Mechanisms of Carcinogenesis and Other Relevant Data

The key events related to toxicity and carcinogenicity are thought to include cellular uptake of cobalt, intracellular release of cobalt ions from particles, and immediate and downstream biological responses related to the proposed modes of action. The first step in the carcinogenicity or toxicity process is the release of cobalt ions in vivo. Water-soluble cobalt compounds release cobalt ions into fluids outside the cell, and the ions enter the cell through ion channels within the cell membrane. In contrast, poorly soluble particulate cobalt compounds are taken up by specific organelles (lysosomes) in the cell via a process called endocytosis; cobalt is then solubilized in the acidic environment in the lysosomes, and the ions are released inside the cell. Evidence for cellular uptake of the different forms of cobalt is provided by studies evaluating their solubility in biological fluids in vitro (e.g., in gastric and lysosomal fluids) (see Properties) and in vitro studies measuring levels of cobalt ions within cells (Ortega et al. 2014; Peters et al. 2007; Sabbioni et al. 1994a; Smith et al. 2014).

Although the mechanism(s) of action for cobalt-induced carcinogenic effects are not completely understood, several key events have been identified that are related to biologically plausible modes of action and are applicable to all cobalt forms that release cobalt ions in vivo. These events include inhibition of DNA repair, genotoxicity, generation of reactive oxygen species (ROS) resulting in oxidative damage, and stabilization of hypoxia-inducible factor 1α (HIF-1α), a protein that increases the expression of genes that promote survival of cells when they receive less oxygen. The proposed modes of action are summarized in Figure 1.

Figure 1

Mechanistic Events in Cobalt Carcinogenicity

Cobalt is considered to be a clastogen, because in in vitro assays in mammalian cells, it primarily causes chromosome damage and DNA strand breaks. Only a few genotoxicity studies in experimental animals were available, but the results were generally consistent with those of in vitro studies. Two potential mechanisms for genotoxicity include (1) direct induction of oxidative damage to DNA by cobalt(II) ions and (2) an indirect effect through inhibition of DNA repair (Lison 2015; Smith et al. 2014).

Cobalt is one of a group of metals (transition metals, like iron and nickel) that promote oxidation and reduction (redox) reactions through transfer of electrons. In vitro studies have shown that cobalt particles and ions can induce ROS in mammalian cells, with cobalt metal and cobalt oxide particles having a greater effect than ions. It has been proposed that ROS can play a role in the tumor development process at several stages, including initiating the process by inducing mutations and promoting proliferation of these mutated cells by deregulating controls on cell growth, leading to tumors. Studies in rats have shown that cobalt causes oxidative stress and oxidative DNA damage in several tissues, including kidney, liver, and lung (Kasprzak et al. 1994), which supports this proposed pathway for cobalt-induced carcinogenicity. Also, a higher frequency of a specific mutation in the K-ras oncogene, a gene with the potential to cause cancer, was found in cobalt-induced lung tumors in mice and rats than in spontaneous lung tumors (IARC 2006b; NTP 1998; 2014). This mutation involves substitution of one nucleotide for another in a G to T transversion, which is a mutation commonly associated with oxidative DNA damage. In addition, cobalt-induced oxidative stress (via the production of ROS) can activate genes and proteins (specifically, the transcription factors NF-κB, AP1, p53, and Nrf2) that in turn regulate the expression of many genes that play a role in carcinogenicity, such as those involved in inflammation and control of the cell cycle (Beyersmann and Hartwig 2008; Davidson et al. 2015; PubChem 2015; Shukla et al. 2012; Valko et al. 2005; Valko et al. 2006).

Finally, a well-established biological effect of cobalt is to mimic oxygen deficiency in cells by stabilizing HIF-1α (Galán-Cobo et al. 2013; Gao et al. 2013; Greim et al. 2009; Maxwell and Salnikow 2004; Nyga et al. 2015; Saini et al. 2010a; Saini et al. 2010b). HIF-1α plays a central role in regulating more than 100 hypoxia-responsive genes and is a major regulator of the adaptation of cancer cells to oxygen deficiency. HIF-1α overexpression has been linked to cancer initiation and progression and is a common characteristic of many human cancers (Cheng et al. 2013; Galanis et al. 2008; Galanis et al. 2009; Paul et al. 2004).

Although most of the toxicological effects of cobalt are attributed to the cobalt ion, direct toxic effects of cobalt particles also contribute, as evidenced by the greater toxicity of cobalt metal than of cobalt sulfate in National Toxicology Program (NTP) rodent bioassays (Behl et al. 2015; NTP 1998; 2014). Differences in the relative toxicity reported for cobalt particles and ions may be partially explained by differences in the mechanisms by which cobalt enters the cell and in the subsequent accumulation and distribution of cobalt within the cell, as well as a synergistic effect between the particles and metal on ROS production (Peters et al. 2007; Sabbioni et al. 2014; Smith et al. 2014).

Cancer Studies in Experimental Animals

Exposure of experimental animals to cobalt metal or cobalt compounds caused tumors in two rodent species, at several different tissue sites, and by several different routes of exposure. This conclusion is based on studies in rats and mice exposed to cobalt metal (five studies), water-soluble cobalt compounds (two studies with cobalt sulfate and one study with cobalt chloride), and poorly water-soluble cobalt compounds (four studies with cobalt oxide). Studies of cobalt alloys and radioactive cobalt in experimental animals were not considered to be informative, because of potential confounding by other carcinogens.

Inhalation exposure of rats and mice to cobalt metal (NTP 2014) or cobalt sulfate (NTP 1998) or intratracheal instillation of cobalt oxide in rats (Steinhoff and Mohr 1991) caused lung tumors (alveolar/bronchiolar adenoma and carcinoma). In addition, inhalation exposure of rats to cobalt metal caused squamous-cell tumors of the lung (primarily cystic keratinizing epithelioma) in females and possibly in males.

In inhalation studies of cobalt metal in rats (NTP 2014), tumors were also induced at sites distant from the lung, including tumors of the pancreas (islet-cell adenoma or carcinoma combined) in males and of the hematopoietic system (mononuclear-cell leukemia) in females, indicating a systemic effect. Increased incidences of kidney tumors (adenoma or carcinoma combined) in male rats and pancreas (carcinoma) in female rats may have been related to cobalt metal inhalation; however, the findings were not conclusive. Inhalation exposure to cobalt metal (NTP 2014) or cobalt sulfate (NTP 1998) induced adrenal-gland tumors (benign and malignant pheochromocytoma), which could have been caused by direct or indirect mechanisms.

In rats, local injection of cobalt at various anatomic locations caused tumors at the injection sites. Although these studies were less robust than the inhalation studies, and sarcomas are common in rats following injection of a variety of compounds, the consistency of the tumor types and findings across different cobalt forms provides supporting evidence for the carcinogenicity of cobalt. Intraperitoneal or intramuscular injection of the poorly water-soluble compound cobalt oxide caused histiocytoma or sarcoma at the injection site (Gilman and Ruckerbauer 1962; Steinhoff and Mohr 1991), and subcutaneous injection of the water-soluble compound cobalt chloride caused fibrosarcoma (Shabaan et al. 1977). Intramuscular or intrathoracic injection of cobalt metal (Heath 1956; Heath and Daniel 1962) or nanoparticles (Hansen et al. 2006) caused various types of sarcoma (primarily rhabdomyofibrosarcoma, rhabdomyosarcoma, or fibrosarcoma). In the study of nanoparticles, no tumors were observed after implantation of substances (e.g., titanium dioxide and silicon dioxide) with the same physical characteristics (i.e., surface-to-volume ratio) as cobalt, suggesting that the tumors were due to carcinogenic properties of cobalt and not just to a reaction to any physical implant.

A few studies in rodents (Gilman and Ruckerbauer 1962; Jasmin and Riopelle 1976; Wehner et al. 1977) found no tumors at certain tissue sites following exposure to the same forms of cobalt that caused tumors in other studies; however, these studies generally lacked sensitivity to detect an effect, because of the use of a less sensitive animal model, shorter study duration, or lower exposure levels.

Cancer Studies in Humans

The data available from studies in humans are inadequate to evaluate the relationship between human cancer and exposure specifically to cobalt and cobalt compounds that release cobalt ions in vivo. The data relevant to the evaluation were from studies primarily evaluating lung cancer in five independent cohorts of workers in different types of industries and two population-based case-control studies of esophageal cancer and other cancers of the respiratory and upper digestive (aerodigestive) tract, one in Ireland (O’Rorke et al. 2012) and the other in the state of Washington (Rogers et al. 1993). Studies of cobalt alloys in humans (primarily joint implants) were not considered to be informative, because they were not specific to cobalt exposure, and the extent of any cobalt exposure was unknown.

Although increased risks of lung cancer were found in most of the cohort studies, it is unclear that the excess risks were due to exposure specifically to cobalt, because of potential confounding from exposures to known lung carcinogens or other study limitations. In the cohort studies, hard-metal (Moulin et al. 1998; Wild et al. 2000) and nickel-refinery workers (Grimsrud et al. 2005) were also exposed to known lung carcinogens. The findings of an increased risk of lung cancer among porcelain painters exposed to cobalt was complicated by a somewhat similar increase in risk among female pottery workers who were not thought to be exposed to cobalt (Tüchsen et al. 1996). In studies of a cohort of cobalt production workers, the excess risk found in the first report of this cohort (Mur et al. 1987) was no longer present in an update of the cohort (Moulin et al. 1993). No association between cobalt exposure and lung cancer was found in a study of stainless- and alloyed-steel workers in France (Moulin et al. 2000). Most of the studies had limited sensitivity to detect a true risk, because of small numbers of lung-cancer cases among exposed workers, crude methods of exposure assessment, or potential healthy-worker-related effects (due to the fact that workers are healthier on average than the general population).

Increased risks of esophageal cancer were suggested in two case-control studies; however, it is unclear whether cobalt exposure contributed to the cancer excess. In both studies, cobalt exposure was assessed from a single sample of toenail clippings taken at or several months after diagnosis of esophageal cancer. Measurements of cobalt in toenails reflect an integrated exposure that occurred 12 to 18 months before clipping, raising the question of whether levels found in toenails close to or, in many cases, after cancer diagnosis reflected the relevant period of exposure for long-latency cancer.

Bioavailability

The carcinogenic and toxic effects of cobalt and cobalt compounds begin with the release of cobalt ions in vivo. The bioavailability of a metal species can be predicted by its solubility in biological fluids, such as synthetic equivalents of gastric and intestinal fluids (for ingestion exposure) or lung (alveolar, interstitial, and lysosomal) fluids (for inhalation exposure), and by studies in cultured cells. Results from studies testing solubility in synthetic biological fluids are shown in Table 1, along with other chemical and physical properties of cobalt metal and these cobalt compounds. These studies demonstrated that cobalt metal and both water-soluble and poorly water-soluble cobalt compounds can dissolve and release cobalt ions in some biological fluids (Brock and Stopford 2003; Stopford et al. 2003, Cobalt Development Institute personal communication 21 Jul and 19 Oct 2015), suggesting that they will release ions in vivo.

Table 1

Physical and Chemical Properties of Cobalt Metal and Some Cobalt Compounds.

Very low values (≤ 2%) for bioaccessibility have been reported for the sulfide and mixed (II,III) oxide (Co₃O₄), and intermediate values (14% to 55%) for stearate and oxalate under the same test conditions. However, other, more informative tests with more physiologically relevant test conditions (e.g., two-week studies with 0.3-μm particles in culture medium in the presence of alveolar macrophages) have reported 50% solubility for Co₃O₄. In addition, Ortega et al. (2014) found that intracellular concentrations of solubilized cobalt ions were similar for Co₃O₄ and cobalt chloride in human lung cells in vitro, suggesting that Co₃O₄ would release cobalt ions in vivo. Results with other biological fluids, such as serum and intestinal, alveolar, and interstitial fluids, indicate that the species of cobalt compound, particle size and surface area, and pH of the surrogate fluid all can affect the solubility of cobalt in biological fluids.

The solubility of cobalt compounds in water depends largely on pH, and cobalt is generally more mobile in acidic solutions than in alkaline solutions (IARC 1991; Paustenbach et al. 2013). Sulfates, nitrates, and chlorides of cobalt tend to be soluble in water, whereas oxides (including the mixed oxide, Co₃O₄), hydroxides, and sulfides tend to be poorly soluble or insoluble in water (Lison 2015). Organic cobalt compounds can be either soluble, as is cobalt(II) acetate, or insoluble, as are cobalt(II) carbonate and cobalt(II) oxalate (CDI 2006). In addition to low pH, solubilization of some poorly water-soluble compounds in biological fluids may be enhanced in the presence of binding proteins (IARC 2006b).

Chemical Characteristics

Cobalt (Co) is a naturally occurring transition element with magnetic properties. It is the 33rd most abundant element, making up approximately 0.0025% of the weight of Earth’s crust. Cobalt is a component of more than 70 naturally occurring minerals, including arsenides, sulfides, and oxides. The only stable and naturally occurring cobalt isotope is ⁵⁹Co (ATSDR 2004; WHO 2006). Metallic cobalt, Co(0), exists in two crystalline forms, hexagonal and cubic, which are stable at room temperature (ATSDR 2004; IARC 1991; WHO 2006). Cobalt predominantly occurs in two oxidation states, Co(II) and Co(III). Co(II) is much more stable than Co(III) in aqueous solution (Nilsson et al. 1985; Paustenbach et al. 2013) and is present in the environment and in most commercially available cobalt compounds (e.g., cobalt chloride, sulfide, and sulfate). Co(III) also is present in some commercially available cobalt compounds, including the mixed oxide (Co₃O₄) (IARC 1991; Lison 2015; Paustenbach et al. 2013) and some simple salts of Co(III) (e.g., Co₂O₃). Important salts of carboxylic acids include formate, acetate, citrate, naphthenate, linoleate, oleate, oxalate, resinate, stearate, succinate, sulfamate, and 2-ethylhexanoate.

Occupational Exposure

The primary route of occupational exposure to cobalt is via inhalation of dust, fumes, mists, or gaseous cobalt carbonyl. Dermal contact with cemented carbide (i.e., hard-metal) powders and cobalt salts can result in systemic uptake. Occupational exposure to cobalt occurs in the following industries: (1) production of cobalt metal or salts, (2) metallurgical-related industries, (3) cemented carbides and bonded diamonds, (4) chemicals and pigments, and (5) electronics, “green” energy, and recycling. Occupational exposure has been documented by measurements of cobalt in ambient workplace air (as shown in Table 4) and in blood, urine (as shown in Figure 2), nails, and hair, and lung tissue from workers or deceased workers (ATSDR 2004; CDC 2013; IARC 1991; 2006b). The highest levels of cobalt in workplace air were generally for hard-metal manufacture involving cobalt metal powders (> 1,000 μg/m³ in some instances) (NTP 2009b), production of cobalt salts, and metallurgical-related industries (> 10,000 μg/m³ in some instances) (IARC 2006b). The highest cobalt levels in urine, blood, hair, and nails also were associated with exposure to cobalt powders.

Table 4

Occupational Exposure to Cobalt in Workplace Air.

Surgical Implants

Total hip implants consist of (1) a femoral head attached to a stem that is inserted in the thigh bone (usually made of ceramic or metal) and (2) a socket or cup that is anchored in the pelvis (made of metal, ceramic, or polyethylene). Cobalt-chromium-molybdenum (CoCrMo) alloy is the predominant alloy used in metal-containing implants, such as metal-on-metal implants (in which both articulating surfaces are metal), polyethylene-on-metal implants, and metal-on-ceramic implants. Other metals, such as nickel, tungsten, iron, aluminum, and titanium, may also be used in implants. Knee implants may also contain cobalt metal; however, unlike some hip implants with metal-to-metal contact, knee implants are designed so that metal surfaces do not contact each other. Cobalt ions may be released into the body throughout the lifetime of a cobalt-containing device (Devlin et al. 2013; Sampson and Hart 2012). Urinary levels of cobalt identified from studies of hip implants reported as stable or that did not specifically address stability ranged from approximately 0.7 to 12 μg/L, compared with a range of 0.01 to 4.2 μg/L for the general population (as shown in the previous graph). Implants may fail because of excessive wear or corrosion by body fluids, increasing the levels of cobalt released from the implants (Sampson and Hart 2012). Dunstan et al. (2005) reported blood cobalt levels of 19 and 52 μg/L in two individuals with unstable (radiologically loose) metal-on-metal implants. In rare cases, high levels of cobalt from failed implants may be associated with toxicity. Recommended levels of blood cobalt for further clinical investigation and action were set at 7 μg/L in the United Kingdom (MHRA 2012) and 10 μg/L in the United States by the Mayo Clinic (2015).

Environmental Exposure

The TRI reported that in 2013, on- and off-site industrial releases of cobalt and cobalt compounds totaled approximately 5.5 million pounds from 723 facilities in the United States (TRI 2014a). Calculations based on media-specific release data from the TRI indicate that releases to land accounted for 82% of total releases in 2013 (TRI 2014b; 2014c). Worldwide, approximately 75,000 metric tons of cobalt enters the environment annually, with similar amounts coming from natural sources (40,000 metric tons) and sources related to human activities (35,000 metric tons) (CDI 2006; Shedd 1993). Recycling of electronic and electrical waste can result in release of cobalt to the environment; however, releases from this source are less of a concern in the United States than in other global regions where recycling is more common and less controlled (Julander et al. 2014).

The average concentration of cobalt in ambient air in the United States has been reported to be approximately 0.4 ng/m³ (ATSDR 2004). Levels can be orders of magnitude higher near source areas (e.g., near facilities processing cobalt-containing alloys and compounds) reported from outside the United States. The median cobalt concentration in U.S. drinking water has been reported to be less than 2.0 µg/L; however, levels as high as 107 µg/L have been reported (ATSDR 2004). Cobalt concentrations have been reported to range from 0.01 to 4 µg/L in seawater and from 0.1 to 10 µg/L in fresh water and groundwater (IARC 2006b). Studies have reported cobalt soil concentrations ranging from 0.1 to 50 ppm. However, soils near ore deposits, phosphate rock, or ore-smelting facilities or soils contaminated by airport or highway traffic or near other source areas may contain higher concentrations (IARC 2006b).

Data for individuals exposed to cobalt from the environment are limited, but a study of metal exposure from mining and processing of nonferrous metals in Katanga, Democratic Republic of Congo, found that geometric mean urinary cobalt concentrations were 4.5-fold higher for adults and 6.6-fold higher for children in urban and rural communities near mines and metal smelters than in rural communities without mining or industrial activities (Cheyns et al. 2014).

Other Sources of Exposure of the General Population

The general population can be exposed to low levels of cobalt primarily through consumption of food and to a lesser degree through inhalation of ambient air and ingestion of drinking water (ATSDR 2004). The daily cobalt intake from food in the United States was estimated to range from 3.4 to 11.6 µg based on analyses of 234 foods in the 1984 U.S. Food and Drug Administration Total Diet Study (Pennington and Jones 1987). Although this amount includes cobalt as part of both vitamin B₁₂ and other cobalt compounds (ATSDR 2004), green, leafy vegetables and fresh cereals generally contain the most cobalt (IARC 1991), and these plant sources of cobalt do not contain vitamin B₁₂. In the 1960s, some breweries added cobalt salts to beer to stabilize the foam (resulting in cobalt exposures of 0.04 to 0.14 mg/kg of body weight), but cobalt is no longer added to beer (ATSDR 2004). Higher cobalt intake may result from consumption of over-the-counter or prescription mineral preparations containing cobalt compounds.

Other potential sources of exposure include consumer products and tobacco smoking. Cobalt is present in only a few consumer products, including cleaners, detergents, soaps, car waxes, and a nickel metal hydride battery (5% to 10% cobalt) (ATSDR 2004; HPD 2014). Various brands of tobacco have been reported to contain cobalt at concentrations ranging from less than 0.3 to 2.3 µg/g of dry weight, and 0.5% of the cobalt content is transferred to mainstream smoke (WHO 2006). However, urinary cobalt levels (unadjusted for creatinine) for cigarette-smoke-exposed and unexposed NHANES participants for survey years 1999 to 2004 did not differ significantly (Richter et al. 2009).

Coast Guard (Dept. of Homeland Security)

Minimum requirements have been established for safe transport of cobalt naphthenate in solvent naphtha on ships and barges.

Department of Transportation (DOT)

Numerous cobalt compounds are considered hazardous materials, and special requirements have been set for marking, labeling, and transporting these materials.

Environmental Protection Agency (EPA)

Food and Drug Administration (FDA, an HHS agency)

Cobaltous salts are prohibited from use in human food.

All drugs containing cobalt salts (except radioactive forms of cobalt and its salts and cobalamin and its derivatives) have been withdrawn from the market because they were found to be unsafe or not effective, and they may not be compounded.

Chromium-cobalt-aluminum-oxide used as a color additive for linear polyethylene surgical sutures used in general surgery must comprise no more than 2% by weight of the suture material, not migrate to surrounding tissue, and conform to labeling requirements in 21 CFR 70.25.

Chromium-cobalt-aluminum-oxide may be used as a color additive in contact lenses in amounts not to exceed the minimum reasonably required to accomplish the intended coloring effect.

Ferric ammonium ferrocyanide and ferric ferrocyanide used to color externally applied drugs (including those for use in the area of the eye) must not contain more than 200 ppm cobalt (as Co) and conform to labeling requirements in 21 CFR 70.25.

21 CFR 369 contains recommended drug labeling statements for over-the-counter cobalt preparations containing ≥ 0.5 mg cobalt as a cobalt salt per dosage unit and which recommend administration rates of ≥ 0.5 mg per dose and ≥ 2 mg per 24-hour period.

An approved new drug application is required for marketing cobalt preparations intended for use by man.

21 CFR 872, 874, and 888 identify class designations (Class I, II, or III) of various cobalt-containing dental prosthetic device alloys, cobalt-chromium-alloy-based facial prosthetics, and cobalt-chromium-molybdenum orthopedic devices that determine the type of premarketing submission or application required for FDA clearance to market.

Cobalt naphthenate may be used in quantities that do not exceed those reasonably required as an accelerator in the production of cross-linked polyester resins used as articles or components of articles intended for repeated use in contact with food.

Cobalt aluminate may be safely used as a colorant in the manufacture of articles or components of articles intended for use in producing, manufacturing, packing, processing, preparing, treating, packaging, transporting, or holding of food at levels not to exceed 5% by weight of all polymers except in resinous and polymeric coatings complying with 21 CFR 175.300, melamine-formaldehyde resins in molded articles complying with 21 CFR 177.1460, xylene-formaldehyde resins complying with 21 CFR 175.380, ethylene-vinyl acetate copolymers complying with 21 CFR 177.1350, and urea-formaldehyde resins in molded articles complying with 21 CFR 177.1900.

Occupational Safety and Health Administration (OSHA, Dept. of Labor)

This legally enforceable PEL was adopted from the 1968 ACGIH TLV-TWA shortly after OSHA was established; it may not reflect the most recent scientific evidence and may not adequately protect worker health.

Permissible exposure limit (PEL) (8-h TWA) = 0.1 mg/m³ for cobalt metal, dust, and fume (as Co).

American Conference of Governmental Industrial Hygienists (ACGIH)

Threshold limit value – time-weighted average (TLV-TWA) = 0.02 mg/m³ for cobalt and inorganic compounds; = 0.1 mg/m³ for cobalt carbonyl and cobalt hydrocarbonyl.

Biological exposure index (BEI) = 15 µg/L for cobalt in urine for cobalt and inorganic compounds, including cobalt oxides but not combined with tungsten carbide, for end of shift at end of workweek.

Consumer Product Safety Commission (CPSC)

The CPSC has issued guidance regarding the potential hazards of specific cobalt- or cobalt-compound-containing art and craft materials (e.g., glazes, glass colorants, paints, toners, pigments, and dyes) and specific precautions to take when using them.

Environmental Protection Agency (EPA)

Regional Screening Levels (formerly Preliminary Remediation Goals): residential soil = 23 mg/kg; industrial soil = 350 mg/kg; residential air = 0.00031 µg/m³; industrial air = 0.0014 µg/m³; tap water = 6 µg/L.

National Institute for Occupational Safety and Health (NIOSH, CDC, HHS)

Recommended exposure limit (REL) (10-h TWA) = 0.05 mg/m³ for cemented tungsten carbide containing > 2% Co (as Co); = 0.05 mg/m³ for cobalt metal dust and fume (as Co); = 0.1 mg/m³ for cobalt carbonyl (as Co) and cobalt hydrocarbonyl (as Co).

Immediately dangerous to life and health (IDLH) limit = 20 mg/m³ for cobalt metal dust and fume (as Co).

Cancer Studies in Humans

Epidemiological studies provide evidence for the carcinogenicity of cobalt-tungsten carbide powders and hard metals based on (1) consistent findings of excess lung-cancer mortality among cobalt-tungsten carbide hard-metal manufacturing workers across studies, (2) higher risks among individuals with higher exposure levels, and (3) positive exposure-response relationships that cannot be explained by confounding with tobacco smoking. However, the epidemiological data are limited, because there are few studies of independent populations.

The published epidemiological literature consists of mortality studies of two independent multi-plant cohorts of cobalt-tungsten carbide hard-metal manufacturing workers, one in France (Moulin et al. 1998) and one in Sweden (Hogstedt and Alexandersson 1990), and cohort studies of two individual factories included in the French multi-plant cohort (Lasfargues et al. 1994; Wild et al. 2000). The French multi-plant cohort included all 10 cobalt-tungsten carbide manufacturing plants in France; in addition, a nested case-control study of lung cancer was conducted within this cohort. The nested case-control study is most informative for evaluating cancer risk, because it used a semi-quantitative exposure scale to evaluate exposure-response relationships and considered potential confounding by exposure to tobacco smoking and other known or suspected occupational carcinogens. The cohort study of the largest French factory shares these advantages; however, because the workers were included in the multi-plant study, it does not provide independent evidence for carcinogenicity. In these two studies, four metrics of exposure were evaluated: (1) exposure level, which was the highest exposure score experienced during an individual’s work history (on a scale of 0 to 9), (2) duration of exposure at a level of 2 or higher, (3) unweighted cumulative dose, which assigned the same level to occasional and full-time exposure, thus favoring peak exposure, and (4) frequency-weighted cumulative dose, which weighted exposure level by the frequency of exposure, thus reducing the effect of occasional exposure. The Swedish study, although limited in size, provides supporting information for an independent population.

Excess lung-cancer mortality (of approximately 30%) was found in both multi-plant cohort studies (Hogstedt and Alexandersson 1990; Moulin et al. 1998); risk estimates were significantly higher among individuals with higher measures of exposure or longer time since first exposure (latency). In the nested case-control study (Moulin et al. 1998), lung cancer risk was significantly higher (odds ratio [OR] = 1.93, 95% CI = 1.03 to 3.62, 35 exposed cases) among workers exposed to cobalt-tungsten carbide (exposure level ≥ 2) than among workers with little or no exposure (exposure level < 2). In exposure-response analyses using workers in the lowest exposure category as the comparison group, lung-cancer risk was significantly higher (by up to fourfold) for workers in the highest categories of both measures of cumulative dose, and an elevated risk of borderline statistical significance was found for workers in the highest exposure-level category. Positive exposure-response relationships were observed for all four measures of exposure: duration (P_trend = 0.03), unweighted cumulative dose (P_trend = 0.01), frequency-weighted cumulative dose (P_trend = 0.08), and exposure level (P_trend = 0.08). Adjustment for tobacco smoking or exposure to known or suspected carcinogens did not change the results. The Swedish study had limited ability to evaluate exposure-response relationships because of small numbers of exposed workers with lung cancer. Nevertheless, the risk of lung cancer mortality was significantly increased for workers with exposure duration of over 10 years and latency of over 20 years (standardized mortality ratio [SMR] = 2.78, 95% CI = 1.11 to 5.72, 7 exposed cases). Analyses restricted to workers with at least 10 years’ exposure or at least 20 years’ latency found somewhat higher SMRs for “high-exposed” than “low-exposed” workers (Hogstedt and Alexandersson 1990).

Excess risks of lung-cancer mortality were also found in studies of the two individual French factories. Wild et al. (2000) reported significantly elevated SMRs (by approximately twofold) for lung cancer among all male workers and among male workers ever employed in presintering workshops or with exposure levels of at least 2. The highest SMRs were observed for male workers in the highest exposure categories of all four exposure metrics (level, duration, and both measures of cumulative dose), although the trends were not statistically significant, and the risk estimates were imprecise. In the study by Lasfargues et al. (1994), the entire cohort had a significantly increased risk of lung cancer, and the risk was highest among workers in the highest exposure-level category. Although small, this study provides supporting evidence that the findings for the French industry-wide cohort were not due solely to the results for the large factory studied by Wild et al. (2000).

Both the French multi-plant cohort study (Moulin et al. 1998) and the larger study of an individual French factory (Wild et al. 2000) found higher risks of lung cancer for exposure to cobalt-tungsten carbide before sintering than after sintering (see Production). The authors stated that exposure was highest during presintering processes; however, there is no evidence of toxicological differences between presintered and sintered materials, and both materials release similar amounts of cobalt ions (see Studies on Mechanisms of Carcinogenesis).

It is unlikely that the excess risks of lung cancer found in the French studies were due to confounding by tobacco smoking or co-exposure to other known carcinogens. In the multi-plant study, the smoking-adjusted odds ratio for cobalt-tungsten carbide exposure (OR = 2.6, 95% CI = 1.16 to 5.82) was similar to the unadjusted risk (OR = 2.29, 95% CI = 1.08 to 4.88). Neither study found increased risks of smoking-related diseases, such as chronic bronchitis and emphysema, and adjustment for smoking or exposure to other occupational carcinogens did not change the findings in the exposure-response analyses (Moulin et al. 1998; Wild et al. 2000). Neither the Swedish multi-plant study (Hogstedt and Alexandersson 1990) nor the small French cohort study (Lasfargues et al. 1994) adjusted for smoking; however, surveys of smoking habits among a subset of workers found smoking rates similar to those in the general population. Overall, the studies are limited by the lack of quantitative exposure assessment and potential confounding; however, exposure misclassification would most likely reduce the likelihood of detecting a true effect.

Studies on Mechanisms of Carcinogenesis

The findings from epidemiological studies are supported by studies on mechanisms of carcinogenesis. Although the mechanism(s) by which cobalt-tungsten carbide causes cancer have not been fully elucidated, it has been shown that (1) cobalt-tungsten carbide releases cobalt ions, (2) cobalt ions affect biochemical pathways related to carcinogenicity, (3) cobalt compounds are carcinogenic in experimental animals, (4) cobalt-tungsten carbide increases the production of reactive oxygen species (ROS) and causes greater cytotoxic, toxic, and genotoxic effects than does cobalt alone, (5) cobalt-tungsten carbide causes key events related to carcinogenesis, including genotoxicity, cytotoxicity, inflammation, and apoptosis (programmed cell death), and (6) the oxidative stress response resulting from increased ROS production may play a role in these key events and may also interfere with cells’ ability to repair damage caused by cobalt-tungsten carbide. The combination of the effects from cobalt ions and the oxidative stress response from ROS production provide plausible modes of action for the carcinogenicity of cobalt-tungsten carbide.

Studies in biological fluids, in vitro systems, experimental animals, and humans have demonstrated that cobalt is rapidly solubilized from cobalt-tungsten carbide. Cobalt dissolution rates were similar for presintered and sintered cobalt-tungsten carbide incubated in various artificial biological fluids (Stopford et al. 2003). Tungsten is not rapidly solubilized from cobalt-tungsten carbide, but can be phagocytized by macrophages (Lombaert et al. 2004). Cobalt was also released from hard-metal dust incubated with plasma and lung tissue (Edel et al. 1990). In experimental animals administered cobalt-tungsten carbide by intratracheal administration, cobalt was solubilized rapidly, cleared from the lung, distributed in the body, and excreted in the urine (Lison 1996). Rats exposed intratracheally to cobalt-tungsten carbide had more cobalt in the urine than did rats administered cobalt alone, suggesting that tungsten carbide increases the bioavailability of cobalt (Lasfargues et al. 1992). Several biomonitoring studies detected elevated levels of cobalt in the urine, lungs, and other tissues of workers exposed to cobalt-tungsten carbide hard metals (Gallorini et al. 1994; Goldoni et al. 2004; Linnainmaa and Kiilunen 1997; Nicolaou et al. 1987; Rizzato et al. 1986; Sabbioni et al. 1994b; Scansetti et al. 1994; Scansetti et al. 1998).

Soluble cobalt compounds are genotoxic and carcinogenic in experimental animals. Cobalt and cobalt compounds that release cobalt ions in vivo are listed as reasonably anticipated to be human carcinogens in the Report on Carcinogens based on sufficient evidence of carcinogenicity from studies of cobalt metal, cobalt sulfate, cobalt chloride, and cobalt oxide in experimental animals and supporting evidence from studies on mechanisms of carcinogenesis. Cobalt ions produce ROS, which cause oxidative DNA damage and act on a number of cancer-related molecular targets. Cobalt ions disrupt cell-signaling pathways (Murata et al. 1999), inhibit DNA repair (Hartwig 2000; Hartwig et al. 2002), regulate genes involved in the response to hypoxia (Beyersmann 2002), replace or mimic essential divalent metal ions, thus altering cellular reactions (Beyersmann and Hartwig 1992; Kawanishi et al. 1994; Lloyd et al. 1998; Nackerdien et al. 1991), and interfere with mechanisms involved in cell-cycle control and modulation of apoptosis (De Boeck et al. 2003b; De Boeck et al. 2003c).

Numerous in vitro studies (reviewed in NTP 2009a) and in vivo studies (Huaux et al. 1995; Lasfargues et al. 1995) have shown greater cytotoxic effects (measured primarily by lactate dehydrogenase release) for cobalt-tungsten carbide than for either cobalt powder or tungsten carbide alone. The mixture’s greater in vitro toxicity to macrophages is not fully explained by greater bioavailability of cobalt (Lison and Lauwerys 1992; 1994). Respirable samples collected at various stages of the hard-metal manufacturing process (including powders for pressing, presintered materials, and powders from grinding of sintered materials) caused cytotoxicity and pathological changes in the lungs of rats after intratracheal injection (Adamis et al. 1997). In addition, cobalt-tungsten carbide causes a type of respiratory toxicity (“hard-metal disease”) that is not observed with exposure to cobalt alone. Hard-metal disease is characterized by a giant-cell interstitial pneumonia that can develop into lung fibrosis (Lison 1996; Lison et al. 1996).

There is some evidence that the greater toxicity of cobalt-tungsten carbide may result from a physicochemical reaction that takes place at the interface between certain carbides and cobalt particles (Lison and Lauwerys 1992). The structural features of the two particles may help to explain the effects. Cobalt metal can reduce ambient oxygen, but only at a low rate of reaction, because of the particles’ surface characteristics. Tungsten carbide is inert and does not react with oxygen but is a good electron conductor. When cobalt and tungsten carbide particles are associated, the cobalt electrons are transferred to the carbide surface, allowing increased oxygen reduction and thus increased production of ROS. Biochemical studies on the production of ROS have shown that cobalt’s capacity to generate hydroxyl radicals is greatly increased by association with tungsten carbide. Formation of the ROS results directly from the interaction of cobalt with tungsten carbide or indirectly from the cobalt ions generated from the Fenton-like reaction of the cobalt metal with the carbide (Lison and Lauwerys 1993; Lison et al. 1995). In oxygen-radical-generating systems, post-sintered powders sampled from final machining (grinding) of cobalt-tungsten carbide products produced higher levels of ROS than did presintered samples of cobalt and tungsten carbide separately or as mixtures (Stefaniak et al. 2010).

Metal-induced generation of ROS in cellular test systems leads to oxidative stress as a result of increased free radicals and insufficient antioxidative defense. Protective mechanisms include cellular antioxidant systems, the stress-protein response, and the involvement of DNA excision and repair enzymes (Kasten et al. 1997; Lombaert et al. 2008; Shi et al. 2004). Fenoglio et al. (2008) studied oxidation of the antioxidant glutathione and cysteine sulfhydryl groups by cobalt-tungsten carbide dust–induced ROS and reported dust-concentration-dependent generation of thiyl radicals at particle surface sites. Depletion of cellular antioxidant defenses could further exacerbate cellular oxidative damage caused by ROS generated by cobalt-tungsten carbide particles.

Regulation of gene expression, including apoptotic, stress-protein, and immune-response pathways, also can be affected by ROS. Lombaert et al. (2008) evaluated the effects of cobalt-tungsten carbide exposure in vitro on patterns of gene expression in human peripheral-blood mononucleated cells and reported statistically significant up-regulation of apoptosis and stress or defense response pathways and down-regulation of immune-response pathways.

Apoptosis has been associated with exposure to a number of known carcinogens (arsenic, cadmium, chromium, nickel, and beryllium) and possible carcinogens (cobalt and lead). Cobalt chloride has been shown to induce apoptosis through formation of ROS in both human alveolar macrophages and a rat pheochromocytoma cell line (PC12); co-administration of antioxidants suppressed ROS production and restored cell viability (Araya et al. 2002; Zou et al. 2001). Cobalt-tungsten carbide, tungsten carbide, and cobalt ions induced apoptosis in human lymphocytes; the effect of the mixture was significantly greater than that of tungsten carbide or cobalt alone (Lombaert et al. 2004).

Cobalt-tungsten carbide is genotoxic in vitro and causes mutations in the lungs of rats exposed in vivo. Its genotoxicity (clastogenic effects) may be caused by increased ROS production from the interaction between cobalt and tungsten carbide, from ionic cobalt, or from both. In addition, cobalt ions inhibit DNA repair, which may also contribute to cobalt-tungsten carbide’s genotoxic effects. Specifically, cobalt-tungsten carbide caused DNA strand breaks in mouse 3T3 fibroblasts and human peripheral-blood lymphocytes (Anard et al. 1997) and micronucleus formation in human peripheral-blood lymphocytes (De Boeck et al. 2003c; Van Goethem et al. 1997). In these studies, cobalt-tungsten carbide was more genotoxic than cobalt alone. In rats exposed by intratracheal instillation, cobalt-tungsten carbide caused DNA damage and micronucleus formation in the lung (type II pneumocytes) (De Boeck et al. 2003a). No increase in DNA damage or micronucleus formation was observed in rat peripheral-blood lymphocytes; however, it is unclear whether circulating lymphocytes are a good reporter for monitoring genotoxic effects from inhaled particles. In humans, neither DNA damage nor micronucleus formation was increased in lymphocytes of cobalt-tungsten carbide hard-metal workers, compared with unexposed workers; however, this study was limited by small sample size (De Boeck et al. 2000). Multiple regression analyses (Mateuca et al. 2005) indicated that both end points were associated with an interaction between tobacco smoking and exposure, and that micronucleus formation was associated with smoking, working in a cobalt-tungsten carbide plant, and having variant forms of genes coding for DNA repair enzymes (X-ray repair cross-complementing group 3 and 8-oxoguanine DNA glycosylase).

In addition, although the pathogenesis of hard-metal disease is not fully understood, it may involve differences in the susceptibility (genetic and/or health-related) of affected individuals to the toxic effects of increased ROS production due to cobalt-tungsten carbide exposure. Further, the mechanisms for fibrosing alveolitis and lung cancer in hard-metal workers may be related, conceivably involving oxidative damage and/or inflammatory events (IARC 2006a).

Cancer Studies in Experimental Animals

No studies in experimental animals were identified that evaluated the relationship between cancer and exposure specifically to cobalt-tungsten carbide powders or hard metals.

U.S. Environmental Protection Agency (EPA)

Occupational Safety and Health Administration (OSHA, Dept. of Labor)

While this section accurately identifies OSHA’s legally enforceable PELs for this substance in 2018, specific PELs may not reflect the more current studies and may not adequately protect workers.

Permissible exposure limits (PEL) (8-h TWA) = 0.1 mg/m3 for cobalt metal, dust, and fume (as Co); = 5 mg/m3 for insoluble tungsten compounds (as W).

Short-term exposure limits (STEL) = 10 mg/m3 for insoluble tungsten compounds (as W).

American Conference of Governmental Industrial Hygienists (ACGIH)

Threshold limit value – time-weighted average (TLV-TWA) = 0.02 mg/m3 for cobalt and inorganic cobalt compounds; = 5 mg/m3 for tungsten metal and insoluble compounds.

Threshold limit value – short-term exposure limit (TLV-STEL) = 10 mg/m3 for tungsten metal and insoluble compounds.

Biological exposure index (BEI) (end of shift at end of workweek) = 15 μg/L for cobalt in urine.

National Institute for Occupational Safety and Health (NIOSH, CDC, HHS)

Recommended exposure limit (REL) (10-h TWA) = 0.05 mg/m3 for cemented tungsten carbide containing > 2% Co (as Co); = 0.05 mg/m3 for cobalt metal dust and fume (as Co); = 5 mg/m3 for tungsten and insoluble tungsten compounds (as W).

Immediately dangerous to life and health (IDLH) limit = 20 mg/m3 for cobalt metal dust and fume (as Co).

Short-term exposure limit (STEL) = 10 mg/m3 for tungsten and insoluble tungsten compounds (as W).