SCIENTIFIC QUESTIONS TO THE CSTEE ON CADMIUM

Note: Following the terms of reference, this opinion focuses on specific questions related to human health issues and does not cover ecosystem protection.

Questions to CSTEE

1. What is the most appropriate study and effect level to be used as the starting point for deriving a limit value for cadmium in air based on non-cancerous renal effects?

2. Accepting that cadmium is a known human carcinogen, and given the uncertainty in the unit risks due to confounding exposure to arsenic, does the committee believe it appropriate to establish a limit value based on cancer effects, and, if so, how?

3. Does the committee consider that an ambient air quality standard based on non-cancer effects would also provide a high level of protection against cancer effects?

Opinion:

1. The CSTEE considers that the studies of Buchet et al. (1990) and Järup et al. (2000) provide a valid scientific basis for the derivation of a limit value for the non-cancer renal effects of cadmium. The appropriateness of these studies, which were based on cohorts of the general population, is further supported by the compatibility of the limit value derived from them with that suggested by other studies carried out on workers. Furthermore, the CSTEE can support the use, as done in the Position Paper, of an overall uncertainty factor of 100 in the derivation of a limit value based from the above two studies.

2. Given the degree of confounding by arsenic exposure in the epidemiological studies, the CSTEE does not believe that it is possible to establish a reliable limit value for cadmium based on the induction of cancer effects.

3. The available data do not permit a reliable assessment of the degree of protection against cancer which would be provided by a limit value based on the non-cancer effects of cadmium. The CSTEE estimates that a limit value of 5-6.5 ng/m3, based on non-cancer effects, could be associated with an excess lifetime risk of up to about 20 cases of lung cancer per million.

Justification of the Opinion:

Question 1

Chronic exposure to cadmium causes kidney damage, starting as impairment of tubular reabsorption which, in more severe cases, can develop to alterations in glomerular function. Early tubular damage is detected as increased urinary excretion of low-molecular weight proteins (β2-microglobulin, α1-microglobulin, retinol-binding protein), enzymes such as N-acetyl-β-D-galactosidase), other small molecules and calcium. Proteinuria resulting from relatively high exposure to cadmium (above 5-10 µg/24 hours urinary cadmium) appears to be irreversible and is associated with the development of more severe kidney malfunction (Roels et al., 1989). On the other hand, there is considerable uncertainty over the pathological significance of small increases in proteinuria found after low-level environmental exposure to cadmium (less than 2 µg/24 hours urinary cadmium). Furthermore, recent evidence suggests that that such increases may recede slowly (over a period of years) if exposure to cadmium is reduced (Hotz et al., 1999). However, the CSTEE notes that the discussion on the risks from cadmium in the ambient air concerns chronic exposure, a condition under which reversibility of the effects does not seem relevant. It is further noted that in most areas of Europe it would take a very long time to decrease the exposure to cadmium via food (mainly cereals, roots and vegetables), even if the emissions to the environment could be stopped completely, as the half-time of cadmium in soil is very long.

Data useful for setting limit values for cadmium have been provided from studies on members of the general population as well as from studies on occupationally exposed cohorts.

Studies in the general population

Many general population studies have utilised urinary excretion of cadmium as a measure of dose. As compared with the use of the air or food concentrations as a dose metric, urinary cadmium has the advantage of being a biological indicator of the total body burden which, because of the long persistence of cadmium in the body (half-life of around 20 years), reflects long-term exposure arising from all sources and routes. Furthermore, urinary cadmium correlates closely with cadmium concentrations in the kidney (target tissue), at least at low exposure levels (Roels et al., 1981; Orlowski et al., 1998).

Two studies are particularly useful for risk estimation in the general population. Buchet et al. (1990) carried out a cross-sectional study involving 1699 subjects aged 20-80 years living in two regions of Belgium with different levels of environmental cadmium contamination. Multiple logistic regression analysis provided estimates of urinary cadmium cut-off levels above which at least 10% of the population is expected to have abnormal values of urinary markers, with the limit of the normal range taken as the 95th percentile of values observed in subjects without diabetes, urinary tract disease or treatment with analgesics. Thus the calculated urinary cadmium cut-off levels were associated with a 5% prevalence over and above a 5% background prevalence. Because the whole of the studied population was exposed to cadmium, it is not known how much of this background prevalence may have been due to cadmium. Among urinary protein markers examined, N-acetyl-β-galactosidase showed the lowest cut-off (2.7 µg Cd/24 hours). An even lower threshold (1.9 µg Cd/24 hours) was noted for increased urinary calcium. While the biological relevance of this to kidney toxicity is uncertain, it is of possible relevance to the recently reported association of increased incidence of osteoporosis and bone fragility with low level exposure to environmental cadmium (Staessen et al., 1999; Alfven et al, 2000).

A recent Swedish study (Järup et al., 2000), which is not considered in the Position Paper; examined urinary excretion of α1-microglobulin in 1021 environmentally or occupationally exposed subjects, aged 16-80 years, living in or near a region close to a cadmium battery plant. Multiple logistic regression analysis indicated that a 10% prevalence of increased proteinuria (on top of a 5% taken as background), was associated with urinary cadmium of 1 µg/g creatinine (corresponding to excretion of 1-2 µg cadmium/24 hours). Exclusion from the calculations of the 222 occupationally exposed subjects did not affect the outcome of the calculations. Although certain factors which have been reported in other studies to significantly influence microproteinuria alone or in association with Cd exposure (primary renal disease, diabetes, hypertension, use of analgesics, smoking and other medications), were not controlled for in the statistical analysis, the conclusions of this study are in agreement with those of the study by Buchet et al. (1990) and are compatible with the possibility of an association of microproteinurea with even lower cadmium exposures than suggested by that study. They are also in agreement with a report (Orlowski et al., 1998) indicating that a urinary cadmium concentration of 1.7 µg/ g creatinine corresponds to a cortical kidney cadmium concentration of 50 µg/g, a concentration currently believed to be the threshold for the induction of kidney malfunction (Järup et al., 1998).

Urinary cadmium excretion has been related to chronic cadmium intake using a toxicokinetic model which takes into account both inhalation and oral intakes (US EPA, 1999). The uptake of environmental cadmium depends on the chemical speciation and the route of exposure. Although the speciation of environmental cadmium has not been fully characterised, cadmium oxide is probably the most common form of human exposure, while cadmium sulphide is additionally found in occupational settings. Airborne cadmium is primarily associated with particulates, those found in occupational settings showing a smaller mean aerodynamic diameter (hence a different fractional lung deposition rate) than those found in ambient settings. Systemic absorption after inhalation is high (up to 90% of the amount deposited in the lung) for cadmium oxide, but only about 10% for cadmium sulphide. Absorption after oral ingestion is much lower (3-8%). In the application of the model to estimate the intake necessary to cause the urinary excretion levels of interest, a series of conservative assumptions was made, including that a) exposure involved cadmium oxide (a relatively well absorbed species), b) the fractional lung deposition was 0.21 (half-way between the values corresponding to ambient and occupational cadmium-containing particulates), and c) that the values for the systemic absorption were 90% for the cadmium deposited in the lungs and 5% for that reaching the GI tract. For a population with an oral intake of 0.14 µg/kg.day (the mean dietary intake of the US general population), the model predicts that a urinary excretion of 2.7 µg Cd/24 hours, identified by Buchet et al. (1990) as the threshold for the induction of cadmium-related proteinuria in 5% of the population, would result from life-time inhalation of 650 ng/m3 cadmium. A correspondingly lower exposure level would be expected to correspond to the lower threshold suggested by the Järup et al. (2000) study.

Uncertainty factors: The data of Buchet et al. (1990) and Järup et al. (2000) suggest that the dose-response curve for the induction of proteinuria has a positive slope down to urinary cadmium excretion levels lower than 2 μg/24 hours. Although the slope below 2 μg/24 hours is smaller than above this concentration, possibly suggesting different underlying biological mechanisms, these mechanisms are currently unknown and therefore it cannot be assumed that they are of different health significance. Given that, even at the selected cut-off level of 2.7 µg/24 hours, a significant fraction of the population (over 5%) would be expected to be affected, a factor of at least 10 should be used to derive a NOAEL.

Subgroups with increased sensitivity to cadmium include persons with degenerative kidney damage (e.g. the elderly and diabetics), as well as persons with lower iron stores (e.g. women) who may take up cadmium more readily (Järup et al., 1998). Additional considerations which need to be taken into account in deriving a limit value are the following:

a) Mean dietary intake of cadmium in some countries is higher than the value of 0.14 µg/kg.day assumed above, appearing to be closer to or even exceed 0.2 µg/kg.day in some European countries (Järup et al., 1998; Biego et al., 1998; Muller et al., 1998; Coni et al., 1991). For subjects near the top of the consumption range, dietary exposure is likely to be 2-3fold higher. Higher dietary exposure would also apply to specific population subgroups, e.g. shellfish eaters and vegetarians. Increased dietary intake would leave a smaller margin for inhalation exposure. For example, the toxicokinetic model employed by US EPA predicts that, for subjects consuming 0.36 µg/kg-day (roughly representing the intake of shellfish eaters), the threshold of urinary excretion would be reached at approximately 2fold lower inhalation exposure than indicated above. A further source of exposure that needs to be considered is tobacco smoking, which can result in an amount of absorbed cadmium approximately equal to that absorbed from the diet.

b) Cadmium concentrations in the kidney cortex of the middle-aged general population appear to be in the range 15-50 µg/g, close to the critical value of 50 µg/g already mentioned as the threshold for the induction of kidney malfunction. This implies that, if the critical kidney cortex concentrations are to be avoided, further exposure needs to be strictly limited.

In view of the considerations discussed above, the adoption of an additional uncertainty factor of 10 to derive a limit value from the NOAEL, as suggested in the Position Paper, seems justified, leading to a limit value of 6.5 ng/m3.

Occupational exposure studies

Many occupational studies utilised cumulative airborne exposure to cadmium as the dose metric, and therefore suffer from the disadvantage of depending on the estimation of past workplace exposures. Nevertheless, the data obtained from a large number of such studies demonstrate clearly increased incidence of low-molecular weight proteinuria at cumulative exposures greater than 500 μg/m3 x years (TWA concentrations over 8 hours per day, 5 days per week), and are compatible with a threshold of not less than 100 μg/m3 x years. While rare reports of effects at a lower cumulative exposure cannot be completely ignored, they are difficult to assess because of the very small numbers of affected individuals and the occurrence of microproteinuria in non-exposed populations.

Uncertainty factors: A cumulative occupational exposure of 100 μg/m3 x years is equivalent to a lifetime, continuous exposure of the general population to an atmospheric concentration of 270 ng/m3. For the derivation of a NOAEL from this, it should be remembered that the figure of 100 μg/m3 x years is not a LOAEL but a threshold below which effects are thought unlikely to occur. Thus an uncertainty factor smaller than 10 may be used. On the other hand, as already indicated, evidence from studies utilising urinary cadmium as a measure of dose suggests that the dose-response curve has a positive slope down to background exposures. For this reason, and having in mind occasional reports of effects occurring at lower occupational exposures (Ellis et al., 1985), it would be prudent to use an uncertainty factor of 5, as suggested in the Position Paper.

The cohorts of the above mentioned studies were frequently limited to healthy males of working age, and may not fully reflect the full range of individual susceptibilities likely to be encountered in the general population. Although a recent attempt (TERA 2000) to compare the quantitative outcomes of studies in workers and the general population did not find evidence of a "healthy worker" effect, the reliability of its conclusions is limited (heterogeneity of data reporting in different studies, absence of consideration of age effects). On the other hand, examination of data obtained from comparable cohorts of workers and members of the general population (Buchet et al., 1980) provides evidence that the latter may have higher sensitivity. For this reason the use of an additional uncertainty factor of 10, as suggested in the Position paper, seems justified. This would lead to a limit value of 5 ng/m3, in agreement with that derived from the general population studies which made use of urinary cadmium excretion.

Conclusion: The studies of Buchet et al. (1990) and Järup et al. (2000) provide a valid scientific basis for the derivation of a limit value based on the non-cancer renal effects of cadmium. The appropriateness of these studies is further supported by the compatibility of the limit value derived from them with that suggested by other studies. Furthermore, the CSTEE supports the use, as done in the Position Paper, of an overall uncertainty factor of 100 in the derivation of a limit value based from the above two studies.

Question 2

Cadmium is classified as a category 1 carcinogen (known human carcinogen) by IARC and US EPA. This classification is based largely on the findings of epidemiological studies in the US showing increased incidence of lung cancer in cadmium-exposed smelter workers (Thun et al., 1985; Stayner et al., 1992). Although the presence of confounding by arsenic and tobacco smoking in these studies was recognised, it has been concluded by the above agencies that such confounding is unlikely to account for all the excess of cancers observed, and a cadmium-specific unit risk factor was calculated by the US EPA using the data of Stayner et al. (1992) [US EPA, 1999].

In order to counter criticisms of inadequate control of arsenic confounding, Thun et al. (1985) and US OSHA (1999) [using the data of Stayner et al. (1992)] made use of estimates of the cumulative exposure of the study cohorts to arsenic, and the known unit risk for arsenic, to calculate the expected contribution of such exposure to the observed lung cancers, and came to the conclusion that it could account only for a small fraction (under 10%) of them. The validity of this conclusion has been criticised on account of inadequacies in the estimation of the arsenic exposures employed in the calculations (Lamm et al., 1992; Doll, 1992; Sorahan and Lancashire, 1994). This criticism has been further developed by Sorahan and Lancashire (1997) who, using improved estimates of exposure to cadmium and arsenic, concluded that elevated lung cancer risks are associated only with mixed exposure to cadmium and high concentrations of arsenic, and that at this time it is not possible to distinguish between the contributions of cadmium and arsenic to the overall cancer risk.

Attempts have also been made to derive unit risk factors from both animal experiments and epidemiological data, using the data of Takenaka et al. (1983) for induction of lung cancers in male Wistar rats exposed by inhalation to cadmium chloride. These data have been quantitatively analysed and used to derive an upper bound 95% value of the unit risk factor which corresponds to a limit value, for <1 in a million risk, of approximately 0.03 ng/m3 (US EPA, 1999). This limit value is nearly 10fold lower than that derived from the epidemiological data [0.2 ng cadmium/m3, US EPA (1999)] which, as already pointed out, are likely to overestimate the risk associated specifically with cadmium. In considering these risk estimates, it should be remembered that, in addition to the difficulties related to inter-species dosimetric extrapolation and susceptibility differences, which normally complicate the use of animal data for human risk assessment, the situation is further complicated in the case of cadmium by questions of chemical speciation and the limited understanding of the mechanistic basis of cadmium carcinogenicity. Therefore the available animal studies are not considered suitable for deriving a reliable limit value for the protection of humans.

Conclusion: It is concluded that the available data do not permit the estimation of a reliable unit risk or the establishment of a limit value for cadmium based on the induction of cancer effects.

Question 3:

Based on the arguments presented in the response to Question 1, a limit value in the range 5-6.5 ng/m3 would be derived, based on protection against cadmium effects on the kidney.

The risk calculations carried out by US EPA (US EPA, 1999) using the data of Stayner et al. (1992) resulted in a unit risk of approximately 4.3x10-3 (µg/m3)-1. At an exposure of 5 ng/m3, this would give rise to an excess life-time risk of about 20 cases of lung cancer per million. To the extent to which the cancers observed in the epidemiological studies on which the calculation of the unit risk was based might have been due in part to confounding factors (arsenic, smoking), this excess would be expected to be lower.

An addition factor that should be considered in assessing low-dose risks relates to the genotoxicity of cadmium. While cadmium is capable of inducing a range of genotoxic effects (including mutations and cytogenetic damage) directly, there is evidence that indirect mechanisms, such as inhibition of DNA repair and induction of oxidative stress, may also play a role in its genotoxicity. This would suggest that cadmium genotoxicity has a non-linear component, and that cancer risks at low doses may be smaller than the above calculation suggests.

References:

- Alfven T, Elinder CG, Carlsson MD, Grubb A, Hellstrom L, Persson B, Pettersson C, Spang G, Schutz A, Järup L (2000) Low-level cadmium exposure and osteoporosis. J Bone Miner Res 15, 1579-86

- Biego GH, Joyeux M, Hartemann P, Debry G (1998) Daily intake of essential minerals and metallic micropollutants from foods in France. Sci Total Environ 217, 27-36

- Buchet JP, Roels H, Bernard A, Lauwerys R (1980) Assessment of renal function of workers exposed to inorganic lead, calcium or mercury vapor. J Occup Med 22, 741-50

- Buchet JP, Lauwerys R, Roels H, Bernard A, Bruaux P, Claeys F, Ducoffre G, de Plaen P, Staessen J, Amery A, Lijnen P, Thijs L, Rondia D, Sartor F, Saint Remy A and Nick L (1990), Renal effects of cadmium body burden of the general population. Lancet 336, 699-702.

- Coni E, Baldini M, Stacchini P, Zanasi F (1992) Cadmium intake with diet in Italy: a pilot study. J Trace Elem Electrolytes Health Dis 6, 175-81

- Doll R (1992) Is cadmium a human carcinogen? Ann Epidemiol 2, 335-7

- Ellis KJ, Cohn SH, Smith TJ (1985) Cadmium inhalation exposure estimates: their significance with respect to kidney and liver cadmium burden. J Toxicol Environ Health15, 173-87

- Hotz P, Buchet JP, Bernard A, Lison D, Lauwerys R (1999) Renal effects of low-level environmental cadmium exposure: 5-year follow-up of a subcohort from the Cadmibel study. Lancet 354,1508-13

- Järup L, Berglund M, Elinder CG, Nordberg G, Vahter M (1998) Health effects of cadmium exposure--a review of the literature and a risk estimate. Scand J Work Environ Health, 24 (Suppl 1), 1-51

- Jarup L, Hellstrom L, Alfven T, Carlsson MD, Grubb A, Persson B, Pettersson C, Spang G, Schutz A, Elinder CG (2000) Low level exposure to cadmium and early kidney damage: the OSCAR study. Occup Environ Med 57, 668-72

- Lamm SH, Parkinson M, Anderson M, Taylor W (1992) Determinants of lung cancer risk among cadmium-exposed workers. Ann Epidemiol 2, 195-211

- Muller M, Anke M, Illing-Gunther H, Thiel C (1998) Oral cadmium exposure of adults in Germany. 2: Market basket calculations. Food Addit Contam 15, 135-41

- Orlowski C, Piotrowski JK, Subdys JK, Gross A (1998) Urinary cadmium as indicator of renal cadmium in humans: an autopsy study. Hum Exp Toxicol 17, 302-6

- Roels HA, Lauwerys RR, Buchet JP, Bernard A, Chettle DR, Harvey TC, Al-Haddad IK (1981) In vivo measurement of liver and kidney cadmium in workers exposed to this metal: its significance with respect to cadmium in blood and urine. Environ Res 26, 217-40

- Roels HA, Lauwerys RR, Buchet JP, Bernard AM, Vos A, Oversteyns M (1989) Health significance of cadmium induced renal dysfunction: a five year follow up. Br J Ind Med 46, 755-64

- Sorahan T, Lancashire R (1994) Lung cancer findings from the NIOSH study of United States cadmium recovery workers: a cautionary note. Occup Environ Med 51, 139-40

- Sorahan T, Lancashire R (1997) Lung cancer mortality in a cohort of workers employed at a cadmium recovery plant in the United States: an analysis with detailed job histories. Occup Environ Med 54, 194-201

- Staessen JA, Roels HA, Emelianov D, Kuznetsova T, Thijs L, Vangronsveld J, Fagard R (1999) Environmental exposure to cadmium, forearm bone density, and risk of fractures: prospective population study. Public Health and Environmental Exposure to Cadmium (PheeCad) Study Group. Lancet 353, 1140-4

- Stayner L, Smith R, Thun M, Schnorr T, Lemen R (1992) A dose-response analysis and quantitative assessment of lung cancer risk and occupational cadmium exposure. Ann Epidemiol 2,177-94

- Takenaka S, Oldiges H, Konig H, Hochrainer D, Oberdorster G (1983) Carcinogenicity of cadmium chloride aerosols in W rats. J Natl Cancer Inst 70, 367-73

- TERA (Toxicology Excellence for Risk Assessment) (2000) Cadmium Nephrotoxicity in occupational and general populations. Report prepared for US EPA.

- Thun MJ, Schnorr TM, Smith AB, Halperin WE, Lemen RA (1985) Mortality among a cohort of U.S. cadmium production workers--an update. J Natl Cancer Inst, 74, 325-33

- US EPA (1999) Toxicological Review: Cadmium and compounds. External Review Draft.

- US OSHA (1999), OSHA Preambles: Cadmium. Available online at http://www.osha.gov/pls/oshaweb/owasrch.search_form?p_doc_type=PREAMBLES&p_toc_level=1&p_keyvalue=Cadmium (revision date: July 30, 1999)