Regulations (Preambles to Final Rules) - Table of Contents Regulations (Preambles to Final Rules) - Table of Contents
• Record Type: Occupational Exposure to Cadmium
• Section: 5
• Title: Section 5 - V. Health Effects

V. Health Effects

A vast amount of literature exists which documents the adverse health effects from acute and chronic exposure to cadmium in both humans and animals. The primary adverse health effects which have been observed are lung cancer and kidney damage. This section on health effects will not attempt to describe every study ever conducted on cadmium toxicity. Instead, the most important studies will be reviewed and the testimony and comments submitted to the record of the rulemaking will be presented. For greater detail, the cited reviews and articles should be consulted.

A. Metabolism

Cadmium enters the human body by inhalation, by ingestion, and perhaps by absorption through the skin. Inhaled cadmium is more readily absorbed into the body than is ingested cadmium. Intake of cadmium by ingestion and skin absorption are considered to be of relatively less importance in occupational settings.

In occupational settings cadmium is inhaled in the form of either small particles of fume or larger particles of dust. The extent of deposition depends on the particle size. It is estimated that ten percent of the particles of approximately 5.0 micrometers mean mass diameter (MMD) are deposited in the lung, whereas 50 percent of the particles of 1.0 micrometer MMD are deposited in the lung. Of the proportion deposited, 20 to 25 percent is systemically absorbed. (Exs. 8-619; 8-086a, p. 107). Limited data on smokers indicate a high rate of absorption (10% to 50%) of inhaled cadmium (Exs. 8-86 and 29). Animal experiments have shown an absorption rate of between 10% to 60% of inhaled cadmium (Ex. 29). Thus, smoking habits and personal hygiene are of great importance as a source of indirect exposure in a cadmium-contaminated environment.

Many of the effects of cadmium are systemic. After initial exposure, inhalation, and absorption, cadmium is transported by the blood plasma, although the majority of the cadmium is bound to the blood cells. According to Clarkson et al.:

"Cadmium in blood is distributed between blood cells and plasma......In persons with industrial exposure to cadmium and in cadmium-exposed experimental animals, cadmium in blood is found mainly in erythrocytes...(Ex. 14-18, pp. 160-161)"

Cross-sectional studies of workers who had varying durations of exposure at varying levels of exposure have yielded little definitive information about the kinetics of blood cadmium and the distribution of cadmium between the plasma and blood cells. Therefore, cadmium in blood is most accurately measured in whole blood.

After a sudden increase in exposure, cadmium in blood increases rapidly during the first three to four months and then reaches an apparent steady level which is likely to reflect the average exposure during those months. When high exposures cease, blood levels of cadmium decrease with two distinct half-time components. One component has a half-time of a few months and probably reflects the turnover rate of red blood cells. The other half-time component is several years long and is likely to reflect body burden, presumably consisting of several components with half-lives varying from about one year in blood to about five years in other soft tissues (De Silva, Ex. 8-718; and Friberg, Ex. 8-068-B). Therefore, the transport of absorbed cadmium in blood is of crucial importance.

Cadmium transported to the liver induces the synthesis of metallothionein, a low molecular weight metal-binding protein. Cadmium becomes bound to this protein, forming a metal-protein complex which is then released back to the blood and transported to the kidney. In the kidney, the cadmium-metallothionein complex passes through the glomeruli and is reabsorbed by the proximal tubules. This complex can then be broken down by lysosomes, releasing unbound cadmium which can induce renal synthesis of metallothionein. In workers with only short-term exposures to low levels of cadmium, the cadmium will be bound again in the kidney to the locally produced metallothionein providing a protective effect from cadmium. However, after prolonged exposure, the binding process in the kidney becomes saturated leading to an increase in unbound cadmium which can result in toxic effects (Dr. Goyer, Tr. 6/6/90, pp. 129, 131).

Unlike other heavy metals such as mercury and lead, cadmium occurs in only one valence state, +2, and does not form stable alkyl compounds or other organometallic compounds of known toxicologic significance (Casarett, 1980, Referenced in Ex. 8-735, Casarett, 1986). Thus, it is elemental cadmium that is the toxic agent. In his testimony, Lars Friberg, M.D., Professor Emeritus at the Karolinska Institute in Stockholm where he was formerly Chairman of the Department of Environmental Health, an international expert and pioneer in the field of cadmium-induced kidney disease, past Chairman of the WHO task group on cadmium, and senior editor of a number of comprehensive monographs on cadmium, discussed the mechanisms by which cadmium is taken up by the human body and how cadmium ultimately affects the kidney. Dr. Friberg stated:

"One has firm epidemiological evidence, thanks to the data from Ellis and Roels......that there is an association between total cadmium in kidneys and effects..."(Tr. 6/6/90, p. 100)

The studies by Ellis and Roels are discussed below.

Regardless of the route of absorption or the type of cadmium compound, approximately one half to one third of the body burden of cadmium is found in the kidneys after chronic low-level exposure, with the highest concentrations found in the renal cortex. (Ex. 8-086a, p. 168). After long term exposure, one sixth and one fifth of the body burden are found in the liver and muscles, respectively. As exposure level increases, a greater proportion of the body burden of cadmium will be found in the liver relative to the kidney. Also, upon the onset of renal dysfunction, the level of cadmium in the kidney will decrease (Friberg, Tr. 6/6/90, p. 84). The half-life of cadmium in the liver, kidney and muscles is 5 to 15 years, 10 to 30 years, and more than 30 years, respectively. (Ex. 8-086a, p. 168).

B. Non-Carcinogenic Health Effects

1. Acute Effects

a. Evidence in Humans. A variety of adverse health effects may result from acute exposure to cadmium compounds. The most widely recognized effects are seen in the respiratory system from the inhalation of cadmium fumes and dust. Acute pneumonitis occurs 10 to 24 hours after initial acute inhalation of high levels of cadmium fumes with symptoms such as fever and chest pain. (Exs. 30; 8-086b, p. 4). In extreme exposure cases pulmonary edema may develop and cause death several days after exposure. Such symptoms have been reported among workers exposed to high concentrations of cadmium. For example, after a day's exposure to cadmium fumes, workers developed severe weakness, dyspnea, coughing and tightness of the chest. Chest radiographs showed signs of pulmonary edema (Ex. 8-41).

As in the case above, the exposure levels at which the adverse effects occurred were not recorded in many investigations of acute health effects from high cadmium exposures. Estimation of the exposure levels associated with acute respiratory effects has relied on measurements of the amount of cadmium found in the lung after death modified by assumptions about the proportion of cadmium fumes retained in the lungs. At one time, a lethal concentration of cadmium was considered to be not higher than (and probably lower than) 5 mg/m(3) over a period of eight hours (Exs. 8-41 and 29). This number was estimated by assuming that 11% of the cadmium fumes were retained in the lungs. Given that the victims were exposed for 5 hours, the average concentration was estimated to be 8.6 mg/m(3). This is equivalent to an 8-hour exposure of approximately 5 mg/m(3). Because this estimate rests on a number of assumptions used to derive this exposure level, there is some uncertainty as to the accuracy of this estimate. Also, the amount of cadmium measured in the lung of the fatal cases may have been higher than the amount necessary to cause death. It should also be noted that this type of estimate is for lethal concentrations, and that lower concentrations, may give rise to acute symptoms and significant lung damage without resulting in death.

Little actual exposure measurement data are available on the level of airborne cadmium exposure that causes such immediate adverse health effects. More recent studies have revealed that an eight hour exposure to 5 mg/m(3) should not be regarded as the lowest concentration that can give rise to a fatal poisoning. It is reasonable to believe a cadmium concentration of approximately 1 mg/m(3) over an eight hour period is "immediately dangerous" (Exs. 8-86B; 29). Such exposures can ultimately be life-threatening. In response to questions during the hearing, George Kazantzis, M.D., Emeritus Professor of Occupational Medicine at the University of London, Honorary Consultant Physician at the Middlesex Hospital in London, and a member of the WHO task group on environmental health, testified that he found a statistically significant excess of worker deaths due to chronic bronchitis in his 17 plant cohort study in the U.K. These deaths were, in his opinion, directly related to high cadmium exposures of 1 mg/m(3) or more (Tr. 6/8/90, pp. 156-157).

Although there are little data on the lowest cadmium exposure level that may trigger an acute effect, it is known that a few days of cadmium exposure in excess of 200 ug/m(3) will result in elevated levels of cadmium in the blood and urine. Based upon his experience, Dr. Friberg said that an eight-hour time-weighted-average (TWA) exposure of 75 - 100 ug/m(3) should never be exceeded in order to prevent adverse health effects (Ex. 144-15).

Experimental animals exposed to various cadmium compounds through various routes of exposure have experienced acute pulmonary effects. In a number of studies reviewed by Friberg (Ex. 8-086b, p. 2), NIOSH (Ex. 4-02) and EPA (Ex. 8-619), cadmium exposures ranging from 5 to 10 mg/m(3) over 15 to 120 minute periods were sufficient to induce significant increases in lung weights indicative of pulmonary edema. Also, rats exposed to cadmium aerosol at 60 mg/m(3) for 30 minutes died within 3 days from pulmonary edema (Ex. 8-402). Multiple experimental studies confirm these findings of acute pulmonary effects. In addition, animals injected with cadmium compounds have exhibited acute effects in the testes, ovaries, liver and blood (Exs. 8-420, 8-86B, 8-370, and 8-668).

2. Renal Effects

a. Evidence in Humans - Introduction. The human kidney is a filtration mechanism for the blood. It has three major functions that are vital to maintaining normal health: removing wastes; preventing leakage of essential elements and chemical compounds from the body; and providing homeostasis. According to Robert A. Goyer, M.D., Professor and Chairman of the Department of Pathology at the University of Western Ontario, Chairperson of the World Health Organization (WHO) Task Force on Environmental Health Criteria for Cadmium in 1989 and coauthor of the WHO Environmental Health Criteria Document on Nephrotoxicity (1989), these:

"......are the ones by which we measure the health of the kidney and the health of the person...."(Tr. 6/6/90, p. 123)

The kidney cortex including the nephrons is the anatomic division of the kidney of most concern when considering cadmium's toxicity. Each nephron is a functional filtration unit of the kidney and is composed of glomeruli and tubuli. Each glomerulus is a collecting system made up of capillaries that filter the blood and prevent leakage of all large particles with a molecular weight greater than 30,000. The molecular weight of a molecule is the weight that is equal to the sum of the atomic weights of its constituent atoms. Normally, as blood passes through the kidney, high molecular weight proteins such as albumin, immunoglobulin G, and a variety of glycoproteins do not cross an intact glomerular basement membrane into the kidney tubule. The function of the proximal tubule is to filter out small organic compounds. These include essential electrolytes, metals, and elements that are essential to life, such as calcium, potassium, sodium, and magnesium, among others. The size of low molecular weight proteins and small essential compounds allows them to cross an intact glomerular basement membrane into the kidney tubule where they are routinely reabsorbed by the proximal tubule of the nephron. Only very small quantities of these are excreted in the urine.

Further down the tubule, in the distal portion, water balance is adjusted and urine is either concentrated or diluted. There are very fine capillaries along the tubule for the exchange of essential substances between the blood and the urine. It is in the tubule that the cells are most active metabolically and where most of the electrolyte exchange occurs. For example, the biologically active metabolites of vitamin D are produced primarily in the tubules, and altered vitamin D metabolism may result from cadmium-associated renal tubular dysfunction (Ex. 8-086).

The earliest form of cadmium toxicity occurs in the proximal portion of the tubule. According to Dr. Friberg:

"The first sign of cadmium induced renal damage is tubular proteinuria with increased urinary excretion of low molecular weight serum proteins, such as beta-2 microglobulin.....At the same time there is a dramatic increase of cadmium in the urine....."(Tr. 6/6/90, p.72)

Any of several small proteins in urine may be monitored as markers of kidney function: Retinol Binding Protein (RBP), Beta-2-Microglobulin (B(2)-M), N-Acetyl-*D-Glucosaminidase (NAG), and Metallothionein (MT). In the absence of elevated levels of these analytes, the kidney is considered to be functioning normally.

Retinol Binding Protein (RBP) is a low-molecular weight vitamin A transporting protein which is cleared from serum into glomerular filtrate and resorbed into renal tubular cells. B(2)-M is a low molecular weight protein that is normally reabsorbed by the proximal tubule. Metallothionein (MT) is a metal-binding protein that correlates with cadmium and B(2)-M levels. N-Acetyl-*D-Glucosaminidase (NAG) is an analyte that may correlate well with cadmium levels under 10 micrograms per gram creatinine (ug/g Cr) (Ex. 30), but this finding is disputed (Ex. 148).

Only B(2)-M is widely used as an early indicator of cadmium-induced kidney dysfunction. Carl-Gustav Elinder, M.D., Chairman of the Department of Nephrology at the Karolinska Institute, testified that measurements of B(2)-M in urine constitute a very sensitive indicator of tubular damage. This is because a small decrease in tubular reabsorption from the normal 99.9% to 99% would result in a tenfold increase in the urinary excretion of this protein (Ex. 55).

Dr. Kazantzis, testifying for the Cadmium Council, stated that the earliest indicator of any effect of cadmium on the kidney is the increased excretion in the urine of a number of low molecular weight plasma proteins, and that this effect can be monitored by estimating the B(2)-M or RBP concentrations in the urine (Ex. 19-43A).

As indicated by Drs. Friberg and Goyer, the finding of excess B(2)-M proteins in urine in conjunction with findings of an elevated cadmium body burden, as indicated by elevated levels of cadmium in urine (CdU) and cadmium in blood (CdB), helps establish that kidney disease exists and that it is probably associated with cadmium toxicity. For example, Dr. Goyer testified that if both B(2)-M is below 300 ug/g Cr and CdU is below 2 ug/g Cr, workers are likely to have no damage. Glomerular proteinuria, another form of cadmium-related kidney dysfunction, refers to the presence of high molecular weight proteins in the urine due to the increased permeability of the glomerulus (a "leaky" glomerulus) which allows the passage of the high molecular weight proteins into the proximal tubule. When high molecular weight proteins are not reabsorbed by the proximal tubule, the proteins are excreted in the urine. Glomerular proteinuria is considered to be a clinically significant form of kidney dysfunction that differs from tubular proteinuria (Exs. 8-86-B, p. 63; 4-54). In some cases, glomerular and tubular damage can occur at the same time resulting in a mixed type of proteinuria.

The concept of a "critical concentration" of cadmium in the kidney has been used repeatedly throughout the medical literature on cadmium-induced kidney dysfunction. A critical concentration is a threshold which when crossed leads to adverse health effects. When the concentration of cadmium in the kidney exceeds the critical concentration, the effects of cadmium-induced kidney dysfunction start to occur in individuals (Ex. 144-3-C).

Dr. Goyer testified that according to his review of the literature, there is a:

".....commonly coded critical concentration of cadmium for people for the so-called [population critical-concentration], where 50 percent of the working population that has a kidney concentration of 200 micrograms per gram of tissue would have detectable renal disease. And this seems to be about the same level in people there's a much wider variation in what that level ... because there are people with different states of ability to make metallothioneine and different essential iron intakes....the variations that are reported by Roels and by Ellis in their human studies by neutron activation are from one to 300 or more micrograms....."(Tr. 6/6/90, pp. 130-131)

Despite individual variability, the concept of an individual's critical concentration of cadmium in the kidney has been well established. There is general agreement that the critical concentration in the renal cortex is about 200 ug/g of kidney cortex wet weight. On the basis of autopsy studies, the WHO task group concluded that the critical level in the renal cortex necessary for the appearance of proteinuria ranged from 100 to 300 ug cadmium (Cd) with the likely estimate being 200 ug Cd/g wet weight, or 200 mg/kg, (Exs. 4-12, 8-440). Similarly, a review of autopsies and in vivo measurements, made through neutron activation analysis of human kidney tissues, showed that adverse effects first occur in the range of 170 to 200 ug Cd/g wet weight (Ex. 8-086b, p. 99). Roels et al., concluded that the "critical concentration" for renal cadmium cortex is found in the range 160 to 285 ug Cd/g. Above 285 ug Cd/g renal cortex, the probability is very high that all persons will show signs of renal dysfunction (Ex. 57-K). In animals, the concentration of cadmium in the renal cortex at which dysfunction first appears ranges from 100 to 300 ug Cd/g wet weight, with most species showing proteinuria at 200 ug Cd/g (Ex. 8-086b, p. 97). According to Clarkson et al.,:

"...the critical concentration of cadmium in renal cortex for a substantial (but unspecified) proportion of an exposed group of individuals has been estimated to be around 200 (ug/g) mg/kg (1.78 mmol/kg) both in animals and in humans.... Calculations for humans based on in vivo neutron activation analyses ...[indicate] that about 10 percent of a population develops renal effects at concentrations of Cd in kidney cortex lower than 200 mg/kg wet weight... (Ex. 14-18, p. 157)."

In a study by Ellis (Ex. 4-27), the cadmium body burden for each worker was determined through the direct measurement of the level of cadmium in the kidneys of workers with and without kidney damage. As noted by C.H. Hine, M.D., Medical Director of ASARCO, this technique, however, is "not of practical use" for routine monitoring ( Ex. 107). A simpler, more commonly used method to evaluate whether a worker has approached this individual "critical concentration" is to use cumulative exposure to cadmium as a proxy for concentration of cadmium in the kidney cortex. Cadmium exposure histories and personal sampling data can be used to calculate a cumulative exposure level, as was done for members of the Ellis cohort. Despite the uncertainty inherent in estimating cumulative exposures from work hi stories, Ellis found a significant correlation between exposure and cadmium body burden as measured in the liver and the kidney.

b. Review of the Literature. The authors of at least seven studies including Ellis sought to examine the relationship between cumulative cadmium exposure and kidney dysfunction. Rather than relate the concentration of cadmium in the kidney cortex to incidence of tubular proteinuria, these authors related cumulative cadmium exposure to incidence of tubular proteinuria. Exposures and responses in all seven studies have been well characterized. The seven major studies reviewed in depth below are: Kjellstrom et al (Ex. 4-47); Falck et al., (Ex. 4-28); Ellis et al., (Ex.4-27); Jarup et al., (Ex. 8-661); Elinder et al., (Ex. L-140-45); Mason et al., (Ex. 8-669-A); and, Thun et al., (Ex. 8-670). The data from these studies provide strong evidence that occupational exposures to all forms of cadmium are associated with kidney damage at similar cumulative exposure levels.

The seven primary studies cover five populations of cadmium-exposed workers. Excesses of low molecular weight proteins in urine were observed in all of these occupationally exposed populations. In each of these studies, the authors defined kidney dysfunction on the basis of biological markers prior to collecting and analyzing the data, and response was a dichotomous variable with workers separated into two categories: those with abnormal kidney function (i.e. dysfunction) and those with normal kidney function.

The findings of each study are dependent on the manner in which each researcher defined renal dysfunction; the tests used to identify the presence or absence of dysfunction; and the reference methodology used by each laboratory. In general, few studies of kidney function among occupationally exposed cohorts have relied upon RBP, NAG, or MT. Of the seven major epidemiological studies reviewed here, only one (Mason et al, Ex. 8-669-A) relied upon RBP as a marker of kidney function. In Mason's study, B(2)-M was also measured. In the other six studies, B(2)-M was the analyte used to distinguish between normal and abnormal kidney function.

The results from the studies are strikingly similar. The authors found that the type of abnormality and the prevalence of abnormality was similar, at similar exposure levels, regardless of the type of cadmium compound to which workers were exposed. These results, from difference worker populations in different countries, demonstrate that the kidney is one of the major target organs of cadmium toxicity. The studies also imply that there is agreement on what constitutes abnormality.

i. Kjellstrom et al (Ex. 4-47). Kjellstrom conducted a study of kidney dysfunction among 240 workers in a Swedish battery factory. The cohort was comprised of 197 men and 43 women. The factory had two separate plants: the material plant that manufactured raw material for the battery electrodes and the assembly plant that made the complete electrodes and batteries. The primary exposures at the factory were to cadmium oxide (CdO) dust and nickel hydroxide [Ni(OH)2] dust. A group of 87 lumbermen and shipyard workers belonging to the same occupational health clinic served as controls.

A spot urine sample was collected from each study participant, and the pH and specific gravity were immediately determined. Twenty of 327 urine samples collected for the dose-response study had pH < 5.6 and had to be buffered with a phosphate buffer with pH of 7.6 to increase the pH over this level. Then the samples were frozen. B(2)-M in urine was measured using the Phadebas B(2)-M microtest from Pharmacia.

Kidney dysfunction was determined by the level of B(2)-M in urine because increased urinary excretion of B(2)-M would be caused by cadmium, not nickel hydroxide dust exposures (Friberg, 1950, referenced in Kjellstrom). Kjellstrom classified a worker as having abnormal proteinuria if the level of B(2)-M exceeded 290 ug/liter, the upper 95% tolerance limit for the control population. Among men in the assembly plant, 25% exceeded this limit, a significantly higher proportion than the 3.4% of controls who exceeded this limit (p< 0.001). The proportion of men in the material plant who exceeded the limit was 50%, also statistically significantly greater than the proportion in the control group (p< 0.001). Among women, only 2.3% of those in the assembly plant exceeded the limit.

Exposure estimates based on area samples were provided by the battery factory management for the years 1949 to 1972. Kjellstrom estimated that cadmium exposures in the assembly plant had been about 50 ug/m(3) since 1963. Area and personal air sampling data in the material plant were sparse. The first data available were from 1968 when two area samplers found cadmium exposures to be 7 and 31 ug Cd/m(3). Some personal sampling data were also available from 1972. Information on the continuity of exposure, time of first employment, and exposure duration was provided by the safety engineer of the factory.

Workers were classified into four categories according to exposure histories. The non-exposed group consisted of six workers who worked on the premises of the plant but were found to have virtually no exposure to cadmium. The intermittently exposed group consisted of forty-nine workers. Finally, there were two groups of workers who had spent time in jobs where exposure to cadmium was continuous. These two groups were comprised of 15 workers from the material plant and 170 workers from the assembly plant.

The 185 people in the continuously exposed groups were the primary subjects of this study. Nine of these workers were from Yugoslavia. Their results were analyzed separately from the other exposed workers because in some parts of Yugoslavia, Bulgaria, and Rumania, tubular proteinuria is endemic (Balkan nephropathy). Forty-eight workers (26%) had started employment in 1974, three years before this study was conducted. Fifty-two workers (28%) had been employed since before 1952. The majority of these 52 had been moved to "cadmium-free" work before the study was conducted.

Workers exposed continuously to cadmium at levels of 50 ug/m(3) for 6 to 12 years showed a higher prevalence of proteinuria than controls (19% vs. 3.4%). For the continuously exposed groups, prevalence of proteinuria increased with length of time since first employment; the prevalence was 6.2%, 18%, and 57% among workers with less than 6 years, 6 to 22 years, and greater than 22 years of exposure, respectively.

Using data on 129 exposed workers who responded to a smoking questionnaire and for whom B(2)-M levels and length of employment information was available, the effects of smoking on kidney function were evaluated. The prevalence of proteinuria among non-smokers was 13.3 percent, while among current smokers it was 32.4 percent. Thus, the prevalence of kidney dysfunction among current smokers was about two and one-half times higher than among non-smokers. The authors noted that the total dust concentrations in workroom air may have been high enough to cause substantial contamination of the workers' hands and bodies, and that tobacco products contaminated by a worker's hands could account for the higher prevalence of tubular proteinuria among smokers. The authors' conclusion that the increased incidence of dysfunction among smokers was not attributable to the cigarettes themselves is supported by findings from the Falck study, (Ex. 4-28), described below.

ii. Falck et al (Ex. 4-28). Falck studied 33 male workers in a refrigeration compressor plant in the U.S. The plant produced refrigeration compressors with silver brazed copper fittings. These fittings were brazed either by hand or on an automated line. Falck reported that according to plant managers, neither process has changed since the automatic assembly line was installed in 1968. Air monitoring data for this plant had been collected by the Michigan Department of Industrial Health since 1961. All samples used for estimating exposures for individual workers were from the worker's breathing zone. The mean calculated exposure on the automatic line for an 11 year period was 39 ug/m(3), + or - 7.8 (SE). Mean exposures on the manual line, 110 ug/(3), + or - 25.5 (SE), were significantly higher.

Work histories were obtained for each subject participating in the study, and time on each of the production lines was calculated for each worker. Cumulative cadmium exposure was calculated for each worker by multiplying the mean cadmium exposure of a line with the number of years a worker spent on that line.

A spot urine sample was collected for each worker, and an aliquot of urine was analyzed for B(2)-M. The pH was recorded and adjusted above 5.5 where necessary. All samples were then frozen. B(2)-M in urine was measured using the Phadebas microtest by Pharmacia. Blood samples were collected, and blood and urine cadmium levels were measured using a flameless atomic absorption spectrophotometer (Varian). Three workers had histories of diabetes, kidney infection, or hypertension and were eliminated from the statistical analyses. Descriptive statistics on the workers are presented in Table V-1.

Table V-1 - Descriptive Statistics for a Cohort of 30 Employees at a
            Refrigeration Compressor Production Plant(a)
                            | Normal Kidney|   Abnormal Kidney  |
                            | Function Mean| Function Mean (95% |   P
                            |  CI)(b)      |          CI(b)     | Value (c)
N ......................... |           23 |                  7 | .........
Age (years)................ |   49 (47, 51 |       53  (51, 55) |       .13
TWE (d) (ug/m(3) years) ... |459 (332, 634)|  1137  (741, 1737) |       .02
Smoking Habits (pack-years) |    14 (9, 19)|       24  (14, 34) |       .07
                          Urine Ratios
Protein/creatinine (mg/g) . |    34 (26,43)|     246 (132, 456) |     < .001
B(2)-Microglobulin/         |              |                    |
Creatinine (u/g)(e) ....... |   53 (31, 90)| 6375 (1115, 36463) |     < .001
Cadmium/Creatinine (ug/g) . |   11 (10, 13)|         16 (8, 36) |      0.07
                          Serum Ratios
Creatinine/Serum (mg/100ml) |  1.1 (1, 1.2)|   1.4 (1.2, 1.1.7) |     0.003
B(2)-Microglobulin/         |              |                    |
Serum (ug/ml) ............. |  2 (1.6, 2.4)|     2.3,(1.8, 2.8) |      0.32
  Footnote(a) Data taken from Falck et al (Ex. 4-28).
  Footnote(b) Mean and 95% confidence intervals are presented.  Means for
age and smoking habits are arithmetic means; all others are geometric
means.  Confidence intervals are constructed from arithmetic standard
deviations for age and smoking; all others are from the geometric standard
  Footnote(c) P-value is associated with a test of differences between
group means.
  Footnote(d) Time-weighted inhalation exposure estimate (i.e. dose).
  Footnote(e) B(2)-Microglobulin measured in urine, based on spot samples.

Forty-one males who were not known to be exposed to cadmium were selected to serve as controls. Reference values calculated from the spot urine and serum samples collected from the controls, were established at: glucose > 130 mg/g Cr; total protein > 173 mg/g Cr; or B(2)-M > 629 ug/g Cr. Eight workers' results exceeded the reference values, and for these eight, a 24 hour urine sample was obtained. Reference values for 24 hour samples were established at glucose > 250 mg/24 hour; total protein > 188 mg/24 hour; or B(2)-M > 400 ug/liter urine from 24 hour samples from 7 controls.

Seven workers out of 33 were found to have abnormal kidney function (21%). This was significantly higher than the 7% of controls who had abnormal spot urine results. The average cumulative time-weighted exposure for workers with abnormal renal function was 1,137 ug/m(3)-yrs, significantly higher than the average cumulative time-weighted exposure of workers with normal renal function (p=.02). Furthermore, as cumulative exposures increased, the prevalence of kidney dysfunction increased.

According to the authors, the change in renal function observed in the workers was not age related since there was no statistically significant difference in age between workers with abnormal and normal renal function (p=.13). Information on smoking habits was collected from questionnaires and medical histories. The authors found that the difference in pack-years smoked between the workers with abnormal and normal renal function was also not statistically significant (p=.07). Despite this the authors stated that ingestion of cadmium was not likely to be significant given the manufacturing process and plant layout. If Kjellstrom's hypothesis that contamination of smoking products was a likely explanation of increased proteinuria among smokers in his study, it would explain why no differences were noted in renal function between smokers and non-smokers in this study by Falck.

iii. Ellis et al (Ex. 4-27). Ellis conducted a study of 82 male workers at a U.S. cadmium smelter. The cohort was comprised of 51 active workers and 31 retired workers employed in production, non-production, office, and laboratory work. Each cohort member completed a health history questionnaire, took a physical exam, gave specimens for blood and urine tests, and provided 24-hour urine samples. Urine samples were pH adjusted and frozen. Kidney function was judged to be abnormal if urinary levels of B2-M exceeded 200 ug/g Cr or if total urinary protein levels exceeded 250 mg/g Cr. These limits were chosen to comply with those reported by Roels et al., (Ex. 57-K). In addition, the cadmium content of the left kidney and the liver was measured directly by the in vivo prompt-gamma neutron activation technique. Information on smoking habits was not available.

Eighteen active workers (35%) and twenty-three retired workers (74%) were classified as having abnormal kidney function. Descriptive statistics for the entire cohort are presented in Table V-2.

 Table V-2 - Descriptive Statistics for a Cohort of 82 Active and
             Retired Cadmium Smelter Employees(a)
                                                  |   Normal   |   Abnormal
                                                  |   kidney   |    kidney
                                                  |  function  |   function
                                                  |mean (SD)(b)|   mean (SD)
                              Active Workers
N ............................................... |          33|         18
Age (years) ..................................... | 42.6 (13.3)| 53.6 (6.8)
Duration of Exposure (months) ................... |   141 (118)|  264 (105)
TWE(c) (ug/m(3)-years) .......................... |   105 (9.0)| 1690 (2.7)
Renal Cadmium(d) (ug/g) ......................... |   125 (2.8)|  230 (2.0)
Liver Cadmium (ppm) ............................. |  11.3 (2.8)| 63.9 (1.5)
                              Retired Workers
N ............................................... |           8|         23
Age (years) ..................................... |  69.0 (8.3)| 67.9 (6.9)
Duration of Exposure (months) ................... |    342 (75)|  329 (103)
TWE(c) (ug/m(3)-years) .......................... |   379 (3.3)| 3143 (3.6)
Renal Cadmium(d) (ug/g) ......................... |   148 (2.1)|  169 (1.7)
Liver Cadmium (ppm) ............................. |  14.0 (3.1)| 33.6 (2.9)
  Footnote(a)  Data taken from Ellis et al (Ex. 4-27).
  Footnote(b)  Mean (Standard Deviation) presented.  Means and SDs for age
and duration of exposure are arithmetic means and SDs.  All others are
geometric means and SDs.
  Footnote(c)  Time-weighted inhalation exposure estimate (i.e. dose).
  Footnote(d)  Renal cortex cadmium concentration; assumes 145 g weight
for the total kidney and a 1.5 ratio between cortex and total kidney

Cadmium exposure histories based on employment records, area monitoring techniques, and personal sampling data were obtained for all 82 exposed workers. The chronological record of each worker's job assignments was obtained from personnel files at the smelter. Cumulative exposure estimates were developed for each member of the cohort from industrial hygiene data provided by Smith (Ex. 4-64). That is, for each worker, the time-weighted inhalation exposure (TWE) was calculated by multiplying the duration of exposure in a given work area (t(1) by the estimated inhalation exposure for that area and year (E(1) and then summing these values to obtain cumulative exposure or TWE = E (E(1)T(1) .

For the actively employed workers, a significant correlation (r=0.70, p< 0.001) was observed between TWEs and liver cadmium burden. Furthermore, whenever a worker's liver burden exceeded 40 ppm and exposure exceeded 400 - 500 ug/m(3)-yr, there was evidence of renal abnormalities. The highest correlation was obtained between the kidney cadmium burden data and TWEs for the active workers without evidence of kidney dysfunction (r=0.83, p< 0.001). The percentage of workers with renal abnormalities was found to increase as exposure increased. (See Section VI-Quantitative Risk Assessment.) iv. Jarup et al (Ex. 8-661). Jarup sought to model the relationship between cadmium exposure and tubular proteinuria. The Swedish battery factory where Jarup conducted his investigation was the same as where Kjellstrom conducted his study of cadmium exposure and kidney dysfunction. Jarup collected data from 326 male and 114 female cadmium battery workers with at least three months of employment between 1931 and 1982. The only other criterion for being included in the study was that at least one measurement of urinary B(2)-M must have been made for each worker.

The response variable used in this study was B(2)-M levels in urine. B(2)-M in urine was measured using the radioimmunoassay (RIA) method, Phadebas, developed by Pharmacia. Since B(2)-M levels had been measured since 1972, the B(2)-M level for any worker was taken as the average of all B(2)-M levels measurements made for that worker. Workers with B(2)-M levels exceeding 310 ug/g Cr were judged to have tubular proteinuria.

Jarup considered three distinct measures of exposure in his analysis. First, he considered cadmium in air which had been measured since 1947. Using information on length of employment obtained from company files for each worker, Jarup estimated cumulative cadmium dose by multiplying the length employment in a particular work area by the average air concentration of cadmium for that area and time period and then summing the product of these for each worker.

The second measure of exposure considered by Jarup was cumulative blood cadmium dose. Data on blood cadmium levels together with other laboratory data had been collected at regular intervals since 1967. Cumulative blood cadmium dose was estimated as the average blood cadmium concentration times the times the number of months employed.

The final measure of exposure considered by Jarup was an alternative method for calculating cumulative blood cadmium dose. The study authors hypothesized that because air cadmium concentrations were much higher in the 1940's and 1950's, workers hired before 1967, particularly those hired before 1950, would have had higher blood cadmium concentrations during those periods, and therefore simply multiplying average blood cadmium concentrations by number of months employed might underestimate the true cumulative blood cadmium concentrations. The alternative measure the study authors devised was to use linear regression to model the relationship between year and blood cadmium levels for each worker and then to extend the observed relationship to the years for which there was no data. Forty workers show evidence of tubular proteinuria. The results of Jarup's analyses indicated that a dose-response relationship exists between kidney dysfunction and exposure regardless of the measure of exposure used, but they also suggested that cumulative blood cadmium level may be a more sensitive indicator of cadmium-induced renal dysfunction than cumulative cadmium in air. Table V-3 shows the prevalence of tubular proteinuria among workers grouped according to cumulative airborne cadmium exposure and cumulative cadmium exposures in blood estimated using the alternative method.

Table V-3 - Cumulative Airborne Cadmium Exposures by Percent
            Prevalence of Kidney Dysfunction
 Cumulative Exposure          |   Mean  | Number |    Percent
(Cum Exp)(ug/m(3)-years)      | Cum Exp | People | Dysfunction(1)
< 359 ....................... |     131 |     3  |       1.1
359 - <  1,710 ............... |     691 |     7  |       9.2
1710 - <  4,578 .............. |   3,460 |    10  |      23.3
4578 - <  9,458 .............. |   6,581 |    10  |      32.3
9458 - <  15,000 ............. |  12,156 |     5  |      31.2
> 15,000 .................... |  21,431 |     5  |      50.0
  Footnote(1) Percent dysfunction = number with dysfunction divided
by number in that exposure category times 100.

In his paper, Jarup noted that on the basis of data from other studies of the relationship between cadmium in air and kidney dysfunction, it has been estimated that exposure to an average workroom concentration of 50 ug Cd/m(3) for 10 years would result in a prevalence of tubular proteinuria of 10%. The corresponding prevalence found in Jarup's study is only 4%, (95% CI=2%-8%), which is also somewhat less than the prevalence reported by Kjellstrom in an earlier study of the same factory (Ex. 4-47). Jarup suggests several possible explanations for why the prevalence of tubular proteinuria is lower in his study than in the others.

First, the earlier studies were all cross-sectional meaning that only workers employed at a certain point in time were included. This study by Jarup included all workers employed for more than three months at the factory. Computation of the cumulative airborne cadmium concentration took into account the varying exposure conditions over time, and the study included workers with low cumulative exposures. In addition, this study had more information about the prevalence of kidney dysfunction among the retired workers, although data on the retired workers were not evaluated separately.

Second, the other studies were relatively small while the Jarup study is one of the largest. This means that the confidence intervals computed for the prevalence observed in the other studies must be wide. According to Jarup, the consequence of this is that even if the variations may seem great between the various studies, there are probably no significant differences between the dose-response relationships reported in the various studies.

Third, the Jarup study was conducted in a battery factory whereas most other studies were carried out in other industrial environments, such as smelters, where the fraction of respirable dust is probably significantly larger. Jarup suggested that this might affect the results of his study. However, given that there are probably no significant differences between the dose-response relationships reported in the various studies, the effect of particle size distributions may be a relatively unimportant factor in the etiology of kidney dysfunction.

Another factor that may have affected the B(2)-M levels among these workers is the method used to collect and handle urine samples. The authors did not report whether the samples were spot or 24 hour samples, or what the urine pH was per B(2)-M sample. If urine samples were collected over a 24 hour period with pH less than 6.0, it is more likely that the B(2)-M would have been degraded and erroneously low B(2)-M levels would have resulted. If spot samples were used and pH was not adjusted, then B(2)-M would also have been degraded.

Dr. Gunnar Spang, who works for a medical organization providing occupational health services to a large Swedish NiCd battery manufacturer and who was a co-author of this study, testified at the hearing. He submitted post-hearing comments on the number of cases, per cumulative exposure category, that, in his opinion, were cases associated with cadmium exposures (Exs. 58 and 80). His comments alter the findings somewhat. (See Table V-4.)

Table V-4 - Cumulative Airborne Cadmium Exposures of Percent
            Prevalence of Kidney Dysfunction
 Cumulative Exposure          |   Mean  | Number |    Percent
(Cum Exp)(ug/m(3)-years)      | Cum Exp | People | Dysfunction(1)
< 359 ....................... |     131 |     0  |         0
359 - <  1,710 ............... |     691 |     2  |       2.6
1710 - <  4,578 .............. |   3,460 |     9  |      20.9
4578 - <  9,458 .............. |   6,581 |     9  |      29.0
9458 - <  15,000 ............. |  12,156 |     5  |      31.2
> 15,000 .................... |  21,431 |     5  |      50.0
  Footnote(1) Percent dysfunction = number with dysfunction divided
by number in that exposure category times 100.

He stated that in the lowest dose group with mean dose = 131 ug/m(3)-yrs, no cases were cadmium-related. Thus, the response rate in this group would be zero, as opposed to 1.1% reported by Jarup et al. For the remaining dose groups, the response rates would be as follows: 2.6% for the group with mean dose = 691 ug/m(3)-yrs; 20.9% for the group with mean dose = 3460 ug/m(3)-yrs; 29% for the group with mean dose = 6581; 31.2% for the group with mean dose = 12156; and 50% for the group with mean dose = 21431. Thus, according to Dr. Spang, a dose-response still exists, but the dose-response curve is less steep than the curve described by Jarup. Also, according to Dr. Spang, for a 45 year exposure of about 77 ug/m(3), one-fifth of all workers would have tubular proteinuria. However, for workers with about 15 ug/m(3) exposures, over a 45 year working lifetime, the prevalence of kidney dysfunction would be about 2.6%, according to Dr. Spang, as opposed to 9.2%, indicated by Jarup. Dr. Spang testified that several workers in this plant had kidney stones, but he was unable to determine whether the prevalence was greater than that in the general population (Tr. 7/17/90, p. 217).

In summary, workers with tubular proteinuria had a proportionately higher serial CdB dose than their fellow workers without renal dysfunction but with the same cumulative air cadmium dose. Some workers will be more sensitive than others, i.e., their cadmium uptake will be greater than other workers, and, at lower exposures, will have tubular proteinuria. (See Section VI - Quantitative Risk Assessment).

v. Elinder et al, (Ex. L-140-45). Elinder conducted a study of 58 male and 2 female cadmium-exposed workers employed at least five years before 1978 at a factory that produced coolers, radiators, and heat exchangers in Sweden. The plant began using cadmium containing solders in 1955, but the plant was demolished and a new plant was constructed in 1973. By 1978, the whole workplace had been seriously contaminated and had to be repainted and renovated to reduce the cadmium exposures at the workplace. Since 1978, no cadmium-containing products have been used in this plant. No exposure data from the plant were available before 1976.

B(2)-M in urine was measured using a radioimmunoassay (RIA) method, Phadebas, developed by Pharmacia. The night before the health examination each person was asked to ingest about four grams of sodium bicarbonate (Samarin R). This was done in order to produce a pH exceeding 5.6 in the morning urine sample the next day because B(2)-M is degraded in acidic urine.

No control group was included; according to Elinder the normal concentration of cadmium and B(2)-M in urine and blood has been well documented. Based on the literature the authors selected 300 ug/g Cr, the upper 95 - 97.5 percentile for the urinary excretion of B(2)-M among persons without tubular dysfunction, as the upper limit of normal.

Each individual's exposure to cadmium was assessed by a group of four people (one representative from the company, one from the employees, and two from the local labor inspectorate) who had knowledge of the previous conditions in the plant and of the type of work carried out by each worker during the whole period when cadmium was in use.

Exposure was classified into four categories; high, medium, low, and no cadmium exposure. Before 1955 and after 1978 there was no exposure to cadmium at all. Based on the measurements in 1976, it was assumed that high, medium, and low exposure was about 0.5, 0.15, and 0.05 mg/m(3), respectively. The number of years each worker had been occupied in activities that gave high, medium, low, or no exposure were recorded. A cumulative dose was estimated for each worker expressed in milligrams of per cubic meter-years (mg/m(3)-yr).

Elinder categorized renal dysfunction based upon B(2)-M levels into "slight proteinuria" ( B(2)-M > 300 ug/g Cr) or "pronounced proteinuria" (B(2)-M levels > 1,000 ug/g Cr). He then used these two categories of degree of kidney dysfunction to evaluate the relationship between years of exposure and prevalence of B(2)-microglobulinuria. Results from this analysis, presented in Table V-5, indicate that as years of exposure increases, the prevalence of slight and more pronounced proteinuria increases.

Table V-5 - Prevalence of B(2)-Microglobulin in
            Relation to Years of Cadmium Exposure
Years of      |   Slight         |  Pronounced
Exposure      | Proteinuria(1)   |  Proteinuria(3)
              | No. (Percent)(2) |  No. (Percent)(4)
4 to 7 ...... |            1 (6) |            0 (0)
8 to 13 ..... |           6 (46) |           3 (23)
14 to 19 .... |           6 (46) |           5 (38)
>20 ......... |          11 (65) |           6 (35)
All ......... |          24 (40) |          14 (23)
  Footnote(1) Number of people in that category; B(2) > 300 ug/g Cr <  1000
ug/g Cr.
  Footnote(2) Percent is the number of cases of slight proteinuria divided
by the total number of people with the corresponding number of years of
  Footnote(3) Number of people in that category; B(2) > 1000 ug/g Cr.
  Footnote(4) Percent is the number of cases of more pronounced
proteinuria divided by the total number of people with the corresponding
number of years of exposure.
  Source:  Elinder et. al., (Ex. L-140-45)

When cadmium in urine and proteinuria were evaluated, a close relationship was found. At a urinary cadmium excretion exceeding 10 ug/g Cr the proportion of cases of B(2)-M excretion exceeding 300 ug/g Cr (slight proteinuria) was 88% and the proportion exceeding 1000 ug/g Cr (pronounced proteinuria) was 75%.

Elinder evaluated the relationship between age and years of exposure. The prevalence of B(2)-microglobulinuria (B(2)-M) was increased in workers aged over 59, suggesting that age was a potential confounder in the assessment of the relationship between dose and response. To examine the role of age in this relationship Elinder fit a multiple regression model relating urinary B(2)-M excretion to age and cumulative dose using the model: log (B(2)-M)=[aXage]+[bXcumulative dose]+[c]. Elinder concluded that dose was related to B(2)-M-U, whereas age was not. In further analysis, when workers over age 59 were excluded, the prevalence of B(2)-microglobulinuria still increased with the numbers of year of exposure, with the estimated cumulative dose, and urinary cadmium. Thus, Elinder concluded that age is not an important confounding factor in this study.

Degree of kidney dysfunction was also related to cumulative cadmium exposure and to CdU. Table V-6 shows that for cumulative cadmium exposure, the prevalence of slight and more pronounced proteinuria both increase as exposure increases. Indeed, Table V-6 suggests that there is a 19% risk of developing tubular proteinuria (urinary B(2)> 300 M/g Cr) at a cumulative cadmium dose of less than approximately 22 ug/m(3) over a 45 year working lifetime (1 mg/m(3) divided by 45 years = 22.2 ug/m(3).) If exposures are between 22 and 44 ug/m(3) over a 45 year exposure period, this risk increases to 32%.

Table V-6 - Prevalence of B(2)-Microglobulin in
            Relation to Cumulative Cadmium Exposure
Cum exposure  |   Slight         |  More Pronounced
(mg/m(3) -yrs | Proteinuria(1)   |  Proteinuria(3)
              | No. (Percent)(2) |  No. (Percent)(4)
< 1 ......... |           3 (19) |            0 (0)
1 to <  2 .... |           7 (32) |           4 (18)
2 to <  3 .... |           4 (44) |           3 (33)
3 to <  5 .... |           5 (62) |           4 (50)
= or > 5 .... |          5 (100) |           3 (60)
All ......... |          24 (40) |          14 (23)
  Footnote(1) Number of people in that category; B(2) > 300 ug/g Cr. <
1000 ug/g Cr.
  Footnote(2) Percent is the number of cases of slight proteinuria divided
by the total number of people with the corresponding level of cumulative
  Footnote(3) Number of people in that category; B(2) > 1000 ug/g Cr.
  Footnote(4) Percent is the number of cases of more pronounced
proteinuria divided by the total number of people with the corresponding
level of cumulative exposure.
  Source: Elinder et. al., (Ex. L-140-45)

Finally, Table V-7 shows that for CdU, like age and cumulative cadmium exposure, the prevalence of slight and more pronounced proteinuria both increase as CdU increases. Table V-7 indicates that CdU above 15 ug Cd/g Cr are associated with very high risks of kidney dysfunction (82% for B(2)-M levels above 1,000). (For further discussion of this study see Section VI - Quantitative Risk Assessment).

Table V-7 - Prevalence of B(2)-Microglobulin in Relation to
            Cadmium Levels in Urine
                       |   Slight         | More Pronounced
    CdU (ug/g Cr)      | Proteinuria(1)   | Proteinuria(3)
                       | No. (Percent)(2) | No. (Percent)(4)
= or <  2.1 ........... |            1 (7) |           0 (0)
2.1 to = or <  5.1 .... |           3 (25) |           0 (0)
5.1 to = or <  10.1 ... |           6 (33) |          2 (11)
10.1 to = or <  15.1 .. |           4 (80) |          3 (60)
>15.1 ................ |          10 (91) |          9 (82)
All .................. |          24 (40) |         14 (23)
  Footnote(1) Number of people in that category; B(2) > 300 ug/g Cr <  1000
ug/g Cr.
  Footnote(2) Percent is the number of cases of slight proteinuria divided
by the total number of people with corresponding levels of CdU.
  Footnote(3) Number of people in that category; B(2) > 1000 ug/g Cr.
  Footnote(4) Percent is the number of cases of more pronounced
proteinuria divided by the total number of people with corresponding
levels of CdU.
  Source:  Elinder et. al., (Ex. L-140-45)

vi. Mason et al, (Ex. 8-669-A). Mason conducted a detailed investigation of renal function among 75 male workers with at least one year of employment as cadmium brazers at a copper-cadmium alloy production facility in the U.K. Seventy-five unexposed workers, matched to the exposed workers on age, sex, and employment status, served as controls. Although not individually matched for smoking history, the proportion of current, former, and non-smokers was similar among the exposed and the controls as was the number of pack-years smoked. Only 11 exposed workers were employed in the production of alloy at the time of the study; for many of the exposed workers, occupational exposure had ceased some years before the study was undertaken.

The medical evaluation for both groups of workers included a questionnaire, a medical examination, blood and urine analyses, and in vivo measurements of kidney and liver cadmium levels by prompt-gamma neutron activation. Retinol binding protein (RBP) was measured using the enzyme linked immunosorbent assay (ELISA) and B(2)-M was measured with a Phadebas B(2)-M microtest kit by Pharmacia.

Measurements of cadmium in air of the factory had been taken between 1951 and 1983. Personal and area sampling data were available from 1964 to 1983. Pre-1964 exposures were estimated after discussion with occupational health physicians, occupational hygienists, representatives from management and the workforce, and took into account changes in production and ventilation. Questionnaires and plant data were used to obtain each worker's exposure history. A cumulative exposure index was calculated by summing the products of length of time employed at a particular exposure level multiplied by exposure level.

Significant increases in the urinary excretion of albumin, RBP, B(2)-M, N-acetyl-B(D)-glucosaminidase (NAG), alkaline phosphatase, gamma-glutamyl transferase, and significant decreases in the renal reabsorption of calcium, urate, and phosphate were found in the exposed group compared with the referent group. Measures of glomerular filtration rate (GFR) creatinine clearance, serum creatinine, and B(2)-M indicated a reduction in GFR in the exposed population. Many of these tubular and glomerular function indicators were significantly correlated with both cumulative exposure index and liver cadmium burden.

RBP was the biological marker chosen to measure renal function. RBP was not age-related in the unexposed population. The upper 95th percentile for urinary RBP calculated for the unexposed population, 40 ug/mmol Cr or 356 ug/g Cr, was the level used to define workers with tubular proteinuria. The frequency of tubular proteinuria in exposed workers, grouped according to a cumulative exposure index, is presented in Table V-8.

Table V-8 - Prevalence of Kidney Dysfunction by Cumulative
            Cadmium Exposure
    Cumulative            |  Number   |   Number    |  Percent
    exposure(1)           | normal(2) | abnormal(3) | abnormal
= or <  500 .............. |       96  |          5  |     4.9
> 500 - = or <  1000 ..... |       14  |          0  |     0.0
> 1000 - = or <  1500 .... |        3  |          5  |    62.5
> 1500 .................. |        4  |         20  |    83.3
  Footnote(1) Cumulative Exposure measured in ug/m(3) - year.
  Footnote(2) Normal measured by Retinol Binding Protein (RPB):
< 40 ug RPB/mmol Cr.
  Footnote(3) Abnormal measured by Retinol Binding Protein (RBP):
> 40 ug RBP/mmol Cr.

Mason used a two phase linear regression model to regress a variety of biochemical markers on cumulative cadmium exposure and on liver cadmium in order to identify an inflection point signifying a threshold level above which changes in renal function occur. (The appropriateness of this and other models is discussed further in Section VI -Quantitative Risk Assessment.) Urinary total protein, retinol binding protein, albumin, and B(2)-M all suggested a threshold at about the same cumulative exposure level of 1100 ug/m(3)-yrs whereas changes in the tubular reabsorption of urate and phosphate suggested a higher cumulative exposure threshold.

In 1991, Hartley and Mason 1991 (Ex 150) extended the Mason study to include a follow-up analysis of the effects of cadmium exposure on changes in glomerular filtration rate (GFR). In this paper, recently submitted for publication, six indicators of glomerular function were evaluated. These included three glomerular filtration variables, (B(2)-M in serum, creatinine in serum, and creatinine clearance), and three different estimates of creatinine clearance. The exposed population was split into five groups of approximately equal size on the basis of cumulative cadmium exposure: 0-250 ug/m(3)-yrs, 500-1000 ug/m(3)-yrs, 1000-3300 ug/m(3)-yrs and greater than 3300 ug/m(3)-yrs. The one-sided 95% interval in the referent population was used as the cut-off point for normal or abnormal results with respect to each glomerular filtration variable. For those variables correlated to age, the equation representing the 95% one-tailed prediction boundaries of regression with age was used to indicate normality or abnormality. Thus a 5% frequency of abnormality would be expected in the referent population. The results are presented in Table V-9.

Table V-9 - Frequency of Abnormality for Glomerular Filtration Variables
            Percent Abnormality by Type of GFR Indicator
  Cum. Exp. ug/m(3)-yrs  | B(2) S | Cr. S | Cr Cl | Gault | Hull | Mawer
Control                  |      6 |     5 |     5 |     3 |    3 |    3
< 250                    |      6 |     6 |     6 |     - |    6 |    -
250 to 500               |      - |     - |     - |     - |    - |    -
500 to 1000              |      - |    22 |    22 |    14 |    8 |   14
1000 to 3300             |     24 |    24 |    28 |    18 |   12 |   18
> 3300                   |     60 |    54 |    32 |    46 |   40 |   46

Dose-response and dose-effect relationships were observed over the range of exposures encountered, with a significant decrease in GFR at cadmium exposures greater than 1000 ug/m(3)-yrs. The results show a trend towards increasing abnormality in the six indicators of GFR used, with exposure to cadmium of greater than 500 ug/m(3)-yrs.

vii. Thun et al, (Ex. 19-43-B). Thun conducted a study of kidney function among workers at a cadmium smelter in Denver, Colorado. The smelter recovers cadmium from "bag house" dust, a waste product of nonferrous smelters. The cadmium from the bag house dust is further refined into cadmium metal or highly purified cadmium oxide and cadmium sulfide. Forty-five male workers participated in the study. Seventeen of these were production workers and two were salaried workers employed at the smelter at the time of the study. Of the remaining 26, 18 were former long-term production workers and 8 were former short-term production workers. Thirty-two male hospital workers of similar age, ethnic and geographic background served as controls.

To estimate each workers's cumulative exposure to airborne cadmium, detailed work histories were linked with industrial hygiene data collected through 1976 and updated company measurements through 1985. The length of time spent working in a particular department was multiplied by the estimated cadmium exposure level in that department. The sum of these products represents the worker's cumulative exposure or dose, adjusted for respirator use as described by Smith (Ex. 4-64). Workers at this smelter were subjects in a number of studies of the health effects of cadmium, and a more detailed description of their exposures may be found in the discussion of the carcinogenicity of cadmium in the Health Effects section of this preamble.

Workers were required to fill out a questionnaire, provide spot urine and blood samples, and take a pulmonary function test. Fewer cadmium workers smoked than did controls (38% vs. 44%, p=0.6), and of those who smoked, the workers reported smoking fewer pack-years than the controls (14.6 vs. 22.1, p = 0.29). Indices of renal tubular function included urinary excretion of B(2)-M, retinol binding protein (RBP), calcium, and phosphate. Additional analyses for evidence of tubular injury and glomerular function rate (GFR) were performed. Urinary B(2)-M was measured by the radioimmune assay (Pharmacia), and in order to standardize for urine volume, results were corrected to microgram of analyte per gram of creatinine. Level of lead in blood were also measured. Urine pH levels were not reported.

Workers were classified as having abnormal renal function if any of the following conditions were met: B(2)-M >486 ug/gr cr; serum creatinine = or >1.4 mg/dl; RBP >321 ug/g Cr; tubular reabsorption of phosphate (TRP) <69.4%; or, tubular reabsorption of calcium (TRC) < 97.56%. The limits for TRP and TRC are the lower 95% confidence level on the geometric mean of these markers estimated from the control population. The limits for B(2)-M, RBP, and creatinine are the upper 95% confidence level on the geometric mean of these markers estimated from the control population. Thun found that 24 of the 45 study subjects (53%) suffered some form of renal abnormality. In addition, Thun reported that two of 32 controls (6%) were abnormal by this definition (personal communication, 11/91). The results from this study are in Table V-10.

  Table V-10 - Seventy-seven Participants, Cadmium Smelter Study(1)
                            |     Cadmium      |                 |
                            |     Workers      | Unexposed       |P value
Number .................... |               45 |              32 | .......
Age (mean + or - SD) ...... | 54.4 + or - 15.5 | 0.1 + or - 13.0 |     0.2
Percent Hispanic .......... |              58% |             16% | < 0.0001
Percent Current Smokers ... |              38% |             44% |     0.6
Pack-years (mean + or - SD) |      14.6 (19.8) |     22.1 (29.4) |    0.29
Years Cadmium Work          |                  |                 |
 (GM, Range) .............. |        19 (1-38) |               0 | .......
Cumulative Exposure         |                  |                 |
 (mg/m(3)/days)(2) ........ |     604 (0-5383) |               0 | .......
Blood Cadmium, ug/l         |                  |                 |
 (GM + or - SD) ........... |   7.9 + or - 2.0 |  1.2 + or - 2.0 | < 0.0001
Urine Cadmium, (ug/gr Cr)   |                  |                 |
 (mean + or - SD) ......... |   9.3 + or - 6.9 |  0.7 + or - 0.7 | < 0.0001
B-2 in Urine, (ug/gr Cr)    |                  |                 |
 (GM + or - SD) ........... |        470 (4.4) |       190 (1.6) |  0.0001
RBP in urine, (ug/gr Cr)    |                  |                 |
 (GM + or - SD) ........... |        266 (7.3) |        88 (1.9) |  0.0012
Systolic BP (mm Hg)         |                  |                 |
 (GM + or - SD) ........... |       134 (1.14) |      120 (1.14) |  0.0004
Diastolic BP (mm Hg)        |                  |                 |
 (GM + or - SD) ........... |        80 (1.13) |       73 (1.13) |   0.002
Blood Lead, (ug/dl)         |                  |                 |
 (GM + or - SD) ........... |  11.9 + or - 1.8 |  8.3 + or - 1.4 |  0.0013
  Footnote(1)  Exs. 8-670 and 19-43-B
  Footnote(2)  Converted to ug/m(3)-years, (Thun, Ex. 83), by multiplying
mg/m(3) by 1000 and dividing 365, or (604 times 1000 divided by 365 = 1655
divided by 45 = 36.8) 1,655 ug/m(3)-years; or 36.8 ug/m(3) for 45 years.

Several medical conditions were reported more frequently by the cadmium workers than by the controls including kidney stones (18% vs. 3%, p=.07), prostatic disease (20% vs. 6%, p=.09), diabetes (18% vs. 3%, p=.07), and hypertension (38% vs. 16%, p=.03).

The relationship between cumulative exposure and the prevalence of various renal abnormalities was examined, and a dose-response relationship was observed. The prevalence of abnormalities increased with cumulative exposure to cadmium with multiple renal abnormalities becoming apparent in persons with cumulative exposure = or > 300 mg/m(3)-day (or 822 ug/m(3)-yrs which is equivalent to an exposure of 18 ug/m(3) over a 45 year working lifetime). (See Section VI - Quantitative Risk Assessment).

viii. Summary. A summary of the major findings from these seven studies follows. For cumulative exposures up to 500 ug/m(3)-yrs, the prevalence of kidney dysfunction ranged from 4% to 32%; for cumulative exposures above 500 to 1,000 ug/m(3)-yrs, the prevalence of kidney dysfunction ranged from 9% to 66%; and for cumulative exposures above 1,000 ug/m(3)-yrs, the prevalence of kidney dysfunction ranged from 21% to 55%.

Taken together, the data from the seven studies would seem to refute the position of a number of commentors regarding what level of airborne cadmium represents a "safe" level. For example, Dr. Kazantzis has stated that exposure to 20 to 30 ug/m(3) of cadmium for 8 hours per day for 45 years (900 to 1350 ug/m(3)-yrs) is the no-observed-effect-level for kidney effects (Ex. 19-43A). On behalf of the Cadmium Council, Dr. Spang testified that workers will not have renal dysfunction if their cumulative cadmium exposures are up to 900 ug/m(3)-years (which is equivalent to 20 ug/m(3) for 45 years). Dr. Spang cited Sweden's standard of 20 ug/m(3) for respirable cadmium as adequate to prevent kidney dysfunction (Tr. 7/17/90, p. 217). Dr. Bond, who directs the medical surveillance program of two cadmium compound manufacturers in the United States, questioned the results from studies that show dysfunction at relatively low cadmium exposures. He did not agree with the definition of pathology used in these studies. Dr. Bond stated that a 20 ug/m(3) PEL would be reasonable and would not result in significant renal disease (Ex. 119).

Despite the opinions of these eminently qualified physicians, other record evidence indicates that the exposure level which they propose as "safe", 20 ug/m(3) for 45 years, (900 ug/m(3)-years) would result in a high prevalence of kidney dysfunction, in excess of 10% (Exs. L-140-50; 4-47; 4-27; L-140-45). OSHA also notes that the Swedish physician Dr. Elinder stated that the proposed OSHA regulations were better than the present regulations in Sweden (Ex. 55). Finally, while the Agency acknowledges Dr. Bond's objection to the definition of pathology used in these studies, OSHA is impressed by the consistency of the definition of pathology used by the various researchers in their studies which have been published in peer-reviewed journals.

Drs. Thun, Elinder, and Friberg reviewed these seven occupational studies together and concluded that the studies could be compared and pooled despite the fact that they varied in size, in criteria used to define kidney dysfunction, and in the amount of available exposure information. Pooling the data from the seven studies allowed the authors to obtain an estimate of the prevalence of dysfunction among workers with low cadmium exposures. In most of the studies, the number of workers with small exposures was too small to obtain a reliable estimate of prevalence, and therefore these workers had to be combined with workers with higher levels of exposure.

Results from the pooled analysis are presented in Table V-11.

Table V-11 - Occupational Studies Relating Kidney Dysfunction to
             Cumulative Exposure to Cadmium(1)
                   |       |         Prevalence(2) at exposures(3)
Study              |  N    |______________________________________________
                   |       | 100 - 199 | 200 - 299 | 300 - 399 | 400 - 499
Ellis, 1985 ...... |    82 |       0/3 |           |       0/3 |       2/3
Thun, 1989 ....... |    45 |       0/2 |       0/4 |       0/2 |       0/1
Falck, 1983 ...... |    33 |           |       1/3 |       1/2 |       0/7
Kjellstrom, 1977 . |(4)240 |           |           |           |
Jarup, 1988 ...... |   440 |     1/110 |      1/35 |      1/25 |      0/20
Elinder, 1985 .... |    60 |           |           |       0/2 |       0/1
Mason, 1988 ...... |    75 |      2/10 |       0/6 |       0/8 |       1/2
Pooled Data (%) .. |       |      2.4% |      4.2% |      4.8% |      8.8%
  Footnote(1) Source:  Thun, M.J., Elinder, C.G., Friberg, L., "Scientific
Basis for an Occupational Standard for Cadmium," Am. J. Ind. Med., 20:
629-642, 1991, (Ex. L-140-50)
  Footnote(2) Prevalence = number diagnosed as having kidney dysfunction in a
specific exposure category compared the total number of workers in that
exposure category
  Footnote(3) Exposures = in units of ugCd/m(3) - years
  Footnote(4) Included in Jarup

The prevalence of kidney dysfunction among workers with cumulative exposures between 100 - 199 ug/m(3)-yrs was 2.4%. For workers with exposures between 200 - 299 ug/m(3)-yrs, the prevalence was 4.2%. For workers with cumulative exposures between 300 - 399 ug/m(3)-yrs, the prevalence was 4.8%. For workers with cumulative exposures between 400-499 ug/m(3)-yrs, the prevalence was 8.8%.

Thun et al also plotted the observed prevalence from each of the seven studies by cumulative exposure. The data show a similar pattern between dose and response for each of these studies. The prevalence of kidney dysfunction increased sharply at cumulative exposures above 500 ug/m(3)-yrs. In all of the studies except Jarup, the prevalence of kidney dysfunction was about 10% when the cumulative exposures reached about 450 ug/m(3)-yrs.

c. Other Studies. Not all of the epidemiological studies on the renal effects cadmium that were submitted to the record had large enough cohorts or adequate dose data to assess the relationship between exposure and dysfunction. Nonetheless, these studies are useful for assessing the relationship between a variety of biological markers and kidney dysfunction. Specifically, these studies can be used to assess the efficacy of a variety of biological markers as a determinant of the critical concentration of cadmium which induces kidney dysfunction.

i. The NIOSH Health Hazard Evaluation of Gates Nickel-Cadmium Battery Plant (Ex. 128). The National Institute of Occupational Safety and Health (NIOSH) conducted two medical surveys of workers exposed to nickel and cadmium dusts in a nickel-cadmium battery plant in the U.S. The first survey was done in February of 1989 and was completed by 39 male workers in the plate-making and pressed plate areas of the plant where there is potential for exposure to high levels of cadmium. A group of 36 males, selected by the company and thought to have no cadmium exposure, served as controls. The second survey was done in October of 1989 and was complete by 91 workers in areas with either low or high exposure to cadmium but minimal exposure to nickel.

Both studies entailed administration of a questionnaire; measurement of height, weight, and blood pressure; and collection of first-voided morning urine samples and fasting serum samples. The questionnaire collected information about age, history of diabetes, hypertension, smoking, non-steroidal anti-inflammation drug use, and previous occupational exposures to cadmium, lead, and solvents. No information was provided on the pH of the urine samples or on how the urine samples were collected and handled.

In the first survey, the biologic makers selected to assess renal function included urinary phosphorous, B(2)-M and RBP which were creatinine-standardized to adjust for variations in urine concentrations. Levels were considered abnormally high if they exceeded the arithmetic mean level plus two standard deviations in the unexposed population, standardized for creatinine.

The biologic markers used in the second survey were similar to those used in the first survey except that urinary excretion of B(2)-M was not measured. Other indices of renal tubular function used in the second survey included urinary excretion of calcium and glucose. The laboratory reference values for normal limits of B(2)-M and total protein were urinary B(2)-M< 300 ug/liter and total protein = or < 135 mg/l. For RBP the reference values were less clear and covered a wide range of levels, e.g. urinary RBP< 30-190 ug/l or 0-406 ug/l.

Exposure data which were provided by the company were noted by NIOSH to raise several questions. One of these was whether exposures, which were reported as 8-hour TWAs, truly represented 8-hour TWAs. The company reported to NIOSH that in order to compensate for their normal 12 hour workshifts, the company modified the OSHA PEL by reducing it by 33 percent. Another problem was that because of a lack of consistent workstation terminology, considerable manual compilation of the computerized data was necessary in order to reconstruct each worker's exposure history. Finally, NIOSH believed that actual work practices may have differed from those reported on the printouts.

No consistent differences in urinary proteins between the cadmium-exposed group and the non-exposed group were observed in either survey. Furthermore, cumulative airborne cadmium levels as calculated from the exposure data provided by the company did not show a significant relationship with any measure of renal function used in this investigation. In both surveys, however, the analysis of cadmium-exposed workers with CdU levels greater than 10 ug/g Cr (23% in the first survey and 28% in the second survey) compared to those with less than 10 ug/g Cr clearly suggested that the group with higher levels of CdU did have modest elevations of the urinary proteins.

In the first survey, 3 of the 39 cadmium-exposed workers (8%) demonstrated evidence of cadmium-induced renal dysfunction. Two of these workers had elevated levels of only albumin in their urine, while the third had elevated levels of B(2)-M and RBP as well. In the second survey, 3 of 91 cadmium-exposed workers (3%) had elevated levels of urinary albumin, but none had elevated levels of urinary RBP. In comparison, of the 69 workers in both surveys with no or low exposure, none had evidence of abnormally high levels of urinary proteins.

The possibility of glomerular dysfunction was suggested in the first survey by a slightly higher mean serum creatinine in the exposed group than in the control group. NIOSH noted that when one considers that the non-exposed group was significantly older than the exposed group, the difference in serum creatinine may actually be larger than reported.

The NIOSH authors concluded that cadmium-induced renal dysfunction is evident in this study population. They also found that subclinical effects such as significant increases in mean levels of the urinary tubular enzymes, N-acetyl-B(D)-glucosaminidase (NAG; p=.05) and urinary alanine aminopeptidase (AAP; p=.02), are apparent in cadmium-exposed workers with CdU levels above 10 ug/g Cr compared to those below this level.

ii. Lauwerys et al, (Ex. 8-718). Lauwerys conducted a study of 11 workers employed in a small factory in Belgium which used or produced cadmium oxide, cadmium metal, cadmium sulfide, and various cadmium salts. Workers were observed for 13 months. Although the factory employed only seven workers on the production side, Lauwerys was able to follow a total of eleven workers because four workers left the plant and four new employees were hired during the study period.

The total airborne concentration of cadmium at the various work locations was very high. The median values ranged from 110 to 2125 ug/m(3). Reliance on personal protective devices was minimal; the authors noted that only one worker wore a mask during work. In view of the hygiene practice of the workers, the authors considered that ingestion of cadmium may have played a role in the overall exposure.

During the observation period, 150 personal air samples were collected. Each sampling period lasted two to nine hours. The airborne cadmium levels sampled in this factory ranged from 88 to 14,232 ug/m(3) with overall median, mean, and standard error of 565, 1119, and 125 ug/m(3), respectively. Omitting the most extreme result (14,232 ug/m(3)), the values ranged from 88 to 6276 ug/m(3) with overall median, mean, and standard error were 563, 1031, and 90 ug/m(3), respectively.

Cadmium concentrations in workers' blood and urine were measured. Because of the employee turnover and the high degree of collaboration requested from the workers (repeated blood and urine sampling) it was not possible to survey all the workers during the same length of time. On the other hand, Lauwerys considered the high employee turnover to be an advantage for this type of survey because it allowed the follow-up of newly employed workers.

An evaluation of renal function was performed once for each of the current workers (n = 8). To do so, urine was collected over a known period of time (usually 4 to 5 hours). The volume of urine was measured, and 10 ml was immediately transferred into a tube containing 1 ml of 0.4 mole/liter phosphate buffer, with a pH of 7.6. These samples were stored at - 20 deg. C and B(2)-M was measured using the radioimmunoassay (RIA), Phadebas from Pharmacia. Aliquots of urine were taken for the determination of creatinine, total proteinuria, amino aciduria, and some enzymatic activities (B-galactosidase, lactate dehydrogenase, alkaline phosphatase, total and tartrate resistant acid phosphatase, and catalase). The remaining volume was then stored at 4 deg. C with 0.1 percent sodium azide as preservative until ultrafiltration for electrophoresis and quantitation of individual proteins (orosomucoid, albumin, transferrin, and IgG). A sample of blood was also taken for determining the creatinine level and the same enzymatic activities and specific proteins in plasma as in urine. Despite the great scatter of the individual results, Lauwerys stated that for most workers exposed for more than 250 days, the average cadmium in urine levels were similar. The overall mean cadmium level was 161 ug/g Cr with a standard error of 9 ug/g Cr, regardless of type of exposure. All but one of these workers were exposed primarily to cadmium oxide. The one not exposed to cadmium oxide was exposed primarily to cadmium sulfide.

According to Lauwerys, the results of this study indicate that an integrated exposure of 1500 to 3000 ug/m(3)-yrs leads to kidney disturbances and/or renal lesions. Lauwerys also concluded that workers whose exposure is such that CdU never exceed 15 ug/g Cr would not develop kidney lesions.

To evaluate the mechanisms by which cadmium is taken up into the blood and urine in new workers, Lauwerys studied four new workers. One worker was followed for only 30 days. This worker, who wore a respirator, was exposed to cadmium oxide dust and salts (mean cadmium concentrations in air were 1829 ug/m(3) + or - 528; the median exposure level was 1167 ug/m(3)). This worker exhibited a much lower increase of cadmium concentration in blood and urine than the other three new workers. Lauwerys indicated that this worker was more motivated to follow better hygiene practices (hand-washing, no smoking at work) than the other workers and this limited his total exposure.

Results of biological monitoring for the other three workers are presented in Table V-12.

Table V-12 - Cadmium in Blood and Cadmium in Urine Levels(1) in New Workers
             Exposed in a Plant Producing Various Cadmium Compounds by
             Exposure Level and Type of Exposure and Smoking Habits of
Worker|     Mean      |Median|Smoker   |Exp.   |CdB- |CdB-  |CdU-   |CdU-
      |     air(2)    |air(2)|Status(3)|Type(4)|20(5)|140(6)|20(7)  |140(8)
      |               |      |         |       |     |      |       |
A(1)  |1329 + or - 438|  613 | S       | A     |  30 |  120 |  10.0 |  17.2
A(2)  |2043 + or - 452| 1926 | S +     | B     |  20 |  120 | 7.5(9)| 20(10)
A(3)  |2031 + or - 263| 1827 | NS      | B     |  50 |  105 |  25.0 |  25
  Footnote(1) From visual review of figures (Ex. 8-718).
  Footnote(2) ug Cd/m(3).
  Footnote(3) S = smoked <  1 pack/day; S + = smoked > 1 pack/day;
NS = non-smoker.
  Footnote(4) Main Type of exposure: A = Cadmium oxide dust and fume;
cadmium carbonate powder; cadmium salts; B = cadmium oxide dust;
  Footnote(5) Cadmium in blood levels after approximately 20 days of
exposure, in ugCd/liter whole blood.
  Footnote(6) Cadmium in blood levels after approximately 140 days of
exposure, in ugCd/liter whole blood.
  Footnote(7) Cadmium in urine levels after approximately 20 days of
exposure, in uCd/gram creatinine.
  Footnote(8) Cadmium in urine levels after approximately 140 days of
exposure, in ugCd/gram creatinine; average of four samples taken on days
130, 140, 142, and 144;
  Footnote(9) Average of two samples taken on days 10 and 24;
  Footnote(10) Average of samples taken on day 138 and 146 of exposure.

Regarding cadmium in blood levels, Lauwerys concluded that after the start of exposure, the concentration of cadmium in blood increases linearly up to 120 days and then levels off. Kjellstrom (1977, referenced in Ex. 8-718) found a similar evolution of cadmium concentration in blood with time among workers newly exposed to a much lower level of cadmium dust, about 50 ug/m(3). Kjellstrom reported, however, that the steady state level (about 3 ug/100 ml) was five times lower than that found in the workers examined in Lauwerys study.

Kjellstrom's findings are not surprising given the fact that the cadmium pollution in the Lauwerys plant was significantly higher than that in the cadmium-nickel battery factory investigated by Kjellstrom (1977). One can conclude that when equilibrium is reached, the cadmium level in blood is a good indicator of the average intake during recent months. This was also confirmed by Lauwerys by the finding that in three workers, the blood cadmium levels measured before and after a leave of absence of 4 weeks were not significantly different.

Dr. Lauwerys calculated the linear regressions between the duration of exposure and the cadmium levels in blood for the various phases identified (two for blood). Since cadmium levels in blood of workers exposed for more than 120 days are mainly a reflection of exposure, Dr. Lauwerys expected that no significant correlation between duration of exposure and cadmium levels in blood during phase 2 would be found. The observed relationship between cadmium in blood and exposure in phase 2 was not statistically significant. Thus, levels of cadmium in blood reflect recent exposures, but if exposures have been very high, will not decrease significantly, at least over four weeks.

Regarding cadmium in urine, Lauwerys (Ex. 8-718) concluded that the interpretation of the urine data is certainly less straightforward than that for blood. For one thing, the kidney function of the workers must be taken into consideration since it is known that kidney lesions may be associated with an increased urinary excretion of cadmium.

Dr. Lauwerys calculated the linear regressions between the duration of exposure and the cadmium levels in urine for the various phases identified (four for urine).

Dr. Lauwerys proposed the following hypothesis to explain the evolution of cadmium concentration in urine found during this survey. In workers newly exposed to high levels of cadmium, phases 1 and 2 are concomitant with a rapid and marked increase of cadmium body burden probably associated with an induction of metallothionein. As this binding process becomes progressively saturated, a sharper increase in cadmium concentration in urine occurs (phase 3) which eventually, if high exposure persists, will be mainly a reflection of recent cadmium intake (phase 4) rather than an indicator of body burden. Dr. Lauwerys noted that the beginning of the second phase, during which the cadmium binding sites become progressively saturated, corresponds to a urinary concentration of cadmium of approximately 15 ug/g Cr. This would suggest that as long as this level is not exceeded in male workers exposed to cadmium the saturation of all the body binding sites is not yet reached. If the binding occurring during phase 2 is a true detoxication process, which Dr. Lauwerys indicated remains to be confirmed, one would expect that workers whose exposure is such that cadmium in urine never exceeds 15 ug/g Cr would not develop kidney lesions. This hypothesis is in agreement with Lauwerys' previous clinical observations of signs of kidney damage in some workers who excreted more than 15 ug Cd/g Cr (Lauwerys et al.. 1974). Hence for adult males occupationally exposed to cadmium, Lauwerys proposed a tentative biological threshold of 10 ug Cd/g Cr in urine. The validity of this proposal is confirmed by the correlation between CdU and cadmium in kidney found in 309 Belgian workers whose cadmium in kidney was measured in vivo by neutron activation. It was found that a cadmium concentration in renal cortex between 200 and 250 ppm, considered as the critical level, corresponds to a CdU concentration of approximately 10 to 15 ug Cd/g Cr (Roels et al..1979).

However, Dr. Lauwerys stated that it should be stressed that this biological threshold is proposed only for adult males occupationally exposed to cadmium and does not necessarily apply to other groups of the general population, e.g., women after menopause and children, whose sensitivity to cadmium could be different.

Lauwerys findings agree with those of De Silva (Ex. 8-716) who noted that the urinary cadmium concentrations rise and fall with exposure, probably with a delay of several months.

iii. Roels et al, (57-K). Roels studied the effects of cadmium on male workers employed in one of two Belgium zinc-cadmium plants. Two hundred and sixty-four workers (264) were included in the analyses. Of these, 236 were active employees (Group A) and 28 were either retired or had been removed from jobs where they were exposed to cadmium (Group R).

To assess the cadmium pollution at the plants, airborne cadmium concentrations were measured with static air samplers at each of the principal worksites. In one plant, cadmium concentrations ranged from 3 to 67 ug/m(3), and in the other plant, they ranged from 5.8 to 168 ug/m(3). When monitoring was conducted at the worker's breathing zone, these levels were much higher.

For each worker, cadmium concentrations in the liver and in the kidney were measured in vivo by neutron capture gamma-ray analysis using the transportable measurement system developed at the University of Birmingham, U.K. Cadmium in the blood and cadmium, B(2)-M, albumin, total protein, and calcium in the urine were also measured. For each urine sample, an aliquot of 5 ml was immediately transferred into a tube containing a buffer with a pH of 7.6 and stored at -20 deg. C until B(2)-M levels and albumin were measured. B(2)-M was measured by radioimmunoassay using the Phadebas microglobulin test developed by Pharmacia.

Workers were considered to have abnormal kidney function if their total urinary proteins exceeded 250 mg/g Cr, if their B(2)-M exceeded 200 ug/g Cr, or if their albumin exceeded 12 mg/g Cr. These criteria were derived from a group of 88 unexposed workers whose urinary cadmium levels were below 2 ug/g Cr.

One hundred and forty-nine (149) of the active workers were engaged in jobs not directly related to cadmium production and these workers were found to have normal renal function. The remaining 87 active workers were involved in cadmium production daily, and of these, 15 (17%) had signs of renal dysfunction.

Examination of the cumulative frequency distributions and the correlations between the various biological parameters in different subgroups led the authors to the following conclusions: (a) calciuria is not much different among the subgroups; (b) CdB mainly reflects recent exposure to cadmium in the absence of cadmium-induced renal damage; (c) CdU follows the body burden of cadmium but increases proportionately much more in workers with renal dysfunction particularly when signs of tubular dysfunction are present; and (d) cadmium in the liver is proportional to duration and intensity of cadmium exposure in workers without as well as with renal dysfunction. The study authors also concluded that renal cortical cadmium does not differ between cadmium workers with and without renal dysfunction, but the observation on which this conclusion is based can be explained by a progressive decrease of cadmium in the kidney cortex after the onset of the renal damage.

The results of this investigation support the concept of a critical concentration of cadmium in the kidney cortex which must be achieved before dysfunction occurs. That concentration was found to range from 160 to 285 ppm in this study. When the critical concentration exceeds 285 ppm, the probability is very high that all persons will show signs of renal dysfunction. This study also demonstrated that in the absence of kidney dysfunction, CdU is correlated with the body burden of cadmium (r = 0.59), but that CdB is not.

On the basis of the interrelationships among cadmium in the liver, cadmium in the kidney cortex, CdU, and the other biological indicators of renal function, Roels concluded that the probability of developing cadmium-induced renal dysfunction in male cadmium workers appears to be very low when the critical CdU level of 10 ug/g Cr is not regularly exceeded. This CdU level corresponded to an average cadmium body burden of 160 to 170 mg.

iv. Roels et al, (Ex. 12-38A). In order to assess the significance of the early renal changes induced by chronic exposure to cadmium, Roels conducted a study of 23 retired workers at two non-ferrous smelters in Belgium. These workers had been removed from jobs entailing exposure to cadmium oxide as dust and fume (Ex. 57-K). They had been removed from exposure either because the level of B(2)-M in urine exceeded 300 ug/l (n=18) or because the level or RBP in urine exceeded 300 ug/l (n=17). In addition, 8 workers had levels of albumin in their urine in excess of 20 mg/l. At the time of removal from cadmium exposure, serum creatinine concentrations were normal in 18 workers (< 13 mg/l), marginally elevated in 3 workers (between 13 and 14 mg/l), and significantly elevated in 2 workers (>20 mg/l).

The average length of occupational exposure to cadmium was 25 years for these 23 workers with a range from 6 to 41.7 years. Workers had been removed from exposure for 6 years on average before they had their first follow-up examination. The mean age at first follow-up examination was 58.6 years with a range from 45.5 to 68.1 years. During each of five follow-up surveys conducted annually, workers were questioned about their health status and drug consumption, and a sample of venous blood (20 ml) and urine (100 ml) was collected. The results of the five annual follow up surveys are included in Table V-13.

Table V-13 - Characteristics and Biological Parameters of 23 Male Subjects
             with Signs of Cadmium Induced Renal Changes, who were removed
             from Exposures for 6 years: Summary of five-year follow up
Character-|   First    |   Second    |   Third    |   Fourth   |  Fifth
istic     |            |             |            |            |
CdU(1)    | 22.2 + or  | 16.0 + or   | 15.5 + or  | 15.6 + or  | 18.0 + or
          |    - 2.9   |    - 2.2    |    - 1.6   |    - 2.0   |    - 2.9
CdB(2)    |     14.3   |     11.8    |     10.1   |      9.3   |      9.7
B(2)U(3)  |     1292   |     1260    |     1684   |     1918   |     1743
RBPU(4)   |     1146   |      801    |      829   |     1396   |     1351
CrU(5)    |     1.37   |     1.23    |     1.52   |     1.34   |     1.48
  Footnote(1) Cadmium in urine, mean + or -  SEM, ug/li.
  Footnote(2) Cadmium in blood, ug/liter whole blood.
  Footnote(3) Beta 2 microglobulin in urine (ug B(2)/g Cr), standardized
to grams creatinine given in row 5.
  Footnote(4) Retinol binding protein in urine (ug RBP/g Cr), standardized
to grams creatinine given in row 5.
  Footnote(5) Creatinine in urine (ug/liter).

This study by Roels confirmed that cadmium-induced proteinuria is irreversible. For the 23 workers, the levels of urinary RBP, B(2)-M, and albumin, which were significantly elevated at the time of the first survey following removal from exposure, had not returned to normal levels five years later. In addition, the study found that serum alkaline phosphatase activity significantly increased during the five year follow up period which may reflect an interference of cadmium with bone metabolism, possibly secondary to a reduction in the conversion of 25-hydroxcholecalciferol to 1,25-dihydroxycholecalciferol by the kidney. The most important finding, however, was a significant increase over time in creatinine and B(2)-M levels in serum which would indicate a progressive reduction of the glomerular filtration rate (GFR) despite removal from exposure.

Roels estimated that the GFR decreased on average by 31 ml/min/1.73 m(2) in the workers he studied during the five years of follow-up. Based on the work of others, Roels stated that in the age range of 45 to 75, the expected decline over five years should normally not exceed 6.5 ml/min/1.73 m(2). For each of the workers in the study, the reduction in their estimated GFR was, on average, about five times greater than this expected level. Interestingly, it was not more pronounced in workers with impaired renal function at the start of the study than it was in those with subclinical signs of renal damage.

v. Roels et al, (Ex. 149). In this study, Roels sought to determine whether an internal dose of cadmium that has not yet induced microproteinuria could affect the filtration reserve capacity of the kidney. Internal dose of cadmium was reflected by cadmium concentration in urine. Microproteinuria was defined as a significantly increased urinary excretion of various plasma proteins: B(2)-M > 300 ug/g Cr or RBP > 300 ug/g Cr, or albumin > 15 mg/g Cr, or a combination of these. The subjects in this study were 108 workers at two zinc-cadmium smelters in Belgium with occupational exposed to cadmium. To be included in the study, the exposed workers must have been exposed to cadmium for one year without interruption and must have been excreting more than 2 ug CdU/g Cr. Also, they could not have been exposed to other known nephrotoxins.

One hundred and seven (107) workers with no occupational exposure to cadmium served as controls. To qualify as a control, a worker should never have been occupationally exposed to any nephrotoxins. The level of cadmium in the urine of a control was required to be below 2 ug/g Cr and there could be no sign of microproteinuria. The control workers were closely matched to the exposed workers on age, and care was taken to see that both groups had similar socioeconomic (education, salary) and environmental (place of residence) characteristics.

A detailed occupational and medical questionnaire was given to each study participant. In addition, data on each worker's dietary habits during the week before the study was also collected. In order to be included in the final statistical analyses, participants in the study must have complied scrupulously with the study protocol that required them to refrain from taking analgesics, and their medical history must not have shown any pathological condition that might have influenced renal function.

During the 30 minutes prior to the baseline test, a spot urine sample (100 ml) was collected. An aliquot of 4 ml was immediately transferred to a tube containing 0.4 ml phosphate buffer, pH 7.6, and kept at - 20 deg. C until the analysis of B(2)-M, RBP, and albumin were performed. These proteins were measured by automated assays relying on latex particle agglutination.

Because early changes in glomerular function cannot be detected by the measurement of basal GFR, Roels developed a test to assess the filtration reserve capacity of the kidney and to detect any early renal changes induced by cadmium. In order to do this, Roels defined the filtration reserve of the kidney as the difference between baseline GFR and the maximal GFR induced by an adequate stimulus such as an acute oral load of proteins or an infusion of amino acids. The maximal GFR obtained during such stimulation would thus represent the maximum filtration capacity. When Roel's test was applied to his study subjects, the results confirmed the observation in previous studies that the age-related decline of the baseline and maximal GFR is accelerated in male workers with cadmium induced microproteinuria.

Another analysis was performed for workers less than age 50 and over age 50. Microproteinuria was present in 20 cadmium workers, all older than 50. It was found, however, that a renal cadmium burden that has not yet caused microproteinuria does not impair the filtration reserve capacity of the kidney.

In conclusion, this study indicates that the age related decline of the baseline and maximal GFR is exacerbated in the presence of cadmium-induced microproteinuria. The investigation supports Roels' previous estimate of the threshold effect concentration of CdU (10 ug/g Cr), which is intended to prevent the occurrence of microproteinuria in cadmium exposed male workers. Roels noted that this conclusion, however, may not be extrapolated to the general population because there are indications that in an occupationally active male population, the influence of the healthy worker effect may lead to an underestimation of the risk of cadmium for other groups of the general population. Also, sensitive workers may not be adequately protected if CdU levels exceed 10 ug/g Cr.

vi. Bernard and Lauwerys (Ex. 35). Bernard and Lauwerys studied 25 male workers who had been removed from jobs with exposure to cadmium when they were found to have elevated levels of B(2)-M, RBP, or albumin in their urine. The serum levels of creatinine and B(2)-M in these workers were found to increase significantly with time. Over a five year period, the average level of creatinine in serum increased from 12 mg/l (SE = 1.1 mg/l) to 15.5 mg/l (SE = 2.2 mg/l), and the average level of B(2)-M in serum increased from 1.89 mg/l (SE = 0.12 mg/l) to 3 mg/l (SE = 0.42 mg/l). The average levels of serum creatinine in two groups of 23 age-matched controls after five years were found to be 11.3 mg/l and 11.2 mg/l, and the average levels of serum B(2)-M for these two groups were found to be 1.9 mg/l for both groups. Thus, age could not account for the increase observed in the exposed workers.

The GFR was estimated according to Wibell at al., as referenced by Bernard. All workers showed a decrease in estimated GFR which ranged from 9 to 78 ml/min/1.73 m(2) over the five year observation period. The average decrease of GFR over that period amounted to 31 ml/min/1.73 m(2), a value which is about five times greater than that observed for the normal population. Investigations of cadmium-exposed workers in Belgium and the U.S. have demonstrated that a low or high molecular weight proteinuria is likely to develop in 10 percent of exposed subjects when the concentration of the metal in the renal cortex reaches about 200 ppm. The corresponding critical levels in urine and blood have been estimated at 10 ug/gr Cr and 10 ug/liter whole blood (lwb), respectively (Ex. 57-K). Once it has appeared, according to Drs. Bernard and Lauwerys, cadmium proteinuria is in most cases irreversible. Follow up studies indicate that the progression of renal dysfunction after cessation of exposure is very slow. Bernard and Lauwerys, in their recent five year prospective study, demonstrated that despite this slow evolution, cadmium nephropathy may progress to renal insufficiency.

vii. Toffoletto et al, (Referenced in 19-43A). Toffoletto presented a paper on the effects of renal function from occupational exposure to cadmium at a conference on heavy metals in Edinburgh in 1989. Toffoletto studied 91 workers exposed to cadmium between 1981 and 1988 in an Italian factory producing and processing cadmium alloys. Periodic measurements of environmental cadmium concentrations had been made for 13 years, and information was available from a biological monitoring program that had been in operation for 8 years. Only subjects with three or more years of exposure or with three measurements of CdU or CdB were included in the study.

The authors used biological monitoring results from a control population to establish the upper limit of normal for CdB and CdU levels: 2.3 ug/liter whole blood (lwb) and 3.0 ug/liter urine (l urine), respectively. At the time of this study, the biological limit values recommended for CdB and CdU were 10 ug/lwb and 10 ug/l urine, respectively. Elevated B(2)-M levels above 260 ug/l were used as an indicator of early tubular damage. For B(2)-M analyses, urine samples were adjusted for pH, and B(2)-M levels were measured using the Phadebas microtest kit.

Among workers whose CdB and CdU levels were always below 10 ug/lwb or below 10 ug/l urine, 3% and 2.7% respectively, had B(2)-M levels above 260 ug/l. Workers with at least one CdB level greater than 10 ug/lwb or one CdU level above 10 ug/l urine (33.3% and 16.7% of workers, respectively) had elevated B(2)-M levels. Forty-eight workers had measurements of CdB, CdU, B(2)-M, RBP, N-acetyl-B(D)-glucosaminidase (NAG), and microalbumin. These workers were evaluated for kidney function. The results are in Table V-14.

Table V-14 - Association Between Tubular and Glomerular Function Indicators
             and Their Distribution Divided into Three Groups of Levels of
             Urine (CdU)
                  |  Total     |  Number
     CdU (ug/l)   |  Number of |  Abnormal
                  |  People    |  (Percent)
< 3 ............. |          7 |      4 (57)
3-10 ............ |         14 |      8 (57)
>10 ............. |         27 |     18 (67)

Four of seven workers with median CdU levels less than 3 ug/l had abnormal levels of NAG, microalbumin, or RBP. Eight of 14 workers with median CdU levels between 3-10 ug/l had abnormal levels of at least one of the four indicators of kidney function (B(2)-M, NAG, RBP, or microalbumin). Eighteen out of 27 workers with median CdU levels greater than 10 ug/l had at least one elevated abnormal kidney function test result. In this latter group, four workers had abnormal levels of all four kidney function measurements.

Among ten workers with elevated B(2)-M levels, half had CdB levels less than 10 ug/lwb. (See Table V-15.) Their mean CdU level was 11 ug/l (range 2.5-13.8). Their mean cumulative exposure was 460 ug/m(3)-yrs (range 260-721). The other five workers had a mean CdU level of 20.8 ug/l (range 12.6-26.2) and mean cumulative exposures of 5982 (1965-8382).

Table V-15 - Mean Cadmium in Urine and Blood Levels Among 10 Workers with
             Elevated Beta 2 Microglobulin Levels by Cumulative Cadmium
Number of Workers |CdU (ug/l)|   CdB (ug/l)     | Cumulative Exposure(1)
5 ............... |     <  10 |11.0 (2.5-13.8    |     450 (250 - 721)
5 ............... |     <  10 |20.8 (12.6 - 26.2)|  5892 (1965 - 8382)
  Footnote(1) ug/m(3)-years

Toffoletto et al. concluded that if CdB and CdU levels are kept constantly below 10 ug/lwb or 10 ug/l urine, the prevalence of kidney dysfunction measured by elevated levels of B(2)-M would be below three percent. However, five workers in this study (5.9 %) with a mean exposure of 460 ug/m(3)-yrs (or 10.2 ug over a 45 year working lifetime) had elevated B(2)-M levels above 260 ug/l.

viii. Buchet et al, (Ex. 8-201). The renal function of workers occupationally exposed to cadmium (n = 148) was compared with that of workers with no occupational exposure to heavy metals (n = 88). The exposed and control populations were employed in two cadmium smelters in Belgium. In order to be a control subject, the worker had to fulfill several conditions: all levels of CdU had to be below 2 ug/g Cr; in the judgement of the plant physician, the worker had to have no occupational exposure to cadmium; and the controls belonged to the same socioeconomic class as the exposed workers.

Five-hour urine samples were collected from the workers while at work. Urine was pH adjusted and samples for B(2)-M analyses were frozen at -20 deg. C. The concentration of B(2)-M was measured by radioimmunoassay using the Phadebas B(2)-microtest kit developed by Pharmacia Diagnostics.

Descriptive characteristics of the exposed and control groups are included in Table V-16.

Table V-16 - Descriptive Characteristics of Cadmium
             Exposed Workers and Controls
           Parameter         | Controls | Exposed
Number ..................... |  88      |     148
Age (years) ................ |  38.6    |    46.5
Duration of Employment       |          |
 (years).                    |  8.4     |    15.4
Mean Level of Cadmium in     |          |
 Urine (ug/100 creatinine)   |  0.88    |   15.76
Mean B(2) in Urine (ug/gr    |          |
 creatinine)                 |  71 (for |     739
                             | N=87     |
                             | workers) |
Prevalence of Kidney         | 6.8 .... | (1)18.2
Dysfunction Based on         |          |
B(2)>200 ug/gl Cr (per cent).|          |
  Footnote(1) p< 0.025

Renal dysfunction was defined by elevated levels of B(2)-M above 200 ug/g Cr, among other definitions. The prevalence of renal dysfunction was significantly different between exposed workers and controls. The prevalence of abnormal results was not different between smokers and non-smokers.

Buchet concluded that excessive exposure to cadmium increased the urinary excretion of both low and high molecular weight proteins and of tubular enzymes. These changes were mainly observed in workers excreting more than 10 ug Cd/g Cr or with CdB levels above 10 ug Cd/lwb. However, among workers whose CdU levels were consistently below 2 ug Cd/g Cr, controls, and among workers whose CdU levels were between 2 - 9.9 ug Cd/g Cr, the prevalence of kidney dysfunction was six percent. The prevalence of kidney dysfunction among workers whose CdU levels were between 10 - 19.9 ug Cd/g Cr and > 20 ug Cd/g Cr was 15 and 40 percent, respectively.

ix. Summary. The eight studies reviewed above demonstrate a consistency in the levels of biological parameters associated with renal dysfunction. In all but one study by Lauwerys, kidney dysfunction was found in workers whose cadmium in urine exceeded 10 ug Cd/g Cr.

The NIOSH Health Hazard Evaluation found modest elevations in low and high molecular weight proteins in the urine of workers whose CdU levels exceeded 10 ug/g Cr. It demonstrated that subclinical effects such as significant increases in mean levels of urinary tubular enzymes, NAG and alanine aminopeptidase (AAP), are apparent in cadmium-exposed workers with CdU levels above 10 ug/g Cr compared to those below this level.

Lauwerys (Ex. 8-718) concluded from his study that workers whose exposure is such that CdU levels never exceed 15 ug/g Cr would not develop kidney lesions. Roels (Ex. 57-K) concluded that on the basis of the interrelationships among levels of cadmium in liver, kidney cortex, and urine, it can be concluded that the probability of developing cadmium-induced renal dysfunction in male cadmium workers appears to be very low when the critical CdU level of 10 ug/g Cr is not regularly exceeded. This CdU level corresponded to an average cadmium body burden of 160 - 170 mg. In his 1989 paper, Roels (Ref. in Ex. 149; also Ex. 12-38-A) concluded that his study indicated that the age related decline of the baseline and maximal GFR is exacerbated in the presence of cadmium induced microproteinuria. The investigation supports Roels' previous estimate of the threshold effect concentration of CdU (10 ug/g Cr), which is intended to prevent the occurrence of microproteinuria in cadmium exposed male workers. Roels noted that this conclusion, however, may not be extrapolated to the general population because there are indications that in an occupationally active male population, the influence of the healthy worker effect may lead to an underestimation of the risk of cadmium for other groups of the general population.

Bernard and Lauwerys (Ex. 35) concluded that the results of their study demonstrated that a low or high molecular weight proteinuria is likely to develop in ten percent of exposed subjects when the concentration of the metal in renal cortex reaches about 200 ppm. The corresponding critical levels in urine and blood have been estimated at 10 ug/g Cr and 10 ug/lwb, respectively.

Toffoletto (Ex. 19-43-A) concluded that if CdB and CdU levels are kept constantly below 10 ug/lwb or 10 ug/l urine, the prevalence of kidney dysfunction measured by elevated levels of B(2)-M would be below three percent. However, five workers in this study (5.9%) with a mean exposure of 460 ug/m(3)-yrs (or 10.2 ug over a 45 year working lifetime) had elevated B(2)-M levels above 260 ug/l.

d. The Biological Significance of Cadmium-Induced Kidney Dysfunction. Prolonged exposure to cadmium may lead to glomerular proteinuria, glucosuria, aminoaciduria, phosphaturia, and hypercalciuria (Exs. 8-086b: 4-28; 14-18, p. 157). These conditions are indicated by excess urinary amino acids, glucose, phosphate, or calcium, respectively. Each of these elements are essential to life, and under normal conditions their excretion is regulated by the kidney. Once low molecular weight proteinuria has developed, however, these elements may dissipate from the body. Loss of glomerular function may also occur, indicated by a decrease in the glomerular filtration rate and an increase in serum creatinine. Severe cadmium-induced renal damage may develop into chronic renal failure and uremia at which point some form of dialysis or kidney operation will be needed (Ex. 55).

Kidney dysfunction persists for years even after cessation of exposure. Loss of calcium and phosphorus may contribute to the increased risk of kidney stones observed in workers. Even in his early study of cadmium workers, Dr. Friberg noted renal stones as a common finding among cadmium exposed workers (Ex. 4-29). Dr. Friberg testified that, in his opinion, kidney stones are a serious sequelae to cadmium-induced renal dysfunction. He and others originally thought that the increased prevalence of kidney stones observed in his studies was confined to Sweden, But later, the increased prevalence of kidney stones was observed in England, and in the U.S. Kidney stones, according to Dr. Friberg, is a very serious disease and is also a sign of a more generalized disorder of the mineral metabolism in the kidney (Tr. 6/6/90, p. 106).

Others held a different opinion about the prevalence of kidney stones among cadmium-exposed workers. For example, Dr. Spang stated that kidney stones are common in the general population of Sweden (20% in men and about 5% in women), and although he observed cases of kidney stones among cadmium-exposed workers, he did not know if the prevalence was different from that of the general population (Exs. 80; 81).

Cadmium may also precipitate clinical osteopathy in persons with inadequate dietary calcium intake (Ex. L-140-50). Diets low in vitamin D and calcium may be a contributing factor to sequelae subsequent to cadmium-induced renal dysfunction.

There are at least two hypothesized scenarios by which cadmium-induced tubular proteinuria can cause other adverse health effects (Ex. 8-086). Under the first of these, cadmium-associated tubular dysfunction causes damage to the production of biologically active metabolites such as vitamin D which occurs primarily in the kidney. Under the second scenario, cadmium may cause atrophy of the gastrointestinal tract thereby reducing its ability to absorb essential elements such as calcium and phosphates. If both scenarios are true, it would lead to loss of essential elements and poor absorption of other minerals to replace those lost.

The gravity of cadmium-induced renal damage is compounded by the fact that there is no medical treatment to prevent or reduce the accumulation of cadmium in the kidney. Dr. Friberg has testified that there is currently no form of chelating agent that could be used without substantial risk (Ex. 29). In contrast to other heavy metals, current chelation therapy does not reduce the body burden of cadmium without producing significant renal damage. When chelated cadmium arrives in the kidneys, the cadmium may still be toxic to renal cells. Thus, large amounts of cadmium may move from the liver or muscle storage sites, overwhelm the kidney's usual attempts to store cadmium in a less toxic form, and accelerate deterioration of renal function.

The kidney cortex contains about three million nephrons. Dr. Goyer testified that:

"....a young, healthy adult uses about half of function is lost because of old age or...diseases...the number of these that are functioning through life continually decreases......"(Tr. 6/6/90, p. 124).

OSHA believes that the loss of function of the proximal tubules as indicated by tubular proteinuria, elevated levels of *2-M in the urine, constitutes material impairment of health.

OSHA acknowledges that the significance of the dysfunction as evidenced by elevated levels of B(2)-M in the urine is controversial. Part of this controversy arises from the fact that a worker with elevated levels of B(2)-M may not experience any symptoms, and although tubular dysfunction can be determined through medical testing, it usually does not manifest itself at first with overt symptoms.

Dr. Goyer testified that the confusion over the interpretation of pathological significance of elevated levels of B(2)-M stems from the fact that injury to the tubuli ultimately affects the functioning of the glomerulus. According to Dr. Goyer, the confusion lies in part in the fact that cadmium's earliest effect is primarily in the tubule, while kidney function is usually measured in the glomerulus (Tr. 6/6/90, pp. 126-127).

While most physicians would agree that glomerular effects and loss of GFR must be taken more seriously than a slight elevation in B(2)-M in urine, the finding of elevated levels of low molecular weight protein in the urine by itself indicates kidney dysfunction in the tubule. As Dr. Friberg stated in his testimony, each part of the nephron is dependent on every other part of the nephron. It is his expectation that if one part of the nephron suffers damage it is more likely that another part will suffer damage (Tr. 6/6/90, pp. 107-108). Ultimately then, cadmium-related tubular effects will be manifested as an effect on the function of the glomeruli, either subsequently to or in association with the onset of tubular proteinuria.

Because of the functional reserve of the kidney, the adaptive increase in a single nephron's glomerular filtration rate, after total or partial loss of other damaged nephrons, tends to obscure injury until a considerable amount of the functional elements of the kidney, the parenchyma, is irreversibly lost. This implies that under normal conditions, the basal GFR is submaximal. If as has been suggested, glomerular balance is very tightly maintained, reduction of tubular function may have repercussions on the glomerular level (Ex. 149). Early changes in glomerular function are not necessarily detectable by the measurement of basal GFR, but such changes may have a significant impact on health (Ex. 149). In a study by Roels (Ex. 149) it was found that a renal cadmium burden that had not yet caused microproteinuria did not impair the filtration reserve capacity of the kidney, but the age related decline of the baseline and maximal GFR is exacerbated in the presence of cadmium induced microproteinuria.

Not all participants in the rulemaking agreed that elevated levels of B(2)-M signified material impairment of health. Mr. Ken Storm, Senior Industrial Hygiene Specialist with Monsanto, stated that elevated levels of B(2)-M may reflect a temporary or permanent change in renal function and tubular proteinuria may result from a biochemical lesion of no clinical significance. According to Mr. Storm, tubular proteinuria would be more appropriately viewed as an early indicator of pre-clinical effects and not, in and of itself, as a material impairment of health. Mr. Storm stated that the intent of OSHA to avoid tubular proteinuria is inappropriate because, in his opinion, tubular proteinuria is not a material impairment of health (Ex. 19-14).

Mr. Storm stated furthermore that urinary B(2)-M and other biological indicators of early tubular dysfunction, such as n-acetyl-d-glucosaminidase (NAG), are nonspecific indicators of tubular proteinuria. Their presence may indicate past excessive cadmium exposure, exposure to another renal toxin, or loss of renal function due to the normal process of aging or other natural causes (Ex. 19-14).

Studies indicate that age alone can not account for the excess of B(2)-M observed in cadmium-exposed workers. Kowal et al. (Ex. 8-642) evaluated the levels of B(2)-M in non-occupationally exposed populations in the United States and found that the average level in the oldest group studied (107 ug B(2)-M/l urine) was only marginally higher than the average level in the groups between age 20 and 70 (69 to 84 ug B(2)-M/l urine) (referenced in Ex. 8-068-B). In addition, Also, several researchers such as Dr. Elinder evaluated the prevalence of B(2)-microglobulinuria by age among occupationally exposed populations and concluded that age was not an important confounding factor (Ex. L-140-45).

The specificity of B(2)-M in urine as a marker of cadmium-induced kidney dysfunction is well established. The only other renal toxins or medical conditions which lead to elevated levels of B(2)-M are anti-cancer drugs, aminoglycosides (antibacterial antibiotics such as streptomycin), anti-inflammatory compounds, and upper respiratory infections (Dr. Friberg, Tr. 6/6/90, pp. 108-109; Ex. L-140-1). As Michael Thun, M.D., Assistant Vice President for Epidemiology and Statistics at the American Cancer Society testified:

"Low molecular weight proteinuria....does occur from other conditions but it's uncommon....part of the reason why the (kidney) data are so consistent is that the studies use a rather specific marker of cadmium renal effects...."(Tr. 6/7/90, p. 174)"

Dr. Bond, medical consultant to SCM Chemicals, testified that:

".... no histological abnormalities [are] seen in the proximal tubules ... when there has been modest increase in urinary B2MG and Cd. .... (people with) .... mild to moderate increases in urinary B(2)MG and Cd do not progress to renal failure if there are no other causes present such as infection, diabetes, etc.(Ex. 77)"

Dr. Friberg, however, stated:

"It should be emphasized that tubular proteinuria may be accompanied by specific histological changes. Sometimes such changes have been reported before the functional changes. There are abundant data from animal studies showing early histological changes (Ref. by Kjellstrom, 1986, pp. 38-43). Experiments from humans are more limited as only a small number of autopsies or biopsies are available. To the extent available, histologic changes were seen first of all in the proximal tubules (Ref. by Kjellstrom, 1986, p. 50-53)."(Ex. 29).

Morphological changes are those that pertain to the form or structure of the organ. Histological changes are those that pertain to the minute structure, composition, and structure of the tissue of the organ. Twenty-three workers were evaluated for whom autopsy or biopsy data on morphological changes in the kidney were available (referenced in Dr. Friberg's written testimony). Of these, 18 workers had proteinuria. Of the 18 workers with proteinuria, all but three had morphological changes in their kidneys. There were no cases of workers with morphological changes without proteinuria (Ex. 144-3, p. 53). In five of the autopsy reports, the morphological changes in the kidneys were mainly confined to the proximal tubules, whereas the glomeruli were less affected.

These results demonstrate that functional changes in the kidney can occur before the microscopic structure of the kidney is severely damaged. The human data on pathological changes are limited, however, and animal data show that in some studies, morphological changes in the tubules emerge before measurable proteinuria. In the absence of a better test, however, it appears that the use of proteinuria as a screening tool for morphological changes in the kidney will identify all cases of workers with histological or morphological changes in kidney tissue as well as identifying those with only functional changes. These results also show that elevated levels of B(2)-M in urine indicate kidney lesions of clinical significance (Ex. 19-14). While a worker with elevated levels of B(2)-M in the urine may not manifest any overt symptoms of illness, nonetheless, the tubuli and glomeruli have lesions that compromise the functioning of the kidney as a filtration mechanism. Any other minor kidney trauma may progress rapidly to serious kidney damage.

It is clear from the testimony of world experts that elevated levels of B(2)-M should be considered to signify material impairment. Dr. Friberg testified that:

"....the beta-2 microglobulin proteinuria....should be regarded as an adverse effect......predictive of an exacerbation of the age related decline of the glomerular filtration rate.......the proteinuria in cadmium poisoning is irreversible and is predictive of more severe effects even if the worker is removed from further cadmium exposure.......It is true that an increased excretion of low molecular weight proteins can be a very early indicator of kidney dysfunction. That's not immediately of the same clinical importance as an overt renal disease. Nevertheless, it is irreversible and the beginning of a process which has a high probability to lead to a progressive disease, a decrease in the glomerular filtration rate which clearly is a serious effect that easily may lead to overt disease. When discussing the kidney damage from Cadmium, it is important that we make it clear that we are talking about serious, but often insidious effects on vital organs. The kidney has a considerable reserve capacity but once this is consumed symptoms may appear in swift succession and the condition of the patient then deteriorates rapidly, and the infection or other, in itself trivial disorder, could be a triggering mechanism. It is our responsibility to prevent this situation even among a small proportion of workers." (Tr. 6/6/90, pp. 73, 82, 86).

According to the American Conference of Governmental Industrial Hygienists (ACGIH):

"Persons excreting 290 ug/L B(2)-microglobulin are not disabled;

indeed they will not experience any symptoms. However, the lesion (from tubular proteinuria) is irreversible and represents a permanent loss of functional reserve. An infection or other condition which compromises renal function, but which would not normally lead to serious illness, could overwhelm the remaining kidney capacity.(Ex. 8-644)"

Dr. Kazantzis did not agree with these positions. He testified that in his opinion, tubular proteinuria alone is not accompanied by any specific histological change, that its pathological significance is unclear, and that renal stone formation has been rare in cadmium workers in recent years (Ex. 19-43A). Dr. Kazantzis stated, however, that in a:

"......small proportion of long-term heavily exposed cadmium workers, tubular proteinuria has been followed by renal glycosuria, abnormal aminoaciduria, phosphaturia, and hypercalcuria."(Exs. 80; 81)

Dr. Kazantzis continued that progressive decline in renal function is a slow process in workers with cadmium-induced nephropathy and that this decline is unlikely to progress to an increased mortality from chronic renal disease. In support of his opinion, he cited his study (Ex. 8-603) in which approximately 7000 cadmium-exposed workers with more than one year of cadmium exposure between 1942 and 1970 were followed up to 1979 (Ex. 8-684). He found an SMR of 65 for all deaths coded as nephritis and nephrosis; the five year update showed an SMR of 85. One worker classified as being in the "ever high" exposure subgroup died from nephritis and nephrosis.

Dr. Elinder indicated, however, that most workers in Dr. Kazantzis' study had such low cadmium exposures that cadmium-associated illnesses would not be induced (Ex. 4-25). By combining 199 workers with high exposures into a group with over 6000 workers with low exposures into one group, the power of the study to find an effect was reduced. Increased mortality from chronic nephritis and nephrosis has been observed in Swedish battery workers (Exs. 4-68 and 8-740). The difference between expected and observed deaths in the Kazantzis study may well be due to local differences in recording certain types of information on death certificates.

Three other epidemiological studies of cadmium exposed workers have shown increased mortality from kidney diseases, genito-urinary tract diseases, or kidney cancer. Thun observed an elevated SMR for genito-urinary cancer (SMR=135, Obs=6) in his total cohort (Ex.4-67); Dr. Elinder (Ex.4-25) reported an elevated SMR for genito-urinary diseases in his total cohort (SMR=300, Obs=3.0); and Holden et al. (Ex.4-39) observed an elevated SMR for genito-urinary cancer in his total cohort (SMR=122, Obs=4.0). Because the number of excess cases in each study is too small to make these findings statistically meaningful, the relationship between cadmium exposure and risk of death from kidney dysfunction is not clear. These three mortality studies, however, provide consistent evidence of excesses of kidney illnesses among cadmium-exposed workers. This suggests the possibility that, at least in some cases, cadmium-induced kidney dysfunction may be associated with excess death.

Death from nephritis, nephrosis or end-stage renal disease is rare. Accurate death rates from kidney disease are difficult to ascertain, in part because such illnesses are uncommon and in part, because they are dramatically underreported by at least 50% [personal communication 4/30/92, National Institute of Diabetic, Digestive and Kidney Diseases]. Dr. Thun indicated that impaired renal function is frequently underreported on death certificates even when the disease is sufficiently severe to require chronic hemodialysis (Modan referenced in Thun; Ex. 4-68). Under-reporting results because deaths from these diseases are coded as deaths due to complications arising from the treatment of these diseases or from sequelae to these diseases such as heart attack, stroke or diabetes.

Treatments for severe kidney diseases such as dialysis or a kidney transplant are available for those who can afford them. As Dr. Friberg indicated, several of his own patients had cadmium-induced uremia and died. If they had had the opportunity for dialysis or renal transplant, they could have been saved (Ex. 29). Such treatments, however, are grave, especially considering that early forms of kidney dysfunctions can be detected and more serious diseases can be prevented.

An additional part of the controversy over the significance of tubular proteinuria is the question of whether it is a reversible effect. In response to this question, Dr. Goyer, citing a study in Japan, stated that half of the people with B(2)-M levels in the range of 500 to 1,000 ug/g, followed for five years, do not show signs that their disease is reversible. Dr. Goyer testified that indeed, "The disease progressively gets worse...(Tr. 6/6/90, p. 136)." Dr. Friberg testified that:

"The continuous release of cadmium from the liver, also after end of the exposure, means that the accumulation of cadmium will take place in the kidneys for a long time after end of exposure. This was shown in animal experiments as early as 1957 by Gunn and Gould ...... Similarly, there is much data showing that the proteinuria in chronic cadmium intoxication is irreversible ..... two studies from beyond doubt that several years after removal of the worker there is either an increase of low molecular weight proteins in the urine or no change at all. There is also an indication that all the subjects with normal levels of beta-2 microglobulin in urine one year before removal can get pathological values....a few years later." (Tr. 6/6/90, pp. 74-75)

The main studies referred to by Dr. Friberg were five-year updates on workers who had been medically removed from occupational exposures due to cadmium nephrotoxicity (reviewed by Bernard and Lauwerys, Ex. 35, Roels, Ex. 12-57K). Among male workers who had been removed from cadmium exposure because of elevated urinary excretion of B(2)-M, RBP, or albumin, the evidence was that kidney dysfunction increased significantly over the five year period. Once it has appeared, Drs. Bernard and Lauwerys concluded, cadmium-induced proteinuria is in most cases irreversible. Bernard and Lauwerys demonstrated that proteinuria slowly progresses. Despite their finding that this evolution was slow, the authors concluded that the onset of proteinuria should be considered to be an adverse health effect, since such cadmium nephropathy may progress to renal insufficiency.

Dr. Bond stated that the clinical significance of slight increases in urinary B(2)-M(for example, 350 ug/l) is uncertain, but that a repeated finding of B(2)-M levels twice that of normal would more likely reflect a permanent effect, based on his experience and the literature. (Tr. 7/18/90, p. 169) Dr. Bond also agreed that cadmium-induced proteinuria must be prevented or minimized in order to prevent material impairment of health (Tr. 7/18/90, pp. 150-258, 175-176). About 20% of the cadmium workers that Dr. Bond has medically evaluated have elevated B(2)-M levels. Dr. Bond removed two of these workers from cadmium exposure in 1986 when their B(2)-M levels in the urine were 3000 to 5000 ug/l. Annual testing after removal indicated that urinary B(2)-M and cadmium levels did not decline appreciably. Dr. Bond stated that in his opinion, these two workers are not sick based on results from tests of their level of serum creatinine and alpha phosphatase which measure kidney function (Tr. 7/18/90, pp. 189-191). Dr. Bond did indicate, however, that he was "concerned" about the welfare of these two individuals because he did not know if they were likely to develop any further problems. (Tr. 7/18/90, p. 229) According to Jarup et al (Ex. 8-661) during the ten-year period of follow-up in his study, none of the cases of elevated *2-microglobulinuria (greater than 310 ug B(2)-M/g Cr) discovered in the high dose groups were reversible. The authors concluded that it was unlikely that any of these cases of tubular proteinuria would disappear after such a long follow-up time and that it was quite possible that more cases of tubular proteinuria would develop with a longer follow-up.

It is clear from the record of the rulemaking that despite some controversy, there is general agreement that renal tubular and glomerular lesions represent permanent loss of kidney functional reserve and that the lesions are irreversible. A worker who does not experience overt symptoms of illness may succumb to other illnesses more rapidly. An infection or other condition which would not normally lead to serious illness but which compromises kidney function could overwhelm the remaining kidney capacity (Ex. 8-644). A worker who has only slightly elevated levels of B(2)-M in urine may later develop proteinuria, even after cessation of exposures, or the worker may develop more severe forms of renal dysfunction. Such dysfunction is of great concern to OSHA. Renal compromise, described above, meets the definition of material impairment as intended in the OSH Act and as defined in this final standard (Sec. 6(b)(5)).

e. The Renal Effect of Cadmium Pigments and Other Less Soluble Forms of Cadmium - A Review of the Literature and Comments. OSHA received substantial comment on the renal toxicity of insoluble cadmium compounds, particularly cadmium pigments (Exs. 19-42-A; 19-14). "Solubility" is the process by which one substance is dissolved in another and is separated into its components by chemical action. Some cadmium compounds like cadmium oxide and cadmium sulfide, are relatively less soluble than others such as cadmium chloride and cadmium sulfate. It has been hypothesized that less soluble compounds may be less bioavailable than more soluble compounds and therefore less toxic. "Bioavailability" of cadmium compounds refers to the degree to which cadmium becomes available to the target tissue after exposure.

Several commentors and hearing participants were of the opinion that insoluble cadmium compounds are less toxic to the kidney (Exs. 19-42-A; 19-41; 14-14). For example, according to the Society of Plastics Industries (SPI), a trade organization of more than 2000 members representing all segments of the plastics industry, the health effects associated with cadmium have been observed when exposure has been to compounds that are not typically used in the coloration of plastics (Ex. 19-41). SPI also stated that a variety of animal studies indicate that cadmium sulfide and other similar cadmium-based pigments are significantly less bioavailable than other cadmium compounds. They cited the Agency for Toxic Substances and Disease Registry's (ATSDR) Toxicological Profile for Cadmium (1989) to support their position:

"The toxicity of cadmium depends on the chemical and physical forms of the element. In general, soluble compounds...are better absorbed and hence more toxic than highly insoluble forms....Studies described here are focused mainly on cadmium oxide or cadmium chloride, and the results cannot be applied equally to all other cadmium compounds" ( ATSDR, as quoted in SPI Ex. 19-41, p. 6).

In a written submission to the record, Richard Bidstrup, Counsel for SCM Chemicals, Inc., also supported such a position (19-42A). He argued that cadmium pigments are less soluble and less bioavailable, and thus less toxic to the kidney. He noted that toxicity to the kidney from cadmium pigments is one to three orders of magnitude less than for other forms of cadmium and that:

"....these results are consistent with the mechanistic and solubility data indicating that cadmium ions--the toxic agent of concern--are much less bioavailable from cadmium pigments than from other forms of cadmium." (19-42A)

Mr. Bidstrup, on behalf of SCM, and other commentors (Ex. 19-42-A), cited separate studies performed by Miksche, Feitz, and Greenberg as sufficient evidence of lower renal toxicity of pigments. These studies are reviewed in more detail below.

i. Miksche (Ex. 12-10-E). In this study published in the Proceedings of the Third International Cadmium Conference, Miami, 1981, by the Cadmium Association, Cadmium Council, and Ilzro, Inc., the effect of cadmium exposure on health was evaluated in a group of 36 workers involved in cadmium pigment production and 21 workers involved in acrylonitrile-butadiene styrene (ABS) plastics production. The location of the facilities was not indicated.

Concentrations of CdB and CdU, and B(2)-M levels in urine were available from periodic medical surveillance examinations conducted since 1980. The B(2)-M in urine was measured using the Phadebas radioimmunological method developed by Pharmacia. Measurement was either completed on the day of urine sampling or the samples were frozen immediately and kept at - 20 deg. C until the analysis was performed; sample pH levels were not presented.

The 36 workers in the cadmium pigment production plant had an average of 11.75 years of employment (range of 1 to 32 years). The average air concentrations reported were for 1977, 1979 and 1980 as 50 ug/m(3), 30 ug/m(3), and 30 ug/m(3), respectively. Among the 21 plastics production workers engaged in the application of cadmium pigments, the average length of employment was 11.3 years (range of 4 to 15 years). No exposure levels were reported for this latter group of workers. Miksche indicated that an age-matched control group of workers without occupational exposure to cadmium was used for comparisons with the group of pigment applicators. The results from this study are in Table V-17.

Table V-17 -  Cadmium Concentrations in Blood and Urine and Beta 2
              Microglobulin Concentrations in Urine of Workers in
              Pigment Production or Application
    Group   |Num-|     CdB(1)      |      CdU(2)      |   B(2)-M U (3)
            |ber |                 |                  |
Pigment     |    |                 |                  |
Production  | 36 | 10.3 + or - 2.4 | 8.78 + or - 2.18 | 77.35 + or - 22.36
            |    |          (2-36) |         (0.5-38) |         (17.6-304)
Pigment     |    |                 |                  |
Application | 21 |1.34 + or - 0.29 | 1.54 + or - 0.29 |
            |    |       (0.4-3.0) |        (0.4-3.1) |
Control for |    |                 |                  |
Pigment     |    |                 |                  |
Applicators |    |                 | 1.26 + or - 0.26 |
            |    |                 |        (0.2-3.0) |
  Footnote(1) Mean cadmium in blood level [(Mean + or - 2SEM, or standard
error of the mean) and range], ug/liter whole blood
  Footnote(2) Mean cadmium in urine level [(Mean + or - 2SEM, or standard
error of the mean) and range], ug/gram creatinine
  Footnote(3) Mean Beta 2 microglobulin in urine, ug/gram creatinine,
[(Mean + or - 2SEM, or standard error of the mean) and range]

Miksche reported no indications of elevated levels of B(2)-M among the exposed workers. Mean levels of CdB and CdU among the pigment production workers were above 10 ug/lwb and 5 ug/g Cr respectively which are the levels judged normal by other researchers (e.g. Ex. 29). Miksche reported finding no correlation between B(2)-M levels in urine and concentration of cadmium in air. Little exposure data were provided, however, to evaluate this conclusion. The only exposure data reported were the average air concentrations for three years while the study covered 32 year of time during which some of the production workers were potentially exposed. In addition, information on the intermittency of the exposures at these facilities or on the degree of respirator usage was not provided.

ii. Fietz et al, (Ex. 12-10-F). The study by Fietz, published in the Proceedings of the Fourth International Cadmium Conference, Munich, 1983, by the Cadmium Association, Cadmium Council and Ilzro, included 67 workers engaged in the production of cadmium pigments and 32 workers engaged in the further processing of those pigments. The location of the plants was not provided. Data for the study were provided by the company(ies) from occupational health screening reports and company exposure monitoring data. Among the information on biological parameters included in the medical evaluations were levels of CdB, CdU, urine creatinine and B(2)-M in urine. B(2)-M in urine was measured using a Phadebas test. Normal or "tolerable levels" for these biological markers were selected based upon principles laid down in BG Principle G 32 (referenced in Fietz). These levels were: CdU < 17 ug/g Cr, CdB < 15 ug/lwb (BAT); and B(2)-M < 300 ug/l urine. According to the authors, smoking data were not analyzed since there were no differences observed between smokers and non-smokers.

Sampling for airborne exposures included both personal and area samples. Air measurements were maximum short-term exposures for a duration of 30 minutes to two hours. Critical workplaces were measured more frequently than non-critical workplaces.

Workers in the study were placed into five different categories based on the type of jobs performed within a specified area. Groups I through III were involved in pigment manufacturing (e.g. raw material mixing, combustion, washing, drying and finishing). Groups IV and V were involved in pigment use and/or processing (e.g. paint formulation and pigment mixing). The authors indicated that exposure to cadmium pigments comprised about half of the working time for each worker. Two subgroups of workers were also identified: Group A included those who, on average, had worked more than ten years, and Group B included workers who, on average, had worked four years.

The average exposures per year were provided, for the time period from 1978 to 1982. Exposures to cadmium for groups I - III ranged from 14 to 201 ug/m(3), with the levels decreasing over time. Some workers in groups I and III had higher exposures, ranging from 175 - 1336 ug/m(3), but the authors stated that respirators were required for these workers.

For group IV, the exposure level was estimated to be 20 ug/m(3) in 1981 and 39 ug/m(3) in 1982. For Group V, only the exposure levels for 1982 were given, ranging from 0.5 to 10 ug/m(3). These values were much lower than those found in pigment production/manufacturing.

The B(2)-M excretion among workers in pigment production is indicated in Table V-18. The levels of CdB and CdU among workers in pigment production are indicated in Table V-19.

Table V-18 - B(2)-Microglobulin Excretion of
             Workers in Pigment Production
             Average Levels of B(2)-M in
             Urine (ug/Gram Creatinine)
Groups  |  1979 |  1980 |  1981 |  1982 |  Mean
        |       |       |       |       |
I-A     |  99.3 |  48.9 | 29.5  |  78.3 |  64.0
I-B     |  28.0 |  43.9 | 30.9  |  51.1 |  39.4
II-A    | 103.1 | 146.6 | 38.24 | 220.8 | 128.3
II-B    |  51.2 |  31.0 | 23.4  |  50.3 |  37.8
III-A   | 727.0 | 457.0 | 67.15 |  47.8 | 279.8
III-B   |  63.6 |  79.5 | 38.2  |  33.6 |  50.6

Table V-19 - Levels of Cadmium in Blood and Urine Among Workers in
             Pigment Production: Average Levels of Cadmium in Blood
             (ug/liter whole blood) and Cadmium in Urine (ug/gram
Group |         Cadmium in blood         |         Cadmium in Urine
      | 1978 | 1979 | 1980 | 1981 | 1982 | 1978 | 1979 | 1980 | 1981 | 1982
      |      |      |      |      |      |      |      |      |      |
IA ...|   20 |   15 |   11 |    7 |    7 |  6.0 |  9.2 | 10.9 |  9.1 |  8.4
IB ...|   12 |    8 |    9 |    7 |    9 |  4.6 |  5.4 |  7.1 |  6.4 |  6.3
IIA ..|   19 |   19 |   13 |   14 |   16 | 16.2 | 15.2 | 14.1 | 10.0 | 12.7
IIB ..|    8 |    7 |    6 |    5 |    6 |  4.3 |  3.2 |  3.1 |  3.1 |  3.1
IIIA .|   16 |   13 |   10 |    9 |    8 |  7.6 | 10.0 | 12.2 |  6.9 |  8.5
IIIB .|   11 |    9 |    6 |    6 |    8 |  4.4 |  4.1 |  5.3 |  4.8 |  4.5

CdU and B(2)-M levels were higher in the production workers, Groups I-III, than for pigment users with lower exposures (Groups IV-V). According to Fietz, workers in Groups IV and V had levels of CdB, CdU and B(2)-M in urine that corresponded to the levels of unexposed normal groups, but these levels were not given.

Each worker in Group A received greater exposure which was evident from their higher CdB, CdU and B(2)-M levels. The CdU levels of workers in Group B were considered to be tolerable levels, based on the criteria stated above. On the basis of these limits, seven production workers had to be removed to a cadmium free workplace as a precaution against further damage.

The authors concluded that the study showed that the use of technical measures such as exhaust ventilation, sealing of machines, enclosure of sources of dusts and consistent use of respirators can reduce the cadmium air levels and the harmful effects from cadmium. The results from the study, however, are limited for evaluating whether the low levels of cadmium and B(2)-M in the urine observed in the pigment users were the result of the lower absorbability of cadmium pigments or lower cadmium air concentrations.

iii. Greenberg et al, (Ex. 12-10-G). The Greenberg study was a follow-up study of 38 workers exposed to both lead and cadmium during the manufacturing of pigments and vitreous enamels. The follow-up was performed in two phases, one in Pittsburgh and one in Cleveland. The results of this study were published in the Archives of Environmental Health, in 1986.

In the Pittsburgh phase of the study, all workers at the plant were contacted through their union representatives, and over a seven month period, a total of 44 workers (40%) volunteered for admission to the Clinical Research Unit of the University of Pittsburgh School of Medicine. During a three day in-patient stay, detailed work and exposure histories were obtained, a physical examination was performed, and detailed laboratory studies were undertaken. Subjects were considered smokers if they had ever regularly smoked.

The second phase of the study was performed at the Cleveland work site during a single session two to nine months after the Pittsburgh hospitalization. Workers who participated in the Pittsburgh phase of the study were evaluated for tibia lead content by x-ray fluorescence and for liver and kidney cadmium content by neutron activation.

In order to evaluate renal function, urine was collected for determination of 24 hour creatinine, B(2)-M, and cadmium excretion. The normal value for B(2)-M was set at - or = 370 ug/24 hour sample. CdU concentrations greater than 5 ug/l and CdB levels greater than 7 ug/lwb were considered to be abnormal (Tsuchiya, referenced in Greenberg). Total urinary protein levels were considered to be abnormal if the levels exceeded 150 mg/24 hour. Maximal urine concentrating ability after an 18 hour water deprivation was determined to establish abnormal urinary osmolality (> 800 mOsm/kg).

As part of the renal function tests performed for these workers, minimum urine pH was determined. This test was performed using the oral administration of ammonium chloride (Wrong and Davies, referenced in the paper). Normal pH, as stated in the paper, was less than 5.4. Urine pH was not reported for the urine samples used in the determination of B(2)-M levels, but the normal level achieved during renal function testing, pH < 5.4 in 30 of 31 workers tested, is a level at which B(2)-M in urine will degrade. It is unclear whether the ammonium chloride test interfered with the accuracy of urinary B(2)-M determinations.

As part of the preliminary screening, environmental exposures to cadmium and lead were determined by measurement of airborne concentrations of the two metals in personal and area air samples. For the 38 workers who provided such data, the average length of employment reported was 20.7 years (11 to 37 years). Cadmium air levels were reported as "single measurements" with a range of 0 to 384 ug/m(3). The mean airborne level of cadmium in maintenance departments was reported as 5 ug/m(3). The mean airborne level of cadmium in "cadmium departments" was reported as 229 ug/m(3). The authors did not state whether these values were time weighted averages (TWA). The authors, however, did state that 31 percent of the values among all workers measured exceeded the NIOSH recommended level of 40 ug/m(3), which is a TWA.

Detailed work history information, reported by the workers during the medical examinations in the hospital, was scored for exposure to cadmium or lead independently according to a protocol established by Greenberg. If a worker was never warned that his/her CdU or CdB levels were elevated as determined by plant safety officials or if a worker never worked in an exposure area, the worker was classified as having no exposure. Workers were classified as having light exposures if they had no elevated levels of CdU or CdB or they worked briefly or transiently in exposure areas. Workers were classified as having moderate exposures if their CdU or CdB levels were normal or moderately elevated, if more than half of their work time was spent in exposure areas, or if they were a smelter operator. Workers were classified as having heavy exposures if their levels of CdU or CdB were known to have been high, they had been removed from their job site because of elevated levels, and/or they were exposed to intense or prolonged exposure. For some workers, this information was unavailable and these workers were classified as having unknown exposure levels. Workers who were unable to recall warnings about levels of cadmium in blood or urine were classified as having been moderately exposed.

Owing to the long duration of employment and the plant policy of switching workers from one area to another, most workers had mixed exposure to both lead and cadmium processing areas. As a rule, subjects were unable to recall details of warnings about toxic blood urine lead or cadmium levels. Thus the bulk of subjects were classified as having moderate exposure to both metals.

The authors reported that normal kidney cadmium burdens were considered to be levels up to 8.6 mg for non-smokers and 12 mg for smokers. Normal liver concentrations of cadmium were up to 7.0 ug/g and 9.5 ug/g for these two groups, respectively. The authors considered that the critical value for renal cadmium content was 30 mg. Workers were classified according to smoking status since cigarette smoking constitutes a significant exposure to cadmium. The mean value of kidney burden for nonsmokers (7.4 + or - 4.4 mg) was significantly lower (P < .02) than the value of 12.3 + or - 7.2 mg for smokers. Four of 18 (22%) nonsmokers and 8 of 20 (40%) smokers had kidney burden values above normal. No subject had a kidney burden above the critical concentration of 30 mg.

The mean liver cadmium value in nonsmokers was 4.5 + or - 2.6 ug/g and was significantly lower than the 7.90 + or - 4.9 ug/g value of their co-workers who smoked (p< .02). Four of 18 nonsmokers (22%) and 5 of 20 smokers (25%) had liver cadmium levels above normal. No worker had a value for liver cadmium above 40 ug/g, the value at which it has previously been shown that renal cadmium levels are associated with renal damage.

Urinary cadmium concentrations averaged 3.4 + or - 3.7 ug/l and ranged from undetectable to 16 g/l. Three of 38 workers (8%) had elevated cadmium in urine concentrations. CdU excretions ranged from undetectable to 24.5 ug/24 hr (mean = 6.43 + or - 7.7 ug/24 hr). CdB levels ranged from below the 1 ug/lwb detection limit to 9 ug/lwb. If the undetectable values were taken to be 1 ug/lwb, the mean value was 3 + or - 2 ug/lwb. Three of 37 subjects (8%) had values above the upper limit of normal for CdB (7 ug/lwb). Three workers had elevated levels of B(2)-M. Three workers had abnormal levels of total protein in urine.

The results of the qualitative dose-response among smokers are presented in Table V-20. It was not possible to perform a similar analysis in the nonsmokers because virtually all these workers gave a history of moderate occupational exposure.

Table V-20 - Cadmium in Urine Levels Among Smoking Workers with Industrial
             Exposure to Cadmium Pigments (Greenberg, 1986)
                  |                  Exposure History
   Cadmium Levels |   High N = 5      | Moderate N = 9   |   Low N = 4
Liver (ug/g) .... | 11.0 + or - 6.1   | 7.1 + or - 4.7   | 6.0 + or - 4.2
Kidney (mg) ..... | 17.2 + or - 9.4   |13.3 + or - 6.0   | 6.4 + or - 2.5
Urine (ug/24hr) . | 11.2 + or - 8.7(1)| 6.6 + or - 4.1(1)| 1.2 + or - 1.7
  Footnote(1) Significantly different when compared to low exposure group.

Few significant renal abnormalities were observed in this cohort of exposed workers. The prevalence of increased B(2)-M or protein excretion was low; only six workers had an increase of one or the other and only two of these workers had excessive cadmium body burdens. When measured body burdens of cadmium were accounted for, however, it was not surprising that the evidence of renal dysfunction attributable to cadmium was low. The highest kidney cadmium burden in any of the subjects in the study was approximately 20 mg, a value below the 31 - 42 mg value that has been shown in previous studies to be the critical level for renal toxicity. A dose-response model used by these authors previously, according to Greenberg et al., would have predicted that less than five percent of the population in the present study should have abnormal urinary B(2)-M or protein excretion. The authors stated that the relatively low renal burdens seem, therefore, to reflect low exposure conditions.

iv. De Silva, (Ex. 8-716). De Silva and Donnan conducted a study of past and present workers at a small cadmium-pigment manufacturing plant in Australia. Cadmium-in-air tests were conducted in July 1977. Personal samplers were worn for at least three shifts by each of the men working on the three main processes: furnace work, crushing, and general duties and cleaning. The results indicated that all TWA exposures were greater than 1 mg/m(3), with about 50 percent of the dust in the respirable range. There was little difference in the exposures experienced by men carrying out different duties. The furnace operated almost continuously whereas exposure during other processes, general duties, milling, crushing, and cleaning, were intermittent. The furnace operation was usually carried out by the same two men whose CdB concentrations confirmed their heavy exposure (54 ug Cd/lwb and 24 ug/lwb).

The three most heavily exposed workers, the two furnace operators and a raw blend operator, experienced respiratory damage and symptoms of kidney dysfunction after only seven years of employment. The estimated weekly average total cadmium dust exposure of these workers was approximately 1.5 mg/m(3) for the furnace operators and approximately 2 mg/m(3) for the raw blend operator. For workers engaged in other duties, dusty work occupied about half their time, and their estimated weekly average total cadmium dust exposure was approximately 0.7 mg/m(3).

The levels of cadmium in blood and urine (CdB and CdU) were measured in the workers. Excessive absorption of cadmium was found in five of nine current workers. All five had CdU levels at or above 10 ug/l, and two, with CdU levels in excess of 25 ug/l, had proteinuria. Three older men, two past employees and the factory manager who had not been engaged in production work for at least ten years, were found to have elevated levels of CdU although exposure had ceased many years earlier. In all three of these cases, proteinuria was present.

Three current workers and the three older men had been employed in production at the plant for at least 7 years. Each of these men experienced chronic cadmium poisoning and showed signs of renal tubular damage. Two of the current employees, one of whom had severe symptomatic emphysema, were excreting more than 40 ug/day of cadmium, had protein in their urine, and had impaired ability to acidify and concentrate urine. The other current employee was excreting 22 ug/day of cadmium in urine, and although he was not passing proteins, there was evidence of damage to his kidney tubules. His respiratory function tests showed moderately severe emphysema. The plant manager who had not been exposed for over ten years was passing cadmium at the rate of 14 ug/day. He had proteinuria and decreased ability to acidify and concentrate urine. Another of the older employees had been examined for cadmium poisoning, and the third older employee had a mild defect in concentration and acidification of urine and a moderate obstruction of his airways.

None of the men with exposures for less than seven years showed symptoms when the initial study was conducted, although one worker did have a mild airway obstruction. When renal function tests were conducted on these workers some two and one-half years later, however, symptoms began to appear. One worker had a decreased ability to concentrate urine, although no proteinuria was detected, while another worker showed normal renal function but intermittent proteinuria. Exposures roughly equivalent to 78 ug/m(3) over a 45 year working lifetime (at 700 ug/m(3) for three to five years, or 3,500 ug/m(3)-yrs for a maximum cumulative exposure) resulted in signs of cadmium poisoning not associated with clinical symptoms.

The results of this study indicate that kidney damage may occur even when concentrations of cadmium in urine are low (i.e. 15 ug/l) in workers with current exposure to cadmium. The authors suggest that if CdU levels exceed 15 ug/day, (the approximate equivalent of 15 ug/l), they should be regarded as unacceptably high, and should be kept below 15 ug/day to avoid the possibility of renal damage In the case of the three people with past exposure to cadmium, cadmium in urine remained below 15 ug/day, according to the authors, but this did not indicate freedom from kidney damage. The authors concluded the respirable fraction of insoluble cadmium dust should not be regarded as merely nuisance dust.

v. Wibowo et al, (Ex. 8-729). In 1982, Wibowo conducted a study of "second degree" users of cadmium compounds in the Netherlands. Thirty-four pairs of workers, one exposed and one control, were matched according to age, smoking habits, ethnic origin and factory. Study subjects were employed in one of five factories: a glass bottle production plant using cadmium pigments for label decoration; a plastic production plant using cadmium stabilizers; a cadmium plating department of an aero-engine factory; a wall-paper decoration plant using cadmium stabilizers and pigments; and a television tube production factory using cadmium sulfide for its flourescence property. Cadmium exposure in each of the factories was characterized as low, but no exposure data were presented.

Workers were evaluated for renal effects. Venous blood and spot urine samples were collected to measure CdB, hemoglobin, hematocrit, CdU, B(2)-M in urine, and creatinine in urine. Measurements of B(2)-M were performed according to the radioimmunoassay method with the Phadebas B(2)-microtest kit (Pharmacia Diagnostics) and were standardized to the creatinine content of the urine. To prevent degradation of B(2)-M, the pH was brought to above 5.5 by adding 0.5 M NaOH. For external control, the laboratory participated in an international comparison program for CdB.

Results of the analysis lead the authors to conclude that in "second degree" cadmium users, low cadmium exposures are reflected in CdU levels but not in CdB levels. This in turn would indicate an increased body burden due to long term, low level "second degree" occupational exposure to cadmium. An observation of statistically significantly elevated CdB and CdU levels among exposed workers in the aero-engine factory relative to controls supported this position. The authors concluded that even

" the very low levels of exposure the body burden increases with the duration of cadmium exposure and indirectly with smoking habits."(Ex. 8-729)

vi. Verschoor et al, (Ex. 19-42-8). In 1987 Verschoor conducted a study of renal function in 27 workers with second degree cadmium exposures. The workers were employed in one of two plants that were in the Netherlands; nineteen worked in a plant where cadmium is incorporated into the production of plastics (plant A), and eight worked in a plant where cadmium was used in the welding of radiators (plant B). Eight workers at a grain elevator located in the same geographic area as the plastics production plant who had no occupational exposure to cadmium served as controls. These eight were of similar age and smoking habits as the exposed workers.

One subject from plant B was known to have clinical renal dysfunction and was excluded from the analysis. (This dysfunction had existed for more than ten years and had not been caused by cadmium exposure.) The analyses, therefore, were carried out on 26 exposed workers and 8 referents.

Spot urine samples and venous blood samples were collected using standard procedures to protect against contamination. Cadmium-free disposable syringes and polythene tubes and boxes were used for the sample collection. CdB and CdU levels were measured for biological monitoring. B(2)-M and creatinine in serum, blood urea nitrogen (BUN), and IgG and albumin in urine were measured to evaluate glomerular function. RBP and B(2)-M in urine and the lysosomal enzyme NAG in urine were measured as indicators of change in tubular function, and total urinary protein was measured as an overall renal parameter. Only urine samples with pH greater than 5.5 were used for the determination of B(2)-M levels. Urinary density was chosen by Verschoor to adjust for the differences in the urinary creatinine concentrations between the exposed groups. All the urinary parameters were adjusted for a urinary density of 1,020 because this was the mean value of the urinary density of the group with level of CdU below 5 ug/g Cr.

Cadmium concentrations were not measured in either plant. Instead, CdU levels were used to estimate relative cadmium exposures. Workers with higher levels of CdU were considered to have been exposed to higher levels of airborne cadmium than workers with lower levels of CdU. The cadmium-exposed workers were thus divided into groups according to their CdU levels: The unexposed group whose CdU levels were less than 3 ug/g creatinine (30 nmol/l); the low occupational exposure group whose CdU levels were between 3 and 5 ug/g creatinine (30 to 50 nmol/l); and the higher occupational exposure group whose CdU levels were greater than 5 ug/g creatinine (50 nmol/l).

Most of the urinary parameters were found to be within the normal range. No significant differences were observed between the exposed workers and the controls for creatinine in serum, BUN, total urinary protein, or albumin in urine.

The levels of CdU and CdB among all exposed workers were elevated over the levels CdU and CdB in the referent population. For workers in plant A, The geometric mean of CdB levels was 1.26, 5.78, and 0.2 ug/lwb for workers in Plant A, Plant B, and the control group, respectively. The geometric mean of CdU levels was 2.3, 11.1, and 0.68 ug/gr Cr for workers in Plant A, Plant B, and the control group, respectively.

Only workers in the higher occupational exposure group (i.e. the group whose CdU levels exceeded 5 ug/g Cr) had excess levels of those biological markers which indicate change in tubular function. For this group of workers, all of whom were employed in plant B, the levels of urinary B(2)-M, RBP, and NAG were statistically significantly elevated over the levels of these parameters in the control group (p < .05).

vii. Kawada et al, (Exs. 8-732). In 1986 Kawada studied 29 workers who were exposed to cadmium pigments either in pigment manufacturing (CdS pigments or cadmium selenide pigments) or in synthetic resin preparation. The purpose of the study was to evaluate the relationship between B(2)-M in urine and NAG activity, in workers with low levels of CdU. Thirty-five non-exposed workers were also evaluated; these were new workers on a rotation system and non-exposed workers.

Area samples found that the highest exposures ranged from 3 to 350 ug/m(3). The mean respirable dust concentrations for the most highly exposed workers ranged from 0.18 to 3.0 ug/m(3). Urine samples were collected stored at 4 - 6 C; pH levels were not specified. Levels of B(2)-M in urine were analyzed using the enzyme immunoassay kit (Fujirebio Inc., Tokyo).

Cadmium, NAG activity, B(2)-M, and creatinine in levels in urine were measured twice in the exposed workers, once in April and again in September. The geometric mean of CdU was found to be 0.7 ug/g Cr in April (range of 0.2 to 9.5 ug/g Cr) and 1.2 ug/g Cr in September (range of 0.5 to 7.0 ug/g Cr). The correlation coefficient between CdU and NAG was 0.261 in April (n=61) and 0.389 in September (n=50). The correlation coefficient between CdU and B(2)-M was 0.241 in April (n=63) and 0.115 in September (n=50). Kawada concluded that NAG is a more sensitive indicator of cadmium absorption than B(2)-M even at CdU levels of less than 10 ug/g Cr.

viii. Kazantzis et al, (Ex. 8-102). In 1962, Kazantzis conducted a study of 12 of the 13 men who produced cadmium pigments for use in paint, plastics and glass in a small British factory. While both cadmium oxide and the cadmium pigments (cadmium sulphide, cadmium zinc sulphide and cadmium seleno-sulphide) were produced in this factory, it was during the production of pigments workers experienced the greatest cadmium exposure due to the repeated handling of the material. The authors observed that

"...for similar quantities handled, the ratio of exposure for pigments is something like two and a half times as great as for oxides."(Ex. 8-102)

No exposure data were available. Instead, workers were grouped based on length of employment at the plant, combined with the type of work the men performed. These data were gathered through in-depth interviews and surveys with the employees. The authors found that three men had been in the factory for 2 years or less, four had 12 to 14 years of exposure, and five men had 25 to 31 years of exposure.

Venous blood samples were taken and analyzed for urea, carbon dioxide capacity, chloride, sodium, potassium, inorganic phosphate, alkaline phosphatase, cholesterol, calcium and uric acid. Random 24-hour urine samples were also collected to measure urinary protein, cadmium, calcium, amino acids, inorganic phosphate and sugar. Freshly passed specimens showed that pH of the urine of all 12 workers was somewhat acidic, falling between 4.85 and 6.55. Chest X-rays taken and comprehensive respiratory function tests were performed.

All five of the workers who had been exposed to cadmium compounds for more than 25 years showed consistent clinical proteinuria. In addition, three workers with fewer years of exposure showed renal tubular disorders and greater than normal protein in the urine. CdU exceeded 30 ug/day in 10 of the 12 workers studied. Several of the workers with clinical proteinuria showed definite respiratory impairment.

In addition to the 12 men studied by Kazantzis, there was one more with over 25 years of exposure who died before the study began, presumably from chronic cadmium poisoning. A great deal of health information for this worker was abstracted posthumously from his medical records and autopsy report. Specimens of lung, liver and kidney were analyzed for cadmium content. The cadmium concentration in the kidney was 55 ug/g wet weight and in the liver was 88 ug/g wet weight. In the lung, the cadmium concentration, 500 ug/g wet weight, was extremely high.

This study is the oldest reviewed by OSHA, and the biological parameters monitored for kidney dysfunction are less specific than those currently accepted as the most reliable indices for cadmium-related health effects. Nonetheless, the evidence suggests appreciable abnormal clearance of amino acids, calcium, glucose, phosphate, urate, water, and faulty acidification of the urine. The study authors concluded that:

"The findings establish that chronic cadmium poisoning can arise in the pigment industry, and suggest that cadmium proteinuria is clinically significant and should be regarded as an early manifestation of renal tubular damage."(Ex. 8-102)

ix. Summary. Epidemiological studies of workers in pigment-production and pigment-using facilities show that cadmium is absorbed into the body and that cadmium pigments should not be regarded separately from other cadmium compounds. Results from the studies submitted by industry do not support the position that cadmium pigments pose less hazard to workers than other cadmium compounds.

SCM cited the study by Miksche (Ex. 12-10-E) as evidence that cadmium pigments are not as toxic to the kidney as other forms of cadmium (Ex. 19-42A). Data from the study, however, indicate that the cadmium production workers were absorbing cadmium from pigments, as evidenced by the fact that the mean CdB and CdU levels of these workers were 10.3 ug/lwb and 8.78 ug/g Cr, respectively. The B(2)-M levels in these pigment production workers were normal. Among pigment applicators, no abnormal levels of CdU, CdB, or B(2)-M were found, but no exposure data were provided. This makes it impossible to evaluate cadmium exposures among the production workers.

The Cadmium Council (Ex. 19-43) questioned OSHA's argument that the low body burdens reported by Miksche might be due to "low exposure" when, according to the Council, workers in the cohort were exposed for up to 32 years and when, according to the Council, airborne cadmium levels were found to be 50 ug/m(3) in 1977 and higher than that before 1977. In response, OSHA notes that 32 years represents the maximum length of exposure for a worker in the Miksche cohort and that the mean lengths of exposure for the two separate plants were 11.75 years and 11.3 years. The Agency also notes that no exposure data are available at all for the application plant, and the exposure level of 50 ug/m(3) represents the average exposure for only one year at the production plant. Exposure levels at the production plant for the other two other years for which exposure data are available were lower than 50 ug/m(3).

The study by Fietz (Ex. 12-10-f) was also cited by industry as evidence that cadmium pigments are far less toxic to the kidney than other forms of the metal. Again, the Cadmium Council rejected the argument made by OSHA that the low levels of urinary analytes observed in this study were due to low levels of exposure (Ex. 19-43). The Council stated that in 1978 these workers were being exposed to 147 to 201 ug/m(3) of cadmium, and they still had "generally low" cadmium body levels. OSHA notes, however, that cadmium pigment production workers did show evidence of over-exposure to cadmium even when of CdU as high as 17 ug/g Cr and CdB as high as 15 ug/lwb (BAT) were considered "safe". Furthermore, the exposure data are insufficient to evaluate the exposures for Groups IV and V, and it is clear from the data that exposures in the pigment-user facility were much lower than in the production facility.

The statements by the Cadmium Council regarding high exposures (147 to 201 ug/m(3)) in the Fietz cohort being associated with low body burdens in 1978 are misleading. The authors reported that in 1978, average cadmium exposures (ug/m(3)) in 1978 for Groups I, II, and II were 147, 194, and 201, respectively but that production of cadmium pigments comprised about half of the working time. This would result in exposures of 73.5, 97, and 100 ug/m(3). In Group IA, IIA, and IIIA, that is among the workers who had worked for an average of ten or more years, the levels of CdU in 1978 were 6 ug/g Cr, 16 ug/g Cr, and 8 ug/g Cr, respectively. In Group IA, IIA, and IIIA, the levels of CdB in 1978 were 20 ug/lwb, 19 ug/lwb, and 16 ug/lwb, respectively. Levels of B(2)-M in urine for these groups were not available for 1978, however, in 1979, for workers in Groups IA, IIA, and IIIA, the levels of B(2)-M were 99 ug/g Cr, 103 ug/g Cr, and 727 ug/g Cr, respectively. Thus, if workers experience cadmium or cadmium-pigment exposures similar to those in Group III and if workers have more than ten years of employment in a plant with exposures similar to those in this study workplace, some (7/67 = 10.4% or more) will have renal dysfunction as indicated by B(2)-M levels greater than 300 ug/g Cr.

According to comments submitted to OSHA during the rulemaking, levels of CdU above 15 ug/g Cr and levels of CdB above 10 ug/lwb are levels that indicate cadmium has been absorbed and systemically distributed throughout the body via the bloodstream; overexposure to cadmium has occurred; and excessive cadmium body burden is present. Based on this definition it is evident that cadmium exposures in pigment production plants can result in uptake of cadmium by the body. If exposure levels are kept close to those observed in this study for 1982 when exposures ranged from 0.5 to 10 ug/m(3), few workers will be at risk of kidney dysfunction, regardless of the type of cadmium compound used.

The Cadmium Council and SCM cited Greenberg's study (Ex. 12-10G) as evidence that among cadmium pigment workers,

".....the prevalence of increased B(2)-microglobulin or protein excretion was low.(Exs. 19-43 and 12-10-G)"

SCM commented that this finding was significant given that the average length of employment for these workers was 20.7 years, and 31% of recent exposure measurements exceeded 40 ug/m(3) (11 of 35 recent samples). The study authors themselves, however, concluded that the relatively low renal burdens reflected low exposures. Furthermore, according to the definition of "normal" test results established by the authors, three of 38 (8%) workers had elevated CdU concentrations; three of 37 (8%) subjects had values above the upper limit of normal for CdB; three workers had elevated levels of B(2)-M; and, three workers had abnormal levels of total protein in urine. Finally, the finding of dose-response relationships between liver cadmium burdens and exposure, kidney cadmium burdens and exposure, and levels of cadmium in urine and exposure indicate that cadmium absorption occurs in pigment production facilities.

SCM referenced the study by Wibowo (Ex. 8-729) and the study by Verschoor (Ex. 19-42-8) to assert that:

"... There is no human evidence indicating that cadmium pigments have caused significant renal effects" (Ex. 19-42-8).

Careful review of these studies, however, indicate that the downstream, or secondary users of cadmium pigments in these studies either had low exposures or the compounds used were insoluble. Nonetheless, low level exposures to insoluble cadmium pigments are absorbed into the bloodstream and result in levels of CdB and CdU that are elevated over background. The Verschoor study in particular suggests that even when second degree usage of cadmium pigments, such as in the plastics plant, are low, small amounts of cadmium are absorbed and systemically distributed to the kidneys. The low levels of CdB and CdU among the plastics workers in Plant A can not be attributed solely to lower absorption of insoluble compounds since no exposure data are available; the low levels of CdB and CdU may be due to lower exposure levels.

De Silva (Ex. 8-716) concluded from his study that the insoluble respirable cadmium compounds are less able to enter the blood stream than the soluble compounds. According to De Silva, this is supported by the case reported by Kazantzis et al, 1963, (referenced in De Silva) which resulted in necropsy and showed an unexpectedly high concentration of cadmium in the lung, compared with other published figures from cases of chronic cadmium poisoning due to soluble cadmium. While less cadmium enters the blood, however, De Silva concluded that more is retained in the lung. Respirable insoluble compounds probably contribute significantly to the lung damage, and, therefore, according to De Silva, there is no reason for differentiating between soluble and insoluble respirable cadmium, due to their effects on the lung.

The lower biological indices of absorption obtained after exposure to insoluble compounds of cadmium may be due to a reduced ability of the respirable fraction to enter the blood stream rather than to the reduced solubility of the non-respirable fraction in the gut. It could perhaps be concluded that non-respirable insoluble compounds may be slightly less hazardous than soluble compounds, but to regard them as merely nuisance dust is not only risky, according to De Silva, but also liable to misinterpretation and to encourage carelessness in their use.

Dr. Friberg was questioned about the relative toxicity of cadmium compounds, in particular cadmium pigments. In his testimony, Dr. Friberg discussed the findings of the study by De Silva as follows:

".... It's a small study but still I think it is, to some extent, impressive. (Workers) were exposed to insoluble cadmium at a pigment manufacturing plant......all....had signs of the tubular the abstract....(the authors)"

"...suggested that urinary cadmium concentration should be kept below....15 micrograms per day, but this was in 1981, to avoid the possibility of renal damage and that the insoluble respirable fraction of cadmium dust should not be regarded as merely a nuisance dust. There was a lot of information of blood levels .....after a couple of months exposure values of cadmium in blood per liter....(were)....something like.......18, 6, 9, 14, 24, 7, 17, 22, 10, 21, and 10.....very high levels. But as they mentioned somewhere, there could, of course, be some form of an exposure also to a soluble dust......(Tr. 6/6/90, pp. 90-91)"

Thus, relatively insoluble cadmium pigments are taken up into the blood and may cause damage; there is medical evidence from studies of workers exposed to cadmium sulfide and cadmium pigments, although limited, that all cadmium compounds have been associated with renal effects at comparable exposure levels, regardless of the chemical compound.

f. Evidence in Animals. Experimental animal studies, some of which are reviewed below, support the finding that cadmium induces proteinuria in humans. Studies also support the concept of a threshold or critical concentration of cadmium in a target organ and the finding that increased concentrations of B(2)-M in the urine constitute biological markers of cadmium-induced tubular proteinuria (Exs. 30; 8-86).

Friberg induced proteinuria in rabbits by exposing them to 8 mg/m(3) of cadmium oxide dust by inhalation for 5 hours/day for 8 months (Ex. 4-29). In the same study, rabbits injected with 0.65 mg/kg cadmium sulfate developed proteinuria after two months of exposure. A number of other experimental studies in which animals were exposed either orally or through injection have also resulted in cadmium-induced proteinuria (Exs. 8-086b; 8-402).

Some studies have induced glomerular proteinuria in experimental animals while others have induced mixed-type proteinuria. For example, Bernard (Ex. 4-20) injected rats with 1 mg/m(3) of cadmium chloride for 5 days/week for 2 months. The induced proteinuria was characterized by increased excretion not only of low molecular weight proteins but also of high molecular weight proteins indicative of glomerular dysfunction. In a similar study (Ex. 4-49), rats injected with cadmium showed mixed-type proteinuria, and after prolonged oral exposure, rats developed glomerular proteinuria.

i. The Renal Effects of Cadmium Pigments and Other Less Soluble Forms of Cadmium in Animals. OSHA received comment from SCM, SPI, and the Dry Color Manufacturers Association (DCMA), a trade association representing small, medium, and large pigment manufacturers in the U.S. and Canada, that studies by Hazleton Laboratories, Glaser, and Rusch demonstrate that absorption of cadmium and the potential for renal effects from pigment is much lower than that from other cadmium compounds (Exs. 19-42-A; 19-41; 120). For example, Richard Bidstrup, Counsel for SCM Chemicals, Inc., stated that these studies "demonstrate that absorption of cadmium and potential for renal effects from pigment is from 10 to 1,000 times less than that from other cadmium compounds (Ex. 19-42-A)." In addition, he referred to other studies with longer exposure times that have reached the same conclusion (i.e., Heinrich et al., 1986; Princi and Geever, 1950; and Oberdorster, undated; all in Ex. 19-42A).

According to the DCMA, in-vitro data also support its position that the cadmium in pigments is less bio-available than the cadmium in other compounds because cadmium pigments are relatively insoluble in the dilute acids (i.e., a pH of about 4) often found in biological systems (Ex. 120). DCMA states that, in contrast to the sulfide, cadmium oxide is highly soluble in dilute acids, which explains, in DCMA's view, the equivalent lung toxicity of cadmium oxide and other water-soluble cadmium compounds. In contrast, the solubility of cadmium pigments at a pH of 4 was much lower. Therefore, the oxide is "thousands of times" more soluble in the dilute acid environment of the lung than cadmium sulfide, and the oxide is therefore much more bioavailable than the sulfide (Ex. 120). DCMA reported that work by Stopford shows a good correlation between weak-acid extraction and serum extraction. In addition, a positive relationship has been demonstrated between the amount of acid-extractable cadmium in pigments and the "resulting body burdens after ingestion" (Ex. 120, p. 10).

OSHA has reviewed the main animal studies submitted by industry representatives in support of their assumption that cadmium pigments are less toxic to the kidney. The Agency review is presented here. Hazelton Laboratories conducted a short term rat feeding study to determine whether or not there was a positive correlation between cadmium solubility and cadmium absorption through the gastrointestinal tract (Ex. 12-10-B). In this study, extraction tests were conducted with distilled water and with acid to determine the solubility of 12 different cadmium pigments. These same pigments were then fed to rats for one week at levels of 10,000 ppm and 50,000 ppm in the diet to evaluate the level of absorption of cadmium from the pigment. For purposes of comparison, rats were also fed a highly soluble cadmium compound, CdCl(2), at a concentration of 10 and 50 ppm in the diet. The proportion of cadmium absorbed was determined by measuring the amount of cadmium found in the urine, kidneys and liver and dividing by the amount of cadmium found in the feces and GI tract contents. The degree of solubility of the pigments was much lower than the degree of solubility of CdCl(2). CdCl(2) was 61% soluble whereas the pigments were from 0.06% to 1.38% soluble. Correspondingly, the proportion of cadmium absorbed from the pigments was also much lower than for CdCl(2); 0.65% of the CdCl(2) was absorbed compared to .0004% to .0060% of the cadmium pigments.

From these data the authors concluded that there was a positive correlation between solubility and absorption; the greater the solubility the greater the amount absorbed by the body. OSHA notes, however, that this feeding study lasted only one week. While the percent of cadmium absorbed from the pigments after one week's exposure is relatively low compared to CdCl(2), the total percentage absorbed after chronic exposure to cadmium pigments (e.g. 18 months) is not known and may be more substantial.

In the study by Princi and Geever (Ex. 8-459), 30 dogs were divided into three groups: ten dogs served as controls; ten dogs were exposed to cadmium oxide; and ten dogs were exposed to cadmium sulfide. Exposure lasted six hours per day for five days per week. Concentrations of cadmium in air ranged from 3 to 7 mg/m(3). Ninety-eight percent of the particles were less than 3 microns in diameter. The average length of exposure for cadmium sulfide was 895 hours, or about 30 weeks (895 hours divided by 30 hours per week is approximately 30 weeks); the longest exposure was 1,417 hours (approximately 47 weeks). The average length of exposure for cadmium oxide was 1,102 hours (approximately 37 weeks); the longest cadmium oxide exposure was 1,319 (approximately 44 weeks). Four dogs exposed to cadmium oxide eventually died of bronchopneumonia.

The authors noted that because of these deaths, exposure times varied. Several dogs had to be killed because of severe injuries received from fighting among themselves. Despite these limitations, the authors reported that a sufficient number of dogs survived the entire experiment to produce significant results. No information was provided on the survival times for any of the animals in this study. The authors concluded that most of the cadmium dust inhaled was stored in the lungs, liver and kidneys. Lesser amounts were stored in bones and teeth. The authors further concluded that cadmium sulfide (CdS) and cadmium oxide (CdO) differ greatly in solubility and absorbability, and must therefore differ greatly in the amounts required to produce intoxication.

For the ten dogs exposed to CdO, the average level of cadmium in blood was 67 ug/lwb, and the average level of cadmium in urine was 130 ug/l. For the ten dogs exposed to CdS, the average level of cadmium in blood was 50 ug/lwb, and the average level of cadmium in urine was 50 ug/l. Clearly, a portion of the CdS dose was systemically absorbed into the body, however, a portion of the CdS may have decomposed into more soluble ionic forms, e.g., CdS04 (Exs. L-140-27-B, L-140-44), increasing the cadmium's solubility. OSHA notes, however, that at exposure levels as high as 3 to 7 mg/m(3), the percent dissolution of CdS to more soluble forms is probably low (see, for example, L-140-27-B). The average level of cadmium in the lungs for the ten dogs exposed to CdO was 2.64 mg/100 gm (range from 1.16 to 4.7 mg/100 gm). For the ten dogs exposed to CdS this level was 3.63 mg/100 gm (range from 1.23 to 6.02 mg/100 gm). The authors concluded that more CdO was absorbed from the lungs, but OSHA notes that more CdS remained in the lung.

The study by Rusch (Ex. 12-10-D) was an acute inhalation study involving the exposure of male and female Sprague-Dawley rats to dusts of cadmium carbonate (CdCO(3)), cadmium yellow pigment, cadmium red pigment and to cadmium fume at concentrations of 100 mg/m(3) for two hours. The animals were then followed for 30 days in order to determine whether there were differences in uptake and distribution of compounds with different solubilities. The "cadmium red" was a finely divided red powder in hexagonal crystal form containing 69.9% cadmium, 16.4% selenium, and 13,2% sulfur (sulfide). Particle size analysis by sedimentation indicated that 99% of the particles were less than 5 micrometers in diameter. Cadmium yellow was a finely divided powder in a hexagonal crystal form produced by high temperature calcination. It contained 77.4% cadmium, 21.7% sulfur (sulfide), 0.28% zinc, and 0.27% selenium. Ninety-six percent of the particles were less than 5 micrometers in diameter. The cadmium carbonate was a reagent grade white amorphous powder which was finely divided by milling. The cadmium fume material was derived from a 10% aqueous solution of cadmium acetate dihydrate. The aerosol was produced by passing the solution through several heating processes and absorption containers prior to administration to the test animals.

No mortality was observed among rats exposed to either cadmium pigment after 30 days follow-up, but 3 out of 52 rats (6%) died from exposure to CdCO(3) and 25 out of 52 rats (48%) died from exposure to cadmium fume. In the pigment-exposed groups, greater amounts of cadmium were eliminated by the feces at faster rates than for the CdCO(3)-exposed rats. The rats exposed to CdCO(3) also showed higher kidney cadmium levels. The authors stated that CdCO(3) followed predicted patterns of uptake, distribution and retention, whereas the pigments showed only minimal uptake and tissue deposition. Therefore, it appeared that inhalation exposures to soluble compounds resulted in more rapid uptake and higher body burdens than did exposure to less soluble cadmium compounds. OSHA notes that in this study, the exposure level was extremely high, lasted only 2 hours, and follow-up was for only 30 days. This makes it difficult to extrapolate the results to occupational settings where exposures may occur over long periods of time and at low doses or where exposures may occur repeatedly at high levels.

In the study by Glaser et al., (Ex. 12-10-C), male Wistar rats were exposed continuously for 30 days to aerosols of cadmium chloride and cadmium oxide at 0.1 mg/m(3) and aerosols of cadmium sulfide (CdS) at 1 mg/m(3). CdS was administered at a higher dose because of its lower solubility.

No clinical signs of intoxication were observed among any of the exposed groups. Mean CdU in the CdS group showed a slight but statistically significant increase (p < 0.05) over the controls at the end of the exposure period. Mean CdU was also slightly statistically significantly elevated for the CdO group (p < 0.05) at the end of the observation period.

Glaser found that cadmium was retained in the lung, liver and kidneys for all three compounds tested. Lesser amounts of CdCl(2) were retained in the lungs of exposed rats compared to the amount of CdO and CdS retained. After one month's exposure approximately 25 ug of cadmium were retained in the whole lung of CdCl(2) exposed rats whereas approximately 50 ug and 140 ug of cadmium were retained in the lung for CdO and CdS exposed rats, respectively. The authors note that a 10 times greater exposure in the form of CdS did not result in a 10 times greater amount of cadmium in the whole lung. Therefore they suggested that there must be a difference in toxicokinetics (i.e. deposition, dissolution, clearance or toxicity) for CdS. OSHA notes, however, that photodecomposition of CdS may have occurred thereby altering the compounds administered to the lung to include more soluble forms of cadmium, (e.g., CdSO(4)), and affecting the amount that would be retained in the lung.

Glaser also observed that for the CdCl(2) and CdO exposed rats, more of the cadmium was distributed to the cytosol fractions of the lung than for the CdS exposed rats, indicating that more of the CdS was retained in the extracellular fractions and was not absorbed into the cell. For a site-of-contact carcinogen, which some evidence suggests cadmium may be, it is entirely possible that the more insoluble the compound, the greater the carcinogenic potential. In fact, there was evidence of a cytotoxic effect to the alveolar macrophages from exposure to CdS equal to that observed from exposure to CdO.

Each of the cytotoxic effects observed in the rats exposed to CdS and CdO were greater than the effects observed in the rats exposed to CdCl(2). In addition, the lung metallothionein-cadmium content for rats exposed to CdS and CdO were similar to one another but greater than the metallothionein-cadmium content in CdCl(2) exposed rats. Metallothionein is produced in response to cadmium ions and, according to the authors, is an indication of cadmium's bioavailability. In the liver and kidney, cadmium burdens were significantly higher for the CdO exposed rats and for the CdS exposed rats than for the CdCl(2) exposed rats. After one months's exposure, approximately 15 ug of cadmium had accumulated in the liver and kidney of CdCl(2) exposed rats compared to 70 ug and 60 ug of cadmium which had accumulated in the livers of CdO and CdS exposed rats, respectively. The authors stated that it was unexpected that cadmium accumulation in the liver and kidney would be lower for CdCl(2) exposed rats than for CdO and CdS exposed rats because of CdCl(2)'s higher solubility; they had thought that cadmium accumulation was correlated to the solubility of the compound. Some of the elevated liver burdens noted by Glaser for CdS exposed rats may be due to photodecomposition of the CdS (Exs. L-140-44; L-140-27-B), but this would not explain the high liver burdens for the CdO exposed rats.

The results of this study suggest that absorption and bioavailability may not be simply equated to the compound's solubility. For example, Glaser found that the body burdens of cadmium in the kidney and liver for CdO and CdS exposed rats were similar despite the fact that ten times more CdS was administered. This would imply that the lower solubility of CdS may be responsible for the lower accumulation of cadmium. It has been suggested that such a result could be due to photodecomposition, but if photodecomposition did occur and a substantial amount of CdS degenerated into more soluble forms, it would be difficult to explain why 10 times more CdS resulted in a liver burden equivalent to that of the CdO exposed rats (Exs. L-140-44; Ex. L-140-27-B). The body burdens of cadmium in the kidney and liver are higher for CdO and CdS exposed rats than CdCl(2) exposed rats despite the fact the CdCl(2) is more soluble than CdO, CdS. CdCl(2) may be more soluble than cadmium compounds into which CdS may have photodecomposed. Thus factors other than solubility influence the systemic absorption and bioavailability of cadmium pigments. These factors could be further influenced by long term exposure (i.e. greater than one month).

According to DCMA, a later study by Glaser et al., 1990, (Ex. 8-694-B) shows that ionic cadmium released by photodecomposition accounts for the apparent increase in availability seen in this study. Gunter Oberdorster, Ph.D., Professor of Toxicology at the Environmental Health Sciences Center at the University of Rochester and formerly of the Fraunhofer Institute for Toxicology and Aerosol Research in West Germany, analyzed the data from this study, however, and concluded that they showed only a 2- to 3-fold reduction in the availability of cadmium from pigments (Ex. 120, p. 20).

OSHA concludes that the Hazleton Laboratory, Glaser, Princi, and Rusch studies do not provide adequate evidence to show that cadmium pigments are not as toxic as other forms of cadmium. (See for example Ex. 12-10.) These studies used short exposure periods that might not be relevant to long-term, low-level occupational exposures or obtained conflicting results that do not indicate a simple relationship between solubility and bioavailability. OSHA has also reviewed the studies by Heinrich and Oberdorster, and has addressed the main comments from these two experts in its discussion of the carcinogenicity of cadmium pigments.

DCMA submitted comments on the mechanisms by which cadmium pigments could be less toxic to the kidney and less bioavailable than other cadmium compounds (Ex. 120). DCMA stated that the chemistry of pigment compounds is substantially different from that of other compounds. Because the cadmium ion is "soft", based on its low charge-to-mass ratio, and the sulfide ion is also soft, sulfide ions stabilize the cadmium ions in pigment compounds. DCMA argued that any pigment-derived cadmium ions that were dissolved in biological fluids would already be complexed by water molecules or chlorides, both of which are hard ligands. According to DCMA, this concept of hard/soft ions would also suggest that the sulfide is less soluble than other chemical forms of the metal (Ex. 120, p. 11). However, soft-acid/soft-base theory refers to stronger bonding associated when large, easily polarized anions coordinate to large, easily polarized cations. This theory is used to explain the enhanced stability compared to ions that are mismatched is size, and does not pertain to chemistry of sulfide (Ex. 152). Since the human body can muster formidable chemistry in defense against foreign material, CdS may be solubilized in the body by a number of mechanisms (Ex. 152). For example, as Dr. Friberg stated, and as mentioned previously:

" seems to be quite clear that you cannot talk about insoluble dust...once this dust comes down to the lungs and is taken up by the macrophages then the solubility could be quite different from the solubility in vitro...."(Tr. 6/6/90, pp. 102-103)

Other commentors agreed with Dr. Friberg. Once an insoluble cadmium pigment enters the lungs, some is absorbed into the body and some is retained in the lungs (Ex. L-140-50). The portion of cadmium that is systemically absorbed is retained for up to 30 years depending upon the part of the body in which the cadmium is stored. Even low atmospheric levels of cadmium accumulate in the body (Ex. L-140-50). Ellen Silbergeld, Ph.D., Adjunct Professor of Toxicology and Pharmacology and Experimental Therapeutics at the University of Maryland Medical School and Adjunct Professor of Health Policy at the Johns Hopkins School of Hygiene and Public Health, testified at the hearing regarding the bioavailability of different cadmium compounds. Dr. Silbergeld testified that solubility in aqueous media is only one aspect "of the physical, chemical, and potential biological behavior of a compound for purposes of estimating toxicity" (Tr. 6/7/90; pp. 228-235). Other factors that must be considered are particle size, portal of entry, chronicity of exposure, and other types of cellular response, which may be "much more important than this particular measure of aqueous solubility" (Tr. 6/7/90, pp. 228-235). Dr. Silbergeld testified further that many studies that have attempted to "exonerate" other metal-containing pigments such as lead pigments have shown that when exposure is chronic and the outcome is chronic (as opposed to acute), the molecular compound of the metal is "relatively unimportant." Dr. Silbergeld stated that "those differences in solubility would be unlikely to confer a difference in terms of tissue dose beyond the range of one or two" (Tr. 6/7/90, p. 229). Dr. Silbergeld continued that when the compound is encountered chronically from the external environment and remains within the body compartments for a long period of time, one factor such as solubility does not make much difference in overall toxicity" (Tr. 6/7/90, p. 229).

Dr. Silbergeld described the process of phagocytosis or endocytosis as a process where intracellular organelles come to the defense of the invaded cell which leads in turn to a "very great increase in the local concentrations of the compound at the site of inclusion, and which results in the cell's secretion of acids that change the chemical by encouraging dissociation of the metal from the salt or incorporation of the metal from the salt or incorporation of the metal onto a carrier protein, etc. Also, the cell may react in a way that produces oxidative stress" (Tr. 6/7/90, pp. 228-235).

Another argument made by DCMA was that metallothionein, a protein that contains 33 percent sulfur organized in seven clusters, "coordinates" metals such as cadmium in a very tight structure. Thus, metallothionein, once bound, may actually protect against cadmium's carcinogenic effects [DCMA cites Testimony at 3-56, 3-58] (Ex. 120, p. 11). DCMA then argued that, because the cadmium in pigments is also "tightly held between layers of sulfur atoms in a hexagonal...packing array,...the sulfur in pigments can be anticipated to protect organisms against the toxic effects of cadmium ions in the pigments" (Ex. 120, p. 12).

Based on its review of the in-vivo and in-vitro studies, DCMA concluded that cadmium pigments are not metabolized by similar mechanisms or at similar rates as other cadmium compounds. Thus, according to DCMA, "OSHA cannot assume that cadmium pigments have toxicity equal to [that of] other cadmium compounds" (Ex. 120, p. 12). This hypothesis was countered by analogies drawn between cadmium with other compounds such as lead, nickel, and asbestos (e.g. M. Costa, Tr. 6/7/90, pp. 56-58). Dr. Silbergeld suggests that endocytotic incorporation of insoluble materials may lead to higher concentrations within the cell which could, in turn, heighten the potential toxicity of "soluble" [sic, insoluble] compounds (Ex. 120, p. 13). DCMA dismissed this idea because studies have shown that most cadmium sulfide was in the extracellular compartments of the lung (Glaser 1986, cited in Ex. 120) and that most of the cadmium from pigments is retained in the lavagable portion of the lung and is not bound to the cells (Oberdorster, cited in Ex. 120). Further, DCMA noted that Oberdorster's work shows that cadmium sulfide is cleared from the lung by the same mechanism used to clear nuisance dust. DCMA believes that these studies, which show that cadmium pigment is retained and cleared outside the lung cells obviate OSHA's position that "the endocytotic assimilation of cadmium pigments is a plausible mechanism for assuming an equal potency for adverse effects from these compounds" (Ex. 120, p. 14).

After reviewing these comments, the Agency believes there are insufficient data to conclude that one hypothesis is more acceptable than another. Furthermore, experimental data appear to show cytotoxic differences even between amorphous and crystalline cadmium sulfide compounds, for example, in the incidence of transformations in Syrian hamster embryo cells (Ex. 92). Crystalline CdS compounds, which perhaps are more similar to the hexagonal crystalline cadmium red pigments used in the Rusch et al. study (Ex. 12-10-D) than amorphous compounds, displayed greater cell-transforming activities at both cytotoxic and non-cytotoxic levels than did the amorphous compound. Qualitative assessment indicated considerably less uptake of amorphous CdS than of crystalline CdS although this phenomenon was difficult to observe quantitatively with light microscopy (Ex.92). OSHA recognizes that in occupational settings, all CdS will not necessarily be in its hexagonal crystalline form. Furthermore, all pigments will be mixtures of more and less soluble forms of cadmium.

Among the studies which have examined cadmium pigments there is some evidence to suggest that cadmium pigments are less soluble than other cadmium compounds such as cadmium chloride. It is possible that due to their relative insolubility the pigments are also less available to the body tissues. The evidence is equivocal, however, with respect to the observable toxic effects. The short term animal tests seem to show fewer adverse effects (e.g. lower mortality and cadmium body burdens) among animals exposed to cadmium pigments. The animals, however, were exposed for only short periods of time. Yet, even in these short term exposure studies there is evidence of accumulation of cadmium in the lung, liver and kidney. There is also positive evidence of tumor formation in rats exposed to a cadmium pigment compound. In epidemiological studies of pigment users, low urinary cadmium and beta-2-microglobulin levels were observed among cadmium pigment workers but, in most cases, the level of exposure was poorly reported or not given, raising the possibility that the lack of effect seen among these pigment exposed workers was simply a result of low exposure (e.g. Ex. 8-729). Among workers employed in the manufacture of cadmium pigments, renal effects were noted, and in one study, it was concluded that "insoluble cadmium" pigments and dusts should not be regarded merely as nuisance dusts because such dusts can cause kidney damage (Ex. 8-716). Although there is some evidence to suggest that cadmium pigments are less soluble than other cadmium compounds, there is not sufficient data to show that this reduced solubility correlates with a reduced toxicity, especially after long term exposure. One study even suggests an increased bioavailability with a less soluble cadmium compound. After long term exposure to cadmium pigments, cadmium may in fact be retained or may accumulate in body tissues and result in adverse health effects in a manner similar to the adverse effects which have been observed after long term exposure to other cadmium compounds.

OSHA concludes that there is insufficient evidence to quantify a difference in renal potency, and there is a lack of agreement between commentors on the presence of renal effects among pigment workers. Given the limitations of these data and the severity of the type of health effects that can result from rather small errors in estimates of risk, OSHA should not regulate cadmium pigments, or CdS pigments, differently from other cadmium compounds. OSHA acknowledges that several commentors hold the opinion that CdS is less bioavailable, but the opinions vary on the degree of difference in bioavailability, i.e., 2-3 fold reductions in some studies; reductions of 10 to 15 in other studies; a reduction of thousands times less in another study. The data are inadequate to develop public policy decisions that would allow some workers to be exposed to higher amounts of cadmium than other workers, based upon type of compound alone. Opinions differ on the type of cadmium compound in use in the workplace. In the absence of data that indicate only relatively insoluble forms of cadmium sulfide will be present, OSHA can not separate one cadmium compound from others for regulatory purposes. OSHA must err on the side of worker-health and believes that mixtures of relatively more and less soluble forms will be present at one time or another in the workplace.

3. Pulmonary Effects

Reduced pulmonary function and chronic lung disease indicative of emphysema have been observed in workers with prolonged exposure to cadmium fume and dust. In a study by Friberg (Ex. 4-29), workers at an alkaline accumulator factory exposed to cadmium dust at estimated concentrations of 3 to 15 mg/m(3) for 9 to 34 years experienced impaired olfactory sensation, shortness of breath, and impaired lung function with associated poor physical working capacity. Rabbits exposed to cadmium dust from that factory exhibited chronic inflammatory changes in the nasal mucosa and signs of emphysema in the lung (Ex. 4-29).

Subsequent studies have confirmed the findings of these initial clinical and experimental studies. Bonnell (Ex. 4-22) and Kazantzis (Ex. 4-42) studied workers exposed to cadmium fume at copper-cadmium alloy factories for 5 to 15 years. The average concentration of cadmium over an 8-hour period was reported not to have exceeded 270 ug/m(3). The workers exhibited shortness of breath and impairment of pulmonary function, which were suggested to have been the result of emphysema. Similarly, a study of workers exposed to cadmium dust at concentrations below 200 ug/m(3) for greater than 20 years showed significantly lower pulmonary function compared to within plant non-exposed controls (Ex. 4-50). Smith (Ex. 4-63) examined workers who were exposed to airborne cadmium at 0.2 mg/m(3) or greater for 6 years or more at a cadmium producing plant. Workers were found to have decreased pulmonary function and mild to moderate interstitial fibrosis. Findings in this study suggested that the lung damage was due to prolonged exposure rather than repeated acute exposures. No worker's medical records showed evidence of acute illnesses which would have occurred if cadmium air levels were 5 mg/m(3). Furthermore, a dose-response relationship between reduced pulmonary function and months of cadmium exposure was observed (i.e. pulmonary function decreased as the months of exposure increased). It should be noted that in many of these studies proteinuria was observed in a number of the workers who experienced adverse respiratory effects indicating that both chronic systemic effects and damage at site of contact result from inhalation of cadmium dusts and fumes.

4. Skeletal Effects.

Workers with progressive forms of proteinuria have exhibited skeletal system effects associated with improper bone mineralization such as osteoporosis and osteomalacia. It is possible that cadmium-induced disturbances in the kidney are associated with these adverse effects (Ex. 8-086b, pp. 111-158). For example, the active metabolite of vitamin D, 1,25-dihydrocalciferol (1,25 DHCC) forms in the kidney and stimulates intestinal absorption of calcium which is required for normal bone mineralization. As cadmium accumulates in the renal cortex it may inhibit the metabolism of vitamin D to its active metabolite. Additionally, cadmium-induced renal damage may decrease the tubular reabsorption of calcium, and thus increase the urinary excretion of this essential element from the body. Recent studies of patients with cadmium-induced bone defects have also shown reduced concentrations of vitamin D metabolites in their blood (Ex. 8-189).

Bone mineralization may also be inhibited when there is interference with collagen metabolism. Cadmium may inhibit the formation of collagen fibers by interfering with the copper-dependent enzymes responsible for the cross linking of collagen molecules into fibrils. These fibrils form collagen fibers which in turn provide the fiber structure necessary for proper mineralization of bone. Improper bone mineralization results in a decreased density and softening of bone, conditions associated with osteoporosis and osteomalacia.

In humans, adverse bone effects have been observed after long-term exposure to cadmium. In a follow-up study of workers exposed to cadmium dust for 28 to 45 years, several workers showed hypercalciuria (an excess of calcium in the urine) with one case advancing to osteomalacia (Ex. 8-9). A case study by Friberg of a battery plate worker exposed to cadmium for 36 years documents the development of renal tubular dysfunction and severe osteomalacia (Ex. 8-170). Friberg notes, however, that relative to the number of workers with reported severe renal tubular damage the reported number of cases of adverse bone effects is low (Ex. 8-086b, p. 140).

One reason adverse bone effects occur infrequently may be that the bone has a reserve of calcium to maintain an adequate level in the body and thus it may take a long period of time for cadmium to induce bone disease. A second reason is that dietary deficiencies, in addition to cadmium exposure, may be necessary to induce bone effects. For example, in cadmium-polluted areas of Japan, cases of Itai-Itai disease, (a condition characterized by osteomalacia and renal tubular dysfunction), have been causally related to cadmium exposure from contaminated rice. However, among the cases there was also a dietary deficiency of calcium and vitamin-D, suggesting that the inadequate consumption of essential food elements and vitamins may have been a contributing factor to the disease (Ex. 8-086b, p. 151-153).

In animals exposed to cadmium either by injection or ingestion, disturbances in calcium metabolism with osteoporotic and osteomalacic conditions have been observed. For example, chicks exposed to cadmium in their feed for 3 weeks showed a decrease in calcium absorption from the intestine suggesting a possible effect on the formation of 1,25-DHCC (Ex. 8-3). Calcium absorption was observed to decrease as levels of cadmium in the feed increased.

Osteomalacia was induced in rats fed dietary concentrations of 10, 50 or 100 ppm cadmium for 19 months (Ex. 8-112). Osteoporotic changes increased as cadmium doses increased. The rats fed cadmium developed osteoporotic changes in bone before the onset of kidney damage indicating that cadmium may possibly have a direct effect on bone rather than an indirect effect through renal damage (Ex. 8-55). Friberg, however, presents a review of experimental studies in which the preponderance of data seem to suggest that chronic exposure to cadmium induces osteoporosis and osteomalacia subsequent to, and perhaps associated with, renal tubular damage (Ex. 8-086b, p. 115-139).

5. Reproductive and Developmental Effects.

a. Reproductive Effects in Male Animals. A number of studies have demonstrated testicular necrosis after systemic administration of cadmium salts in animals such as rats, rabbits, monkeys, guinea pigs, golden hamsters and calves. (Ref. in Exs. 8-420; 8-107; 8-337; 8-338; 8-86B). Parizek and Zahor have also reported regressive changes of the seminiferous epithelium in rats within 4 to 6 hours after subcutaneous injection of cadmium chloride or lactate. These changes progress to total necrosis within 24 to 48 hours (Ref. in Ex. 8-420). Morphological changes in the spermatozoa of the ductus deferens and proximal parts of the epididymis occur, but changes in the spermatozoa in distal parts of the epididymis are sometimes also observed. White observed that cadmium is extremely toxic for sperm cells in vitro (Ref. in Ex. 8-86). Schmid et al. have shown that the normal sperm motility is inhibited at cadmium concentrations exceeding 1.6 uM (=180 ug/l) (Ref. in Ex. 8-86B). It has been suggested by Chiquoine that cadmium-induced testicular necrosis is common in species possessing scrotal testes and absent from those possessing abdominal testes (Ref. in Ex. 8-86B).

The vascular bed and the blood flow of the testicles are affected very rapidly following injection of cadmium. Cadmium increases permeability of the testicular capillary blood system. Francavilla et al. showed that capillary damage leads to massive vascular escape of fluids and blood substances into the interstitium which subsequently results in edema and circulatory stasis (Ref. in Ex. 8-86B). Organ-specific carbonic anhydrase was suggested by Hodgen et al. to be the primary target of cadmium toxicity in the testicles (Ref. in Ex. 8-86B).

Piscator and Axelsson did not observe any pathological changes in the testicles of rabbits exposed to subcutaneous injections of cadmium (0.25 mg Cd/kg, 5 days/wk) for as many as 24 weeks and then followed for 30 weeks before sacrifice (Ref. in Ex. 8-95). It is suggested that the cadmium accumulated from long-term exposure is mainly bound to metallothionein and that this protein has a protective effect. Zenick et al. did not see any effect of cadmium exposure on testis weight, sperm count, number of abnormal sperm and testis morphology in male Long Evans hooded rats exposed to 0, 17.2, 34.4 and 68.8 mg Cd/l in drinking water over a period of 70 days (Ref. in Ex. 8-86B). Also, reproductive outcome was completely normal in these rats.

Krasovskii et al., however, showed that with chronic dosing, adverse effects on spermatogenesis occurred at lower doses than with acute dosing (Ref. in Ex. 8-86B). They found significant reductions in sperm number and motility and a significant increase in desquamation of spermatogenic epithelium in rats given 0.005 or 0.0005 mg Cd/kg orally for six months. There were no adverse effects at 0.00005 mg/kg. Dwivedi et al. demonstrated a depression of spermatogenesis, increased production of abnormal sperm and atrophy of the seminal vesicles with daily doses of 0.001 m mol/kg (0.2 mg/kg) given intraperitoneally for one month which was similar to that produced by a intraperitoneal dose of 0.01 m mol/kg (2 mg/kg) (Ref. in Ex. 8-86B). They also noted marked inhibition of choline acetyl transferase activity in the spermatozoa, a change known to be associated with impaired sperm function and sterility. Senczuk and Zielinska-Psuja noted damage to the spermatogenic tubules and interstitial tissue hypertrophy following administration of 8 or 88 mg cadmium chloride/kg in the diet for 12-15 months (Ref. in Ex. 8-86B). These changes, however, were not seen at 3 or 6 months on the diet.

Battersby et al. demonstrated that administration of cadmium chloride (5 and 50 ppm) in drinking water for up to 40 weeks did not alter the ultrastructural appearance of the prostate gland in rats of varying ages (Ref. in Ex. 8-86B). The testosterone concentration also did not differ significantly from controls. Low levels of cadmium (< 5 ppm) were accumulated by the ventral lobe of the prostate, although the metal was not detectable subcellularly.

Changes in blood androgen and gonadotropin levels have been shown to parallel the extent of histological damage to the interstitial tissue of the testes by several investigators such as Favino et al., Saksena et al., Lau et al., and Dutt et al. (Ref. in Exs. 8-86B; 8-206). The human chorionic gonadotropin (HCG)-induced serum testosterone concentration was reduced to less than five percent of that in control rats even at comparably low doses (0.18, 0.34 and 0.83 mg Cd/kg) resulting in decrease in the weight of the testes, seminal vesicles and epididymis of rats. Changes in secondary sex organs following cadmium injection are thought to be due to these hormones, but Timms et al. have evidence from rat studies suggesting cadmium may itself have a direct effect on the prostate gland (Ref. in Ex. 8-86B).

b. Reproductive Effects in Female Animals. Kar et al. demonstrated that the ovaries of prepubertal rats undergo morphological changes after injection (route unspecified) of 10 mg CdCl(2) (6 mg Cd/kg) (Ref. in Ex. 8-86B). One week after the administration, recovery was complete. Massive ovarian hemorrhage was induced by injection (route unspecified) of cadmium chloride or acetate (2.3-6 mg/kg) by Parizek et al. (Ref. in Ex. 8-246). Similar results were also reported by Watanabe et al. (Ref in Ex. 8-370).

Parizek noted that, in contrast to the good survival of nongravid rats and those injected after giving birth, administration of cadmium salts (0.02 mmole/kg; during the last four days of pregnancy) to gravid rats resulted in high mortality (76% for first pregnancies, slightly less for second pregnancies) within one to four days after injection (Ref. in Ex. 8-86B). The first sign of illness in some of the injected gravid rats was the appearance, within six hours of injection, of blood in the urine. Occasionally, when rats were observed at the time of death, violent convulsions were seen. Generalized visceral venous congestion, intense pulmonary congestion, hemorrhagic edema and sometimes massive pleural effusion were seen at autopsy in rats dying within 24 hours after injection. At this stage the kidneys were swollen and hyperemic, with focal or diffuse hemorrhages predominantly situated in the renal medulla.

Copius Peereboom-Stegeman et al. reported that cadmium exposure seems to increase the thickness of the basal lamina in the blood vessels in a dose-related manner in the uterus of female rats subcutaneously injected with 0.36 and 0.18 CdCl(2)/kg for 8 to 60 weeks. But these are also indications of an increased thickness of the basal lamina with age. Possibly, cadmium exposure was accelerating the age-related changes in these blood vessels (Ref. in Ex. 8-86B).

Exposure of female rats to cadmium sulfate (3 g/day for 4 months) prolonged the estrous cycle in a study by Tsvetkova (Ex. 156). In four months, average length of diestrous phase in the experimental females was 6.2 + or - 0.02 days, and in the control 1.2 + or - 0.02 days. Der et al. have also shown altered estrous cycles in rats given 250 ug/day of cadmium chloride by intramuscular injection for 54 days (Ref. in Ex. 8-86B). After 25 days, regular cycles ceased and the animals went into persistent diestrus. The dose of 250 ug/day caused other signs of toxicity viz. lower weight gain, coarse hair coat, sluggish movements and significant reductions in uterine, ovarian and pituitary weight, but no histological changes in the uterus or ovary. Injection of 50 ug/day produced few toxic signs and had no effect on estrus cycles or reproductive organ weights.

c. Reproductive Effects in Humans. There is only limited data on reproductive effects in humans, there is some evidence in animals. There is no evidence of cadmium-induced testicular necrosis in humans, most likely because extremely high doses would be required to induce such an effect. Friberg suggests that if the absorbed oral dose required to produce a testicular effect is proportional to the doses administered in the injection studies, a dose of 70 mg to a 70 kg man would be required to elicit the same response as the 1 mg/kg dose studied in animals (Ex. 8-086b, p. 185). The lack of data on testicular function following cadmium exposure in humans makes it difficult to draw any conclusions on possible acute testicular effects in man.

OSHA reviewed two reports which addressed the effect of cadmium on the male reproductive system in humans. Exposure data were insufficient in these reports to adequately evaluate dose-response effects.

Smith et al. (1960; Ex. 155) reported five cases of fatalities as a result of chronic cadmium poisoning. The workers were exposed repeatedly to brief, intermittent, but high concentrations of cadmium fume in the manufacture of a copper-cadmium alloy. In the process, a 50 percent cadmium master alloy was first prepared and small quantities added to crucibles of molten copper which were stirred manually. Large amounts of cadmium fume were released. Histopathological examination at necropsy revealed that all five cases had emphysema but little evidence of bronchitis and kidneys with little damage except for slight hyaline arteriosclerosis. In four cases the testes were also examined and all exhibited normal tubules with many mitoses. In all, however, there was very infrequent maturation to spermatids or spermatozoa. The authors stated that the plentiful mitotic activity in the spermatocytes suggests that the depression of maturation to spermatids and spermatozoa is an effect of terminal illness rather than of chronic cadmium toxicity.

The androgen function of men occupationally exposed to cadmium during the manufacture of alkaline storage batteries has been studied by Favino et al. (1968, as cited in Friberg et al., 1986/Ex. 8-86B). Ten cadmium-exposed workers and ten controls matched for age and body weight were examined. The cadmium-exposed subjects worked in one of two processes in the plant. In the first area some chemical and physical processes were carried out to prepare the material for the negative electrode of the storage battery: from a sulphonitric mixture cadmium was precipitated as Cd(OH)4 by NaOH at 70 deg. C, filtered under pressure and dried at about 140 deg. C. Then a fine cadmium powder was prepared which was mixed with kerosene and water. In the second area the technical processes were carried out to supply the elements of the storage battery and to complete the manufacture of the battery by connecting the negative with the positive electrode, which is a mixture where nickel is the active component. Some cadmium-exposed subjects were working at the time of the study and some had left the plant several years before. Of those currently employed at the plant, those in the first area had, at the time of the study, been continuously working in that area for about two to five years or more without interruption. Those in the second area had been continuously exposed to cadmium until some years before the study initiation and then were employed in other areas of the plant. The control workers had other jobs in the same plant. The majority worked in lead storage battery manufacture.

Androgenic function was assessed by the measurement of basal urinary excretion of 17 ketosteroids, androsterone, etiocholanolone, testosterone, and epitestosterone. There was no significant difference between the exposed and control groups in the mean level of these hormones. Epitestosterone was higher in the exposed versus control groups, but this difference was not significant. None of the workers had proteinuria. The presence or absence of symptoms such as impotence as well as the number of children was reported in cadmium-exposed but not control workers. One cadmium-exposed worker claimed impotence and inability to have children after starting to work with cadmium. This worker had low urinary levels of the 17 ketosteroids, androsterone, etiocholanolone, and testosterone. This worker also had urinary cadmium levels above the normal range. Another worker who also claimed impotence had normal urinary hormone levels. The authors concluded that more information was needed about the first affected worker before it could be determined whether cadmium was toxic to his genital function. OSHA is of the opinion that the above mentioned studies provide no evidence of any adverse effect of cadmium on human testicular function.

d. Developmental Effects. There are few studies of the developmental effects of cadmium in humans. Tsvetkova (1970; Ex. 156) studied 106 female workers exposed to various cadmium compounds in alkaline storage battery, chemical reagent and zinc casting plants. The exposed workers were from 18 to 48 years of age. They had worked from 2 to 16 years. Workers in the alkaline battery factory were exposed to cadmium oxide in concentrations of 0.1 to 25 mg/m(3). Those in the chemical plant were exposed to a number of soluble cadmium salts in concentrations of 0.16 to 35 mg/m(3). Workers in the zinc casting factory were exposed to cadmium sulfate, cadmium sulfide and metallic cadmium in concentrations of 0.02 to 25 mg/m(3). A control group of workers not exposed to cadmium was also examined, but no further description of this group was provided.

The authors reported that they were unable to show changes in the menstrual cycle of women in contact with the indicated cadmium compounds. It was reported, however, that there were isolated changes in the menstrual cycle which were a function of endocrinal-gynecological illnesses said to develop from working with cadmium compounds. Further details of these illnesses were not provided.

The course and time of pregnancy of the cadmium exposed workers were reported to correlate with physiological norms. The average neonatal weight for either boys or girls born to cadmium-exposed workers employed in either the battery or zinc casting plants was significantly lower (p <.01 for each group) than that of the controls. For boys born to control, battery, and zinc casting plant workers, average weights were 3.719 kg + or - 0.120 (11 newborns), 3.217 kg + or - 0.036 (13 newborns), and 3.388 kg + or - 0.028 (17 newborns), respectively. The average weight of girls born to control, battery, and zinc casting plant workers were 3.544 kg + or - 0.82 (9 newborns), 2.918 kg + or - 0.032 (14 newborns), and 3.106 kg + or - 0.031 (10 newborns), respectively. Neonatal weights of children born to workers in the chemical plant were lower but not significantly than those of controls. Thirteen children (8 boys and 5 girls) were examined. Additionally, 4 out of 27 children of the zinc factory workers were reported to have clear signs of rickets, one had retarded eruption of teeth and two had dental disease. The children of controls were not similarly affected.

It is OSHA's opinion that the limited amount of information on methodology, including selection of controls, provided in this study does not allow one to interpret the findings.

Exposure of animals during pregnancy to cadmium, at doses in the order of mg/kg, gives rise to fetal death as well as severe malformations. Teratogenic effects of cadmium and death of the embryo occur as a consequence of cadmium given in early pregnancy, whereas fetal death is the dominating effect when cadmium is administered shortly before delivery.

Lauwerys et al. and Roels et al. have shown correlations of maternal, placental, and fetal blood levels of cadmium indicating that cadmium accumulates in the placenta (Ref. in Ex. 8-668). Parizek et al. have reported that cadmium salts, given in small amounts to pregnant rats, evoked rapidly progressive changes in the placenta, resulting in destruction of the pars fetalis (Ref. in Ex. 8-555). The cadmium administration appeared to cause hemorrhagic changes and necrosis in the placenta which could lead to embryonic death and uterine hemorrhage as shown by Parizek and Chiquoine (Ref. in Exs. 8-555; 8-86B). The studies by Ferm et al. and Dencker indicated that the transfer of cadmium across the placenta to the fetus varies during gestation. (Ref. in Ex. 8-668) During early organogenesis, in mice and hamster, cadmium reaches the embryo. Berlin and Ullberg observed that after closure of the vitelline duct, cadmium uptake (after parenteral administration) was diminished and remained low throughout the remainder of gestation in mice (Ref. in Ex. 8-668). Similar results were also seen in rats by several investigators such as Sonawane et al., Ahokas and Dilts, and Levin and Miller (Ref. in Ex. 8-86B; 8-88; 8-86B). After a single oral dose of cadmium to rats on day 17 of gestation, little cadmium was detected in the fetus in a study by Ahokas and Dilts (Ref. in Ex. 8-88).

Tsvetkova observed lower fetal weight when pregnant rats were exposed to cadmium sulfate (3 mg/m(3)) by inhalation but there was no evidence of increased fetal mortality (Ex. 156). In a study by Prigge, exposure of pregnant rats to Cadmium chloride aerosols at three dose levels (0.2, 0.4 and 0.6 mg Cd/m(3) continuously for 21 days) also resulted in reduction of weight gain in all three groups of exposed pregnant rats. Fetal weights were also significantly reduced in animals exposed to the highest levels. Fetal alkaline phosphatase activity was also elevated in the most highly exposed group. A marked dose dependent decrease in the activity of alkaline phosphatase was observed in exposed pregnant as compared to exposed nonpregnant animals. Hemoglobin levels and hematocrits were increased in both pregnant and non pregnant animals. However, no changes in fetal erythropoiesis were seen (Ex. 154).

Cadmium administered parenterally during organogenesis caused fetal malformation which varied with the time of administration and strain of animals as shown by Chang et al. and Ferm and Hanlon et al. (Ref. in Ex. 8-668). For example, in the hamster, intravenous administration of 2 mg/kg cadmium sulfate on day 8 of gestation induced a high percentage of fetal deaths and facial deformities in survivors. Exencephaly and skeletal defects were also observed by Ferm and Carpenter (Ref. in Ex. 8-668).

A dose dependent increase in teratogenic effects was also seen by Ishizu et al. after subcutaneous administration of cadmium chloride (0.33, 0.63, 2.5 or 5.0 mg/kg) to pregnant mice on day 7 of gestation (Ex. 8-195). At the dose of 5.00 mg/kg, exencephaly, lack of tail, rachischisis and vaginal atresia were noted. When the fetuses were treated and stained with alizarin, skeletal malformations in the skull region, vertebral parts and ribs were observed. When external malformations were combined, the total malformation appearing rate exceeded 80 percent. When dosage of cadmium was reduced to 2.5 mg/kg, 0.63 mg/kg or 0.33 mg/kg, both external and skeletal malformations were reduced significantly, malformation appearing rate was less than one percent with 0.63 mg/kg and zero with 0.33 mg/kg. The concentrations of cadmium in the liver, kidney and placenta of mice exposed to cadmium were 800, 450 and 10 times higher than the control mice. There was no measurable amount of cadmium in the fetuses of cadmium chloride injected mice. In authors opinion, administered cadmium stayed mostly in the placenta at least in the late pregnancy and was not transferred to the fetuses (Ex. 8-195).

A dose-dependent rise in the fetal death rate, decrease in fetal weight, and increase in the rate of anomalies, which included micrognathia, cleft palate, clubfoot and small lungs were noticed in rats after daily subcutaneous injection of 4, 6, 8 or 12 mg cadmium chloride (2.6 to 7.7 mg Cd)/kg given for 4 consecutive days beginning on day 13 and extending to the 16th day of gestation were reported by Chernoff (Ex. 8-155).

A high incidence (as many as 80%) of the fetuses with hydrocephalus were seen by Samarawickrama and Webb when pregnant rats were given a single intravenous injection of 1.25 Cd/kg between days 9 and 15 of gestation. Other defects observed were anophthalmia, microphthalmia, gastroschisis and umbilical hernia. It is mentioned in the report that 1.1 mg Cd/kg produced no malformations, while 1.35 mg Cd/kg killed all embryos (Ex. 8-157).

In a study by Schroeder and Mitchener, oral administration of cadmium (10 ppm in doubly deionized water) to breeding mice resulted in a loss of strain in two generations (Ex. 153). Sharp angulation of the distal third of the tail, a congenital abnormality, was seen in 16 percent of the F1 and F2A generations. Of the offspring that lived beyond weaning 13 percent were runts and 30 percent died. Inability to breed in F2B generation by some pairs was also noted.

In contrast, no growth, reproduction, or frequency of malformations were noted by Wills et al., after four descendant generations of rats were exposed to very low cadmium concentrations in the diet; 0.07, 0.10 and 0.125 Cd/kg (Ref. in Ex. 8-86B). However, exposure levels were so low, in fact close to the natural background levels, that toxic effects were hardly to be expected.

A lower placental and fetal weight was also seen by Tsvetkova after exposure of female rats to cadmium sulfate (3 g/day for 4 months by inhalation) (Ex. 156). The cadmium content in the liver of the experimental group embryos was 17.64 + or - 0.02 ug/g, and the control group 7.99 + or - 0.04 ug/g. The progeny of the experimental rats exposed to cadmium were less viable than the controls.

Ferm and Layton suggested that teratogenic effects of cadmium in the hamster or mouse could be prevented by pretreatment with small amounts of cadmium, indicating a protective mechanism involving induction of metallothionein (Ref. in Ex. 8-668). Several investigators such as Lucis et al., Arizono et al., and Hanlon et al. have shown that cadmium administered to pregnant animals binds to a metallothionein-like protein in the placenta (Ref. in Ex. 8-668).

Samarawickrama and Webb demonstrated a reduced zinc uptake in the rat fetus after cadmium treatment on day 9 to 15 of gestation, suggesting that the inhibition of DNA synthesis, by inhibiting zinc-dependent thymidine kinase activity, as the mechanisms of damage in the fetus. Dose was found to be critical; 1.1 mg/Cd/kg was not teratogenic in rats, while 1.25 mg Cd/kg caused high incidence of terata and 1.35 mg Cd/kg killed all fetuses (Ex. 8-157).

e. Conclusions. In conclusion, based on the studies presented above, OSHA believes that there is strong evidence that acute systemic administration of cadmium leads to testicular necrosis in various species. Regressive change in seminiferous epithelium, morphological changes in spermatozoa and inhibition of sperm motility after acute exposure are well substantiated. OSHA is of the opinion that acute administration of cadmium leads to morphological changes in ovaries and ovarian hemorrhage in female rats. However effects of cadmium on the reproductive organs after long-term exposure to low levels of cadmium are usually mild or absent. The lack of toxic effects seems to be associated with protective effects of metallotheionein that binds cadmium. Nevertheless, OSHA assumes that adverse effects on spermatogenesis occur after chronic dosing with cadmium at levels as low as 0.0005 mg/kg. Significant reduction in sperm number and motility and significant increase in desquamation of spermatogenic epithelium have been shown in rats given 0.0005 mg Cd/kg orally for six months. Chronic administration of cadmium leads to an increase in the thickness of the basal lamina in the blood vessels in a dose-related manner in the uterus of female rats. Based on Parizek study, OSHA believes that pregnant rats are more sensitive to the adverse effect of cadmium than the nonpregnant female rats. OSHA also concludes that there is sufficient evidence to indicate that chronic administration of cadmium can prolong the estrous cycle in rats. OSHA regards cadmium to be teratogenic and fetotoxic during early pregnancy as well as fetotoxic when administered shortly before delivery in rats.

6. Other Effects

Other adverse effects have been reported in experimental animals chronically exposed to cadmium. There are scattered reports of chronic effects on the gastrointestinal tract, peripheral nervous system and endocrine organs. More commonly documented effects in animals include anemia, changes in liver morphology, immunosuppression, and hypertension. For example, various experimental animals fed or injected with cadmium have commonly exhibited anemia, possibly due to cadmium's influence on the absorption and distribution of such metals as zinc and iron (Ex. 8-086-B, p. 167). Similarly, rats chronically exposed to cadmium oxide dust by inhalation developed anemia (Ex. 4-29). Animals exposed to cadmium by various routes of administration have shown morphological changes in the liver as well as disturbances in hepatic enzyme concentrations (Ex. 8-086-B, p. 161). Chronic oral exposure of mice to cadmium through drinking water decreased antibody synthesis (Ex. 8-24) and induced immunosuppression (Ex. 8-35).

There is conflicting evidence with respect to cadmium induced hypertension. Several studies have shown an increase in blood pressure after exposure to cadmium. Hypertension has been induced in rats orally exposed from 3 to 24 months to 0.1 to 10 mg cadmium/liter drinking water (Ex. 8-14). In this study, levels as low as 0.1 mg/l for 3 months increased systolic blood pressure. The renal cortical level was 5 to 30 ug cadmium/g wet weight, which is below the critical concentration at which proteinuria is commonly detected. There are also studies, under similar experimental conditions, which have shown no hypertensive effects (Ex. 8-086b, pp. 170-173). It has been suggested that dietary differences may have caused the different responses, because rats on rye-based diets exhibited increased blood pressure whereas rats on other non-rye-based diets did not.

7. Conclusions about Non-Carcinogenic Effects

There is an abundance of data clearly indicating that exposures to cadmium in the industrial environment can cause serious toxic effects in human beings. Not only are there many experimental studies showing the acute and chronic effects of cadmium exposure, but there is also a great amount of human evidence among cadmium-exposed workers indicating adverse effects from chronic exposure to cadmium.

In humans, one of the earliest observable adverse effects from chronic exposure to cadmium is tubular proteinuria, the presence of an excess amount of low-molecular weight proteins in the urine (Exs. 12-07, 4-27, 4-28). This condition indicates impairment or loss of kidney function. Because of the body's ability to accumulate and store cadmium over long periods of time this condition may develop even after a reduction in or cessation of external cadmium exposure. Upon prolonged exposure tubular proteinuria may progress to more severe forms of renal dysfunction such as glycosuria, aminoaciduria, phosphaturia, and glomerular proteinuria or chronic nephrotoxicity. There is no specific treatment for chronic cadmium poisoning or for restoring kidney function. Persons with cadmium induced renal disease are at increased risk for developing kidney failure if additional renal insults occur such as exposure to other nephrotoxins including medications, infections of the renal-urinary system, obstruction of the urinary system, or reduced volume of blood flow to the kidneys due to reduced blood volume or vascular disease. In cases of cadmium-induced kidney damage, rigid control of diet, water intake and electrolyte balance in addition to medical treatment is required. Since other environmental sources of cadmium such as water, food, and ambient air may contribute to the total body burden, it is necessary to minimize all exposure to cadmium to prevent additional adverse health effects.

As noted in the lead standard (43 FR 52952), diseases resulting from exposures to heavy metals proceed in five stages: (1) normal, (2) physiological change of uncertain significance, (3) pathophysiological change, (4) overt symptoms (morbidity), and (5) mortality. Within this process there are no sharp distinctions, but rather there is a continuum of effects. Categories overlap due to the variation in individual susceptibilities and exposures in the working population. Although step 2 is of uncertain significance, by step 3 (pathophysiologic changes) significant adverse health effects have occurred. Tubular proteinuria is considered by OSHA to represent pathophysiologic changes of consequence, and such changes represent material impairment, given OSHA's current understanding of the progressive stages of cadmium effects.

Other adverse health effects of concern to OSHA include reproductive effects, liver and spleen effects, and, non-carcinogenic lung effects, such as bronchitis. OSHA is of the opinion that the animal and human data are remarkably consistent. Human studies show high acute cadmium toxicity in the form of renal, liver, and pulmonary effects, and the animal studies show similar effects. There is renal damage and lung disease (bronchitis) in chronically exposed humans and in chronically exposed animals. There is also good correlation between ITAI-ITAI disease in humans and demineralization of the bone in animals, and liver damage is seen both in humans and animals.

C. Mutagenicity

A wide range of tests have been conducted to determine the mutagenic effects of cadmium. The mutagenicity of cadmium has been tested in bacteria, plants, insects, and mammalian cells, including human cells, in vitro and in vivo. Comprehensive reviews of these various investigations have been provided by Friberg (Ex. 8-086b, p. 223), Degraeve (Ex. 4-24), and EPA (Ex. 4-04). Both positive and negative results have been reported from these studies. This has lead to a somewhat confusing picture as to the mutagenicity of cadmium. The following section will give an overview of the more pertinent studies covered in the above reviews.

Cadmium has been shown to modify the metabolism of both RNA and DNA. Evidence has been obtained both in vitro and in vivo in microorganisms, plants, and mammalian cells showing enhancement and inhibition of RNA synthesis, degradation of DNA repair, inhibition of DNA synthesis, and inhibition of thymidine incorporation into DNA.

Gene mutation studies on microorganisms, yeasts, and mammalian cells have given mixed results on cadmium's mutagenic effects. For example, positive and negative mutagenic responses were observed in histidine reverse mutation assays using the bacteria Salmonella typhimurium. Some of these studies were considered inconclusive because several protocols were used in the assays. For example, different strains of S. typhimurium were tested using different dose regimens (e.g. single doses and doses with other chemicals). Conflicting and inconclusive results were also observed in gene mutation studies using yeast. For example, in a test for the induction of petite mutations, p-mutants were induced at the high and low doses but not at the middle dose. In a similar yeast assay, no p-mutants were induced at all, however; the dose was so toxic that only one percent of the yeast cells survived. Gene mutation assays using mammalian cell cultures of mouse lymphoma and chinese hamster cells have shown increased mutation frequencies with cadmium treatment.

Conflicting results were also reported in mutagenicity tests on fruit flies. Negative results were observed in sex-linked recessive lethal mutation tests, but positive results were observed in dominant lethal mutation tests. However, among the negative results it was noted that in one case too few chromosomes were tested while in another case the number of chromosomes tested was not reported. Thus, the scope of the tests may have been too small to detect a positive response.

In higher order plants, the mutagenicity tests have been mostly positive. Aberrations such as chromosomal lesions and breaks were induced in several different species of plants.

In mammalian cells, in vitro studies on human lymphocytes, have shown increased incidences of structural chromosomal aberrations after treatment with cadmium. Among the observed aberrations were chromatid breaks, symmetrical and asymmetrical translocations, and deletions. In vitro tests on other mammalian cells in culture, such as Chinese hamster cells, displayed no increase in structural chromosomal aberrations with cadmium treatment but did show an increase in numerical chromosomal aberrations (e.g. hyperploidy and diploidy).

Numerical aberrations were also observed in vivo in the oocytes of mice and hamsters treated with cadmium. In these studies no structural chromosomal aberrations were noted. Numerical aberrations were also observed in the blastocytes of cadmium treated mice, indicating that aberrations induced in the oocytes may be transferred to the embryo. Other in vivo tests on mice have shown negative responses. For example, in micronucleus assays, the frequency of micronuclei in experimental groups did not increase compared to control groups. Also, in dominant lethal assays no increase in mutants was observed in mice injected with cadmium chloride compared to controls. Heritable translocation assays revealed no observable translocations in the spermatocytes of the F1 progeny of mice injected with cadmium chloride.

As in other test systems, in vivo studies on humans have produced conflicting results. For example, lymphocytes from the blood samples of some patients suffering from Itai-Itai disease showed a high rate of chromosomal aberrations such as chromatid breaks and translocations; however, a similar examination of other Itai-Itai patients showed no aberrations. Similarly, positive and negative results were observed in vivo among cadmium exposed workers in two different smelter plants. It was noted that for the positive effects these workers may also have been exposed to other metals such as lead and zinc which might have induced or contributed to the observed aberrations.

Thus, although a number of positive mutagenic responses have been observed, there are also a number of conflicting negative responses. It is difficult to make comparisons or to make conclusions about these conflicting results since the studies investigated different endpoints, and often used different protocols. Thus, until more conclusive mutagenicity studies are conducted and reported, cadmium may be considered to be a potential mutagenic agent.

D. Carcinogenic Health Effects

Cadmium has been shown to induce cancers in laboratory animals and is associated with lung and prostate cancer in man. Cancer is the second most common cause of death in the U.S. today. Lung cancer claims the largest share of cancer deaths among males and the second largest share of cancer deaths among females. The National Center for Health Statistics reports that in 1980, the lung cancer death rate was 68.8 per 100,000 for males and 24.4 per 100,000 for females.

Mortality and incidence are related by the case fatality rate, the proportion of incident cancer cases that terminate as cancer deaths. When a cancer is well defined and the mortality rate is similar to incidence, as with lung cancer, the case fatality rate is close to 100 percent. Few cases of lung cancer are curable, despite advances in medical and surgical oncology. Survival rates for lung cancer patients are poor with about 10% surviving five years or more after diagnosis (Ex. 8-62). OSHA considers lung cancer to represent the gravest material impairment of health because it is almost certainly fatal. For prostate cancer, the case fatality rate is lower. Prostate cancer does not always lead to death. Males may have prostate cancer for some time without any clinical manifestation of the disease. Some of these tumors lack the capacity for rapid growth, while others invade surrounding tissue and metastasize to distant organs and cause death. In 1980, 22,881 men died of prostate cancer; the prostate cancer death rate was 20.8 per 100,000 men. Early diagnosis and treatment have reduced the mortality rates associated with prostate cancer. Nevertheless, because workers who work with cadmium are found to be at higher risk (Ex. 8-683) of prostate cancer, OSHA has evaluated the relevant epidemiological studies of prostate cancer among cadmium exposed workers. Prostate cancer also represents the gravest material impairment of health.

1. Evidence in Animals

Cadmium has been shown to be a carcinogen in animals when administered by inhalation. The strongest evidence of carcinogenicity comes from a rat bioassay by Takenaka et al (Ex. 4-67). In this well conducted study, cadmium was found to induce lung carcinomas in exposed Wistar rats. Incidence in the exposed groups was statistically significantly elevated over the incidence in controls, and a statistically significant dose-response was observed.

Takenaka exposed three groups of male rats continuously for 18 months to cadmium chloride aerosols with nominal cadmium concentrations of 12.5, 25, and 50 ug/m(3). An additional group of 41 rats served as controls. The animals received water ad libitum during the experiment but were fed only 8 hours per day to minimize food contamination. The rats were observed for 13 months after the last exposure, at which time all surviving rats were sacrificed. There was no statistically significant difference in mean survival times among the four groups of rats, although the mean survival time for the high dose group was slightly shorter than the mean survival time for the other groups.

A histopathological examination was given to all rats surviving the exposure phase of the study unless their bodies were too autolyzed to allow such an exam. Cadmium concentrations were measured in the lungs, liver, and kidneys of a subgroup of each exposure group. Concentrations in the lung were nearly as high as in the liver. In all organs concentrations were observed to increase with dose except that only the low dose rats were found to have a slightly higher concentration in the lung than was found in the middle dose rats.

The incidence of lung carcinomas was 0/38 (0%) in the controls, 6/39 (15.4%) in the low dose group, 20/38 (52.6%) in the middle dose group, and 25/35 (71.4%) in the high dose group. The majority of carcinomas were adenocarcinomas; however, epidermoid carcinomas, mucoepidermoid carcinomas, and combined epidermoid carcinomas and adenocarcinomas were observed. The incidence of each of these tumors is presented in Table V-21.

Table V-21 - Incidence of Lung Carcinomas in Male Wistar Rats Exposed to
             Cadmium Chloride Aerosols(a)
Tumor Type 50 ug/m(3)              |  Controls  | 12.5 ug/m(3) | 25 ug/m(3)
                                   | (percent)  |  (percent)   | (percent)
Adenocarcinoma 15/38 (39%) ....... | 14/35 (40) |     0/38 (0) |  4/39 (10)
Epidermoid 7/35 (20%) Carcinoma .. |   0/38 (0) |     2/39 (5) |  4/38 (11)
Mucoepidermoid 0/38 (0%) Carcinoma |   3/35 (9) |     0/38 (0) |   0/39 (0)
Combined 1/35 (3) Epidermoid       |            |              |
Carcinoma and Adenocarcinoma ..... |   0/38 (0) |     0/39 (0) |   1/38 (3)
   Total 25/35 (71%) Carcinomas .. |   0/38 (0) |     6/39 (15)| 20/38 (53)
  Footnote(a) From Takenaka et al (Ex. 4-67)

The Takenaka study appears to have been the first animal study to conclusively document a lung cancer response from inhaled cadmium. Takenaka noted that a number of prior experimental study results had only raised the possibility of lung cancer being induced by cadmium inhalation. Other studies, however, have shown the induction of lung cancer and other cancers as a result of either inhalation or subcutaneous injection of several different cadmium compounds. The Risk Assessment Guidelines published by the Office of Science and Technology (OSTP) call for taking account of negative as well as positive studies in assessing the weight of evidence.

Since 1980, OSHA has not published guidelines nor a standard concerning how it will assign weight of evidence in the qualitative evaluation of carcinogenicity in experimental animals. Other Agencies have published guidelines, however, including OSTP and EPA. In EPA's guidelines, five conditions are identified that, if present, may lead to a relatively high degree of confidence in the results of animal bioassays:

(1) Biologically independent tumors were found at a large number of sites;

(2) Independent experiments have demonstrated carcinogenic responses in both genders and in multiple species or strains of animals;

(3) There is a clear-cut and statistically significant dose-response relationship;

(4) There is a dose-related shortening of time-to-tumor occurrence; and

(5) There is a dose related increase in the proportion of tumors that are malignant.

Of these five conditions, four appear to exist for cadmium. OSHA requests comments concerning the degree of confidence that should be placed on the experimental study results related to cadmium in light of these five criteria.

The Takenaka study grew out of a pilot study by Heering et al (Ex. 4-04). In that study, 10 rats were exposed for 18 months to cadmium chloride aerosols with a nominal cadmium concentration of 20 ug/m(3). The animals were sacrificed when exposure ended and four adenomas and one adenocarcinoma were observed.

Results from a study of intratracheal instillations of cadmium oxide are more equivocal. In a study of male Fisher-44 rats, Sanders and Mahaffey found no evidence of cadmium-induced lung carcinomas, but they did observe an increased incidence of mammary fibroadenomas (Ex. 4-61). In that study, three groups of rats were given intratracheal instillations of 25 ug cadmium oxide. Forty-eight rats received one treatment at 70 days of age; 46 rats received two treatments at 70 and 100 days of age for a total dose of 50 ug cadmium oxide; and 50 rats received three treatments at 70, 100, and 130 days for a total dose of 75 ug cadmium oxide. Forty-six rats serving as controls received one intratracheal instillation of 0.9% sodium chloride solution.

The observed incidence of mammary fibroadenomas was 3/45 (7%) in the controls, 7/44 (16%) in the low dose group, 5/41 (12%) in the middle dose group, and 11/48 (23%) in the high dose group. Using the Fisher Exact Test, only the high dose group had a statistically significantly elevated incidence over incidence in the controls (p=.027). Two (5%) adenocarcinomas of the lung were observed in the middle dose group. The average number of tumors per tumor bearing rat were 1.4, 1.5, 1.6, and 1.8 for the control, low dose, middle dose, and high dose groups respectively. The authors reported that this difference was significant (p=.044) in a chi-square test for independence between number of tumors and treatment groups. Slightly more rats in the control group were found to have no tumors (16%) than treated rats (5 to 7%).

Additional evidence of the carcinogenicity of inhaled cadmium is provided by the results from a long term bioassay by Oldiges et al (Exs. 12-10i, 12-10h. and 12-35). In this study, groups of 20 male and female Wistar rats were exposed to cadmium chloride concentrations at 30 ug/m(3) or 90 ug/m(3), cadmium oxide dust at concentrations of 30 ug/m(3) or 90 ug/m(3), cadmium oxide fumes at concentrations of 10 ug/m(3) or 30 ug/m(3), cadmium sulfate at a concentration of 90 ug/m(3), cadmium sulfide at concentrations of 90 ug/m(3), 270 ug/m(3), 810 ug/m(3), or 2430 ug/m(3), or a combination of cadmium oxide and zinc oxide dust at concentrations of 30 and 300 ug/m(3) respectively or 90 and 900 ug/m(3) respectively. Twenty male rats and 20 female rats served as controls.

Most groups of animals were exposed for 22 hours per day for 7 days per week. For each of these groups, exposure continued for 18 months or until 25% of that group had died. Other groups of animals were exposed to their cadmium compound for 40 hours per week for 6 months. This shorter exposure protocol was chosen to determine whether a brief exposure period would induce primary lung tumors. Animal groups were followed through month 31 of the study or until 75% of a group had died. At many of the exposure concentrations, doses proved to be too toxic and many animals did not survive the 31 months of study.

Preliminary results from this study are presented in Table V-22. The primary tumors observed in these rats were bronchio-alveolar adenomas, adenocarcinomas, and squamous cell tumors.

Table V-22 - Incidence of Primary Lung Tumors in Male and Female Wistar
             Rats Exposed to Four Cadmium Compounds
                      |          |   |  Months     |  Months  |   Lung
                      |  Dose    | S |     of      |    of    |   tumor
      Exposure(c)     | ug/m(3)) | E | exposure(a) | Study(b) | incidence
                      |          | X |             |    (b)   |
Controls .............| .......  | M | ........... |       31 |  0/20
                      |          | F | ........... |       31 |  0/20
Cadmium ..............|      30  | M |          18 |       30 | 15/20
Chloride .............|      30  | F |          18 |       31 | 13/18
Cadmium ..............|      90  | M |           6 |       30 | 11/20
Chloride .............|      90  | F |           6 |       29 |  3/18
Cadmium ..............|      90  | M |          14 |       31 | 11/20
Sulfate ..............|      90  | F |          18 |       29 | 18/20
Cadmium ..............|      90  | M |          18 |       30 | 17/20
Sulfide ..............|      90  | F |          18 |       31 | 15/20
Cadmium ..............|     270  | M |          16 |       30 | 14/19
Sulfide ..............|     270  | F |          16 |       30 | 16/19
Cadmium ..............|     810  | M |           7 |       30 | 11/20
Sulfide ..............|     810  | F |          10 |       29 | 13/20
Cadmium ..............|    2430  | M |           4 |       30 |  7/16
Sulfide ..............|    2430  | F |           3 |       31 |  6/19
Cadmium ..............|   270(d) | M |           6 |       27 |  3/20
Sulfide ..............|   270(d) | F |           6 |       29 |  3/20
Cadmium ..............|      30  | M |          18 |       31 | 15/20
Oxide Dust ...........|      30  | F |          18 |       31 | 15/20
Cadmium ..............|      90  | M |           7 |       31 |  9/17
Oxide Dust ...........|      90  | F |          11 |       31 | 11/16
Cadmium ..............|    90(d) | M |           6 |       31 |  4/20
Oxide Dust ...........|    90(d) | F |           6 |       31 |  3/20
Cadmium ..............|    30(e) | M |          18 |       29 | 13/18
Oxide Dust ...........|    30(e) | F |          18 |       29 | 12/20
Cadmium ..............|      10  | M |          18 |       31 |  0/19
Oxide Fume ...........|      10  | F |          18 |       31 |  0/19
Cadmium ..............|      30  | M |          18 |       31 |  3/19
Oxide Fume ...........|      30  | F |          18 |       31 |  4/17
Cadmium Oxide and     |          |   |             |          |
  Zinc Oxide Dust ... | 30/300(f)| M |          18 |       31 |  0/20
                      |          | F |          18 |       31 |  0/20
Cadmium Oxide and     |          |   |             |          |
  Zinc Oxide Dust ... | 90/900(f)| M |          18 |       31 |  8/20
                      |          | F |          18 |       31 |  7/20
  Footnote(a) Study protocol called for 6 or 18 months of exposure, but
exposure was terminated when 25% of an animal group died.
  Footnote(b) Months of study includes months of exposure.  All animals in
a group were sacrificed when mortality in that group exceeded 75%.
  Footnote(c) Incidence is number of animals with at least one primary
tumor divided by the number of animals at risk.  Primary lung tumors are
bronchio-alveolar adenomas, adenocarcinomas, and squamous cell tumors.
  Footnote(d) Exposure was for 40 hours per week.
  Footnote(e) Rats were fed a zinc-reduced diet.
  Footnote(f) Dose was 30 ug/m(3) of cadmium and 300 ug/m(3) of zinc or
90 ug/m(3) of cadmium and 900 ug/m(3) of zinc.

The extremely high mortality rates seem to make this study unsuitable for quantitatively assessing the risk associated with each of the cadmium compounds studied or for assessing their relative carcinogenic potency. The study results indicate, however, that while zinc oxide dust may mitigate the carcinogenic potential of lower doses of cadmium oxide, each of the cadmium compounds alone is carcinogenic in animals exposed to these levels through inhalation.

In an inhalation study by Heinrich (Ex. 12-42), male and female Syrian golden hamsters and female mice were exposed to cadmium chloride, cadmium sulphate, cadmium oxide, or cadmium sulfide using exposure concentrations between 10 and 270 ug/m(3) for 19 hours per day, five days per week, for 50-70 weeks. An abstract of the study has reported that no cadmium-related significant increase in lung tumor rate was observed in either species. However, complete experimental data were not included.

There have been numerous studies involving the subcutaneous or intramuscular injection of cadmium into both rats and mice. The U.S. Environmental Protection Agency's "Updated Mutagenicity and Carcinogenicity Assessment of Cadmium" presents a summary of many of these studies (Ex. 4-4, p. 62-64).

A short summary of several of these studies in provided in the following section. Several studies have failed to demonstrate a carcinogenic effect from cadmium. In a series of studies, rats and mice were given 5 ppm cadmium acetate or oxalate in drinking water throughout their lives (Exs. 8-308; 8-121; 8-196). Compared to controls, there were no significant differences in the incidence of tumors in animals treated with cadmium, although mortality was increased in rats and male mice. In a study of prostatic changes due to cadmium, Levy et al. (Ex. 8-194) treated rats by subcutaneous injection of cadmium sulphate into the flank once weekly for two years in doses of 0.2, 0.1, and 0.05 mg. A low incidence of sarcomata at the injection site was seen in the treated groups. Levy stated that this finding was not unexpected, having been previously reported by Haddow et al. in 1964, Kazantzis, in 1963, and Health et al., in 1962 (Ex. 8-117). No neoplastic changes were seen in the prostate gland, and there was no treatment-related increase in the incidence of neoplasms at other sites.

In two further studies of the effect of cadmium on the prostate gland by Levy et al. (Ex. 8-034; 8-117), mice and rats were treated with cadmium sulphate by gastric instillation. Dosing regimens were 0.35, 0.18, and 0.087 mg/kg body weight once weekly for two years for rats, and 1.75, 0.88, and 0.44 mg/kg body weight once weekly for 18 months for mice. Concurrent dosing regimens of mice and rat controls were run using gastric instillation of equivalent amounts of distilled water. In both studies, no neoplastic lesions of the prostate or urinary tract were seen. Tumors seen in other organs could not be related to cadmium treatment.

Loser (Ex. 8-643) treated rats with cadmium chloride in the diet for two years at doses of 1, 3, 10, and 50 ppm. Fifty male and fifty female rats were used for each level; 100 rats of each sex served as concurrent controls. Cadmium treatment was not associated with an increased incidence of total numbers of tumors or any specific type of neoplasia.

Other studies (Exs. 4-55; 4-57; 8-253) show that the injection of cadmium metal or certain salts of cadmium produce sarcomas at the site of injection as well as interstitial and Leydig cell tumors of the testes in experimental animals. The simultaneous administration of zinc and cadmium has been found to reduce the incidence of cadmium-induced testicular tumors (Ex. 8-253). For a discussion of these studies, please see Elinder (Ex. 8-086B p. 206).

OSHA has not relied upon the injection and peroral studies for assessing carcinogenic risk, nor upon the preliminary data on inhalation. The reasons for this are set forth below in the Significance of Risk section of the preamble.

OSHA relied, in part, upon the review by the International Agency for Research on Cancer [(IARC) Ex. 8-656] using IARC's criteria for categorizing animal data. IARC states that CdCl(2), CdO, CdSO(4), and CdS produced local sarcomas in rats following injection. CdCl(2) and CdSO(4) produced testicular tumors in mice and rats after subcutaneous administration. IARC concluded that the animal data are "sufficient", that is, a causal relation-ship has been established between exposures to cadmium and an increased incidence of malignant neoplasms or a combination of benign and malignant neoplasms in two or more species or in two or more independent studies in one species. IARC classifies cadmium as a probable human carcinogen because it is biologically plausible and prudent to regard agents for which there is "sufficient" animal evidence of carcinogenicity as if they presented a carcinogenic risk to humans. OSHA received several comments on the study by Takenaka et al., and on the study by Oldiges et al., which have been addressed in the section on quantitative risk assessment. However, OSHA received substantial comments on the issue of photodecomposition in the study by Oldiges et al., which was subsequently re-published by Glaser et al. with essentially the same data, based upon the same experiments. These comments are addressed below.

2. The Carcinogenicity of Cadmium Pigments

DCMA has submitted comments that there is no evidence that cadmium pigments, per se, are carcinogenic and that even if they were to be considered carcinogenic, they are less potent as carcinogens. These arguments are based upon the opinion that the carcinogenicity of cadmium sulfide (CdS) pigment as administered in the chronic rat inhalation study by Oldiges et al. was the result of CdS undergoing photodecomposition to cadmium sulfate (CdSO(4)), the latter compound being responsible for the cancers observed in the experimental animals. Secondly, DCMA argued that even if cadmium pigments were determined to be carcinogenic, they are less potent as carcinogens than the more soluble forms of cadmium such as CdCl(2) and CdSO(4) because they are less soluble and hence less bioavailable in human tissue. DCMA points out that there are applications where cadmium pigments are the only source of exposure to cadmium and that in these situations cadmium pigments should be given a different permissible exposure limit from other cadmium compounds (Ex. 144-20).

DCMA requested that OSHA reopen the record in the cadmium rulemaking on the issue of the toxicity and carcinogenicity of cadmium pigments in order to give the DCMA the opportunity to cross examine OSHA's witnesses on this issue (Ex. 144-20). OSHA reopened the record to receive the results of new studies and comments on these studies and to allow interested parties to comment on opinions of Dr. Oberdorster and Dr. Heinrich, who were requested by OSHA to addressed the issue of cadmium pigment carcinogenicity.

[Dr. Oberdorster is currently Professor of Toxicology at the University of Rochester and formerly of the Fraunhofer Institute of Toxicology and Aerosol Research in Germany, where the carcinogenicity studies of cadmium compounds and some of the CdS photodecomposition studies were conducted. Dr. Heinrich is a toxicologist with the Fraunhofer Institute of Toxicology and Aerosol Research in Hannover, Germany.] OSHA denied a request to commence a new set of hearing on the issue of cadmium pigment carcinogenicity. OSHA has considered all comments received during the period the record was reopened. OSHA has also considered all comments that were received prior to the reopening of the record.

For background information, a substance is defined as a cadmium pigment, if (1) the contained cadmium is chemically bound to either sulfur (CdS) or selenium (CdSSe) (Sic), and (2) it is used as a colorant, and (3) it contains less than 0.1 percent acid extractable cadmium, as determined by the EN-71 extraction method (DCMA Ex. 120). "Solubility" is the process by which one substance is dissolved in another. Some cadmium compounds, like cadmium oxide (CdO) and cadmium sulfide (CdS), are almost insoluble in water, whereas cadmium chloride (CdCl(2)) is highly soluble in water. CdO, on the other hand, while insoluble in water is highly soluble in the lung (Ex. 142). In addition to the medium in which a substance can be dissolved, solubility also depends upon the form of the material that is being dissolved. For example, finely divided particles because of their larger surface area/mass ratio are more soluble than larger particles (Ex. 152). More soluble compounds may be more bioavailable and thus more toxic than less soluble compounds, but bioactivity may also be related to mechanical and surface properties (e.g., fibers, Sio(2)). "Bioavailability" of cadmium compounds refers to the degree to which cadmium becomes available to the target tissues after exposure.

As presented in OSHA's proposal, data from the Takenaka et al. and Oldiges et al. studies suggested that all of the cadmium compounds administered by inhalation demonstrated a similar qualitative and quantitative carcinogenic response. However, during the OSHA hearing, the issue was raised that the cancer response resulting from inhalation exposure of CdS to Wistar rats in the Oldiges et al. (1989) study may have been the result of photodecomposition of the CdS to cadmium sulfate, the latter compound causing the lung cancer (Ex. 8-694-D).

Note: Virtually the same study was published in 1989 by Oldiges et al. (Ex. 8-694) and in 1990 by Glaser et al. (Ex. 8-694-B). In this section of the preamble, the chronic rat inhalation study for carcinogenicity of four cadmium compounds will be referred to as the Oldiges et al. study (Ex. 8-694-D).

Based upon preliminary results in his laboratory, Mr. Leonard Ulicny of SCM Chemicals suggested that the particles of CdS pigment in the aqueous suspension used in the Oldiges et al. study may have been solubilized under the influence of light (photodecomposition). Thus, OSHA reopened the docket to receive comments on the issue of the photodecomposition of CdS in the Oldiges et al. study and the role it may have played in the carcinogenic response observed in the study. If complete solubilization only of cadmium pigments and formation of ionic Cd ++ is responsible for the carcinogenicity of these compounds, the results of the Oldiges et al. study for CdS need to be reconsidered in terms of the carcinogenic response observed in the animals. If, on the other hand, an intrinsic toxic particulate effect contributes also to the carcinogenic effect of cadmium pigments such as CdS, then in vivo solubility may not be a good indicator of the carcinogenic potency.

Data from several studies related to insolubility, stability and ionization of CdS were submitted into the record. The data indicate that under proper conditions of light and aqueous suspension of CdS, CdS decomposes and forms cadmium sulfate (CdSO(4)), and the percent of CdS ionized under such conditions depends upon the concentration of CdS in the aqueous suspension and the intensity of light present.

Mr. Ulicny first raised the issue that CdS particles in suspension may be solubilized under the influence of light (photo-decomposition) based on preliminary results in his laboratory. He performed a solubility study of CdS in aqueous suspension under lighting conditions of ~1000 lux for about a week. Gagliardi and Ulicny (1990) presented detailed results of their in vitro experiments (Ex. 144-1) and questioned the validity of the carcinogenic response observed in rats exposed to CdS in the study by Oldiges et al. since a significant fraction of CdS could have been solubilized to soluble CdSO(4). They referred to photodecomposition or photo-oxidation of CdS as having occurred under similar conditions in the Oldiges et al. study. However, no evidence was provided at the time of cadmium hearings that such photodecomposition could occur under the relatively low lighting conditions that were present during the Oldiges et al. inhalation study (approximately 50 Lux for 12 hours/day only).

Photodecomposition of CdS was further studied by Glaser et al. in 1991 (Ex. L-140-14). They replicated the aerosol generation procedure employed in the long-term rat inhalation study by Oldiges et al. (1989), the only exception being that the type of CdS used by Glaser et al. was slightly different from that used in the Oldiges et al. cancer study (Ex. 8-694-D). The rat inhalation study had been performed with CdS type E which consisted of 77.1% Cd, 22.0%, S, 0.024% and 0.64% BaSO(4), whereas in the solubility study CdS type E1 was used, consisting of 76.5% Cd, 22.2% S, 0.14% Zn, and 0.8% BaSO(4). Solubility according to DIN53770 was slightly different, being 0.07% for type E and 0.04% for type E1, and also the specific surface area was 11.8 m(2)/g for type E and 8.7 m(2)/g for type E1. Type E and E1 had been manufactured by the same process and the same producer, Bayer AG, F.R. Germany, yet they were taken from different batches. Dr. Oberdorster stated that the very small physicochemical differences between the two types should be irrelevant as far as biological effects are concerned (Ex. 141).

In the solubility study by Glaser et al., the CdS aerosol was generated from two different concentrations of the suspension, i.e., 0.38 mg/ml Cd and 1.25 mg/ml. These concentrations of CdS suspension were used to generate aerosol concentrations of 90 and 270 ug per m(3) in the long-term rat inhalation study. The aerosols were generated by nozzle atomizers with attached cyclones to eliminate larger-sized particles. The volume that was used daily to generate the aerosol was added each day and amounted to about 300 ml per day. This was adjusted with respect to total cadmium to maintain constancy of the overall cadmium concentration. The light intensity inside the CdS suspension reservoir was measured to be about 50 Lux and the aerosol was generated into the inhalation chambers after being electrically discharged by an 85Kr source before entering the chambers in order to simulate as closely as possible the conditions of the long-term cancer study. The solubility experiment lasted for 64 days; samples of the suspension were taken on days 2, 4, 8, 16, 32 and 64 for measurement of Cd ++ ions by voltametry and SO(4)-- by ion chromatography. Aerosol samples were collected on filters taken from the inhalation chambers on day 15 and 30 and were analyzed separately for Cd ++ and total cadmium. CdS particles were separated from Cd ++ ions by centrifugation (1400 g for 30 min) and subsequent three fold filtration of the supernatant.

The results showed that under these lighting conditions (50 Lux) a significant solubilization rate of CdS particles in the aqueous suspension occurred. During the 64-day study period, 0.24 mg Cd ++ per ml was found in the lower concentration of the suspension (0.38 mg/ml), equivalent to 63% being solubilized. In contrast, CdS suspension in the higher concentration (1.25 mg Cd per ml) showed a solubilization of only 11% of total Cd. There was a steady increase of soluble Cd throughout this period which had not quite reached equilibrium at the end of the study period of 64 days. Concurrently with the increase of Cd ++, an increase in sulphate could be measured which appeared to reach an equilibrium for the lower concentration (0.38 mg/ml) after about 40 days in the study.

Examination of the filter samples from the collected aerosols showed similarly that the lower Cd concentration (generated from 0.38 mg Cd/ml suspension) had a higher percentage of soluble Cd content (50.6% of total cadmium) compared to the higher concentration (generated from 1.25 mg Cd/ml suspension) which showed only a soluble Cd fraction of about 15% (Ex. L-140-14). According to Dr. Oberdorster, the aerosols are more relevant to actual exposure than CdS concentrations in the suspension because it is the aerosol to which the test animals are actually exposed (Ex. 141). These results indicate that the low lighting conditions present during the chronic rat inhalation study could have resulted in about 50% of the lower concentration of 90 ug CdS/m(3) and about 15% of the higher concentration of 270 ug CdS/m(3) being converted into a soluble form of cadmium. (Ex. L-140-14).

Konig et al. also evaluated the potential for photodecomposition of pigment CdS using the same aerosol generating equipment and protocol as was employed in the long term carcinogenicity study by Oldiges et al. (Exs. L-140-3 and L-140-27-B). The CdS aerosol concentration was kept at approximately 90 ug/m(3). In about four weeks after starting the experiment, the concentration of cadmium ions in the suspension and in the aerosol reached a plateau at 43.5% and 35.8%, respectively.

These results indicate a slightly lower percentage of photodecomposition as compared to the Glaser photodecomposition study (L-140-14).

Konig et al. (Ex. (L-140-27-B) have also shown that CdS suspended in physiological saline (0.9%) solution at concentrations of 3.33 and 0.83 g Cd/l as applied in the intratracheal instillation studies with rats by Pott et al. (1987) led to solubilization of CdS (Ex. 8-757). The solubilized fraction of Cd within 24 hours was about 3% for the lower Cd concentration (0.83 mg Cd/ml) and about 1% for the higher Cd concentration (3.33 mg Cd/ml). The samples had been illuminated with 800 Lux from fluorescent lamps for 24 hours to approximate the conditions in the study by Pott et al. (1987). The lowest concentration of CdS suspension of 0.16 mg Cd/ml in these same experiments with saline gave the highest solubilization of about 13% within 24 hours.

As a result of the above findings, several questions related to the carcinogenicity of CdS are raised: Can the photodecomposition of CdS to CdSO(4) be solely responsible for the carcinogenic effect in rats observed in the Oldiges et al. study? Is inhaled CdS a pulmonary carcinogen? And, if so, how does its carcinogenic potency compare to that of other Cd compounds? Ulicny and Gagliardi (Ex. 141-1) concluded that the Glaser et al.

photodecomposition study demonstrated that inadvertent co-exposure to ionic cadmium (CdSO(4)) was sufficient to explain the carcinogenic response in the Oldiges et al. study and that the latter study could not be used to define a carcinogenic potency for cadmium pigment.

On the basis of their study results, Konig et al. were of the opinion that the observed lung tumor response in the Oldiges et al. study:

"cannot be attributed solely to the effect of CdS ....The real contribution of the CdS particles to the observed lung tumor rate after inhalation of a mixture of CdSO(4) and CdS depends exclusively on the bioavailability or solubility of CdS particles retained in the lungs" (Ex. L-140-27-B).

Konig et al. then cited studies by Klimisch et al., whereby they used a dry dispersion technique to exposed rats to CdS aerosols. This technique of preparation and administration of CdS make it unlikely that photodecomposition will occur because moisture and light are not present together. The study showed that Cd was retained in the lungs and was bioavailable since it was found in the kidneys.

Konig et al. also cited inhalation studies with rats and monkeys that measured pulmonary retention (Oberdorster and Cox, 1989) and concluded that the study "clearly showed" the bioavailability of CdS retained in the lungs. CdS in these latter experiments was suspended in ethanol and the exposure to the aerosol lasted for only 10 minutes. Since CdS in ethanol is much more stable than in water, Konig et al. assumed that almost no CdSO(4) was inhaled by the animals in the study and concluded:

"because of the high carcinogenic potency of Cd ++ ions in the lungs also very small amounts of dissolved CdS could lead to a tumor response....and therefore inhalable CdS has to be regarded as a probable human carcinogen." (L-140-27-B)

"because of the low solubility of CdS in the lungs and the relatively long biological half-time of inhaled particles in the human lung compared to the rat lung the carcinogenic risk of CdS dust for humans could be higher than expected on the basis of the rat data." (Ex. L-140-27-B)

The study by Ulicny and Gagliardi shows low in vitro solubility of CdS pigment following 30 days of suspension in a solution with a ph of 4.0--a ph similar to that in alveolar macrophages of the lung (Ex. 144-1). The study by Klimisch and Gembardt evaluated the clearance and excretion of CdS and its bioavailability in terms of renal uptake in Wistar rats (Ex. 151). They did not measure liver uptake. A dry dispersion technique was used to expose the animals by inhalation to 0.3 mg/m(3) CdCl(2), 0.2 mg/m(3) CdS, 1 mg/m(3) CdS and 8 mg/m(3) CdS. Including controls, five groups of 60 animals were exposed for up to 10 days and followed post exposure for up to three months. For CdCl(2), renal accumulation was 35% of lung clearance. For CdS, renal accumulation was 1% of lung clearance. This study again shows the bioavailability of cadmium as a result of inhalation exposure to CdS, however, the amount of accumulation is low compared to the that of CdCl(2). Nevertheless, the solubilization and bioavailability of cadmium from inhalation of CdS has been demonstrated in several studies.

Although the results by Konig et al. show slightly less photodecomposition as compared to the results by Glaser et al. from an aqueous suspension and aerosol related to 90 ug/m(3) of CdS, they nevertheless confirm the findings in the study by Glaser et al. (1991) who observed that CdS will be solubilized to a significant degree when kept in an aqueous suspension even under low light conditions over an extended period of time and that lower concentrations of CdS suspensions will lead to relatively higher solubilization rates.

Dr. Oberdorster considered the issue of CdS carcinogenicity partly from the standpoint of evaluating the lung cancer response in rats observed in the Oldiges et al. inhalation study (Ex. 141). Because of high mortality necessitating cessation of exposure in the groups of animals exposed to the three highest concentrations of CdS ( 270, 810 and 2430 ug/m(3) ), Dr. Oberdorster evaluated lung cancer response in the lowest CdS exposure group (90 ug/m(3)). In order to maximize the estimate of CdS pigment that may have become solubilized in the carcinogenicity study by Oldiges et al. (Ex. 8-964-D), Oberdorster used the data from Glaser et al. (Ex. L-140-14) rather than data from Konig et al. (Ex. L-140-27-B) and therefore assumed that 50% (rather than 38%) of CdS would photodecompose to CdSO(4) at the 90 ug/m(3) exposure level. Thus, the possibility exists that the lung cancer response rate in the Oldiges et al. study (75% in males and 85% in females) may have resulted from combined exposure to CdS and CdSO(4) of 45 ug/m(3) each. When these results were compared to the dose response data for CdCl(2) and lung cancer in the Takenaka study, Dr. Oberdorster was of the opinion that this high tumor response rate in the CdS exposed animals could have resulted from the 45 ug/m(3) of CdSO(4) that was produced by photodecomposition. On the other hand, he noted that animals exposed to 90 ug/m(3) of CdSO(4) had a tumor response rate similar to those exposed to 90 ug/m(3) CdS, i.e., 90 ug/m(3) of either cadmium compound resulted in the same cancer response. For this reason, he was also of the opinion that the tumor response rate in the group of rats exposed to 90 ug/m(3) of CdS could have also been the result of the remaining exposure of 45 ug/m(3) of CdS in combination with CdSO(4) if the effects of both compounds were additive. In other words, Dr. Oberdorster was also of the opinion that both cadmium compounds may have been responsible for the high tumor rate observed in the Oldiges et al. study since the high tumor rate placed this value in the flat part of the exposure response curve. Thus, he felt that evidence for the carcinogenicity of CdS could not be determined conclusively from this study.

Dr. Oberdorster also considered the issue of CdS carcinogenicity by evaluating the tumor response observed in rats as a result of administration of various cadmium compounds through intratracheal instillation in the Pott et al. (1987) study. Three separate dose levels for each cadmium compound were used in this study (Ex. 8-757). Animals were exposed to a total of 20 ug, 60 ug or 135 ug of CdCl(2) or CdSO(4) and to a total of 630 ug, 2500 ug and 10,000 ug of CdS. These doses of CdS would result in approximately 60 ug, 70 ug and 100 ug of CdSO(4) being formed through photodecomposition (Ex. 141). In his analysis, Dr. Oberdorster estimated the amount of CdS that would have photodecomposed to CdSO(4) in the Pott study based upon the Konig et al. (1991) study.

Note: Dr. Oberdorster overestimated the actual amount of CdS that would photodecompose to soluble CdSO(4) in his analysis because he assumed that the CdS suspension was exposed to light for 24 hours when he had information that Dr. Pott had kept the suspension in a dark refrigerator over night.

Dr. Oberdorster compared cancer response in the low and middle dose groups because there was early mortality in the high dose groups - over 50% of the animals in the high dose groups had died before the first tumor was observed in the study.

The tumor response (5.1%) in the low dose group of CdS exposed animals was not significantly different from the tumor response of the pooled low dose groups exposed to CdCl(2) and CdO (2.7%) or to the tumor response (6.2%) observed in the pooled middle dose groups of CdCl(2) and CdO exposed animals. Dr. Oberdorster then calculated that the tumor response (22.2%)in the mid-dose group exposed to CdS ( equivalent to about 70 ug of soluble CdSO(4)) was: (1) significantly greater than the tumor response (5.1%) in the low dose CdS group (equivalent to 60 ug of soluble CdSO(4)) and (2) significantly greater that the tumor response (6.2%) with data combined for the mid-dose groups exposed to 60 ug of soluble CdCl(2) or CdO.

Thus, in the presence of CdS, an increase of slightly less than 10 ug of exposure to CdSO(4) (60 to 70 ug) significantly increased the tumor response from 5.1% to 22.2%. In addition, exposure to slightly less than 70 ug of CdSO(4) in the presence of a high dose of CdS yielded a significantly higher tumor response (6.2% versus 22.5%) than observed with exposure to 60 ug of soluble CdCl(2) or CdSO(4). In Dr. Oberdorster's opinion, this significantly increased tumor response as a result of only a slight increase in exposure to CdSO(4) in the mid-dose CdS group as compared to the low dose CdS group, or in comparison to the mid-dose CdCl(2) or CdO group could not have been the result of exposure to only an additional 10 ug of CdSO(4). Therefore, in Dr. Oberdorster's opinion, exposure to CdS was the most likely cause of the increase in lung cancer in the mid-dose group exposed to CdS in this study. Thus, the Pott study provided evidence of a qualitative carcinogenic response as a result of exposure to CdS. Dr. Oberdorster stated:

"no firm conclusion about the pulmonary carcinogenicity can be drawn from the results of the study by Glaser et al. (1990). However, when the inhalation study and the instillation study are viewed together the evidence for carcinogenicity of CdS becomes stronger." (Ex. 141).

The Pott study, however, in Dr. Oberdorster's opinion could not be used to provide quantitative evidence of cancer potency because the cadmium compounds were administered by intratracheal instillation. Dr Oberdorster was of the opinion that CdS is probably less potent as a carcinogen in rats than the other cadmium compounds tested. He stated "to determine how much less this carcinogenic potency is requires a far better knowledge of the underlying molecular and cellular mechanisms of cadmium carcinogenicity than we have at present."

On the basis of their experimental results as presented above, Glaser et al. stated that:

"CdS solubilization may have contributed to the high lung tumor incidences in the CdS exposed rats. However, CdS particles still have a carcinogenic potential to the lung of rats as indicated by the results of Pott et al. (1987), who found an increased lung tumor incidence in rats after repeated intratracheal instillation of CdS particles." (Ex. L- 140-14)

Dr. Heinrich of the Fraunhofer Institute in Hannover also commented on the carcinogenic potential of CdS (Ex. 142). Based upon two lines of reasoning, he was of the opinion that CdS is carcinogenic. First, he noted that rats given two intraperitoneal injections of 0.125 mg of Cd as CdO developed a tumor rate of only 6.4% in the 1987 Pott et al. study (Ex. 8-757). He then stated that the high tumor rate (67%) resulting from 50 mg of cadmium as CdS in the study by Pott could only be explained by a high amount of CdS becoming solubilized in the peritoneal cavity over time. He reasoned that if the 50 mg suspension of cadmium as CdS had already contained appreciable quantities of cadmium ions at the time of administration, the animals would have shown toxic effects. Second, the findings of Konig et al. (Ex. L-140-27-B) of about 40% Cd ++ ion formation in aerosol generated from the 90 ug/m(3) CdS exposure in relation to the cancer incidence in rats exposed via inhalation to 90 ug/m(3) of CdS in the Oldiges et al. study (8-694-D) indicated to Heinrich that about 60% or 56 ug/m(3) of CdS was inhaled by rats in the Oldiges et al. study. As a result, he estimated that the amount of cadmium deposited in the rat lung per day as CdS would have been 2.1 ug. According to Heinrich, only 25% of this amount of cadmium has to be biologically available to reach the same lung burden with cadmium ions and possibly the same tumor rate (15.4%) as the rats exposed to 13.4 ug Cd/m(3) as CdCl(2) in the Takenaka et al. study. As a result of the above observations, Heinrich concluded:

"Thus there is no doubt that CdS retained in the lung will dissolve to some extent depending on the residence time or biological halftime of the inhaled CdS particles in the lung. Therefore, inhalable CdS aerosol has to be regarded as a probable human carcinogen. The longer the residence time of the CdS particle in the lung the more CdS becomes dissolved and the higher will be the carcinogenic potency. As we know that the biological halftime of particles with low solubility in the human lung is about ten times longer compared to the rat lung, the carcinogenic potency of inhaled CdS aerosols is expected to be higher for humans than for rats." (Ex. 142)

In other words, the same dose of CdS to the human lung may cause a greater carcinogenic response than in the rat lung because of longer retention time in the human lung. However, this does not imply that CdS would be a more potent carcinogen than other more soluble cadmium compounds based on the ionic theory of carcinogenesis. Indeed, DCMA (Ex. 144-20) pointed out that all of the cadmium compounds had a similar retention time in the rat lung and that the study by Dr. Oberdorster and Cox (Ex. 31-A) indicated that all of the cadmium compounds are retained in the lungs of primates 10 times longer than in rodents.

Thus, it seems reasonable to assume that the steady state concentration and the resulting retained dose of CdS particles from a comparable exposure concentration will be higher in the human lung than in the rat lung. Moreover, the longer residence time of CdS particles to be released in the human lung means that there is more time than in the rat lung for cadmium ions to be released by dissolution from the CdS particles (Ex. 142). Both of these biological factors lead Dr. Heinrich to conclude that from a comparable exposure concentration, the resulting carcinogenic effect of CdS will be higher in the human lung than in the rat lung (Ex. 142) even though the overall carcinogenic effect of CdS may be less as compared to other more soluble cadmium compounds.

Based upon the ionic theory of carcinogenesis for cadmium compounds and data related to solubility and bioavailability of CdS in rats, Dr. Heinrich (Ex. 142), Dr. Oberdorster (Ex. 141) and Drs. Konig et al. (Ex. L-140-27-B) were all of the opinion that CdS-exposed rats are likely to develop a lower lung tumor rate than CdCl(2) or CdO-exposed rats. Neither Dr. Heinrich nor Dr. Oberdorster, however, could give an estimate of the carcinogenic potency of CdS to humans because of lack of knowledge of the mechanisms involved. For example, Dr. Heinrich stated that we do not:

".... know how many cadmium ions are actually necessary to induce the observed carcinogenic effect. We also do not know whether metallothionein-bound cadmium ions in the lung can be remobilized and are thus also available for a carcinogenic effect." (Ex. 142)

With regard to carcinogenic potency of CdS, Dr. Oberdorster stated:

"To determine how much less this carcinogenic potency is requires a far better knowledge of the underlying molecular and cellular mechanisms of cadmium carcinogenicity than we have at present. I can think of several hypothetical mechanistic scenarios which would implicate CdS as being a direct or indirect pulmonary carcinogen." (Ex. 141)

Dr. Oberdorster went on to say that if ionic cadmium is the ultimate carcinogen, then long term in vivo solubilization rates in the lung may permit the estimate of a potency factor for bioavailability of Cd ++. If, on the other hand, an intrinsic toxic particulate effect contributes to the carcinogenic effect of CdS as it may do with respect to toxic effects in the lungs, then in vivo solubility is not a good indicator of carcinogenic potency. He also stated the possibility of a combined effect of solubilized cadmium on further in vivo CdS solubilization and retention due to effects on lung cell function as a possible mechanism.

In summary, DCMA requested a separate health standard for cadmium pigments (Ex. 144-20). The basis for this opinion is that the scientific studies of carcinogenicity by Oldiges et al. and by Pott et al. are flawed because the CdS administered to the animals was subjected to light and as a result the material photodecomposed to CdSO(4), which was responsible for the cancer response.

After reviewing the new photodecomposition studies and comments about them as mentioned above, OSHA is of the opinion that the photodecomposition of CdS to CdSO(4) may have played a role in the cancer response observed in the Oldiges et al. and Pott et al studies. However, photodecomposition of CdS to CdSO(4) could not account for the tumor response observed in the mid-dose group of the Pott et al. study as pointed out by Oberdorster (Ex. 141) and discussed above. If CdS was responsible for the significant increase in the cancer response observed in the mid-dose group, it is reasonable to conclude that CdS has a carcinogenic potential and contributed to the cancer response observed in the inhalation cancer study by Oldiges et al.

After reviewing all of the studies and comments on the issue of CdS carcinogenicity, OSHA agrees with the commentors that the CdS preparations used in animal carcinogenicity studies photodecomposed to varying degrees which depended upon the concentration administered and the amount of light and moisture present. The exact role of this photodecomposition in the quantitative carcinogenic response observed in the various cancer studies, however, cannot be determined. Nevertheless, evidence was presented during the rulemaking that CdS is a probable human carcinogen. This opinion of OSHA is derived from a combination of the following observations:

(1) The analysis of Dr. Oberdorster showing that an increase of only 10 ug of CdSO(4) in the presence of CdS resulted in a statistically significant increase in lung tumor response in the Pott et al. study;

(2) The Oldiges et al. study demonstrating a significant increase in lung cancer in rats exposed to CdS by inhalation;

(3) The above mentioned analysis by Heinrich indicating that the lung tumor response among animals given 50 mg of cadmium as CdS in the Pott et al. study can only be explained by a high solubility of CdS in the peritoneal cavity over time;

(4) study results indicating that administration of CdS leads to absorption into the body;

(5) lung retention of CdS aerosol is estimated to be 10 times greater in the human lung than in the rat lung and increases the likelihood of systemic absorption in humans.

With the exception of Mr. Ulicny, all of the investigators involved in the actual research related to CdS photodecomposition or carcinogenicity who offered comments (Glaser et al.; Konig et al.; Heinrich; Oberdorster) were of the opinion that CdS was carcinogenic even though photodecomposition to CdSO(4) may have played a role in the carcinogenic response observed in the animal cancer studies (Exs. L-140-14; L-140-27-B; 142; 141). None of the latter group of investigators was of the opinion that CdS was not carcinogenic.

The animal inhalation studies with the various cadmium compounds by Oldiges et al. indicate a similar carcinogenic response even though the photodecomposition studies raise the possibility that part of the cancer response with CdS may have been due to the photodecomposition of CdS to soluble CdSO(4). The analysis of the Pott et al. data by Dr. Oberdorster indicates that a large part of the carcinogenic response in the animals had to be attributed to CdS since it could not be attributed solely to CdSO(4) (Ex. 141). Thus, it is also possible that a large portion of the carcinogenic response observed in the Oldiges et al. inhalation study of CdS could be attributed to CdS. Therefore, it is also reasonable to conclude that the cancer response of rats in the Oldiges et al. inhalation study could not be attributed entirely to the formation of ionic cadmium through photodecomposition. How much of the cancer response may have been due to the photodecomposition, however, cannot be determined with the data currently available because information on molecular and cellular mechanisms involved in the carcinogenicity of the various cadmium compounds is not known.

It was pointed out that an intrinsic toxic particulate effect could contribute to the carcinogenic effect of CdS and if this is so, in vivo solubility is not a good indicator of carcinogenic potency of cadmium pigments. It was also pointed out that CdS may have acted in an additive manner with the amount of CdSO(4) formed through photodecomposition to induce cancer.

Evidence also indicates that CdS will be retained in the human lung 10 times longer than in the rat lung making it likely that the potency from inhaled CdS will be greater for humans than for rats. If one accepts the ionic theory of cancer for cadmium compounds, CdS pigments may be less carcinogenic, but it is not possible to determine the magnitude of the difference in potency when the results are extrapolated to humans. An intrinsic particulate effect could play a role in the development of cancer with the cadmium pigments and there could also be a combined effect of solubilized cadmium on further in vivo CdS solubilization and retention in relation to lung function.

Thus, OSHA is of the opinion that CdS is an occupational carcinogen. With the data currently available, however, it is not possible to determine whether cadmium sulfide has a different carcinogenic potency from other cadmium compounds though it is possible that it may be less potent. Therefore, with regard to carcinogenicity, CdS will be treated similar to other cadmium compounds.

DCMA (Ex. 144-20) has argued that a conclusion that CdS should be regulated as an occupational carcinogen does not conform with OSHA's Cancer Policy. However, in addition to the evidence of carcinogenicity presented above, Sec. 1990.111 (c) of the Cancer Policy (45 FR 5002-5296, Jan 22,1980) allows the Agency to regulate groups of substances, or combinations of substances, or mixtures of substances found in the workplace. Thus, OSHA may regulate a group of substances on the basis of the scientific evidence available on a single member of the group. In the arsenic standard, OSHA regulated all pentavalent arsenic compounds along with all trivalent arsenic compounds as carcinogens even though evidence of carcinogenicity for the former compounds was based on a single study of pentavalent arsenic exposure which was limited in study design and methodology. In the arsenic standard, OSHA relied upon expert opinion about the study results for the carcinogenicity of pentavalent arsenic, the evidence on the carcinogenicity of trivalent arsenic compounds, as well as on the Supreme Court's benzene decision (I.U.D. v. A.P.I. 448 U.S. 607) that OSHA was free to use conservative assumptions to err on the side of worker protection. Inclusion of pentavalent arsenic compounds in the arsenic standard was upheld by the Court of Appeals in ASARCO, Inc. v. OSHA (746 F. 2d 483 (1984)), which agreed that it was appropriate to utilize evidence of the carcinogenicity of trivalent arsenic compounds in determining the carcinogenicity of pentavalent arsenic compounds. Thus, OSHA believes that it has a scientific basis as well as a judicial basis for including CdS in the current standard and for establishing the same PEL for cadmium pigments that will be set for all other inorganic cadmium compounds. In any event, OSHA notes that all of the inorganic salts of cadmium, including CdS, tested for carcinogenicity by inhalation in the rat produced a highly carcinogenic response.

As discussed in the section of this preamble dealing with CdS carcinogenicity, record evidence and expert opinion leads the Agency to conclude that CdS, in and of itself, is an occupational carcinogen. The final issue raised is whether or not CdS has the same carcinogenic potency as the other cadmium salts. Although the cancer test results showed a similar tumor response for CdS as for other cadmium compounds tested, because of the possibility of photodecomposition, several experts testified that it is not possible with scientific certainty to determine the relative carcinogenic response that could be attributed to CdS as distinguished from CdSO(4). Dr. Oberdorster, for example, stated that an additive effect between CdS and CdSO(4) in the carcinogenic response cannot be excluded and that until more and better data become available, it would not be advisable to establish a different standard for CdS (Ex. 141). Thus, record evidence and expert opinion leads the Agency to conclude that CdS should be considered an occupational carcinogen and have the same PEL as that established for other cadmium compounds.

3. Evidence in Humans - Introduction

Extensive study of five cohorts with occupational exposure to cadmium has found an excess of lung cancer among cadmium exposed workers (Ex. L-140-50). The mortality experience of these workers has been studied repeatedly. The five cohorts are comprised of workers at a cadmium smelter in the U.S. (Thun, Exs. 4-68; 8-658a; Lemen, Ex. 4-51 and Varner, Ex. 8-649); workers at two cadmium battery plants in the United Kingdom (Armstrong and Kazantzis Ex. 8-603; Kipling and Waterhouse, Ex. 4-45; Sorahan, Ex. 4-65; and Sorahan, Ex. 12-12-A); workers from 17 different plants using cadmium in the United Kingdom (Armstrong and Kazantzis, Ex. 8-603; Armstrong and Kazantzis, Ex. 8-565; and Kazantzis et al., Ex. 8-684); workers at a cadmium-alloy plant in the United Kingdom (Holden, Ex. 4-39; Armstrong and Kazantzis, Ex. 8-603); and workers at a nickel-cadmium battery plant in Sweden (Kjellstrom et al, Ex. 4-48; Elinder et al, Ex. 4-25; Jarup et al, referenced in L-140-50). Evaluation of these data is complicated by the fact that the same populations have been examined repeatedly, sometimes by different groups of investigators. For example, the cohort of workers from 17 plants included members of the the cohort at the two cadmium battery plants (Exs. 8-603 and 12-12-A).

Elinder et al. evaluated 13 studies of these cohorts and concluded that in several studies, workers with high exposures were combined with workers with low exposures into one exposure group. This would reduce the ability of these studies to detect an effect due to cadmium exposure (Ex. 4-25). In addition, Elinder concluded that in the largest study (Ex. 8-684), most workers had such low cadmium exposures that cadmium-associated cancer would not be induced. After evaluating the studies, Elinder combined the data from them and found an overall lung cancer SMR of 121 (Obs.=195; Exp.=161.4; p=0.008, two-tailed).

Elinder also found that 12 of the 13 studies reported excess cancers of the prostate, and in 4 of these, the excesses were statistically significant. Elinder noted that the median SMR for prostate cancer from all of the studies was 167, but when the number of observed and expected cases are combined for the most recent updates of the 6 independent studies, (7 of the 13 studies were updates of earlier studies), the statistically significant SMR for prostate cancer for all cohorts is 162 (Obs=28, Exp=17.2, p< .02, two-tailed).

Elinder concluded, "Our interpretation is that the accumulating data on the mortality of cadmium workers with high exposure levels in the past (above 0.3 mg/m(3)) support an association between lung cancer and cancer of the prostate and exposure to cadmium", (Ex. 4-25). Thun et al. subsequently evaluated studies of these these same five cohorts including updates of studies of four of the five cohorts (Ex. L-140-50). The authors' analysis of the prostate cancer data lead them to conclude that while mortality from prostate cancer is slightly increased in several of these industrial cohorts, the number of excess cases is small and there is no clear dose-response relationship with exposure. Several other researchers have concluded that the evidence for an association between cadmium exposure and prostate cancer is limited or decreasing (Exs. 19-43-A and 19-29).

Analyzing the lung cancer data from these studies, Thun et al found that these updates reported a statistically significant increase in mortality from lung cancer in cadmium smelter workers, with two or more years of employment (SMR=137, Obs=24, Exp=10.76, 95% CI= 143-332 (Thun, Ex. 33)); a statistically significant increase in mortality from lung cancer in nickel-cadmium battery workers in the U.K. (SMR=130, Obs.=110, Exp.= 84.5, 95% CI=107-157 (Sorahan, Ex. 12-12-A)); a statistically significant increase in mortality from lung cancer in workers at 17 plants combined in the U.K. (SMR=115, Obs.=277, Exp.=240.9, 95% CI=101-129 (Kazantzis, Ex. 8-684)); and a statistically significant increase in mortality from lung cancer in Swedish nickel-cadmium battery workers (SMR=241, Obs.=14, Exp.=5.8, 95% CI=132-405 (Jarup et al, referenced in Ex. L-140-50)). Thus, in each of these updates a statistically significant excess of lung cancers was observed.

These studies provide qualitative evidence of the carcinogenic effects of cadmium on the human lung. In several of these studies, there are indications of a dose-response relationship between cadmium exposure and risk (Ex. L-140-50). For example, in three cohorts (Exs. 33, 12-12-A, and 8-684), the SMR for lung cancer increases either with length of employment or cumulative exposure to cadmium (Ex. L-140-50).

4. Studies of the U.K. Nickel-Cadmium Battery Factory Cohort

One of the earliest cohort studies was by Kipling and Waterhouse who observed four cases of prostate cancer among a cohort of 248 men employed in a British nickel-cadmium battery factory (Ex. 4-45). Exposure was to cadmium oxide dust. The observed number of prostate cancers was more than seven times greater than the expected number of prostate cancers (Exp=0.58, p=.003) calculated using incidence rates from a regional cancer registry.

In a subsequent study of these workers, Sorahan and Waterhouse observed a statistically significant excess of respiratory cancer (Obs=89; Exp=70.2; SMR=127; p< .05)(Ex. 4-65). An excess of prostate cancer was again observed, but this time was not statistically significant (Obs=8; Exp=6.6; SMR=121).

To assess the relationship between cadmium dose and mortality, the authors devised two measures of cadmium exposure. The first measure was "cumulative duration of employment in high exposure jobs," and the second measure was "cumulative duration of employment in high or moderate exposure jobs." Using the method of regression models in life tables, the authors found that cumulative duration of employment in high exposure jobs was significantly related to prostate cancer mortality but only when the four original cases described by Kipling and Waterhouse were included in the analysis. The measure cumulative duration of employment in high exposure jobs was not statistically significantly associated with lung cancer mortality, but the measure cumulative duration of employment in high or moderate exposure jobs did show a statistically significant relationship to lung cancer mortality. The authors cautioned, however, that this observed effect could be confounded by oxyacetylene fume exposure.

Workers at this factory were studied once again by Armstrong and Kazantzis, who conducted a case-control study of workers who had died of prostate cancer, renal cancer, bronchitis or emphysema, or nephritis or nephrosis (Ex. 4-19). Cases were selected from three cohorts of British workers exposed to cadmium. All of the cohorts had been studied previously. Cohort C1 was comprised of workers from a lead-zinc-cadmium smelter previously studied by Armstrong and Kazantzis (Ex. 8-565). Cohort C2 was comprised of workers from the nickel-cadmium battery factory studied by Sorahan and Waterhouse (Ex. 4-65). Cohort C3 was comprised of workers from a copper-cadmium alloy plant previously studied by Holden who had found statistically significant excess of prostate cancers (Ex. 4-40). Cases consisted of workers who had died of prostate cancer, chronic respiratory disease or renal disease. Only men born before 1940 with at least one year employment before 1970 were included. For each case, 3 controls were selected matched by plant, age, and, as nearly as possible, date of birth.

The authors divided these cohorts into three groups: always low cadmium exposure; ever medium cadmium exposure; and ever high cadmium exposure. They found that the odds of prostate cancer for the ever medium or ever high exposure groups were elevated relative to the always low exposure groups (1.55 and 1.35 respectively), but neither of these odds ratios were statistically significant. The authors noted, however, that the small number of prostate cancer cases makes interpretation of this finding difficult.

In 1987, Sorahan updated his study of the nickel cadmium battery workers (Ex. 12-12A). Twenty-two additional deaths from lung cancer were reported. According to the author, there was some evidence of an association between risk of death from lung cancer and duration of employment in high or moderate (or slight) exposure jobs for "early workers", (i.e. first employed before 1946), but none for "late workers" (i.e. first employed after 1946). A significant increase in lung cancer was observed for the entire study cohort (110 Obs., 84.5 Exp., p< .01). Sorahan did not report a statistically significant increase in lung cancer for his cohort when workers were divided into "early workers" and "late workers", but OSHA's analysis shows that there was a significant excess of lung cancers for the "late workers" (45 Obs., 33 Exp., p< .05 - one tail).

Among "late workers", the SMRs for lung cancer were observed to increase with years from first employment. Because this trend was not observed for "early workers", Sorahan suggested that the there might be selection bias for the "early workers" and that this sub-cohort may be incomplete. The study's inability to demonstrate a significant relationship between duration of employment and lung cancer risk, however, does not mean that there is no association between cadmium exposure and lung cancer risk. Duration of exposure may not be a surrogate for dose, particularly when the length of exposure periods are not adjusted for the particular years in which the exposure occurs. The observed excess of lung cancer deaths among the "late workers" supports an association between cadmium exposure and lung cancer.

5. Studies of the 17 U.K. Plant Cohort

Ades and Kazantzis conducted a study of lung cancer in non-ferrous smelter workers (Ex. 12-14C). This cohort of men employed in a lead-zinc-cadmium smelter was part of Cohort C1 in the Armstrong and Kazantzis study described above (Ex. 4-19). The authors found a significant excess of lung cancer deaths among the entire cohort (182 Obs., 146.2 Exp., p< .005). In subcohorts of workers, a significant excess of lung cancer deaths was observed for workers with 20 to 29 years of employment (44 Obs., 23.1 Exp., p< .005) and for workers with 40 or more years of employment (8 Obs., 2.74 Exp., p< .02).

SMRs for lung cancer death were observed to increase with duration of employment for the cohort. This linear trend was statistically significant. The risk of lung cancer for workers with more than five years of employment relative to the risk for workers with less than five years of employment was also observed to increase with duration of employment. Using a matched logistic regression analysis, the authors were able to associate this increasing risk with exposure to arsenic and lead but not cadmium. This finding, however, could be due in part to the study protocol for choosing controls. Cases and controls were matched by date of hire, but because controls were required to have ten years of follow-up and to survive the matched case, cases and controls may have been inadvertently matched on cadmium exposure as well.

The entire Armstrong and Kazantzis cohort was studied again by Kazantzis and associates (Ex. 8-684). In this update, the authors followed the workers for an additional five years. Seventy-five additional cases of lung cancer were observed, resulting in a significant excess of mortality due to lung cancer for both the additional five year period (SMR=134; 95% CI=103-164) and the entire study period (Obs=277; Exp=240.9; SMR=115; 95% CI=101-129).

The increased lung cancer risk occurred mainly among those first employed before 1940, and the risk increased with length of employment and length of follow-up. The majority of lung cancer deaths were among workers employed in the non-ferrous smelter studied by Ades and Kazantzis. This worksite provided over 60% of the total study population, but its workers' exposures were characterized only as low or medium. No exposures in the smelter were characterized as high.

Over the entire study period, there was a statistically significant excess of mortality due to stomach cancer (Obs=98; Exp=70.6; SMR=139; 95% CI=111-166). Of the 98 deaths observed, 22 occurred during the five years of added follow-up, giving a statistically significant excess of stomach cancer mortality for that five year period (SMR=179; 95% CI=112-271).

Dr. Kaztanzis testified that there were other major illnesses, in addition to cancer of the lung, considered a priori to be possibly related to cadmium exposure that needed to be evaluated in this large cohort study. These included chronic bronchitis (chronic obstructive airway disease) and emphysema, among others (Tr. 6/8/90).

According to Dr. Oberdorster, who testified at the hearing, OSHA did not discuss the important finding of the Armstrong and Kazantzis study, that there was a high risk of dying from bronchitis in the group of "ever high" Cd-exposed workers.

According to Dr. Oberdorster, this is an important health effect of Cd exposure, reconfirmed in an updated study by Kazantzis, and it deserves more attention (Tr. 6/6/90, pp. 153-246) because: "Increased cell proliferation rates which can be assumed to be present in the workers with chronic bronchitis could indeed be an important risk factor in carcinogenesis (Ex. 31, Attachment D)." Furthermore, Dr. Oberdorster stated, "Dying from bronchitis is comparable to dying from lung cancer." (Ex. 31, p. 4) The update included the diseases considered to be important by Dr.

Kazantzis. The results of the study confirmed a significant excess risk from bronchitis related to intensity of exposure. The SMR for workers classified as having "ever high" exposures was 382 (Obs.=13; Exp.=3.4, 95% CI=203-654). The SMR for workers classified as having "ever medium" exposures was 146 (Obs.= 25, Exp.=17.1, 95 % CI = 94-215. The SMR for workers classified as having always low exposures was 123 (Obs.=140, Exp.=114.3, 95% CI = 102-143). The bronchitis SMR for the entire cohort was statistically significantly elevated (SMR = 132, Obs.= 178, Exp.= 134.9, 95% CI = 113-151).

In response to questions during the hearing about the marked excess of mortality from bronchitis which showed a strong relation to both intensity and duration of exposure that was observed in his follow-up study, Dr. Kazantzis agreed that the dose response is cadmium-exposure-related. Dr. Kazantzis stated that:

"I find it very difficult to account for that in any other way other than the cadmium exposure.....our study has...a very high proportion of low exposed workers, nevertheless, we have found this very marked dose response relationship. (Tr. 6/8/90)"

The Cadmium Council has argued that the studies by Kazantzis "failed to establish a clear association between cadmium exposure and lung cancer," in part because the studies failed to adequately control for exposure to other carcinogens such as arsenic and nickel (Ex. 19-43). The Council questioned the link between cadmium exposure and prostate cancer because no cases of prostate cancer were observed in the medium or high-exposure groups. The Kazantzis 5-year update also found no prostate cancer in these exposure groups (Ex. 8-684, p. 18).

A new update of this cohort is currently underway (Ex. L-140-50, pp. 701-702). According to Dr. Kazantzis:

"The mortality experience of the cohort is currently being followed up for an additional 5-year period. Preliminary analysis of mortality from lung cancer for the first three of five years confirmed a significantly increased lung cancer risk for the total study period from 1942-1987, but a significant excess lung cancer risk was no longer seen in the high exposure group (Kazantzis, 1990)."

At present, however, this study has not been published and the paper has not been submitted into the record for analysis of the methodology used in the study.

6. Studies of the Swedish Cadmium-Nickel Battery Factory Cohort

In an update of an earlier study by Kjellstrom et al (Ex. 4-48), Elinder et al examined the mortality experience of 545 male workers at a Swedish cadmium-nickel battery factory (Ex. 4-25). While no statistically significant excess of mortality due to any type of cancer was observed, the authors reported that the SMRs for cancers of the lung, prostate, and bladder increased with time since initial exposure (i.e. latency) among workers with at least 5 years of exposure. Thus, for lung cancer, the SMR, was 133 for the entire cohort, but for workers with at least five years of exposure, the SMR was 163 after 10 years latency and 175 after 20 years latency. For prostate cancer, the SMR was 108 for the entire cohort, but for workers with at least 5 years of exposure, the SMR was 125 after 10 years latency and 148 after 20 years latency. For bladder cancer, the SMR was 181 for the entire cohort, but for workers with at least 5 years of exposure, the SMR was 222 after 10 years latency and 250 after 20 years latency.

This study was updated in 1990 by Jarup et al., (referenced in Ex. L-140-50). A statistically significant increase in lung cancer mortality was reported among nickel-cadmium battery workers, using regional rates and 20 years latency (SMR = 232).

7. Studies of Two U.K. Copper-Cadmium Alloy Plant Cohorts

A mortality study of 330 men employed in two factories manufacturing copper alloys was conducted by Holden in 1980 (Ex. 4-39). Plant A was in operation from 1922 to 1966, and plant B was in operation since 1925. Holden reported that 104 men in his cohort had been medically evaluated over time, and in 1953, 22% were found to have chronic cadmium poisoning; eleven workers had both emphysema and proteinuria; eight had proteinuria alone; and, four had emphysema alone. Holden's mortality study indicated that in plant B, the respiratory cancer SMR [International Classification of Diseases, 8th Revision (ICD-8); ICD 160-163] was 222 (Obs.= 8; Exp.=4.5) using the population of England and Wales as a comparison population. When death rates for the urban district in which the plant is located were used, the SMR was 167 (Obs.=10; Exp.=6).

The mortality experience of these workers was subsequently followed-up by Armstrong and Kazantzis in 1982 in a report to the International Lead Zinc Research Organization (Ex. 8-603). In that report, the authors included those cases from the study performed by Holden which satisfied the selection criteria of the Armstrong and Kazantzis study. The lung cancer SMRs for this group of workers were 87 for the control group, 114 for the medium exposure group, and 72 for the high exposure categories. The mortality experience of these workers was combined with that of workers from 16 other plants. The results of the follow-up study are included in the discussion of the studies of the 17 U.K. plant cohort above.

8. Studies of the U.S. Cadmium Smelter Cohort (Globe)

One of the strongest studies supporting the evidence of the carcinogenicity of cadmium in humans comes from the mortality study of cadmium smelter workers at the Globe plant in Denver, Colorado. Previous studies of workers in this plant were conducted by Varner (Ex. 8-649) and by Lemen et al (Ex. 4-61).

This population was first studied by Lemen et al (Ex. 4-61) who found a statistically significant excess of deaths due to malignant neoplasms (i.e. cancer). Lemen's study population consisted of 292 white male workers with a minimum of two years employment between 1940 and 1969. Of these workers, 27 died of malignant neoplasms whereas only 17.57 deaths from this cause were expected (SMR=154; p=.05). Twelve of the 27 deaths were due to respiratory cancer whereas only 5.11 were expected (SMR=235), and this excess was also statistically significant (p< .05). Focusing of the lung cancer incidence, Lemen found that the rates increased with time since first exposure and that the highest rates were observed in workers with more than 30 years of follow-up. Lemen also found an excess of deaths due to prostate cancer, but this excess was statistically significant only when the analysis was restricted to workers with at least 20 years since first exposure (Obs=4; Exp=1.15; SMR=347; p< .05). Lemen's study was followed by a study by Varner (Ex. 8-649). The Varner cohort consisted of 644 workers with at least six months of employment at the Globe smelter between 1940 and 1969. The cohort was followed through 1981. Mortality data was analyzed using Standardized Cause Ratios (SCRs).

Statistically significant excesses of mortality due to lung cancer, rinary tract cancer, specific bladder cancers, and total cancers were observed. Mortality due to prostate cancer was elevated, but the excess was not statistically significant.

Cumulative cadmium exposures were estimated for each member of the cohort using personal monitoring measurements made from 1973 through 1976. Exposures measured during this period were assumed to be constant for the entire period of study. The cohort was divided into a low exposure group (0-4 mg/m(3)-years), a middle exposure group (5-15 mg/m(3)-years), and a high exposure group (16+ mg/m(3)-years). The observed SCRs for lung cancer deaths for each exposure group were: 95 for the low dose group, 159 for the middle dose group, and 332 for the high dose group. The observed SCRs for all cancer deaths for each exposure group were: 108 for the low dose group, 123 for the middle dose group, and 168 for the high dose group. A dose-response relationship was observed between cadmium exposure and lung cancer and between cadmium exposure and total cancers.

Varner attributed the observed excess of lung cancer deaths to arsenic exposure and cigarette smoking but did not present any data on smoking and arsenic exposures by dose group. There was, therefore, little reason to assume that these confounders did not affect all three exposure groups to at least some, if not the same, degree. There was no evidence that smoking was more common among cadmium workers than among the general population and therefore that the observed lung cancer excess was associated with increased smoking.

Varner's study was followed up by Thun et al, (Ex. 4-68). Originally, Thun followed 602 white males who had spent at least 6 months in a production area of the smelter between 1940 and 1969. Workers were followed through 1978. The mortality status of all but 12 workers (2%) was determined; 411 were still alive (69%) and 179 had died (29%). Deceased workers for whom no death certificate was located were assumed dead, as specified in the protocol, with the cause of death unknown. Persons lost to follow-up were assumed to be alive.

From 1886 to 1919, the Globe plant was a lead smelter, and from 1920 to 1926, it was an arsenic smelter. Twenty-six of the 602 workers had been hired prior to 1926. These workers were omitted from subsequent analyses. Most analyses were limited to the remaining 576 workers.

Worker exposures were estimated by Smith et al., who based his estimates on historical area monitoring data adjusted to reflect the actual exposures of workers wearing respirators (Ex. 4-64). Using Smith's exposure estimates and company personnel records, Thun calculated cumulative dose estimates for each worker in his cohort.

Thun analyzed his data using a modified life-table method developed by National Institute of Occupational Safety and Health (NIOSH). Expected rates were calculated from the U.S. population and were adjusted for age, sex, race, and calendar time. Both standardized mortality ratios (SMRs) and standardized risk ratios (SRRs) were examined. To analyze his data by cumulative exposure, Thun divided his cohort into three groups chosen prior to analysis of the data. These groups corresponded to the recommended exposure limits that were in existence at the time the study was conducted. The low dose group consisted of workers with cadmium exposure at 584 u/m(3)-days or less, the equivalent to forty years of exposure at less than or equal to 40 u/m(3). This cumulative dose corresponded to the NIOSH recommended exposure level at that time. The upper limit of the middle dose group, i.e., 40 years of exposure between 41 and 200 u/m(3), was chosen to correspond to the upper limit allowed by OSHA. The high dose group was chosen as 40 years of exposure at greater than 200 u/m(3). Thun also identified for separate analysis a subset of the low exposure group of his cohort in which the 40-year TWA equivalent exposures ranged from 21-40 u/m(3).

Thus, the update of this cohort by Thun et al., (Exs. 4-68 and 8-658a) included estimates of cadmium exposures and an evaluation of the mortality experience of the workers in the cohort by SMRs per dose group. An excess of lung cancer mortality was observed in relation to cadmium exposure.

Forty-three percent of the workers had less than 2 years of employment. Follow-up time was long; 82.5% had more than 20 years of follow-up and 66.3% had more than 30 years of follow-up. Among the entire cohort of 602 workers, a statistically significant excess of deaths due to respiratory cancer (Obs=20; Exp=12.15; SMR=165; CI=101-254) and deaths due to non-malignant gastrointestinal disease (Obs=9; Exp=2.35; SMR=383; CI=175-727) was observed. All deaths due to lung cancer occurred in workers with more than two years of employment. When the analysis was restricted to the 576 workers hired after 1926, the excess of lung cancer death was no longer statistically significant (Obs=16; Exp=10.88; SMR=147). When the analysis of these 576 workers was further restricted to those workers with two or more years of employment, the observed excess was statistically significant (Obs=16; Exp=7.00; SMR=229; CI=131-371).

Analysis of all the 576 workers hired after 1926 indicated that the incidence of lung cancer death increased with dose. A statistically significant dose-response relationship existed between cumulative exposures to cadmium and lung cancer mortality. Among the low dose group, there was a non-significant deficit, i.e. lower than expected, of lung cancer deaths (Obs=2; Exp=3.77; SMR=53; p=0.28; SRR=0.48; See Table VI-14). Among the subset of the low exposure group with exposures equivalent to 21-40 u/m(3) over 40 years, the lung cancer SMR was 100 and the SRR was 96. For the middle dose group, a non-significant excess of lung cancer was observed (Obs=7; Exp=4.61; SMR=152; SRR=1.55). For the high dose group, the excess of lung cancer deaths was statistically significant (Obs=7; Exp=2.5; SMR=280; CI=113-577; SRR=3.45). Thun reported that this dose-response trend was also observed when the analysis was restricted to workers with more than 20 years since first exposure. The regression slope of the SRR for lung cancer was statistically significant indicating that an increase in cadmium exposure is producing a real increase in the risk of lung cancer.

Thun et al. also observed a significant increase in death from non-malignant gastrointestinal disease (NMGID), 9 observed versus 2.35 expected. The death certificates for six of these individuals suggested peptic ulcer disease. For those hired after 1926, there was a significant linear trend between increased cadmium exposure and the SRR from NMGID. The authors thought this observation was noteworthy in light of previously reported associations between cadmium exposure and severe gastrointestinal irritation in humans.

A non-statistically significant excess of genitourinary cancer was observed for the entire cohort first employed after 1926 (Obs=6; Exp=4.45; SMR=135; CI=49-293). Three of these deaths were from prostate cancer. The observed mortality from prostate cancer exceeded the expected, but the excess was not statistically significant (Obs=3; Exp=2.2; SMR=136). There were two other cases of prostate cancer, however, which Thun did not include in his analysis. One of these was a death from prostate cancer which occurred in a guard who had not spent 6 months in a production area of the smelter. The second case was not included because prostate cancer was not the underlying cause of death.

Subsequent to the publication of his study in 1985, an analysis based on an updated cohort was presented by Dr. Thun at a workshop on cadmium and cancer in London, England (Ex. 8-658a). The updated study of the same Globe cadmium smelter included an evaluation of the mortality experience of the cadmium workers through 1984 (Ex. 8-658a). The extended follow-up study included 625 white male workers with six or more months in a production area between 1940 and 1969. The cohort was essentially the same as that previously described except for the addition of 23 new workers who met the eligibility criteria. These additional new workers included in the follow-up through 1984 were identified through further examination of the records of the Globe facility (Tr. 6/7/90, p. 108).

Forty-three percent of the workers had less than 2 years of employment. Follow-up time was long; 85% had more than 20 years of follow-up. By December, 1984, 234 workers were deceased; this represented fifty-five additional deaths since the previous NIOSH study. Thun et al found that, among workers with two or more years of employment hired after 1926, a statistically significant excess of deaths due to respiratory cancer persisted (Obs=24; Exp=10.7; SMR=223; CI=143-332). No lung cancer deaths occurred among workers who were employed for less than two years. The excess of deaths from lung cancer was statistically significant for the new period of observation, 1979-84 (Obs=8; Exp=3.19; SMR=251; 90% CI=108-494) (Ex.8-658a). The excess of deaths due to prostatic cancer was statistically non-significant (Obs=4; Exp=2.35; SMR=170; CI=46-436).

Vital status was known for all but fifteen members, or for 98%, of the updated population study (8-658a). Deceased workers with no death certificates were assumed dead, as specified in the protocol, with the cause of death unknown. Persons lost to follow-up were assumed to be alive. Vital status determination was, therefore, adequate for epidemiological analyses.

With the updated cohort, Thun again analyzed his data using a modified life-table method developed by NIOSH, with expected rates being calculated from the U.S. population, and adjusted for age, sex, race, and calendar time. To analyze his data by cumulative exposure, Thun again divided his cohort into the same three groups. The results, like the earlier study, indicated that the incidence of lung cancer death increased with cadmium exposure. The follow-up study had eight additional cases, more stable exposure estimates, and its findings were consistent with the earlier study. Among the low dose group, there was a non-significant deficit of lung cancer deaths (Obs=2; Exp=6.06; SMR=33; p=0.06). For the middle dose group, with cadmium exposures that ranged from 41 u/m(3)-200 u/m(3) for a forty year TWA equivalent, there was a statistically significant excess of lung cancer (Obs=13; Exp=6.80; SMR=191, p< .05). For the high dose group, the excess of lung cancer deaths was also statistically significant (Obs=9; Exp=3.32; SMR=271; p< .02). The observation that frequency of lung cancer increased with the dose or level of exposure lends support to a causal interpretation between lung cancer and exposure to cadmium.

Thun included two additional analyses in the updated report. The first of these was a comparison of the three original dose groups with SMRs that were calculated using local Colorado lung cancer death rates as the referent population. Among the low dose group, there was a non-significant deficit of lung cancer deaths (Obs=2; Exp=4.37; SMR=46; p=0.19). For the middle dose and high dose groups, the SMRs are 35-40% higher than in the analysis using the U.S. rates. In the middle dose group, the SMR was 263 (Obs=13; Exp=4.95; p< 0.005, one-sided), and in the high dose group the SMR was 373 ((Obs=9; Exp=2.42; p < 0.005 one-sided). This indicates that the lung cancer risks estimated using national death rates as the comparison population may be an underestimate of the true risk.

In its comparison of results from different epidemiological studies, OSHA relied upon the SMRs calculated using expected death rate derived using national death rates. Thun did the same in most of his analyses, primarily to facilitate comparability of results between the different epidemiological studies.

In the second analysis, Thun provided a more detailed breakdown of dose-response, including six exposure strata developed by dividing each of the three previous strata in half. SMRs for lung cancer were developed using a comparison with the U.S. mortality; the resultant slope was linear and was comparable to the results of original analyses.

Studies of the Globe workers are particularly useful for assessing the carcinogenicity of cadmium in humans for four reasons. First, the cohort was highly exposed to cadmium. Second, information was available for the prolonged follow-up of lung cancer (85% of the workers had 20+ years of follow-up). Third, extensive exposure information was available from good industrial hygiene records which could be linked to work histories. This allowed the computation of cumulative individual exposures to cadmium for each worker included in the cohort. Finally, additional information was available on tobacco smoking habits of workers and on arsenic exposures at the workplace. This allowed some control for the potential confounding effects of these known risk factors for lung cancer.

Dr. Thun noted that: ..for a small study, the strength of the exposure, the intensity of the exposure is one of its major advantages. The second has been the opportunity for prolonged follow up, and....85 per cent of the cohorts have had at least 20 years observation since first exposure to cadmium, and the entire cohort has had 15 years. (Tr. 6/7/90, p. 89)

Because Thun's analyses of the Globe workers showed the clearest association between cadmium exposure and lung cancer, it was his work which drew the most comments during the rulemaking.

a. Power. In commenting on the proposed cadmium rule, Environ noted that from the standpoint of the size and power, Thun's study does not compare favorably with some of the other epidemiologic studies that have been conducted (Ex. 12-41). The power of a study relates to its ability to detect an effect. In so far as Thun's cohort was relatively small, Environ's observation is correct. The power to detect a true excess relative risk of 2.0 or greater was 0.893 or 89% when the experience of the entire cohort was considered. When the analysis was limited to the experience of workers employed from 1926 on, the power to detect a doubling of risk was 0.86 or 86%. The probability of detecting a smaller increase in risk is even smaller. For example, the power to detect a true excess risk of only 1.5, the power was only 0.44.

In general, power calculations are used to evaluate possible reasons why a study failed to show an increased risk of a particular cause of death. When a study is conducted and the results show an increased risk of a particular cause of death, as did Thun's study, the hypothetical probability of the study to detect an excess is irrelevant and a power calculation is not necessary. Therefore, the power of the Thun study is not an issue.

b. Case Status. Another issue raised by Environ was whether two deaths observed in the Thun update should have been considered lung cancer cases based upon a recording of information on the death certificate (Ex. 19-43; 12-41). Lung cancer was listed on the part of the death certificate under "other significant conditions" for both of these cases, but the immediate cause of death was coded as metastatic brain disease for one case and pneumonia for the other case (Ex. 8-658a).

As Dr. Thun indicated, the death certificate information was somewhat ambiguous for these two cases in the update. The hospital records for both of these individuals indicated that lung cancer was the underlying disease. Two of three nosologists were of the opinion that the underlying cause of death on the death certificate should have been coded to lung cancer. Thus, there is some disagreement as to the correct coding of the cause of death for these two individuals for purposes of epidemiologic study. Thun considered the coding of these two deaths to be ambiguous because metastatic brain disease could theoretically originate in the lung or, theoretically, in the brain, even though metastases from a primary cancer in the lung are common while metastases from a primary brain tumor are exceedingly rare (Ex. 8-658a).

Regarding the appropriateness of the inclusion of these two cases in a study of lung cancer, another commentor stated that no one was, "...questioning the fact that (these) two men...died as a consequence of lung cancer (SCM Chemicals; Ex. 12-33-D)." The question is one of comparability of lung cancer cases with the referent population.

Thun et al. concluded that, although some disagreement existed among nosologists about appropriate coding methodologies for cause of death, in the analyses of the data by exposure group it would be appropriate only to include these two cases of lung cancer in internal comparisons within the cohort unless similar cases were included in the compilation of vital statistics upon which expected rates of death were based (8-658a). However, as Thun also reported, if the two lung cancer cases were subtracted from the appropriate dose group, the dose-response relationship and the trend would remain basically the same (Tr. 6/7/90 p. 94). In comparisons with the external U.S. population (Ex. 33) as Thun indicated (Ex. 8-658a) even if these two cases are excluded, the excess of lung cancer is still statistically significant among workers in the update employed for two or more years (Obs=22; Exp=10.76; SMR=204; CI=128-310).

c. Deficit of Lung Cancer in the Low Dose Group. Several commenters stated that the finding of a deficit of lung cancer deaths in the low dose group demonstrated that there was no carcinogenic risk from cadmium exposures below 40 u/m(3), since:

"...the statistically significant excess occurred only for a cumulative exposure equivalent to 40 years of exposure above the current OSHA PEL of 200 u/m(3) (SCM Chemicals; Ex. 19-42A)."


"This (deficit) would indicate that there is a difference between workers exposed to high levels of cadmium and workers exposed to low levels of cadmium (Big River Zinc Corporation; Ex. 19-30)."

Mr. George M. Obeldobel, Vice President and General Manager of Big River Zinc Corporation (BRZ), the third largest zinc producer in the U.S. and the largest cadmium producer in North America, took exception to the Thun study as evidence of the carcinogenicity of cadmium in humans:

"....the data showed a strong difference between workers exposed to low levels and high levels of cadmium, i.e., workers exposed to the equivalent of 40 years exposure at:

(a) 21-40 ug/m(3) respirable cadmium, showed only two cases of lung cancer versus an expected 3.77 cases (again, less than the general population).

(b) 41-200 ug/m(3) respirable cadmium, showed seven cases versus an expected 4.61 cases.(Ex.19-38)."

Dr. Thun addressed this issue in his testimony:

"The deficit is an artifact because this population of being compared with the U.S. population. The deficit is...not from any protective effect of low levels of cadmium exposure (6/7/90, p. 187)."

Furthermore, Dr. Thun continued:

"This finding...should not be interpreted as showing a `safe' level of cadmium exposure (Ex. 33 p. 13)."

Thun indicated that there are at least three factors which could explain the finding of a deficit in the low dose group:

(1) The healthy worker effect, (2) Race status of the referent population, and (3) Smoking.

These factors are discussed in detail below as they pertain to both the finding of a deficit in the low dose group and the finding of excess lung cancer deaths among cadmium exposed workers in general. As discussed in section VI (QRA), OSHA notes that the deficit is not statistically significant and may, therefore, be attributed to random fluctuation.

i. Healthy Worker Effect. The "healthy worker effect" is evidenced by studies which show that active workers experience a mortality risk less than that of the general population which includes sick, disabled, and institutionalized persons. Workers tend to be healthier than the general population. Comparison of mortality among workers to that of the general population would bias the results toward an underestimation of risk.

OSHA received comment on the appropriateness of attributing a "healthy worker effect" (HWE) to cancer SMRs. The Cadmium Council (Ex. 119) stated that the healthy worker effect (HWE) could not sufficiently explain the deficit in the low dose group since this effect has little application to cancer. In the post hearing brief by the Cadmium Council, the studies on the HWE put forth by OSHA (Ex. 8-677; Ex. 50a and b) were questioned as to their relevance to cancer mortality or were faulted for being studies of veterans, whose comparability with workers had not been established.

According to a study by McMichael (Ex.8-677), the HWE may not apply equally to all causes of death. As McMichael stated, ".....if one attempts to improve the meaningfulness of an SMR by adjusting for the HWE, allowance must be made for variation between different... causes of death (Ex.8-677)."

McMichael conducted analyses of data generated by Milham (1974) in a study of carpenters and joiners. McMichael's conclusion was that the HWE does tend to create lower SMRs (60-90%) among cohorts of workers when the U.S. population is used as a referent population. In two other submissions to the docket on the HWE, the issue of the applicability of the HWE to cancer SMRs, in particular, was further examined. The first of these, a study by Dr. Monson (Ex. 50-A), suggests that a HWE lasts for about 15 to 25 years after first exposure, depending on the cause of death and the occupational group being followed.

Dr. Monson stated:

" has long been observed that groups of employed persons have mortality rates that are lower than the general population. This favorable mortality experience has been termed the healthy worker effect (HWE). Uncertainty about the strength and extent of the HWE has led to uncertainty in interpreting data from studies in which the mortality rate of an employed group is compared with the mortality rate of the general population.(Ex. 50-A)."

Monson studied ten groups of workers in order to evaluate the strength of the HWE on various causes of death. He noted that while the healthy worker effect was relatively weak in comparison to causal excesses, and that there was a difference between the strength of the HWE for different causes of death, there was, nonetheless, no evidence of a lack of HWE on cancer SMRs.

A second study by Drs. Sterling and Weinkam (Ex.50b) examined the extent, persistence, and constancy of the HWE by selected causes of death including cancer. Sterling and Weinkam chose to evaluate the data from the Dorn Study of Mortality Among U.S. Veterans because: (1) Veterans who were selected to serve in the armed forces qualified for such duty because they had health status comparable to that of workers seeking employment; and (2) Subsequent to termination of service, not all veterans would enter the same occupation. Thus, the high risks of disease among veterans from occupational exposures in hazardous occupations in each cause-specific-death group would tend to be balanced by the inclusion, in the same group, of veterans with low risks of disease from non-hazardous occupations. As a result, the influence of confounding from job exposures on the evaluation of the HWE would be minimal. Sterling and Weinkam observed a persistent HWE for all causes that did not substantially weaken over time. For lung cancer specifically, the HWE persisted at least through age 74 years. For ages 70-74, the lung cancer mortality ratio was 0.72; it was less than this for younger aged veterans.

For the reasons stated by Sterling and Weinkam, OSHA considers the choice of the study group by Sterling and Weinkam to be an adequate and suitable reference group for assessing the influence of the HWE on specific causes of death, including cancer.

Overall, based on the above mentioned references, the HWE appears to be relevant to cancer SMRs, and thus is relevant to Dr. Thun's study of cadmium exposed workers though the strength of the effect may differ for different cancer diagnoses and depends upon other factors relevant to the cohort under study such as age at time of hire.

OSHA agrees with the Cadmium Council that, in and of itself, the HWE may not be sufficient to explain the finding of a deficit of lung cancer in the low dose group in the Thun study. However, the HWE in conjunction with the lower rates of smoking and lower background lung cancer death rates of the exposed population, plus statistical variability may all combine to account for the deficit of lung cancer mortality.

ii. Hispanic surname. Reduced tobacco use among workers is another plausible explanation for the deficit of lung cancer SMR observed in the low dose exposure group. This does not mean that the low dose group smoked less than the other exposure groups. Rather, smoking in all of the exposure groups was less than the general population, and, as Dr. Thun concluded, the HWE was therefore much more extreme (Tr. 6/7/90; p. 102). Dr. Thun based this conclusion on the fact that slightly less than 40% of the workers at the Denver cadmium plant had Hispanic surnames. Hispanics, in general, are known to smoke less than other white males and thus to have lower rates of lung cancer than other U.S. white males. As Dr. Thun stated, "Several studies have shown low rates of lung cancer among Hispanic males in the Southwest, and particularly in Denver (Tr. 6/7/90, p. 98)."

Dr. Thun's conclusion is supported by a study by Dr. Savitz, (Ex. 57-N), who found that in Denver between 1969 and 1971, Hispanics had a lung cancer incidence that was less than that of other white males. Cancer incidence rates for persons of Spanish surname and other whites in the Denver area were derived for two time periods, 1969-71 and 1979-81. Lung cancer death rates among males with Spanish surnames increased from 23.1 per 100,000 person-years in 1969-71 to 45.6 in 1979-81, as compared to other white males whose rates increased from 57.0 to 68.9 in the same time period. While lung cancer was substantially less common among those with Spanish surnames compared to other white men, there was a convergence of the rates. The convergence of the rates could be attributed, according to the author, to, among other things, acculturation of persons with Spanish surnames, the imperfection of Spanish surname as an indicator of ethnicity, or systematic error due to miscounting of illegal immigrants. While acculturation could explain the change in trends, the authors concluded that Spanish surname was not likely to explain the convergence of rates. Spanish surname identifiers appeared to be consistent for both cases and the reference population across the two time periods. The systematic error in enumeration of immigrants, however, would tend to inflate the cancer rates, since the referent population would be less likely to be counted correctly than would the cancer cases.

By comparing the death rates among the Globe cadmium workers, a group that included a mix of workers with Hispanic and non-Hispanic surnames, to U.S. white male lung cancer death rates, the expected number of lung cancer deaths would be overestimated, and the occupational effect of cadmium on lung cancer would be underestimated. The percent of the workers with Hispanic surnames in the low dose group was 38.5%; the percent of the workers in the middle dose group with Hispanic surnames was 41.5%; and, the percent of the workers in the high dose group with Hispanic surnames was 32.2% (Ex.33). The mixture of cadmium workers with Hispanic and non-Hispanic surnames was similar between the three dose groups. To the extent that Hispanic surname reflects ethnicity, and that Hispanics smoke less and have lower lung cancer death rates than U.S. white males, it is unlikely that differences in ethnic group would have caused a significant impact on the SMRs between dose groups, i.e., in the low dose group alone. Thus, it appears as though Dr. Thun's suggested use of Hispanic surname is a plausible method by which to estimate the effect of ethnicity on the lung cancer SMRs in his cohort.

An additional study of the impact of differences in lung cancer mortality between Hispanic and non-Hispanic workers at the Globe plant was carried out by the National Institute for Occupational Safety and Health (NIOSH) (Ex. 79). NIOSH presented a lung cancer mortality analysis based upon the most recent follow-up of the Thun et al., cadmium cohort through 1984. The life table analysis used in this assessment was stratified into four cumulative dose categories, i.e., = or < 584, 585-1460, 1461-2920, and = or > 2921 mg/m(3)-days) and three "time-since-first-exposure" categories, i.e., = or < 10, 10-19, and 20+ years. Separate life-table analyses were performed for members of the cohort with Hispanic and non-Hispanic surnames. For the lung cancer SMRs, U.S. rates for white males were used as the referent group.

The results from the life-table analysis indicated that lung cancer mortality was similar to Dr. Thun's findings, i.e., lung cancer SMRs were not significantly elevated for the entire cohort (SMR=149; Obs=24; Exp=16.07; 95% CI=95, 222; p=.076, 2 tailed test). However, among white males, lung cancer mortality was significantly elevated (SMR=211; Obs=21; Exp=9.95; 95% CI = 131,323; p< .01). A deficit of lung cancer mortality was found among Hispanics (SMR=49; Obs = 3; Exp = 6.12) as would be expected if the referent rates for U.S. white males reflect people who smoked more than people with Hispanic surnames. Lung cancer mortality increased with cumulative exposure to cadmium, and was significantly elevated in the highest exposure group for the combined cohort (SMR=272; Obs=9: Exp=3.3; p< .05) and for the three highest exposure groups among non-Hispanics. A significant excess of lung cancer mortality was also observed among workers in the longest "time-since-first-exposure" category for the combined cohort (SMR = 161; Obs=21; Exp=12.97; p < 0.05, two-tailed test) and for non-Hispanics (SMR = 233; Obs=19; Exp=8.13; p< .01, two-tailed test). The finding of a statistically significant excess of lung cancer among non-Hispanics, and a deficit among Hispanics, further strengthens Dr. Thun's conclusions that the deficit of lung cancer in the low dose group could, in fact, result from an excess of workers with Hispanic surname in the cohort in comparison with the referent population.

The issue of the use of the percent of Hispanic workers in the exposed population to evaluate the confounding effects of smoking was questioned by the Cadmium Council during the reopening of the record (Ex. 144-16). According to the Cadmium Council, during an IARC Symposium in 1991, information was provided that Hispanics who smoke do so with the same frequency as non-Hispanics. The difference is only that workers with Hispanic surnames tend to underreport smoking habits (Ex. 144-16). This opinion by the Cadmium Council was apparently based upon a document prepared by Dr. Lamm, in which Dr. Lamm summarized the results of the IARC International Symposium (Ex. 144-7). No document which represented IARC's formal conclusions was submitted to the OSHA docket. However, if people with Hispanic surnames do in fact underreport their habits, it is not possible to determine whether the underreporting itself could account for the differences in smoking observed between white males and males with Hispanic surnames in the Thun cohort. If smoking habits between white males and males with Hispanic surnames are indeed similar, nonetheless, lung cancer rates in general are lower among males with Hispanic surnames than among white males (Ex. 57-N). The important factor in the analysis by Thun et al., is that the proportion of males with Hispanic surnames included in each dose category was approximately equal. Thus, any potential confounding effect of the inclusion of a large number of males with Hispanic surnames was equally distributed over dose categories, yet the potential confounding effects did not override the dose-response relationship observed between lung cancer excesses and cadmium exposures.

iii. Smoking. Thun provided direct and indirect evidence for his position that the workers in his study smoked less than the general U.S. population by noting the absence of elevated rates of death for other smoking-related diseases in his cohort. As Dr. Thun stated, "(Cardiovascular disease).... is often a good marker of smoking habits (Tr. 6/7/90, p. 101)." Dr. Thun found that death rates from cardiovascular disease were about two-thirds that of the general population (SMR=65; Obs=56, Exp=85.7).

Dr. Kazantzis, who testified as an expert witness for the Cadmium Council, agreed that indirect evidence such as the finding of a significant deficit of major diseases related to cigarette smoking, such as cerebral-vascular disease and other cardiovascular diseases, would be a factor that must be taken into consideration in evaluating the relationship between smoking and lung cancer excesses. In the absence of direct evidence on smoking habits of members of a study cohort, Dr. Kazantzis agreed that such indirect evidence might argue against cigarette smoking as the major factor associated with observed excess lung cancer (Tr. 6/8/90).

Furthermore, the HWE could have affected, and thus lowered, the SMRs for ischemic heart disease in this occupational cohort because heart disease is more likely to be selected against in the recruitment and retention of active workers (Ex. 8-677). Thus, it would appear that smoking was not so prevalent among cadmium workers in the Thun study that its effects would override the HWE on heart disease rates. Dr. Thun stated that :

"... one of the implications of the lower smoking habits of this population [the Denver cadmium workers] is that one should not interpret SMRs below 100 in the low dose group as ... a safe or threshold level for cadmium ... [W]hat those levels reflect is a lower background risk of lung cancer among these workers due to ... less smoking than the U.S. general population ..."(Tr. 6/7/90, p. 102)

Dr. Thun also evaluated direct evidence on smoking habits obtained from a 1982 questionnaire. Responses were provided by surviving workers or their next of kin, and were supplemented by medical records. These responses showed that although the majority of Hispanic workers smoked, most were light smokers. These light smokers reported smoking fewer cigarettes per day than white males. As for other white male workers at the plant, if anything, they smoked slightly less than white males in the general U.S. population (Tr. 6/7/90, p. 100). As Dr. Thun indicated, use of the data from the questionnaire was somewhat limited because the questionnaire obtained smoking information on only 43% of the overall cohort, and the missing information pertained primarily to the low and medium dose groups (74% and 47%, respectively). Without comparable smoking information for each of the dose groups, controlling for smoking based on the direct evidence from the questionnaire might produce biased results, either towards over or underestimating the effects of smoking on the SMRs for each dose group.

There have been differences of opinion as to the value of smoking data and different estimates of the number of packs smoked (Tr. 6/7/90, p. 183; Ex. 8-658a). Lack of data on smoking in the low dose groups, however, made these data less valuable than data on Hispanic surname because workers with Hispanic surnames were evenly distributed in the three exposure groups.

Dr. Thun stated that the use of an internal analysis would yield a valid estimate of the effect of occupational exposure (Tr. 6/7/90, p. 72). Such an analysis was conducted by NIOSH for risk assessment purposes (Ex. 79). In this study, NIOSH presented results from modeling of the dose-response relationship between cadmium exposures and lung cancer mortality and projected risks associated with varying levels of cadmium exposure. (See Poisson regression models fitted to the "5-year lagged analysis.") The parameter estimates for the categorical model represented the effect of each category relative to the low dose category (< 584 mg/m(3)-days). The issue of the use of internal controls by NIOSH will be dealt with further under the Quantitative Risk Assessment Section.

OSHA received several comments on NIOSH's analysis. During the 1992 IARC International Symposium, the NIOSH analysis was questioned as to whether it provides a biologically reasonable explanation of the data (Ex. 144-16). However, this comment was not from a formal document of the IARC International Symposium (Ex. 144-7).

A similar issue was raised by Dr. Starr during the hearing (Ex. 38). Dr. Thomas Starr commented that "OSHA should redo its analysis with the individual person-year information collected by Thun et al.," noting that the latter data were only very crudely characterized by the median cumulative exposure for each category (Ex. 38). He also recommended that "OSHA should reanalyze the Thun et al. (Ex. 4-68) data with the multistage dose-response approach model utilizing an approach similar to that described by Crump and Howe (1984)." The NIOSH risk assessment and subsequent addendum (Exs. L-140-20, L-163) incorporated both of these recommendations. OSHA's use of additional models is further discussed in its section on quantitative risk assessment. (See Section VI.) iv. Conclusions. In summary, OSHA notes that the finding of a deficit in lung cancer in the low dose group may have resulted from other factors, such as the healthy worker effect, lower smoking rates, and Hispanic ethnicity as noted by Dr. Thun. The deficit should not be interpreted as an absence of lung cancer risk at doses of cadmium less than 40 u/m(3).

d. Smoking and Arsenic in the Thun Cohort. The roles of smoking and arsenic as causes for the excess lung cancers observed in the Thun study were the focus of many of the comments received by OSHA. These concerns were primarily related to the findings of a study by Lamm et al. 1988, which purported to show that there was no link between cadmium exposure and cancer (Exs. 19-43E, 12-13a, 12-33d, and 12-33e). For example, the Cadmium Council reiterated its earlier argument that the Thun study failed to control for smoking, citing the case-control study by Dr. Lamm which, in the Council's view, demonstrated an association between smoking and cancer but not between cadmium and cancer. The Cadmium Council also stated that ENVIRON peer-reviewed the Lamm study and "endorsed its conclusions". Several other commenters made observations similar to those of the Cadmium Council (Ex. 19-43; 12-33e).

i. Lamm's Case-Control Analysis. Dr. Lamm conducted a case-control study (Exs. 19-43e and 144-7b), in which he attempted to replicate the data used in the mortality study by Thun et al., (Ex. 144-7-b). Dr. Lamm identified 599 of the 602 white males in the Thun et al., study (144-7-b), which included all of the lung cancer cases previously identified in the Thun et al., study. Eligibility for both the Thun et al., and the Lamm case-control studies was the same. The workers had to have been employed over six months in the cadmium smelter production areas "...during 1940-1969" (144-7a). Dr. Lamm used a "nested" case-control design of 25 cases of lung cancer matched with 75 controls from the entire cohort of 602 workers included in the Thun et al., study.

The case-control study was used to evaluate lung cancer and the contribution of arsenic and smoking as well as cadmium to the lung cancer deaths (Ex. 19-43e). Dr. Lamm matched cases with controls on age, race, sex, and date-of-hire (Ex. 144-7-b). Cadmium exposures for both cases and controls were estimated using NIOSH's original exposure estimates,which Thun calculated for each employee at his worksite. Smoking histories were obtained by company personnel from interviews and company records for about 43% of the cohort (Ex. 19-43-e).

Based on data from 597 workers in this plant, Lamm identified three time periods of potential arsenic exposure (Ex. 12-33-e), or period-of-hire risk factors (Ex. 144-7b). Based on 25 cases of lung cancer and 75 controls, and using the cadmium exposures developed by Thun et al., the arsenic feedstock data, and the previously identified period-of-hire risk factors, the relationships between lung cancer deaths and cadmium exposures, smoking, arsenic, and period-of-hire, respectively, were independently evaluated.

ii. The Smoking Argument. Based on the amount of tobacco consumed (pack-years per employee), Dr. Lamm found that the relative case-to-control smoking history, for workers with known smoking history, was 2.5, with a range from 1.6-8.3 (Ex.12-33-e). When Dr. Lamm categorized workers as ever/never smoked, the odds ratio for smoking as a lung cancer risk factor was 8.2 (Fisher's exact test two-tailed; p=0.047; 95% CI: 1.04 to 367.05). Based upon these analyses, Lamm concluded that cigarette smoking was responsible for the elevated lung cancer risk observed among the cohort (Ex. 12-33-e).

Dr. Thun stated that the missing data on smoking (available, for example, for only 57% of controls; see Section VI - QRA.) made it invalid to use these individual smoking histories in case control analyses (Tr. 6/7/90; p. 101). While smoking clearly affects lung cancer rates in general, it does not account, in and of itself, for the lung cancer excesses observed in Thun's study, based on Dr. Lamm's analysis. Dr. Thun indicated that the preferred evaluation of smoking status and its effects on lung cancer death rates would be to use Hispanic surname or an internal analysis to control for both the confounding due to the HWE and smoking.

OSHA is of the opinion that the smoking data from the questionnaire survey were limited and should not have been used alone in the case-control study to evaluate the effects smoking. If smoking were a main cause of lung cancer, one would not expect a deficit of lung cancer in the low dose group. Smoking would elevate these rates, too.

iii. The Arsenic Argument. A second analysis by Dr. Lamm entailed examining lung cancer deaths (SMRs) for the entire Thun et al., cohort, including workers employed before January 1, 1926 (Ex. 19-43e; Ex. 12-33e; Ex. 144-7-b). For this analysis, Dr. Lamm merged data from the Thun et al., study with data from the plant owners. Mean arsenic concentrations in the feedstock for various time periods were calculated based on a 10% sample of arsenic feedstock records (Ex. 19-43e). Dr. Lamm indicated that there were three critical calendar time periods of potential arsenic exposure based on the arsenic exposure concentration in the feedstock. These three periods were characterized as: prior to 1926, very high; 1926-39, high; and 1940-69, moderate to low. The reductions in exposures by calendar time period resulted from changes in plant processes. Arsenic exposures, it was assumed by Dr. Lamm, occurred uniformly throughout the plant.

Dr. Lamm then examined the SMRs for lung cancer by period of hire. He found that for workers hired prior to 1926, the SMR was 492 (Obs=3, Exp=0.61); for workers hired between 1926 and 1939, the SMR was 283 (Obs=6, Exp=2.12); and, for workers hired between 1940-69, the SMR was 88 (Obs=8, Exp=9.08). A full discussion of the lung cancer response among workers hired prior to 1940 is included in Section VI-QRA. Dr. Lamm concluded from these findings that the lung cancer risk appeared to reflect arsenic exposure confounded by cigarette smoking. Addressing Dr. Lamm's conclusions, Dr. Thun stated that:

"The conclusions (by Dr. Lamm) ... go well beyond the data presented ... the first conclusion implies that the average arsenic concentration in feedstock being processed by the Globe plant is 5%. This statement is incorrect ... in actuality, the mean annual arsenic concentration (geometric mean) of the feedstock after 1927 averaged between 2-3% with the exception of the year 1930 (5.6%) and 1931 (4.9%). Lamm and co-workers also imply that arsenic exposure is generalized throughout the plant. In fact, arsenic exposure is limited to the early steps of the process, in the sampling, calcine, and roasting areas .... In our opinion, arsenic exposure may have contributed to the excess of lung cancer deaths at the Globe plant, but this contribution has been overstated by the spokesmen for the cadmium industry.(Ex.8-645)."

Furthermore, Dr. Thun commented that the feedstock data that Dr. Lamm used was:

".... based .... upon visual impressions of records of arsenic in feedstock entering the plant. Dr. Lowell White, a former employee of the company, compiled the estimates of arsenic concentration, by year ..... We (NIOSH) have obtained the records of arsenic in feedstock from the company and have analyzed the data through 1958 ... The calculated geometric means do show slightly higher values during the years 1930-38 than thereafter, but the averages are lower than those estimated by the company scientists. (Ex. 8-658a)."

In other words, as Dr. Thun indicated during the hearings:

"(Dr. White) sort of eyeballed the sample of records. But we obtained the actual records and analyzed them and determined the geometric mean and the percentage of arsenic in feedstock that was actually present is shown .... under NIOSH calculations. ( Tr. 6/7/90, p. 104)."

Mr. George M. Obeldobel, Vice President and General Manager of Big River Zinc Corporation (BRZ), the third largest zinc producer in the U.S. and the largest cadmium producer in North America,took exception to the Thun study on the carcinogenicity of cadmium in humans:

"Without including the ... 26 workers [who had worked at the smelter prior to 1926 when the smelter functioned as an arsenic smelter] ... the excess of lung cancer death was not statistically significant, suggesting that arsenic played a large confounding role in the data. (Ex. 19-38)."

Although the lung cancer SMR was not statistically significant for the total cohort hired after 1926 (SMR=147; Obs=16; Exp=10.87 95% CI=84, 239), according to the Thun study, lung cancer mortality was significantly elevated among workers with two or more years of exposure.

The fact that the lung cancer SMR for the total cohort was not significant, however, does not indicate that the arsenic exposures had a more significant effect than anticipated by Dr. Thun. Dr. Thun excluded workers hired prior to 1926 to control the effects of arsenic exposure within the cohort. The true effect of excluding these workers hired prior to 1926, as Dr. Thun indicated, would be to reduce the potential confounding among the cadmium workers for arsenic, not to entirely eliminate its effect.

The continuing controversy about the exposure assessment used by Dr. Thun and the potential for arsenic confounding, centered, in part, around whether or not the arsenic exposures in the plant were similar for all cadmium exposed workers, (i.e. whether the amount of arsenic per worker was similar between dose groups in the dose-response analysis).

As Dr. Thun has stated:

"... the arsenic exposure occurred only in departments that processed incoming feed material: sampling, mixing, roasting, and calcine furnace area. Other stages of the process are housed in separate buildings where workers were exposed to cadmium but not to arsenic. It's important to emphasize the localized areas of arsenic exposure at this plant which differ from the more generalized exposure in arsenic and copper smelters. Company representatives have drawn analogies between the Globe plant and the Tacoma copper smelter, but the difference is that arsenic exposure was generalized in ... these other plants and is highly localized in [the Globe] plant (Tr. 6/7/90, p. 105)."

Thun's evaluation of the entry level and long-term jobs, and their associated arsenic exposures, was indirectly confirmed by Mr. Robbins, of the Globe ASARCO plant. The areas of the plant where the highest exposures to cadmium occurred were in the retort furnace and pre-melt areas (Tr. 6/12/90, p. 55).

As Mr. Robbins indicated:

".... the area of the retort ... [primarily cadmium exposures] ... is one of the areas that we're very concerned about product quality. So it's important to have experienced people .... working in that job. And it's my sense that the turnover rate is fairly low ... (Tr. 6/12/90, p. 47)

The highest arsenic exposures would have occurred in the solution charging area, which was in the "front-end" of the process. Workers in these areas, as well as the solution area, would historically, have been considered to be in entry level jobs according to Mr. Robbins (Tr. 6/12/90, p. 51). The area in which the "quality" of the cadmium compound was of most importance would not have been entry-level positions, with some arsenic exposures.

Furthermore, when the Globe plant had the Godfrey roasting and calcine operations, those were the highest areas for arsenic exposure and would have been entry level positions as well, although there would have been some cadmium exposures in these areas. Mr. Robbins stated:

"Thun was looking at historical exposures at the Globe plant and there was a time when Globe had Godfrey roasters...that are no longer there....[T]here would have been job categories that aren't here that would have been here 20 years ago..(Tr. 6/12/90, p. 47).

In response to questions, Mr. Robbins indicated that in all areas that workers had exposures to high levels of arsenic, the workers would also have some exposure to cadmium. However, the high cadmium exposure operations, which require long-term, skilled professionals, generally do not have arsenic exposures. Thus, there clearly were work areas which were predominantly cadmium exposure areas where there was no arsenic exposure (the retort furnace and pre-melt sections). In addition, there were work areas with high arsenic exposure and some cadmium exposure (currently the solution charging areas and historically both the solution charging and high purity production areas.) This would mean that while all the workers in the Thun et al., cohort could have had some exposure to arsenic in entry level job positions, the arsenic exposures would tend to be equal among the workers and across exposure categories as Dr. Thun indicated. In each of the cumulative cadmium exposure dose groups, the arsenic exposure per worker would not increase as cadmium exposures increased.

When questioned about whether arsenic and cadmium exposures were truly independent, Dr. Thun responded:

"They're not completely independent. There is some independence because when workers would move to other departments like retort and foundry in which cadmium exposures were substantial, they would be exposed to cadmium but no arsenic. (Tr. 6/7/90, pp. 167-168)."

If all the jobs with cadmium exposure were the same ones that had arsenic exposures, the finding of a dose-response relationship between lung cancer and cumulative doses of cadmium would be more likely to be confounded by arsenic. However, there does not appear to be a perfect correlation between cadmium and arsenic exposure. Thus, the finding of a dose-response relationship between lung cancer and cadmium exposure should not be attributed to arsenic exposure, either in part or in total. (Tr. 6/8/92, p. 147). Thun et. al. conducted an analysis of the potential magnitude of the effect of arsenic and concluded that no more the 0.77 cancer cases could be attributed to arsenic (Ex. 4-68) (See Section VI-QRA).

OSHA received the comment that as late as 1979, this plant received an OSHA citation for levels of arsenic exceeding 100 u/m(3) (Ex. 12-41). As Dr. Thun indicated:

"Only six industrial hygiene measurements were made in these (arsenic) areas before 1975. In 1950, airborne arsenic concentrations ranged from 300 to 700 u/m(3) near the roasting and calcine furnaces, the areas of highest exposure. Measurements ..... in 1979 ... show that arsenic exposures in these areas had decreased to about 100 u/m(3) (Ex. 4-68)."

It was therefore argued that arsenic should be considered as the major factor affecting the observed elevated lung cancer deaths in the Thun study (L-19-59). The main issue was that Dr. Thun's assumptions for calculating cumulative exposure contained a number of errors that would tend to underestimate the risk due to arsenic and overestimate cadmium's risk per unit of dose.

In Thun's analysis, the estimated number of cancer deaths that would result from the arsenic exposures in this cohort by was calculated by assuming:

(1) An average arsenic exposure of 500 u/m(3) in the "high arsenic"


(2) A respirator protection factor of 75%, similar for both cadmium and arsenic;

(3) An estimated 20% of person-years of exposure spent in high-arsenic jobs; and, (4) An average employment duration of 576 post-1926 workers of 3 years for 1,728 person-years of exposure. As described in section VI-QRA, Dr. Thun concluded that no more than 0.77 lung cancer cases could be attributed to arsenic exposures, based on estimated arsenic exposures and using the risk assessment model developed by OSHA.

Dr. Lamm (Ex. L19-59) proposed the following alternative assumptions and arguments regarding exposures:

(1) The appropriate midrange of arsenic exposure levels is 2,650 u/m(3);

(2) The respirator protection factor should be 1.00, based on observations of respirator use at various smelters (Globe, Tacoma, and Anaconda);

(3) Employees spent 67% of their cadmium exposure time in high cadmium exposure areas and 47% of that time was spent concurrently in high arsenic exposure areas, meaning that an estimated 32% of person-years of exposure were spent in high-arsenic jobs; and, (4) A total of 3,941 person-years of cadmium exposure. When these factors are taken together, the estimate of arsenic-induced lung cancer is high. Dr. Lamm stated,

"... we conclude that the NIOSH [Thun] estimate of 0.77 cases ... was based on a vast underestimate ... (Ex. L-19-59)"

Dr. Thun indicated that Dr. Lamm's analysis included several misconceptions and faulty statistical procedures which caused an overestimation of arsenic exposure at the Globe plant (Ex. L-140-23). First, the exposure data from the 1940's to which Dr. Lamm refers come from a survey by the University of Colorado, Division of Industrial Hygiene, dated 1945. According to Dr. Thun, Dr. Lamm incorrectly reported the mid-range of these exposure data as being 8,700 u/m(3) rather than 500 u/m(3) used by NIOSH. Dr. Lamm reduced this level as a "midrange" value of exposures by eliminating the highest level and then taking the midrange, resulting in his estimate of 2,650 u/m(3). Dr. Thun, on the other hand, indicated that the median of these exposure measurements is 205 u/m(3); the geometric mean is 213.6 u/m(3); and, the arithmetic mean is 392.2 u/m(3).

Dr. Thun recommended that perhaps the best measure of the midpoint of these samples would be the geometric mean, weighted for the number of workers in each operation. This value equals 156.9 u/m(3) without taking respiratory protection into account.

An unspecified "midrange value", used in Dr. Lamm's report, is less useful than the mean and median values given by Dr. Thun, and mathematically is meaningless. It would appear that the usual level of exposure to arsenic in areas with potential exposure to arsenic was considerably less than Dr. Lamm estimated.

Furthermore, according to the same survey, cited above, by the University of Colorado, the departments where arsenic exposure potentially occurred were only operational for four months of the year. The intermittent nature of these exposures should have been taken into consideration by Dr. Lamm, according to Dr. Thun. If the intermittent nature of these exposures had been taken into consideration, it would have resulted in further lowering of cumulative arsenic exposures.

Dr. Lamm claimed that the respirator usage and exposure data on arsenic exposures in three plants (the Anaconda, ASARCO Tacoma, and the ASARCO Globe plants) should be used to represent the respirator use and the arsenic exposures at the Globe plant. In this way, according to Dr. Lamm, a resulting protection factor of 1.00 for respiratory protection should be used instead of 0.26 used by Thun. This, however, ignores the fact that Smith's study of respirator usage, cadmium exposures, and arsenic exposures, was conducted at the Globe plant. The respirator protection factor used by Thun is thus more appropriate for use in studies of workers from the Globe plant.

Dr. Thun indicated that Dr. Lamm's assertion that the controls in his case-control study spent 68% of their time in high cadmium areas does not contradict Thun's estimate that an estimated 20% of person-years of exposure were spent in high-arsenic jobs. Dr. Thun's estimate is an estimate of time spent in high arsenic areas, not high cadmium areas. The high arsenic areas were sampling, mixing, roasting, and calcine. Furthermore, Thun's estimate was based on the entire cohort, covering the entire post-1926 period, while Dr. Lamm's estimate was based on a small sample of workers in his case-control study (N=75, or 12% of the cohort) who did not die of lung cancer. Thus, the estimate of time spent in arsenic exposure areas in the Thun study is more likely to be an accurate estimate in that it was based on a larger sample, utilized more complete employment records than were available to Dr. Lamm, was confirmed by several sources, and did not exclude workers who died of lung cancer.

Dr. Thun indicated that the major problem encountered by Dr. Lamm in his analysis was that he lacked sufficient work history data to identify departments with potential arsenic exposure in adequate detail to analyze these data by duration of employment; Thun used person-years of employment in high arsenic areas instead.

For several reasons, Dr. Thun believed that the estimate of 0.77 cancers may have been an overestimate rather than an underestimate. These included: (1) the high arsenic exposure jobs were frequently staffed with short-term employees who were not in the study cohort, thereby excluding workers with less than six months employment; and, (2) urinary arsenic levels of workers in the high areas from 1960-80 averaged only 46 u/L, which would equal an inhaled arsenic of 14 u/m(3), far lower than the estimated average arsenic exposure of 125 u/m(3), cited by Thun (Ex. L 140-23).

As Dr. Thun indicated, it is clear that Dr. Lamm's model overestimates the number of deaths attributable to arsenic:

"On the basis of Dr. Lamm's risk assessment mode, 143 deaths from lung cancer were expected based on arsenic exposure. Since only 24 lung cancer deaths were observed, it is apparent that Dr. Lamm's model greatly overestimates the number of deaths. On the other hand, Dr. Thun's model predicted 0.77 deaths, which is more consistent with the observed deaths (Ex. L-140-23)."

However, according to Dr. Thun:

"If we took our worst case scenario and the estimated exposure to arsenic was twice what we said it was (representing a two-fold improvement in respiratory efficiency over time), then in the cohort over their entire life time, one would expect twice 0.77 cases of lung cancer attributable to arsenic. A very small number, less than two, which would by no means explain the excess that's observed. So the two fold underestimation that may have occurred due to differences in respirator efficiency does not account for the excess of lung cancer.(Tr. 6/7/90, p. 172)."

Furthermore, Dr. Thun stated that if arsenic was causing the excess of lung cancers in this cohort, one would have seen a higher SMR for lung cancer in the low dose exposure group (Tr. 6/7/90, p. 177).

OSHA has recalculated the estimated number of arsenic lung cancer deaths that may have resulted from arsenic exposures, in the quantitative risk assessment section. (See Section VI-QRA.) OSHA is of the opinion that the original analysis by Dr. Thun was adequately justified. It is fair to state, as did Dr. Thun, that while arsenic may have contributed to the excess lung cancers observed in this plant, it would not totally account for the excess lung cancer.

iv. Lamm's Period-of-Hire Argument. One primary objective of Dr. Lamm's case-control study was to examine the role of cadmium exposure when controlling for period-of-hire (Ex. 144-7-b). Relative cadmium exposure was expressed as a ratio of the mean cadmium exposure of the cases to that of the controls. Based on the case-control study, Dr. Lamm observed that the relative case-to-control cadmium exposure (mg-yrs/m(3)) was 1.0, with a range from 0.9-1.1 (Exs. 12-33-e; 19-43-e; 144-7). Dr. Lamm concluded that the lung cancer risk in this workplace was:

"... more related to the period of hire, not to the cumulative cadmium exposure. The period of hire appears to be a surrogate for arsenic exposure as related to feedstock. The measures used here seem to indicate that exposure to arsenic and cigarette particulates, rather than cadmium particulates, may have caused the lung cancer increase of these workers. (Ex. 144-7-b)"

OSHA has reanalyzed Dr. Lamm's results in the quantitative risk assessment section. In summary, the results of Dr. Lamm's case-control analysis could be explained by methodological errors in the study design or by random statistical fluctuations. (See Section VI-QRA). One example of methodological errors, Dr. Thun pointed out, would be the use of the case-control design in Dr. Lamm's study which is at variance with the common use of this method. Typically, "nested" case-control studies are conducted when there is an opportunity to classify exposure more accurately than is possible in a cohort study in which all workers are considered exposed to some degree. However, in the case-control study, the original cadmium exposure data from the Thun study was used for a subset of workers, and no additional information was collected and/or used for the whole cohort. Thus, the choice of a case-control study design introduced a major disadvantage in that the controls were not representative of the study population. That is, only a portion of the total information was used as opposed to a cohort study where all information on all workers is used. At the same time, this case-control study design applied to these data offered few of the advantages of the case-control design, (i.e., the addition of new data on the total cohort) (Ex. 8-645).

Random statistical fluctuations, which were perhaps exacerbated by "overmatching", could explain the results. Overmatching occurs when controls are matched to cases on a variable that is related to exposure but is not a risk factor for disease independent of exposure. Such overmatching can actually reduce study precision. (See Section VI-QRA overmatching.) In general, when rules are developed for selecting controls, factors known or strongly suspected of being related to disease occurrence should be taken into consideration. Once a factor is matched, it is eliminated as an independent study variable and the control group can only be used for the study of other factors. This suggests that caution is required regarding the amount of matching attempted in any study. If the effect of a factor is in doubt, the preferable strategy is not to match for that factor but to control for it in the statistical analysis. If in a case-control study, for example, one matches cases to controls on exposure, no information on the association between exposure and disease will be obtained.

In general, several criteria should be met if one is to consider designing a "matched" study: the purpose of the study design should not pertain to the factor matched; if the factor is not "matched", one should be reasonably certain that the factor will be confounding; such confounding should be expected to be more than just trivial; and, there is no possibility that the factor is part of the causal pathway linking the exposure and disease under study.

According to Dr. Thun, in the case-control study by Lamm controls were inappropriately matched, or "overmatched", to cases on the variable of "date of hire" (Ex. 8-645). While OSHA agrees that cadmium exposure is very likely to be correlated with year-of-hire, OSHA believes the correlation is not perfect. (See Section VI-QRA). However, in terms of the Lamm case control study, overmatching is a potential flow that would lead to the finding of no difference in cadmium exposures between cases and controls. According to Dr. Thun:

"Overmatching is most severe for the eight cases and 24 controls hired before 1940. Workers hired before 1940 were only enrolled in the study if they continued to work in cadmium production for at least six months after 1940. Thus, cases and their controls hired in 1930 typically remained in cadmium production for 10.5 years (until July 1, 1940) to be enrolled in the study (Ex.8-645)."

The approach used by Lamm, according to Dr. Thun, almost guarantees that there will be little difference between cadmium exposures among cases and controls. Dr. Thun continued:

"Still another problem is that the investigators overestimate the relevant cadmium exposure of the controls. In the Lamm analysis, controls are allowed to accumulate exposure after death of the case. Thus, the cases' potential exposure ends at death, but the controls' exposure is not similarly truncated. This approach inappropriately inflates the exposure of the controls and obscures the potential ill effects of exposure. A less biased method of analysis would exclude the experience of controls following the death of the case (Ex. 8-645)."

Thus, the main difference between cases and controls, if overmatching on controls for arsenic and cadmium occurred, would be, as Dr. Lamm noted, that there could be differences in smoking, but no differences in exposures to either cadmium or to arsenic.

e. Thun's Exposure Assessment. In general, for epidemiological purposes, exposure data should contain information on other substances in the workplace to which workers may have been exposed. For a given work area, records should be kept on the substances used, the quantity used, and the period of their use. The work area should be some identifiable area that is small enough to have relatively uniform usage and exposure to industrial substances but large enough to have a reasonable number of workers. Further, a work area should be convertible to some personnel code, so that data on work areas can be linked to data on individual workers.

Dr. Thun's exposure assessment for the members of his cohort was the subject of comments in the cadmium rulemaking. Several commentators agreed with the Environ study in its comments on the Thun study:

"A further strength of this study for the purpose of risk assessment is that quantified ranges of cadmium exposure were estimated. From plant and personal hygiene measures, exposure assessments for individual workers have been made....for many of the other epidemiological studies exposure data are limited to the identification of an average exposure level for the plant. Dose-response relationships cannot be postulated for these other studies. (Environ, Ex. 12-41)."

The exposure estimates provided to Thun were developed by Smith after extensive communication with plant personnel who were familiar with the operation of the plant. Thun computed individual cumulative cadmium exposures for each worker included in the cohort. Smith (Ex. 4-64) described the plant and the work areas in great detail:

"Cadmium production was performed in a complex of ten buildings, with each phase in a physically isolated building or section of a building ... Cadmium enters the process principally as cadmium oxide dust, recovered as a by-product of air pollution control at nonferrous smelters, especially zinc smelter...Each shipment of raw material is assayed for cadmium when received. The cadmium-bearing materials are roasted, mixed with sulfuric acid, calcined, and dissolved in a sulfuric acid solution that is purified by precipitation and filtration. Highly purified cadmium metal is plated out of the solution in an electrolytic refinery (tankhouse) and melted and cast into shapes at the foundry. Some of the metal is reoxidized in gas- heated retorts to make high purity oxide, and some is redissolved in sulfuric acid and treated with hydrogen sulfide to make yellow cadmium sulfide pigment. Dry materials are transported between buildings and process and then pumped to the electrolytic refinery.

Substantial levels of airborne cadmium have been observed over a long period in several areas of the plant ... Exposures to cadmium oxide (CdO) dust occurred during sampling, loading, transporting of dust between roasting, mixing, and calcine operations, and during the loading of the purified oxide. Exposures to CdO fume occurred during the roaster, calcine, retort, and foundry operations. Exposures to cadmium sulfate occurred during the solution and tankhouse operations (mist). Job designations are associated with the operating departments ... with little overlap between them. The plant also has facilities to produce small amounts of lead, arsenic, thallium, and indium. These operations are performed sporadically by a few individuals in three isolated buildings. High purity arsenic, thallium, and indium are produced in facilities on a laboratory scale.

The cadmium production process has not changed during the past 40 years;

however; over this period the company has added a number of emission control systems to reduce exposures in the work areas. Ventilation controls have been added to the roasting, mixing, calcine, foundry, and retort areas ...

Plant offices and laboratories are located in separate buildings outside the production area. However, they have some cadmium contamination from dust carried in on the footwear and clothes of supervisory personnel and from general air contamination from the plant's activities."(Ex.4-64)

From his description, Smith provided information on areas in the workplace where cadmium exposures existed, the relative cadmium exposure levels in these areas, the length of employment of workers in each area, and workers' exposures to other chemicals. When used with existing historical exposure measurements, individual worker's personnel records, and information provided by plant personnel, Dr. Smith was able to quantify exposures for individual workers in the cohort. The true advantage of this study over other epidemiological studies is the estimate of dose for each member of the cohort.

In using the exposure data to assess the risk of cancer from exposure to cadmium, a major assumption underlying Dr. Thun's analysis is that one year of exposure to cadmium at 10 u/m(3) is equivalent to 10 years of exposure at 1 u/m(3). This assumption was questioned during the rulemaking process (Ex. 19-13). Epidemiological data upon which to make such an evaluation are usually lacking. Analyses performed in the quantitative risk assessment section (Section VI) of this preamble indicate virtually the same potency estimate for cancer in animals exposed to CdO dust regardless of whether exposure was continuous of intermittent. Thus, dose rate effects in relation to cadmium exposure and lung cancer were not observed. This observation supports the use of a cumulative dose concept for occupational exposure to cadmium and lung cancer.

In addition to the above issue, comments were submitted to OSHA on other assumptions in Thun's study (Ex. 12-41). Cumulative exposure was estimated using length of employment, jobs within the plant, and an estimate of each worker's time-weighted-average inhalation exposure to cadmium calculated from personal sampling data (1973-1976) and area sampling data (1943-1976). These estimates were originally made by Dr. White (Ex. 4-64).

Because many of the personnel records specified general work areas rather than single departments, Thun categorized each period of a worker's employment into one of seven broad job categories. The average exposure to airborne cadmium for each of these composite categories was calculated on the basis of the industrial hygiene data reported by Smith (Ex. 4-64), with each department contributing to a weighted average according to the proportion of workers usually employed there. Each worker's cumulative exposure over time was computed as the sum of the number of days worked in a given job category for the relevant time period. Cumulative exposure was expressed in milligrams per cubic meter-days (mg-days/m(3)).

In the Environ report, the authors summarized the main questions about the method by which Drs. Thun and White estimated exposure. The Environ report called into question: (1) the ratio used between area and personal exposure geometric means for six of the eight general plant areas; (2) the weighting factor; and, (3) the factor used to calculate the effects of respirator usage. Regarding the ratio between area and personal exposure geometric means, area sampling data were available for the time period between 1943 and 1976. Personal exposure measurements were also available for the time period 1973-1976. Dr. Smith calculated a multiplier, or ratio of personal exposure data to area exposure data, based upon data collected in six of eight work areas in the Globe plant between 1973-76. This ratio was then used to adjust the remainder of the historical area exposure data to reflect personal exposures.

The Environ report indicated that there was the possibility that the ratios might not have been the same in the earlier years before the introduction of many of the industrial hygiene controls. If, as the Environ report pointed out, the personal measures in 1973-76 were in fact unrepresentative of the other time periods for which actual personal monitoring data were not available, the ratio between the air concentrations and the personal measures could be different from those calculated and used by Smith. The end result of this assumption could be that the Thun data underestimated dose and thus overestimated the risks associated with cadmium exposure.

Smith acknowledged that:

"... the accuracy of the estimates may change with increasing time into the past because conditions have changed in some of the work areas ... [and] as a result ... the precision was probably not as good for the older exposures as for the more recent ones (Ex. 4-64)"

Smith stated that:

"... we could not construct personal exposure histories based on each individual's own sampling ... [W]e are uncertain about the degree of error introduced by the use of group data. In spite of these difficulties, our estimates were reasonable attempts to consider all of the factors that affect inhalation exposures and have produced exposure-effect relationships that are consistent with the known toxicological behavior of cadmium (Ex. 4-64)."

OSHA agrees with Dr. Smith that the true direction of the bias in the exposure assessments can only be guessed. Despite the concerns raised, even the Environ report concluded that of all the assumptions that could operate in the misclassification of exposures, the use of the ratio measure was the "exception" (Ex. 12-41).

Dr. Smith clearly indicated that he was basing his assessments of earlier exposures on area data. The calculation of personal exposures from area measurements was based on the assumption that the ratio between them was approximately constant. He stated that this assumption was reasonable because:

(1) The same sampling locations had been used for area samples for the past 20 years;

(2) The ratios were calculated from the averages of several measurements for both area and personal data (individual samples can vary quite widely), and (3) The jobs were routine, well-defined, limited by area, and had not changed significantly during the past 40 years, all of which would tend to make the ratios extremely stable (Ex. 4-64).

The ratio calculated by Dr. Smith between the 1973-76 personal and area sampling data is neither unexpected nor implausible, and would not necessarily change for historical area data regardless of ambient airborne levels, particularly for heavy metal exposures.

In general, in epidemiological studies, even though personal data are preferred for heavy metals and best reflects the worker's environment if sampling is conducted closer than 30 cm from nose (8-668), exposure data of any kind are often lacking. The exposure information is rarely of adequate quality or based upon individual workers personal sampling. Thus, it is plausible that area data could be used to approximate the personal exposures. Since area data were available to Drs. Thun and Smith, it is appropriate that they should have been used.

Regarding the use of area data, the Environ report raised one additional recommendation which was that Smith use the area sampling measures that existed for the time period 1943-54, with or without the ratio multiplier, to calculate the exposures for those periods. If cadmium levels at the smelter were in fact much higher during this period, as the report indicated, then assigning measures taken from 1955 to 1959 to earlier years would underestimate the actual exposure conditions, perhaps considerably, and thus overestimate risk.

In his original report, Smith addressed this issue, stating that he did not use these data for several reasons:

"Although sampling data were available for most of the work areas before 1955, these data were not used because (1) different sampling techniques were used-impingers and electrostatic precipitators instead of filters-and (2) a different sampling strategy was used-breathing zone sampling with hand-held collectors instead of fixed-location samplers ... Therefore, 1955-59 conditions were used to estimate pre-1955 conditions ... (Ex. 4-64)."

However, during questioning at the cadmium hearings, Dr. Thun further clarified this issue. Dr. Smith had data for the five year period from 1945-50, and these data were used by NIOSH for estimating exposures per worker. Prior to 1945, exposure data were estimated based upon existing data and the knowledge that major engineering controls or changes in the process during those earlier years were absent (Tr. 6/7/90, p. 128).

A second objection raised in the Environ report was that each period of a worker's history was categorized into one of seven broad job categories whose weighted average exposure was based upon the departmental measures from the Smith estimates. The weight used for this average was the number of workers usually employed in the department.

As Dr. Thun indicated, in order to link the personnel records to the industrial hygiene data, it was assumed that the average exposure in the high exposure category, (category 1 which included sampling, roasting/baghouse, mixing, calcine, foundry, and retort areas) and in the low exposure areas (category 7 which included solution, tankhouse, and pigment areas) would be a weighted average of the various departments within these two categories. The average exposure in the high exposure category was assumed to be a weighted average based on the average number of worker-shifts per week in each department.

Environ stated that this was "peculiar" since the number of workers is not an inherent part of the estimate of exposure. Thus, use of the number of measures as a weight had the possibility that the exposure would be overestimated because more measures may have existed for an area where a problem existed. Further, the area of high exposure might not be one where many workers were employed (Ex. 12-41).

The weighting system, or number of eight-hour shifts completed in each department per week, used by Drs. Smith and Thun was based upon telephone conversations with Dr. Lowell White, an ASARCO hygienist, and with Mr. Ernie Lovato, the former president of the local union and a long time employee. The estimates were generally consistent with a listing of job titles in each department at the Globe plant in a 1980 doctoral thesis by Dr. Jeffrey Lee (Ref. in Ex. 33; Smith, 1976).

Thun divided his cohort into homogeneous exposure subgroups on the basis of each individual's occupation, place of work, and shift, and the basic sampling unit was the "worker-shift." The weighted effort was allocated to certain "worker-shifts" based on several factors: the standard deviation in exposure for that particular homogeneous group, the labor turnover of the group, and the number of men belonging to the group. An underlying assumption of this strategy was that exposure to an individual worker is supposed to be indistinguishable from the shift average of the whole group.

It is clear that the measure of the exposure that is applied to any individual in a group should be the mean exposure level. An individual's cumulative exposure should be derived as the mean exposure level times the duration of each individual's employment in that part of the plant.

OSHA agrees with the Environ assessment that Thun's use of average exposures would stabilize the estimates and make the assessments relevant even without a weighted scheme based on the number of measures. However, OSHA did further evaluate the magnitude of the effects of such assumptions relative to arsenic exposure and substantially verified the Thun results for weighted averages (see Section VI-QRA). Thus OSHA concluded weighted averages are more appropriate than unweighted averages.

The last major issue raised by Environ about assumptions in Smith's study was the reduction of the calculated personal exposure by using a factor of 3.9 (approximately 75%). This factor was judged by Smith to be the degree of protection afforded by respirator use. In a series of papers, (Exs. 4-64; 8-281), Smith evaluated the average reduction in inhalation exposures produced by the intermittent use of filter cartridge respirators by cadmium workers in the Globe plant.

Questions were raised about the actual degree of protection afforded by respirator use when applied to correcting for arsenic exposures (Exs. 12-41;. L-19-59). The main problem was the application of Smith's factor of 3.9 to correct for respirator use as it pertained to arsenic exposures. It was noted that respirators were not routinely worn, which would, in turn, affect the cadmium exposure assessment, as well.

According to Environ and others (Exs. 12-41; 19-43) exposures were underestimated due to the use of the respirator protection factor. Respirator use before 1940, it was stated, was likely to have been minimal at best and the respirators available were not very protective (Ex.19-43). In the case of arsenic, the dose was underestimated thus understating the risk; more lung cancer risk should have been attributed to arsenic. In the case of cadmium, the exposure was underestimated, and the risk overstated.

According to Mr. Bidstrup, Counsel for SCM Chemicals, Inc., who cited Dr. Lowell White, formerly of ASARCO, it was not until barrier cream was developed that workers wore respirators in arsenic areas because workers who used respirators before that were afflicted with "... a `horrible face rash' or dermatitis (Ex.19-42-b-3, pp. 10-11)." Workers often did not wear respirators for this reason. In addition, Dr. White noted that a different type of respirator (a Jesse James respirator) was worn until the 1940's - not the style of respirator used when the protection factor for cadmium was calculated. Furthermore, if respirators were worn at all, they were often improperly fitted. These considerations make it:

"... very difficult to accept Thun's assumption that the same respirator protection factor of 3.9 which was purportedly operative for cadmium exposures was similarly protecting against arsenic exposures. These and other reservations exist as to whether arsenic has been adequately controlled for as a potential confounder (Ex.19-42-b-3, pp. 10-11)."

Dr. Smith described respirator use in this plant as follows:

"The company has had a policy since the 1930's that requires the use of filter cartridge respirators in work areas with the potential for fume and dust exposures. This policy has been tightly enforced in recent times, but the level of enforcement in the past is unknown, although workers and management have been well aware of the hazard of acute cadmium exposures (Ex. 4-64)."

Thus, both Dr. Smith and the commentors were aware of the same issues. However, as Dr. Thun indicated at the hearing:

"... I think that the impression that has been communicated that there was a major arsenic problem historically in the calcine and roaster area of this plant ... has been constructed rather recently ... when I have spoken about this with Professor Smith ... he says that Globe was not recognized to have an arsenic problem, that the symptoms of arsenic skin irritation were not being brought forth to the environmental group, and that the speculation about what past exposures may have been is based to a large extent on speculation. (Tr. 6/7/90, pp. 149-150)."

However, Dr. Thun did agree that Dr. Smith's impression of the respirator usage among arsenic exposed workers at this plant did differ from that of Dr. White. And while Dr. Thun is of the opinion that Dr. Smith is probably more accurate, OSHA acknowledges that the true degree of respirator use may not be known. An error in estimates of respirator protection leads to uncertainty in exposure assessments which could result in either overestimates or underestimates of exposure. OSHA developed three sets of exposure data for arsenic exposures, one one of which was respirator adjusted. The original respirator-adjusted estimates were within the range of the two estimates thereby supporting the respirator adjustment further (see Section VI-QRA).

The protection factor that Dr. Smith calculated was based on intermittent usage measurements that included times when the respirator was both worn and when the respirator was hanging around the neck of the worker during the workday shift. This would cover those situations in which workers were unable to wear respirators due to facial rashes. Dr. Smith estimated the inhalation exposures for nine workers by measuring the cadmium concentration inside the respirator while it was worn or hanging around the worker's neck. Air concentrations of cadmium were measured simultaneously inside the respirator and at the worker's lapel using a dual sampling system, and workers were sampled for three consecutive days over a full days' workshift. On the average the inhalation exposure was 26% of the lapel concentration, i.e., the average effective protection factor, or the ratio of personal exposure to inhalation exposure, was found to be a geometric mean factor of 3.9. This factor was then used to calculate the inhaled cadmium exposure per worker. Personal exposures were divided by 3.9 to estimate the median inhalation exposures of workers in those department where respirators were worn, and these values were used in the development of each workers' cumulative exposure assessment.

Thus, Smith made three corrections to the existing area exposure level data base. The first correction was to adjust the historical area sampling data by a correction factor to reflect personal sampling. The second correction was the weighting of high exposure areas by number of worker-shifts, and the third was the division of the estimated personal exposures, in those departments and calendar periods in which workers wore respirators, by 3.9 -the geometric mean respirator protection factor measured in the Globe plant.

OSHA has substantially verified the latter two adjustments, (see Section VI-QRA), and believes that the ratio adjustment for area samples did not operate towards misclassification of exposure (Ex. 12-14). OSHA agrees with Dr. Smith's conclusions that he attempted to address all of the factors that could affect the exposures to which workers were exposures. It was obvious, for instance, that the effects of emission control systems and improved ventilation were taken into consideration in each of Smith's estimated exposures for each department as a function of time. (See pg. 323, Table I, Ex 4-64.) Through his study design, Dr. Smith tried to estimate the actual inhalation exposures for individual workers, i.e., the actual dose of cadmium that a worker received to the lung, rather than the ambient cadmium levels, taking into consideration variations in respirator use.

Further, according to Dr. Roth:

"... regarding the exposure estimates ... we have no reason to believe that any of the measurements are biased in either one direction or the other and from a statistical standpoint, that's the key issue. (Tr. 6/6/90, pp. 55-56)."

It is OSHA's opinion that despite the issues raised regarding the exposure assessment, the forethought by Smith and Thun in applying this exposure assessment to the individual's in the cohort, taking into consideration the use of available data and using advice from the representatives of the workplace, would tend to minimize the errors in the exposure assessments. The underlying strength of the Thun study was the availability of unusually good industrial hygiene records that allowed for the computation of cumulative individual exposures, and from that, some quantitative dose response data. OSHA agrees with the Environ report which stated that despite questions, the quantified exposure estimates were: "... a further strength of this study for the purpose of risk assessment ... (Ex. 19-42b)." Furthermore, the validity of these assessments has been determined, to some extent, by comparison of Thun's results with those of other independent researchers and in other studies.

Regarding this last point, Dr. Thun stated that:

"The calculations of cumulative exposure have been validated, to some extent, by comparing them to in-vivo data collected by the Brookhaven National Laboratory in 1979 (Ex. 4-27), A strong correlation was found between the calculated cumulative exposure, using these data, and the Brookhaven measurements of liver cadmium ... The strength of this approach is that it considers both the intensity and duration of exposure ... this gives you an actual estimate of cumulative exposure ... our analysis grouped individuals into categories ... to correspond to the level of detail that was available in the personnel records.(Ex. 33)"

Dr. Thun was referring to a study by Ellis et al (Ex. 4-64). For Dr. Ellis' analysis, cumulative exposures for individual workers were estimated in the same way as for Thun's analysis, on the basis of the same plant personnel records. A chronological record of each worker's job assignments was obtained. The time-weighted inhalation exposure (TWE) was calculated by multiplying the length of time (t(1)) in a given work area by the estimated inhalation exposure for that area and year (E(1)) and then summing these values for the total exposure history, or

               TWE = SUM E(1)t(1),

where E(1) was obtained from the original estimates reported by Smith (Ex 4-64).

Another partial validation of the exposure assessments of these workers was done by Smith who calculated the quantity of cadmium that would be retained in the kidneys of the workers in his study, based on his calculated time-weighted exposure measurement of cumulative cadmium exposure per individual worker, and compared this with the "critical concentration" of cadmium that would result in kidney dysfunction, indicated by a number of other researchers (Ref. in Ex. 4-68). Smith found that:

"The quantity 41 mg per kidney is somewhat higher than the 15-30 mg per kidney reported as the critical value for cadmium by several investigators (WHO Task Group, 1975; Nomiyama et al, 1979: Ellis et al, 1980); however, it is quite close, considering the assumptions and the errors in the exposure estimates. These findings provide further support for the quality of the exposure estimates.(Ex4-64).

To the extent that levels of cadmium in urine reflect cumulative cadmium exposures in air, and are not influenced by the presence of kidney damage, these results by Dr. Smith appear to confirm Dr. Thun's exposure assessments.

Thus, the cumulative exposures in Thun's study were somewhat validated by two independent researchers who further evaluated the health status of a subset of the same workers included in Thun's study. Using in-vivo measurements of cadmium in liver and kidney tissues and kidney cadmium content, the two researchers, Smith (Ex.4-64) and Ellis (Ex 4-64), found associations between cumulative cadmium exposures and effects of cadmium that were both similar to those of other researchers and were plausible, given the available medical data on cadmium's effects. These "validity checks," of the exposure assessment provides another important advantage of the Thun study over other studies. The confidence that can be placed in the results obtained from the use of such exposure assessments, for both cadmium and arsenic exposures, is increased.

As Dr. Thun stated:

"... all of the cohorts have potential exposure to other occupational lung carcinogens besides cadmium. For the metallurgical groups ... there is potential exposure to arsenic and for the nickel-cadmium battery plants, there is potential exposure to nickel ... but the exposure data on this study are really exemplary. There are few occupational cohorts for which historical exposures are as well documented."(Tr 6/7/90, p. 87)

f. OSHA's Conclusions Regarding the Thun Study. The major issues regarding the Thun study include potential confounding from arsenic exposure and from cigarette smoking. (See also Exs. 19-43; 66.)

Regarding the multiplicative or synergistic effects of smoking together with occupational cadmium exposures suggested by some commenters (Ex. 66), OSHA is aware of no evidence that supports the hypothesis of such synergistic effects. Of far greater concern to the Agency is the contamination of cigarettes by cadmium in the workplace (Ex. 29). Both cadmium and smoking are associated with lung cancer. Arsenic does not appear to have played a major role in the excess risk of lung cancer observed in the study. With data indicating that confounding from arsenic and cigarette smoking did not play a major role in the lung cancer excess, it would be imprudent to overlook the epidemiological data showing that cadmium appears to have been responsible for the excess lung cancer risk.

The obvious strength of the Thun study comes from the finding of a dose-response relationship. Dr. Kazantzis acknowledged that the study "...has shown evidence of the carcinogenicity of cadmium with a dose-response relationship" (Tr. 6/11/90, p. 142), and Dr. Rodericks stated that he would still point to the dose-response information from the Thun study "as a most important set of data"(Tr. 7/18/90, p. 26).

It is OSHA's opinion that the findings of the Thun study demonstrate a significant dose-response between cadmium exposure and lung cancer that could not be explained by confounding from cigarette smoking and arsenic exposure.

9. Summary and Conclusions

Complete in-depth review of the rulemaking record and careful analyses of the available epidemiologic data have led OSHA to conclude that a significant association has been demonstrated between occupational cadmium exposure and lung cancer. Evidence for this association is further strengthened by studies demonstrating the induction of lung cancer in experimental animals exposed by inhalation to several different cadmium compounds. The data also have revealed a significant association between cadmium exposure and prostate cancer, though recent studies show less of an association than earlier study results. IARC (Ex. 8-656) concluded that cadmium is probably a human carcinogen. The data presented in this rulemaking record and the analyses conducted and presented in the quantitative risk assessment section of this standard rule out any major potential confounding in the Thun et al. study from cigarette smoking and arsenic exposure on the excess risk of lung cancer observed among the cohort members. Thus, the association between cadmium exposure and lung cancer as noted by IARC is further strengthened by these new data and analyses. The epidemiologic findings demonstrating the carcinogenicity of cadmium establish cadmium as an occupational carcinogen.

Some commenters questioned the consistency of the epidemiologic data (Ex. 19-42b-3; The Environ Corporation, March 17, 1989). For example, Environ stated that "the results of lung cancer mortality studies in industrial settings where cadmium is present are mixed: the majority of reports do not show an excess lung cancer incidence in association with cadmium exposure and several provide strong evidence of alternative causes of the cancer excess observed." According to Environ, its conclusions are consistent with those reached by IARC, "... although our evaluation includes more recent studies than those reviewed by this agency (IARC)" (Ex. 19-42b-3). Environ continued:

"On the basis of the commonly-applied criteria for causation based on epidemiological evidence that: (1) a positive dose-response relationship be observed, (2) the association is not explicable by bias in recording, detection, confounding or chance; and (3) the association is observed repeatedly in different circumstances, it appears evident that the epidemiologic data on the potential lung carcinogenicity of cadmium are insufficient to establish causality for humans (Exs.12-41 and Ex. 19-42b, p. 19)."

SCM Chemicals, citing a review of epidemiologic studies relating to cadmium by a 12-member panel of expert epidemiologists, criticized the epidemiologic studies relating to cancer for failing to consider the effects of confounding factors in drawing their conclusions (Ex. 19-42A).

OSHA noted earlier that for all five major cohorts of cadmium-exposed workers, the lung cancer SMRs are elevated, and in updated studies of four of the cohorts, the SMRs for lung cancer are statistically significantly elevated (Ex. L-140-50). These four cohorts were U.S. cadmium recovery workers with two or more years of employment (Thun et al., Ex. 8-658a); nickel-cadmium battery workers in the U.K. (Sorahan, Ex. 12-12-A); nickel-cadmium battery workers in Sweden (Jarup et al., L-140-50); and workers in 17 British plants (Kazantzis et al., Ex. 8-684). Not only were the SMRs elevated for four of the five cohorts, but according to Drs. Thun, Elinder, and Friberg, in three of the five cohort studies, (Exs 8-658a, 12-12-A and 8-684), a dose-response relationship was evident (Ex. L-140-50). Regarding the other commonly applied criterion cited by Environ, bias in recording, detection and chance confounding, analyses by Thun et al. and by Stayner show that arsenic contamination and confounding from cigarette smoking could not have accounted for the excess lung cancer risk in the Thun study. Furthermore, the study demonstrated a dose-response for lung cancer in relation to cadmium exposure.

Thus, contrary to the opinion stated by Environ, all three of the commonly applied criteria for causation are satisfied by the existing epidemiological data for occupational cadmium exposure and lung cancer:

(1) A positive dose-response relationship has been demonstrated;

(2) The association does not appear to be explicable by chance, bias in selection of subjects for study or by confounding from other exposures;

(3) The association has been observed repeatedly in different occupational settings.

A large number of epidemiologic studies demonstrate an association between occupational exposure to cadmium and kidney dysfunction. Exposure data were of high quality from six epidemiologic studies that allowed estimates of a dose response between cadmium and kidney dysfunction. No participant in the rulemaking challenged the etiologic basis for cadmium and it role in kidney dysfunction.

Commenters have testified that in contrast to toxins such as Cr VI and lead, the correspondence between human and animal data for all adverse health effects of cadmium is "very good" (Tr. 6/7/90; pp. 3-71). Human studies indicate acute and chronic cadmium toxicity in the form of renal, liver, and pulmonary effects, and animal studies have demonstrated similar effects. Chronic animal studies by inhalation show cadmium-induced lung cancers in rats and some cases of prostatic cancers; the same effects have been observed in human studies. Both renal damage and lung disease (bronchitis) have been observed in chronically exposed humans and animals. There is also good correlation between ITAI-ITAI disease in humans and demineralization of the bone in animals. Liver damage is also seen both in humans and animals. However, it is noted that some effects have been seen only in animals (6/7/90; Tr. 3-3 to 3-71).

OSHA has received information indicating that additional epidemiologic studies are being conducted and some participants feel that the final standard should be delayed until these studies are completed. (Caza, Ex. 19-29; Kazantzis, L-140-50, pp. 701-701; DCMA, Ex. 19-40). OSHA believes, however, that delaying the final standard is not warranted and that the record supports this conclusion. OSHA agrees that the human-animal correlation for cadmium:

"... is probably one of the best sets of correlations that one sees between experimental animals and humans for a toxic agent." (M. Costa, Tr. 6/7/90, p. 8)

[57 FR 42102, Sept. 14, 1992; 58 FR 21778, April 23, 1993]

Regulations (Preambles to Final Rules) - Table of Contents Regulations (Preambles to Final Rules) - Table of Contents