The carcinogenicity of chromium.

The carcinogenicity of chromium compounds is reviewed with specific attention to the gaps in knowledge for risk estimation and research needs. The most important problems at present are whether trivalent chromium compounds cause cancer, and whether there is a difference in cancer causing effects between the soluble and the slightly soluble hexavalent compounds in the practical exposure situation. Dose estimates for risk estimation based on epidemiological investigations are also lacking. Present evidence indicates that the trivalent chromium compounds do not cause cancer although high concentrations in some in vitro systems have shown genetic toxicity. Hexavalent chromium compounds cause cancer in humans, in experimental animals and exert genetic toxicity in bacteria and in mammalian cells in vitro. Epidemiological evidence and animal experiments indicate that the slightly soluble hexavalent salts are the most potent carcinogens, but proper identification and characterization of exposure patterns in epidemiological work are lacking. Workers also tend to have mixed exposures. Soluble and slightly soluble salts are equally potent genotoxic agents in vitro. Further work for establishing dose estimates for risk evaluation in epidemiological work is important. In vitro systems should be applied for further identification of the mechanism of the carcinogenic effects, and animal experiments are urgent for comparison of the carcinogenic potency of the different hexavalent salts. Hexavalent chromium salts must be regarded as established carcinogens, and proper action should be taken in all industries with regard to such exposure. At present the carcinogenic risk to the general population caused by chromium compounds seems to be negligible, chromium in cigarettes, however, is an uncertainty in this respect. The amount of chromium and the type of chromium compounds inhaled from cigarettes is not known.

The carcinogenicity of chromium compounds is reviewed with specific attention to the gaps in knowledge for risk estimation and research needs. The most important problems at present are whether trivalent chromium compounds cause cancer, and whether there is a difference in cancer causing effects between the soluble and the slightly soluble hexavalent compounds in the practical exposure situation. Dose estimates for risk estimation based on epidemiological investigations are also lacking.
Present evidence indicates that the trivalent chromium compounds do not cause cancer although high concentrations in some in vitro systems have shown genetic toxicity. Hexavalent chromium compounds cause cancer in humans, in expernmental animals and exert genetic toxicity in bacteria and in mammalian cells in vitro. Epidemiological evidence and animal experiments indicate that the slightly soluble hexavalent salts are the most potent carcinogens, but proper identification and charactenrzation of exposure patterns in epidemiological work are lacking. Workers also tend to have mixed exposures. Soluble and slightly soluble salts are equally potent genotoxic agents in vitro.
Further work for establishing dose estimates for risk evaluation in epidemiological work is important. In vitro systems should be applied for further identification of the mechanism of the carcinogenic effects, and animal experiments are urgent for comparison of the careinogenic potency of the different hexavalent salts. Hexavalent chromium salts must be regarded as established carcinogens, and proper action should be taken in all industries with regard to such exposure. At present the carcinogenic risk to the general population caused by chromium compounds seems to be negligible, chromium in cigarettes, however, is an uncertainty in this respect. The amount of chromium and the type of chromium compounds inhaled from cigarettes is not known. ral toxicology of chromium compounds differences in cancer causing effects between solu-.viewed by the National Academy of ble and slightly soluble salts of hexavalent chromiand by the National Institute of Occuum. The specific compound or compunds which 'ety and Health (2). The careinogenic have caused cancer in chromate workers are not of chromium compounds have been known, and an important problem is also the lack of reviewed by the International Agency adequate dose registration for dose-response evalh on Cancer (3). More recently, several uatioons and risk estimation. The importance of -eviews of the toxicology of chromium smoking in chromium carcinogenesis is not known. ablished (4)(5)(6)(7)(8), and possible mechanisms ogenic effect have been reviewed based Chemistry-Biological Interactions evelopments in in vitro tests (9-11). r does not present an extensive histori-Chromium is a transition metal with an atomic of the carcinogenicity of chromium. weight of 52. The most common valences are 0 +2, s which are important for discussing +c3, and +6. The only important chromium ore is gaps in knowledge of the carcinogenic chromite, FeOCr2O3. Chromates are produced by a gromium are presented. The most imsmelting, roasting and extraction process. Soda ash lems at present pre wheTher trivalent or soda ash and lime are reacted with chromite. are insoluble in water but soluble in acid. The residue from both processes contains an unknown mixture of mostly insoluble chromium compounds (12). In the ferroalloy industry and for refractory work, chromites are used, and only limited amounts of hexavalent chromium compounds are assumed to be present. Chromium metal is produced from chromates by reduction.
Both trivalent and hexavalent chromium are found in nature, but the trivalent is the more common form. Divalent chromium compounds wrill oxidize unless protected from air. Chromium in biological materials is probably always trivalent (13). When animals are exposed to hexavalent chromium, the hexavalent fonn exists for some time indicated by a different distribution pattern than after exposure to the trivalent form (4). After some time only trivalent chromium is found. When added to tissue in vfitro, the reduction takes place within minutes (14). The so-called glucose tolerance factor, which has not been fully identified, is a trivalent chromium complex with niacin (15). The trivalent chromium ion has a strong tendency to form coordination complexes and chelates in aqueous solutions and does not exist in the free form (Table 1). Water may serve as a ligand, but in complex solutions water may be replaced by other anion (halides, sulfates, organic acids). Under alkaline conditions, foming of polynucleate complexes which may precipitate due to a replacement of water with OHtakes place (13).
The hexavalent form of chromium is almost always linked to oxygen and is a strong oxidizing agent ( Table 1). The hexavalent form is readily reduced to the trivalent form by contact with organic matter. Oxidation oftrivalent chromium (or chromium metal) to the hexavalent form in organic matter seems unlikely (13).  (13,16). Once in the cell, the hexavalent form is reduced to the trivalent. The site in the cell for this conversion has been suggest. ed to be both the plasma membrane, the endoplasrnic reticulum, the mitochondria or the cell nuclew (17)(18)(19)(20). The biological effects of the hexavalent form may thus either be related to the reduction process, to the subsequent binding of the trivalent chromium to macromolecules, or possibly to both.
Trivalent chromium complexes with a number of ligands within the cell. High concentration of chro. mium is normally found in nucleic acids (21). Its function, if any, is not known, but a stabilizing effect of chromium on RNA has been demonstrated in vitro (22). Chromium in RNA may be in an organic form; orgnochromium in compounds of toxicological significance has not been described, but chromium carbonyl causes cancer in a transplantation animal technique (23).

Environmental Exposure
Chromium exposure to the general public has important health consequences because of its skin sensitizing effect. Chromium is among the most common sensitizers in allergic eczema (24). The daily intake from food has been estimated to be in the range of 0.03-0.1 mg (1,4). The chromium content in municipal drinking water has been estimated to be in the same range as in rivers and lakes (1)(2)(3)(4)(5)(6)(7)(8)(9)(10) limgll.), occasionally somewhat higher (1).
Urban air concentrations have been reported from less than 10 ng/m3 to about 50 ng/m3, for rural stations seldom above 10 ng/m3 (1 using local air or water pollution (12).
Tobacco has been reported to contain up to about mg/kg of chromium, but most values are rerted to be low 5 mg/kg (25). No estimates of the ed amount or ofthe chemical fornm of chromium tobacco smoke have been found. With the possiexception of chromium in tobacco, where furiernvestigations are needed, no associatiol has t been made between chromium exposure to the neral public and human ancer.
coupational Exposure Potentially hazardous exposure to chromium compounds may take place in the metal refining/chromate uction industry, in the metallurgic industry and in the refractory brick industry. Hazardous chromium exposure is also reported in industries which are secondary users of chromium chemicals n pigments production and plating. Welders and grinders in the secondary metal industry (steel welding and grinding, anticorrosive paints) are also exposed to chromium compounds, and chromium may be found in small amounts in a variety of industrial settings (catalysts in chemical industry, paint production and use). Chromium is extensively in tanning (2). Mancuso and Hueper (26) and Mancuso (27) Wrported exposure levels of up to i mg/M3 in a chromate producing plant, but most values were in the range of 0.26-0.52 mg/m'* A five-shift mean value of 1.35 mg/m3 for a sack filling operation was eported by Langard and Norseth (28) in an old chromate pigment plant, while in another modern Wlant levels were mostly below 0.2 mg/m3. In a 0 ferrochromium plant most average shift values (personal sampling) were below 0.05 mg/m3, but occasional values up to 1.3 mg/m3 (total chromium) ,.were recorded (29). In a recent review a chromium posure up to 5 mg/m3 in the plating industry was mentioned, but most values were in the range 0.1-0.2 mg/i3 (2). Similar values for chromium ,exposure im different industries have been indicat-Ped in a review by Hayes (30). The exposure levels seem to have been fairly constant during the last 20 N.years in older plants, but lower values are found in new plants.

Epidemiological Studies
The epidemiology of cancer in chromium exposed workers has recently been reviewed by Hayes (30). s g f Even if the first cases of cancer were described in g.,secondary users of chromates (30), the most exten-.sive epidemiological results stem from the chromate producing industry. An inereased risk of August 1981 respiratory cancer in this industry has been irrefutably established from a number of countries. In a review paper from 1950, Baetjer (31) deserbed 109 cases of cancer in the chromate producing industry, 11 cases in the chrome pigment industry and 2 cases im other industries. Already in 1936 German health authonrties recognized lung cancer as a possible occupational disease associated with chromate dust exposure. Other papers reviewing the number of cases in different countries have also appeared (6,(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34). Most cases have been reported from the chromate production industry, but also reports from the chrome pigment industry have appeared (28,35).
Hueper (36) reviewed the histological classification of 123 cases of lung cancer in chromate workers and found 46 squamous cell carcinomas, 66 anaplastic types and 11 adenocarcinomas. Great vanrations have been reported for the period from the first exposure to chromates and the diagnosis of cancer (6). A time period as long as 47 years has been found, mean value from different authors seems to be about 20 years with observed to expected ratios of up to about 40. Cancer of the nasal cavities has been reported in chromate workers, as well as laryngeal cancer (6). Mancuso (27) and Mancuso and Hueper (26) stated in 1951 that inhalation of chromite ore dust characterized as insoluble chromium might be an important factor in the production of lung tumors. None of their lung cancer deaths were found in workers exposed predominantly to soluble chromium compounds (hexavalent). Mancuso (37) recently published a followup of this group and stated that both exposure to trivalent and hexavalent chromium constitute a cancer risk. The original group was divided into workers exposed to predominantly insoluble or soluble compounds. Mancuso found that the lung cancer death rates increased by a gradient level of exposure both to insoluble and to total chromium. None of the groups had exclusive exposure to either type of compund. There is no information about a possible gradient exposure to soluble (hexavalent) compounds in the group exposed mainly to iMsoluble compounds. Nor is there information about a gradient exposure to insoluble, nor to slightly soluble hexavalent compounds in either group. The simiklAr increased risk in both groups as estimated from the death rates may indicate that such exposure is reasonable. If not, trivalent chromium should be an equally potent carcinogen as are the hexavalent compounds.
In a report by the United States Public Health Service the exposure pattern for 10 cases of lung cancer based on weighted avterage exposures by jiob classification has been reported (30). A low risk was contributed to chromite exposure and to exposure to soluble compounds. It was concluded that exposure to acid-soluble, water-insoluble compounds in the roast and residue constituted the excess nrsk. Most epidemiological reports contain no classification of exposed workers to specific jobs or exposure patters, but in a recent paper, Hayes (38) indicates that the high risk is related to the "wet" part of the proces where mainly water soluble hexavalent compounds are handled. The author states, however, that most workers had not been exclusively employed in a single part of the plant.
The major secondary use of chromium is in chromate pigment production, in the plating industry and for taining. Gross and Kolsch in 1943 (39) reported 8 cases from the pigment-producing industry, and Baetjer (31) reported 11 cases, but some of the latter had been reported before. An increased risk of respiratory cancer was demonstrated by Langard and Norseth (28) (risk ratio 38) in a pigment-producing plant, but the risk estimate was only based on three cases. Assuming a Poisson distribution, the inreased risk is, however, significant at the 0.05 level. The workers were exposed to lead and zinc chromate at levels up to 1.35 mg/im3 as chromium. In a survey of U.S. pigment plants (40), an increased risk was found for workers exposed to both zinc and lead chromate, and a corresponding increased risk was found for workers with exposure classified as high for lead and zinc chromates by Davies (35). Maltoni (41) has reported 21 cases of lung carcinoma in a group of 200 chromium pigment workers. Cytological examinations were performed on a group of 116 workers and the incidence of abnormal cells (class III-VI Papanicolaou) was about 25% as compared to 6% in the PVC industry, 1.6% in the chemical industry in general, and 0.6% in heavy smokers.
Of specific interest for the evaluation of soluble versus insoluble, or slightly soluble, chromate compounds as cancer causing agents is the cancer risk in the plating industry because of the uniform exposure to soluble compounds. Royle (40, 41) compared a population of platers exposed to chromium (VI) oxide to a reference population matched with respect to age, sex and smoking habits. He analyzed 246 certified causes of death and found a higher number of malignancies than expected for lung and pleura, for gastrointestinal tract and for other unspecified sites, but only total deaths from malignant neoplastic disease were significantly increased (51 compared to 24 expected, significant at the 1% level). Waterhouse (43) in a preliminary communication, found total deaths from malignant disease above expected number 'in a population of chromium platers, but the difference was not sta-124 tistically significant. He found, however, 49 deaths from lung cancer among male employees compared to 34.88 expected (p < 0.05). The significance of these reports is uncertain. Okuba and Tuschiya (44) have reported the results from an epidemiological study of 589 plating firms in Japan. No increased risk for respiratory cancer was found, but the conclusions are uncertain because of short observation time and a high loss (10%) to follow up. If exposure to hexavalent chromium in the plating industry constitutes a cancer risk, the risk is certainly much less than in the chromate producing or in the chrome pigment industry with the expo. sure levels which have been prevailing in these industries for the last 20 years.
No epidemiological investigations have been published for the large group of chrome pigment users m various other industries, but based on experience from pigment production, these workers must be regarded as being at risk if there is a significant exposure. A large group with a possible risk is welders and gnnders exposed to welding aerosols from stainless steel or from working with chromate primed materials. With stainless steel welding, about 70% of the total chromium in the aerosol is in the hexavalent fonn as soluble sodium or potassium monochromates (45), and fume particles have been shown to be mutagenic (46). Both the relative amount of hexavalent and the solubility of these salts seem, however, to vary with the type of welding (47, 48). Exposure levels for total chromium up to 1 mg/i3 as chromium have been reported for welders (47, 49, 51). In a recent epidemiological investigation of workers in a ferrochromium plant Langard (29) reported a suspect increase of respiratory cancer. Cancer incidence among 976 workers employed before 1960 and employed in ferrochromium or ferrosilicon production was investigated. Nine cases of lung cancer were found in the total population, seven in the subpopulation of 325 employed in ferrochromium production. The expected number of respiratory cancers was 3.1 and 1.8 using national and local expected incidence rates, respectively, and less than 1 compared to an internal reference popula'tion. Dust concentrati'ons (total chromium) of up to 1.3 mg/r3, were recorded at the time of the survey, from 11 to 33% were water-soluble components. Axelson (50) could not demonstrate an increased nrsk of respiratory cancer in a corresponding investigation, but the industrial i process mn this case did not produce significant amounts of hexavalent chromium compounds (solu-

Animal Experiments
A review of animal data up to 1972 has been published by the International Agency for Reearch on Cancer (3). The results support the nassumption that the most potent carcinogenic chrocompounds are the slightly soluble hexavalent ts, specifically calcium chromate. This compound is carcinogenic by tracheal instillation, and by gintramuscular and intrapleural injection. Calcium thromate is the only chromium compound which has given respiratory cancer after inhalation in ,1an imal experiments (56), although tests in progress with suggested positive results have been reported for unspecified mNtrmediates from chromate produc--tion (30 Using the tracheal implantation technique of dLaskn (57, 57), Levi (59) demonstrated a carcinogenic "'effect of calcium chromates. Also zinc chromate gave positive results. One cancer was reported in a ,..group of 100 rats exposed to chromic acid compared -to 8 with calcium chromate and 3 with zinc chromate. The chromic acid results should therefore be -regarded with some caution. Implantation of chro-August 1981 mic chromate alone did not give any cancers, but after the dispersion of the same substance im silica, three cancers were found. No explanation is given for this result. Intermediates from chromate production have also been shown to be carcinogenic in tracheal implantation and possibly as mentioned after inhalation (30,60). The specific active agent cannot be identified in the latter experiments.
In Vitro Studies Important recent developments in the understanding of chromium carcinogenicity have come from the study of mutagenic activity in bacteria and from work with isolated cell systems. Metals as mutagens have recently been reviewed by Flessel (10), and Sunderman (11) has revi'ewed the effect of metals on genetic material both in isolated biochemical systems, in microbiological systems and in in vitro cell systems.
Chromates are mutagenic in Escherichia coli and in Salmonella typhimurium (60-72) and influence DNA repair in Bacillus subtilis (69,73). Venitt and Levi (60) and Petrilli and de Flora (62) in a comparative study found no significant differences in dose-response between the soluble and slightly soluble compounds in either the Salmonella or the Coil system. Chromates seem to be directly acting mutagens as no activation by microsome fractions is necessary. De Flora (19), on the contrary, found a deactivation of sodium dichromate as a mutagen in the Salmonella system by adding liver microsomal fraction. This has been confirmed by Lofroth (18) and PetriUi and de Flora (67). Nishioka (73) found that the mutagenic effect disappeared when the hexavalent salts were reduced to the trivalent form in the experimental system using sodium sulfite. Oxidation of the trivalent form to the hexavalent in the test system, on the other hand, increases the activity (68). Trivalent chromium salts seem in general to be without or with a very weak mutagenic activity in bacterial systems.
Of particular interest is a theory set forward by Lofroth (18) that the causative agent in chromium mutagenesis is trivalent chromium bound to genetic material after reduction of the hexavalent form by a NADPH dependent microsomal enzyme system. The decrease in activity reported by the reduction of the hexavalent form to the trivalent form (19,66,73), might then be explained by a decreased uptake of the trivalent form by the cells.
Potassium dichromate is mutagenic in yeast (74), and soluble hexavalent salts induce morphological transformation and chromosomal aberrations in manunalian cells in vitro (69,(75)(76)(77)(78)(79)(80)(81). Chromosomal aberrations are also found in workers engaged in 125 chrome production (82). Nakamuro et al. (69) found that both hexavalent and trivalent soluble salts had this effect, but the hexavalent salts were the most potent. Newbold et al. (79) could not, however, demonstrate effects of lead chromate (insoluble hexavalent). Soluble trivalent salts had no such effect (79,80). In hamster embryonic cells the effect of potassium dichromate disappeared when chromium was reduced to the trivalent form in the experimental system (77). Sirover and Loeb (83) found increased error frequency in DNA synthesis in vitro with both trivalent and hexavalent chromium with a 30% error with chromium concentrations of0.64 mM (trivalent) and 16mM (hexavalent). These results support the trivalent form to be the active agent at the site of action. Trivalent chromium salts have also been shown to give positive results in other experimental model systems by altering the physiochemical properties ofmacromolecules (84)(85)(86)(87).
Various chromium compounds may influence the metabolism of genetic materials (88)(89)(90)(91)(92)(93). Levis et al. (88)(89)(90)(91) claim, based on studies of DNA/RNA synthesis in hamster fibroblasts (BHK line) and human epithelial-like cells (HEp line), that the hexavalent form is reduced by passing the plasma membrane, but assume that trivalent chromium is the active agent within the cell. Both the soluble hexavalent and trivalent salts decreased the DNA synthesis, but after adding either form only the trivalent was found inside the cells. The necessary concentration was a 100 times higher for the trivalent forn than for the hexavalent fonn to get an effect. Hexavalent chromium first stimulated nucleoside uptake by the cells, then -inhibited the uptake; amino acid uptake was only inhibited. The trivalent form had no such effects (91,92). Also an effect of potassium dichromate on mitosis in mammalian cells has been demonstrated (93).

Evaluation of Present State of
Knowledge; Chromium as the glucose tolerance factor has an essential function in humans. High concentration of chromium is found in RNA and a stabilizing function of chromium has been suggested. Only the hexavalent and the trivalent chromium compounds have biological importance. The hexavalent form is a strong oxidizing agent and is almost always found together with oxygen. The hexavalent form is rapidly transformed to the trivalent form in contact with organic materials. Hexavalent chromium salts readily pass biological membranes, trivalent chromium does not seem to do so. Once in the cell, the hexavalent form is rapidly reduced to the trivalent 126 form. The biological effects of chromium, including the carcinogenic effect are results of this transfor. mation, subsequent binding of the trivalent form to ligands within the cell or both. The trivalent fornm has a strong tendency to form coordination complexes and chelates in aqueous solutions, it may precipitate at physiological pH because of polynucleate complex formation with OH-ions.
An increased risk of respiratory cancer has been irrefutably established for workers in the chromate industry. A corresponding risk has been shown in the chromate pigment producing industry. An increased risk of respiratory cancer has not been conclusively established in ferrochromium production, in the refractory brick industry or in the plating industry. Secondary users of chromate pigments, including welders and grinders, must be regarded as being at risk on theoretical grounds, but epidemiologically an increased risk of cancer has never been shown. The number of epidemiological investigations are, however, limited for all groups except for workers in the chromate producing industry. An increased risk of cancer at other sites than the respiratory tract (gastrointestinal tract) has been suggested, but not finally established.
Negative results from the plating industry indicate that soluble chromates either do not cause cancer or are less potent carcinogens than slightly soluble compounds (calcium chromate). Based on results from the chromate producing industry is it not possible to either verify or dismiss this theory because of insufficiently described exposure patterns. Because of analytical procedures, slightly soluble compounds seem mostly to be regarded as trivalent. This may not be correct as also slightly soluble hexavalent salts may be present in the workingenvironment(calcium-zinc-leadchromates).
Animal experiments support the assumption that the slightly soluble hexavalent salts (calcium chromate) are the active agents, but inhalation expenments are few. Tracheal instillation experiments are, however, highly suggestive that this is the case. Soluble and slightly soluble hexavalent salts (calcium chromate) are equally potent for causing mutations in bacterial systems, and both soluble and slightly soluble salts cause chromosomal damage or cell transformation in isolated celLs. Trivalent chromium may cause mutation and cellular transformation in titro in high concentrations. The differences in activity of the soluble and slightly soluble hexavalent salts in the practical exposure situation may be related to retention time on the mucous membrane in the respiratory tract, to a time/concentration factor caused by slow solubilization and removal, or to differences in the intracellular Environmental Health Perspectives tics of the two differentt classes of compunds.
ere is at present no epidemiological evidnece .trivalent chromium to be carcinogenic. Epidelogical evidence is supported by animal experits and particularly by in intro results and emical considerations. When tested in syswhich require that chromium passes biological branes to exert an effect, reduction of the valent form to the trivalent reduces or abolishes effect. Similarly, trivalent chromium which is hout effects im such systems when oxidized to e hexavalent form in the system has genotoxic kets. The trivalent form has a more potent otoxic effect than the hexavalent in systems thout biological membrnes, and as the hexavalent is rapidly reduced in contact with biological terial, the active substance at the site of action is bably the trivalent form. There is no agreement to in what part of the cell this reduction takes e, probably at several sites. It is neither known what extent the effects caused by hexavalent hrouum are directly related to the reduction pocess or to the subsequent binding of the trnvaent form. Available evidence indicates, however, tat the mechanism of chromium carcinogenicity fvolves a direct interaction of trivalent chromium with genetic material or genetic processes, though bdirect interaction by virus or other chemical iubstances has been suggested (95-97).
No risk estimation of cancer in workers exposed o chromium in the chromate producing industry ased on reliable dose-response information is posabe at present. Some exposure data have been ,ecorded, but the chemical characterization of data n the very complex exposure situation in the hromate producing industry is insufficient. A comgicating factor is that most workers have probably i more than one job. It should, however, be ioted that even if there are relatively few investirations reported from the plating industry, refracory brick industry and ferrochromium industry, he risk of cancer must have been less than in the bromate producing industry and in the chrome pigment industry with the exposure levels which ave been prevailing for the last 20 years. Levels .r exposure and risk cannot, however, be estabhshed.

3aps in Knowledge and Research 4eds
Chromium carcinogenesis is in good accordance with the somatic mutation theory for cancer, but why and how the mutant cell develops into a cancer s as for other chemical carcinogens, not known.

Wugust 1981
Once this has been mentioned, the most important gaps in our knoweldge are data for risk estimation based on dose-response for specified industrial processes or for specific chromium compounds.
This problem calls for research with model systems and for further epidemiological research. Further studies on the mechanism of the carcinogenic action is necessary for the evaluation of the interrelationship between the trivalent and hexavalent form in carcinogenesis. It is not known if trivalent chromium might penetrate the biological membrane under specific ci'rcumstances. Animal experiments with defined chemical compounds of chromium are urgently needed. Specifically the relationship between soluble and slightly soluble compunds should be evaluated in such systems. This problem is difficult to solve in epidemiological investigations because of the mixed exposures in many industries. Inhalation or intratracheal instillation should be the preferred methods.
Further epidemiological investigations should be started in all populations where exposure to chromium takes place. Not only cancer in the respiratory tract should be looked for, a possible increased incidence of cancer in all organ systems should be investigated. Cancer in the gastrointestinal tract may be of particular interest. This is also the case for exposure to trivalent chromium. Mixed exposure with smoking is an important problem in risk estimation. The amount of chromium inhaled from cigarettes, and the chemical form are not known. At present we have recorded an increased number of workers with cancer in some high risk groups. It is unrealistic to assume that low risk groups may be evaluated in a similar way. For low risk estimation both in epidemiological work and when using experimental animals, the necessary number of individuals under study soon becomes prohibitive.
Extrapolation from high exposure in relati'vely small groups for short time periods to low level, long-time exposure in large groups for a disease like cancer is hazardous.
Recent developments in applying model systems for mutagenicity studies are promising for evaluation of the mechanisms of cancer caused by metals. Furthermore, epidemiologic studies should be refined with a more detailed work history for the exposed groups. This is difficult for retrospective studies, especially for metal carcinogenesis. Care should be taken to record relevant dose-indicators for prospective and for retrospective work. Intemnational collaboration in epidemiological work should be encouraged to inerease the size of the exposed groups, and registration of workers and exposure levels should be introduced in industry as recoi-127 mended by ILO (98). Epidemiological follow-up of the benefits of early diagnosis is important. Early diagnosis must, however, not be sold as a cure for industrial cancer.
Lastly, let us not forget that even if there are definite gaps in our knowledge in chromium toxicology, our knowledge is sufficient for us to work for an improvement of the worldng conditions where excessive and unnecessary, exposure to chromium takes place in our iMdustry, or in the general environment. Thus, epidemiological investigations as recommended are not necessary for proving that some chromium compounds cause cancer, but they are needed for estimating a maximum risk based on inevitable exposure of populations at hand.