited squamous cell carcinomas. When cobalt(II) oxide was administered by intratracheal instillation at doses of 2 mg/kg body weight (total dose 78 mg/kg body weight) or 10 mg/kg body weight (total dose 390 mg/kg body weight), there were two benign pulmonary tumours among the 100 rats in the low-dose group and two benign and four malignant pulmonary tumours among the 100 rats in the high-dose group. Steinhoff & Mohr (1991) also administered subcutaneous doses of 5 × 2 and 1 × 10 mg of cobalt(II) oxide per kilogram body weight per week, and 5/10 and 4/10 rats, respectively, developed local malignant tumours in a lifetime study. In a related study (Steinhoff & Mohr, 1991), groups of 10 male and 10 female rats were administered three intraperitoneal injections of saline or cobalt–aluminium–chromium spinel powder at 2-month intervals for a total dose of 600 mg/kg body weight. The rats were observed for their natural life span or were sacrificed when moribund. Malignant peritoneal tumours occurred in 1/20 controls (histiocytoma) and 2/20 spinel-treated rats (one histiocytoma and one sarcoma). After intraperitoneal administration of 3 × 200 mg of cobalt(II) oxide per kilogram body weight, 14/20 rats developed malignant intraperitoneal tumours.

Heath (1954, 1956, 1960) treated groups of 10 male and 10 female hooded rats with a single intramuscular injection of 28 mg of cobalt metal powder. The cobalt metal particles ranged in size from 3.5 μm Χ 3.5 μm to 17 μm Χ 12 μm, with large numbers of long narrow particles 10 μm Χ 4 μm. The rats were injected in the thigh. The observation period was 122 weeks, during which 4/10 male and 5/10 female rats developed sarcomas, mostly rhabdomyosarcomas, at the injection site. In a related study, 80 female hooded rats (divided into three groups of 16, 14, and 50) were intramuscularly injected with 28 mg of wear particles (ground artificial hip or knee prostheses composed of a cobalt–chromium–molybdenum alloy) (Heath et al., 1971; Swanson et al., 1973). No control group was reported. Animals were observed for up to 29 months. The incidences of sarcomas at the injection site were 3/16, 4/14, and 16/50. Half the tumours were rhabdomyosarcomas, and the remainder were fibrosarcomas. In a related study, Heath & Daniel (1962) injected two groups of 10 female hooded rats with 28 mg cobalt metal powder (3.5 μm Χ 3.5 μm to 17 μm Χ 12 μm, with large numbers of long narrow particles 10 μm Χ 4 μm) through the right dome of the diaphragm (first group) or through the fourth left intercostal space (second group). Animals were observed for up to 28 months. Of the diaphragm-treated rats, 6/10 died within 3 days, and in the rats injected through the intercostal space, 2/10 died within 3 days. Of the 12 rats that survived the injection, 4 developed intrathoracic sarcomas. Three of these sarcomas were of mixed origin and included rhabdomyosarcomatous elements, while the fourth rhabdomyosarcoma arose in the intercostal muscle.

Meachim et al. (1982) conducted a follow-up study to Heath et al. (1971) and Swanson et al. (1973). Female Wistar rats (n = 51) received intramuscular implants of 28 mg coarse particles (100–250 μm diameter) of ground cobalt–chromium–molybdenum alloy, and 61 Wistar and 53 hooded rats received implants of fine particles (0.5–50 μm). The rats were observed for life. Survival at 2 years was 11/41 for Wistar rats receiving coarse particles, 7/61 for Wistar rats receiving fine particles, 0/53 for hooded rats receiving fine particles, and 5/50 for Wistar controls. No tumours were observed at the implantation sites. Meachim et al. (1982) conducted a similar study in a group of 46 female Dunkin Hartley guinea-pigs that received intramuscular implants of 28 mg of fine particles of ground cobalt–chromium–molybdenum alloy. At 3 years, 12/46 animals were alive. No tumours were reported; however, nodular fibroblastic hyperplasia was observed in eight animals at the implantation site.

Mitchell et al. (1960) implanted a cobalt–chromium–molybdenum pellet (Vitallium alloy) subcutaneously into five male and five female Wistar rats. Rats were observed for up to 27 months, and no sarcomas were reported.

Memoli et al. (1986) implanted seven different test materials containing cobalt alloyed with chromium and nickel, molybdenum, tungsten, and/or zirconium into the femoral bone of groups of 10–17 male and 8–15 female Sprague-Dawley rats. The test materials were small rods (1.6 mm diameter and 4 mm length), powders, or porous compacted wire. The rats were observed for up to 30 months. Untreated and sham-operated controls, consisting of groups of 13 male and 13 female rats, were also studied. Sarcomas at the site of implantation were observed in 1/18 rats given a cobalt alloy powder (41% cobalt) and 3/26 rats given a nickel–cobalt-based powder (51% cobalt). No tumours were observed in two groups of 25 rats given rods with 69% or 47% cobalt, in two groups of 26 rats given rods with 0.11% or 33% cobalt, or in the untreated and sham-operated controls.

Vollmann (1938) implanted metallic cobalt dust into the femoral cavity of two groups of 15–20 rabbits. No tumours were observed at 3 years post-implantation. A follow-up study of these survivors at 6 years revealed sarcomas at the site of implantation in two cobalt-treated rabbits (Schinz & Uehlinger, 1942).

Jasmin & Riopelle (1976) injected groups of 20 and 18 female Sprague-Dawley rats with 5 mg of metallic cobalt powder or cobalt sulfide powder, respectively, into each pole of the right kidney. After 12 months, necropsies were conducted, and no tumours were observed in the kidneys of treated or control rats.

There are no available studies on genotoxic effects in animals exposed by inhalation. Male Swiss mice administered a single oral dose of cobalt (as cobalt chloride) at 0, 4.96, 9.92, or 19.8 mg/kg body weight exhibited a dose–response increase in percentages of chromosomal breaks and chromosomal aberrations in bone marrow cells (Palit et al., 1991a, 1991b, 1991c, 1991d). A single intraperitoneal injection of cobalt (as cobalt(II) chloride) at 12.4 or 22.3 but not 6.19 mg/kg body weight in BALB/c mice caused an increase in micronucleus formation after 30 h (Suzuki et al., 1993). F344 rats injected intraperitoneally with cobalt at 3 or 6 mg/kg body weight exhibited increased levels of oxidatively damaged DNA bases in the liver, kidney, and lung at 2 and 10 days following injection (Kasprzak et al., 1994).

Cobalt, in compounds with a valence state of +2, was mostly negative in mutagenicity tests conducted in Salmonella typhimurium, Escherichia coli, and yeast, but weakly positive in Bacillus subtilis (Kanematsu et al., 1980; Tso & Fung, 1981; Fukunaga et al., 1982; Singh, 1983; Arlauskas et al., 1985; Kharab & Singh, 1985; Ogawa et al., 1986). The only positive report for cobalt(II) is from S. typhimurium TA100 both with and without liver S9 metabolic enzymes (NTP, 1998). S. typhimurium strains TA98 and TA1535 were negative. Cobalt(II) compounds caused genetic conversions in S. cerevisiae (Fukunaga et al., 1982; Singh, 1983; Kharab & Singh, 1985). The reasons for this dichotomy in yeast are unknown. Cobalt in compounds with a valence state of +3 was positive in S. typhimurium and E. coli (Schultz et al., 1982).

In mammalian test systems, many cobalt compounds and metals are genotoxic. Cobalt compounds and cobalt metals have been reported to cause clastogenic effects in mammalian cells such as human lymphocytes (Painter & Howard, 1982; Hamilton-Koch et al., 1986; Anard et al., 1997), transformation in hamster cells (Costa et al., 1982), sister chromatid exchanges in human lymphocytes (Andersen, 1983), and micronucleus formation in mouse bone marrow cells (Suzuki et al., 1993), human lymphocytes (Capomazza & Botta, 1991; Olivero et al., 1995; van Goethem et al., 1997), and rat type II epithelial lung cells (DeBoeck et al., 2003). Cobalt particles are genotoxic in vitro in human peripheral blood mononucleated cells (Anard et al., 1997; van Goethem et al., 1997; De Boeck et al., 1998, 2003). In general, hard cobalt metal is more genotoxic than other cobalt compounds in in vitro test systems.

A study by the NTP that examined the carcinogenicity of cobalt sulfate heptahydrate by inhalation in B6C3F1 mice (NTP, 1998; Bucher et al., 1999) (see section 8.4) also evaluated K-ras mutation frequency and spectra in lung tumours. A higher frequency (5/9; 55%) of G to T transversions was detected in codon 12 of K-ras compared with chamber controls (0/1) or historical controls (1/24). G to T transversions are common DNA changes associated with active oxygen species. This provides supportive evidence that cobalt sulfate heptahydrate may indirectly damage DNA by oxidative stress.

8.6.1 Effects on fertility

Both rats exposed to cobalt (as cobalt chloride) at 13.3–58.9 mg/kg body weight per day for 2–3 months in drinking-water or diet (Nation et al., 1983; Domingo et al., 1984; Corrier et al., 1985; Mollenhauer et al., 1985; Pedigo et al., 1988; Pedigo & Vernon, 1993) and mice exposed to cobalt (as cobalt chloride) at 43.4 mg/kg body weight per day for 13 weeks in drinking-water exhibited testicular degeneration and atrophy (Anderson et al., 1992, 1993).

In an abstract reported by Elbetieha et al. (2004), sexually mature male mice exposed to cobalt(II) chloride at 200, 400, or 800 mg/l in their drinking-water for 12 weeks were assessed for effects on fertility by breeding these exposed males to unexposed females. Fertility, as measured by successful matings, was reduced in mice exposed to cobalt chloride at 400 and 800 mg/l (internal doses of 46.91 ± 4.78 and 93.01 ± 6.76 mg/kg body weight per day, respectively). The number of implantation sites was significantly reduced in females mated with exposed males at 400 and 800 mg/l. The number of viable fetuses was decreased in females mated with males at all three exposure levels. In the 800 mg/l males, absolute epididymal weight was significantly decreased, whereas relative and absolute testes weights were decreased in males exposed to both 400 and 800 mg/l. Epididymal sperm count was decreased in males of all three exposure levels. At 400 and 800 mg/l, males also exhibited reduced testicular sperm counts and daily sperm production. The testes displayed severe abnormalities, including hypertrophy of the interstitial Leydig cells, congested blood vessels, degeneration of the spermatogonial cells, and necrosis of seminiferous tubules and interstitial tissue.

In a study in which B6C3F1 mice were exposed by inhalation to cobalt sulfate heptahydrate (0, 0.3, 1, 3, 10, and 30 mg/m³; equivalent to cobalt concentrations of 0, 0.11, 0.38, 1.14, 3.80, and 11.38 mg/m³) for 6 h/day, 5 days/week, for 13 weeks, testicular atrophy in males and increased estrous cycle length in females were observed at 30 mg/m³. Sperm motility was decreased in mice exposed to 3 mg/m³ or higher (lower exposures not assessed), and increased abnormal sperm and decreased testis and epididymal weights were observed in mice exposed to 30 mg/m³ (Bucher et al., 1990; NTP, 1991) (see also section 8.3).

8.6.2 Developmental toxicity

Oral exposure of female rats to cobalt (as cobalt chloride) at doses of 5.4 or 21.8 mg/kg body weight per day from gestation day 14 to lactation day 21 caused newborn pups to exhibit stunted growth and decreased survival. However, these effects occurred at exposures that also caused maternal toxicity, such as reduced body weight, reduced food consumption, and altered haematological measurements. No teratogenic effects were observed (Domingo et al., 1985). Another study reported that exposure of pregnant rats to cobalt (as cobalt sulfate) at 0–38 mg/kg body weight per day did not affect fetal death rates, maternal body weight gain, average litter size, or average fetal and placental weights. However, a dose-related increase was noted in the percentage of fetuses with retarded body weights (Szakmary et al., 2001). In contrast, Paternain et al. (1988) found no effects on fetal growth or survival after exposing rats to cobalt (as cobalt chloride) at 24.8 mg/kg body weight per day during gestation days 6–15. Exposure of pregnant mice to cobalt (as cobalt sulfate) at 19 mg/kg body weight per day also did not affect litter size, postimplantation loss, or average fetal and placental weights (Szakmary et al., 2001). Rabbits exposed to cobalt (as cobalt sulfate) at doses of >38 mg/kg body weight per day exhibited complete maternal lethality and fetal loss. At 7.6 mg/kg body weight per day, rabbits had increased mortality, fetal resorption, and number of fetuses with retarded body weight (Szakmary et al., 2001).

Dermal exposures on 3 consecutive days to cobalt(II) chloride (in dimethylsulfoxide) caused an increase in cellular proliferation in the lymph node assay in mice (10.8, 27, or 54.1 mg of cobalt per kilogram body weight per day), rats (9.6 or 19.2 mg of cobalt per kilogram body weight per day), and guinea-pigs (14.7 mg of cobalt per kilogram body weight per day) (Ikarashi et al., 1992a, 1992b).

Several studies have demonstrated that a hard metal alloy of tungsten carbide and cobalt matrix is more toxic than either tungsten carbide or cobalt alone. In a proposed mechanism, tungsten carbide facilitates the oxidation of cobalt metal to ionic cobalt (Co²⁺) by transferring electrons from the cobalt atom to molecular oxygen (Lison et al., 1995, 1996). This causes an increase in the solubility of cobalt, relative to cobalt metal, and the generation of reactive oxygen species. The ionic cobalt may be transported by blood throughout the body, causing adverse effects by the generation of reactive oxygen species. In vitro evidence consists of the ability of hard metal particles to generate substantial levels of oxidant species and cause lipid peroxidation (Lison et al., 1995; Zanetti & Fubini, 1997), which does not occur by cobalt or tungsten carbide alone. In addition, hard metal particles have been shown to increase inducible nitric oxide synthase levels, which is responsive to oxidant stress (Rengasamy et al., 1999).

Cobalt toxicity may also be caused through oxidant-based and free radical-based processes. Exposure to soluble cobalt leads to increased indices of oxidative stress, diminished levels of reduced glutathione, increased levels of oxidized glutathione, activation of the hexose monophosphate shunt, and free radical-induced DNA damage (Lewis et al., 1991; Kasprzak et al., 1994; Zhang et al., 1998; Hoet et al., 2002). In the presence of hydrogen peroxide, cobalt(II) stimulates in vitro formation of 8-hydroxy-2’-deoxyguanosine (Ivancsits et al., 2002). A Fenton-type mechanism causes cobalt to generate oxygen radicals, such as superoxide, in both in vitro and in vivo studies (Moorhouse et al., 1985; Kadiiska et al., 1989; Kawanishi et al., 1994; Lloyd et al., 1997). Exposure of rats and guinea-pigs to cobalt results in liver lipid peroxidation and reduced levels of glutathione, superoxide dismutase, catalase, haem oxygenase, and glutathione peroxidase (Sunderman & Zaharia, 1988; Christova et al., 2001, 2002). Cobalt accumulation in cardiac tissues is believed to stimulate carotid body chemoreceptors, which mimics the action of hypoxia (Di Giulio et al., 1990, 1991; Hatori et al., 1993; Morelli et al., 1994). Cobalt exposure also affects genes that are sensitive to oxidant status, such as hypoxia-inducible factor 1, erythropoietin, vascular endothelial growth factor, catalase, and monooxygenase enzymes (Yasukochi et al., 1974; Dalvi & Robbins, 1978; Legrum et al., 1979; Goldberg et al., 1988; Di Giulio et al., 1991; Goldberg & Schneider, 1994; Ladoux & Frelin, 1994; Semenza et al., 1994; Ho & Bunn, 1996; Bunn et al., 1998; Daghman et al., 1999; Hoet et al., 2002). These effects may also lead to the induction of apoptosis, through either these genes or other pathways (Zou et al., 2001).

Soluble cobalt has been shown to block inorganic calcium channels (Henquin et al., 1983; Moger, 1983; Yamatani et al., 1998). This has been shown to reduce steroidogenesis in isolated mouse Leydig cells (Moger, 1983). Calcium influx in liver cells, pancreatic beta cells, and isolated rat islets is altered by soluble cobalt (Henquin & Lambert, 1975; Henquin et al., 1983; Yamatani et al., 1998). By antagonizing calcium, cobalt may also affect neuromuscular transmissions (Weakly, 1973).

In the past, cobalt used to be added to beer as a defoaming agent. Cobalt was found to accumulate in the hearts of heavy beer drinkers and result in cardiomyopathy (see section 9 below). Microscopic analysis found fragmentation and degeneration of myofibres and aggregates of abnormal mitochondria (Ferrans et al., 1964). Mitochondrial effects result in disturbances in energy production and utilization and may be related to the irreversible chelation of lipoic acids under aerobic conditions by cobalt (Webb, 1962). Lipoic acid is a cofactor for the oxidative decarboxylation of pyruvate to acetyl CoA and of alpha-ketoglutarate to succinate (Lehninger, 1982). In rats treated with cobalt, the myocardium exhibits an impairment of pyruvate and fatty acid oxidation (Wiberg, 1968).

Cobalt ions, in the presence of oxidants such as UV radiation or hydrogen peroxide, can cause increased levels of DNA damage in vitro (Hartwig et al., 1991; Nackerdien et al., 1991; De Boeck et al., 1998). Cobalt is hypothesized to inhibit DNA rep