-------
                                                                    Review Draft
                                                                    EPA-600/8-83-025A
                                                                    June  1983
                                     Draft
                              Do Not Quote or Cite
            UPDATED MUTAGENICITY AND CARCINOGENICITY ASSESSMENT OF

                                   CADMIUM
             Addendum to the Health Assessment Document for Cadmium
                          (May 1981) EPA-600/8-81-023
                                     NOTICE


This document is a preliminary draft.  It has not been formally released by the
EPA and should not at this stage be construed to represent Agency policy.  It is
being circulated for comment on its technical accuracy and policy implications.
                 Office of Health and Environmental Assessment
                       Office of Research and Development
                      U.S. Environmental Protection Agency
                            Washington, D.C.  20460

-------
                                  DISCLAIMER


    This report is intended for review purposes only and does not constitute

Agency policy.  Mention of trade names or commercial products does not

constitute endorsement or recommendation for use.
           The Health Assessment Document for Cadmium  (May  1981;
           EPA 600/8-81-023) is available only  from:
                   National Technical  Information Service
                   5285 Port Royal Road
                   Springfield,   VA  22161
                   Telephone:  703-487-4650
           Order No.:  PB-82-115163
           Cost:  $28.00  (subject to change)
                                       11

-------
                                    CONTENTS


Authors and Reviewers	iv

Summary and Conclusions   	 1

    Summaryo  ...........•••••1
    Conclusions    	 3



Introduction  	 5



Mutagenicity          .           	 6

    Gene Mutations in Prokaryotes 	 6
    Gene Mutations in Yeast	13
    Gene Mutations in Mammalian Cell Cultures 	15
    Studies in Drosophila and Other Insects   	    .16
    Chromosomal  Aberrations in Humans and Other Mammalian Systems .   .    .21
    Chromosomal  Aberrations in Plants	42
    Other Indirect Evidence	42
    Summary.	44



Carcinogenicity	46

     Animal  Studies	46
     Epidemiologic Studies	62



Ouantitative Estimation   	    100

    Introduction	100
    Procedures for Determining the Carcinogenic Potency   ....    104
    Cadmium Risk Estimates.   	    113
    Relative Potency  	    123



Appendix A - Comparison of Results by Various Extrapolation Models    .    129


Appendix B - International Agency for Research on Cancer (IARC)
             Classification for Weight-of-Evidence for Carcinogenicity
             of a Suspected Carcinogen	133


References.	136
                                     in

-------
                            AUTHORS AND REVIEWERS


    The Carcinogen Assessment Group, Office of Health and Environmental

Assessment, was responsible for preparing this document.  Participating members

are as follows (principal authors are designated by asterisks):
    Roy E. Albert, M.n. (Chairman)
    Elizabeth L. Anderson, Ph.D.
   *Larry D. Anderson, Ph.D.
   *Steven Bayard, Ph.D.
   *David L. Bayliss, M.S.
    Chao W. Chen, Ph.D.
    Maragaret M. L. Chu, Ph.D.
    Herman J. Gibb, B.S., M.P.H.
    Bernard H. Haberman, D.V.M., M.S.
    Charalingayya B. Hiremath, Ph.D.
    Robert McGaughy, Ph.D.
    Dharm V. Singh, D.V.M., Ph.D.
   *Nancy A. Tanchel, B.A.
   *Todd W. Thorslund, Sc.D.
    The Reproductive Effects Assessment Group, Office of Health and

Environmental  Assessment, was responsible for preparing the section on

mutagenicity.   Participating members are as follows (principal  authors are

designated by  asterisks):
    John R. Fowle III, Ph.D.
    Ernest Jackson, Ph.D.
   *K.S. Lavappa, Ph.D.
    Sheila Rosenthal, Ph.D.
    Carol Sakai, Ph.D.
    Vicki Vaughan-Dellarco, Ph.D.
    Peter E. Voytek, Ph.D.
                                       i v

-------
                           SUMMARY AND CONCLUSIONS

 SUMMARY
 Qualitative Assessment
    As was noted in the Office of Health and Environmental Assessment Health
 Assessment Document for Cadmium (May 1981), cadmium and various cadmium salts
 have produced injection site sarcomas in rats and interstitial  cell  testicular
 tumors in mice and rats after subcutaneous injection.  Specifically,
 injections site sarcomas were produced by cadmium powder, cadmium sulfide,
 cadmium oxide, cadmium sulfate, and cadmium chloride, and interstitial  cell
 tumors were produced by cadmium chloride, cadmium sulfate, and ferritin
 containing cadmium.  However, three drinking water, two gavage, and two
 dietary studies using cadmium acetate, cadmium sulfate, or cadmium chloride
 have shown no excessive risk of cadmium carcinogenicity in rats and mice.  It
 was also noted that human exposure to high levels of cadmium dust and/or fumes
 in smelter and battery plants produced a slight, but statistically
 significant, increase in prostate cancer in several different small-scale
 epidemiologic studies.
    After the above document was prepared, a lifetime rat inhalation study was
 published showing a dose-related induction of lung carcinomas by cadmium
chloride aerosol.  The carcinogenic potency of cadmium by chronic inhalation
 in the rat appears to be at least two orders of magnitude greater than  by
chronic oral  administration.  In another recent study, lifetime observation  of
 rats given cadmium oxide by intratracheal  instillation did not  reveal  a
carcinogenic effect;  however, only three or fewer treatments were given during
the study, which involved a relatively low dose.
    Seven epidemiologic studies reviewed after the May 1981  OHEA document have
not appreciably changed the earlier limited evidence of prostate cancer in

                                      1

-------
humans.  In addition to cancer of the prostate, the risk of lung cancer was
found to be significantly correlated with cadmium exposure in two studies
(Lemen et al. 1976, Holden 1980).  The author of one attributes the excess
risk to the presence of arsenic (Holden 1980),  although arsenic was present in
both studies.  Because of the presence of arsenic along with-cadmium in these
studies, the evidence that cadmium is a causitive agent of lung cancer is
inadequate.
    A clear statement on the mutagenicity of cadmium is rather difficult
because of the conflicting results and lack of adequate test protocols used.
However, gene mutation studies in mammalian cell  cultures, rec-assays  in
bacteria, chromosomal nondisjunction studies in intact  mammals, and other
indirect evidence reported in the text suggest  that cadmium is weakly
mutagenic.

Quantitative Assessment
    Since humans are exposed to cadmium dust or fumes and rats were exposed to
cadmium chloride aerosol, a limitation inherent in the  use of the rat  study in
estimating human risk is the possible difference between humans and rats in
the lung retention of particulates or between the biological  effectiveness of
cadmium chloride aerosol administered to rats and the dust and fumes inhaled
by workers.  Since data are not clear on this point, assumptions of equal  lung
uptake and equal effectiveness were made in order to get an idea of the human
risk.
    Given these assumptions, combined with other assumptions and conventions
used in quantitative risk assessment procedures, the Takenaka et al. (1982)
data on lung carcinomas in rats during lifetime inhalation exposures to
cadmium chloride aerosol were analyzed.  The result of  the analysis is that

-------
the upper-bound cancer risk to humans who continuously breathe 1  ug/m3 of
elemental cadmium for a lifetime is 0.15.
    Based on respiratory and prostate cancer rates from the Lemen et  al.
(1976) study of cadmium smelter workers, the upper-bound cancer risk  from
lifetime exposure to 1 ug/m3 of cadmium in the air has a range of
1.9 x 10" 3 to 2.5 x 10~3.  These estimates are about 100 times less than
the estimate of lung cancer risks from the rat experiment.  Because only
fragmentary information is available concerning cadmium exposures of  the
workers, the extreme range of risk, based on the highest and lowest possible
assumptions about several exposure parameters, is a factor of 100 higher  and
lower than these estimates.  Further detailed analysis and laboratory studies
are needed before the large difference between the estimates based on animal
and human data are resolved.

CONCLUSIONS
    Applying the International Agency for Research on Cancer (IARC) approach
(Appendix B) for classifying the weight-of-evidence for carcinogenicity in
experimental animals, the injection site and testicular tumors in mice and
rats given cadmium metal  or cadmium salts and the lung carcinomas in  rats
exposed to cadmium chloride aerosol by inhalation provide sufficient  evidence
for the carcinogenicity of cadmium and certain cadmium compounds  in
experimental animals.  Although there is a possibility that orally-ingested
cadmium is carcinogenic in rats based on the finding of testicular tumors
induced by injection at a remote site, no response has been observed  with
ingested cadmium, and the potency via the oral  route is at least  200  times
less than via inhalation.

-------
    The available human epidemioiogic data provide limited evidence,  according
to the IARC criteria, that cadmium and certain  cadmium compounds  are
carcinogenic in humans.
    The overall evidence for carcinogenicity, applying the IARC criteriai
places cadmium and certain cadmium compounds  in the 2A category,  meaning that
they are probably carcinogenic in humans.
    The upper-bound unit risk estimate for continuous  inhalation  exposure  at a
cadmium concentration of 1 ug/m3 ranges from  1.9 x 10~3 to 2.5 x  10~3
based on lung clnd prostate cancer, respectively, from  one  smelter worker
study, although there is considerable uncertainty in these estimates  due to
uncertain exposure of the workers.  Nevertheless, these estimates are regarded
as more realistic than the estimate based  oh  the rat inhalation study, which
is about 100 times higher.

-------
                               INTRODUCTION

    This document is a review and assessment of the current information
relating to the mutagenicity and carcinogenicity of cadmium.  It contains a
detailed discussion of information on those subjects which became available
since the earlier Health Assessment Document for Cadmium was prepared by the
Office of Health and Environmental Assessment (OHEA) in May 1981.  It includes
all pertinent material from the 1981 document but does not attempt to repeat
details of the animal  carcinogenicity studies discussed there.

-------
                                MUTAGENICITY

    Cadmium has been investigated for its mutagenic potential  in both
prokaryotic and eukaryotic systems.  The prokaryotic systems include assays
for gene mutation and reparable genetic damage in bacteria.   The eukaryotic
systems include gene mutation studies in yeast, Drosophila,  and mammalian
cells and chromosomal aberration studies in human and other  mammalian cells
exposed to cadmium both in vitro and in vivo.   The following is an analysis of
the literature pertaining to the mutagenic effects of cadmium.

GENE MUTATIONS IN PROKARYOTES
    Gene mutation studies in prokaryotic systems have been  summarized in
Table 1.  A discussion of each study follows.

Salmonella Assay
   Meddle and Bruce (1977) and Bruce and Meddle (1979) tested  the mutagenic
effects of cadmium chloride in the histidine reverse mutation  assay using
Salmonella typhimurium tester strains TA100, TA98, and TA1537.   The test
compound (purchased from ICN Pharmaceuticals,  Plainview,  New York) was
dissolved in water and used at concentrations  of 0.05, 0.5,  5,  50, and 500
ug/plate with and without the application of a metabolic  activation system (S9
mix) derived from phenobarbital-induced rat liver homogenate.   According  to
these authors cadmium chloride did not induce  a significant  mutagenic response
over the control value.  The criterion set for a positive response was 50% or
1.5-fold increase in the revertant frequency over the negative  control  or
spontaneous frequency.  Revertant counts were given only  for strain TA100; the
spontaneous frequency of revertants in this strain was 140  colonies per plate.
The purity of cadmium chloride was not given in this report.

-------
TARLE 1.   MUTAGENICITY EVALUATION OF CADMIUM:   RENE  MUTATIONS  IN PROKARYOTES
Test System
Salmonella
typhimurium

Salmonella
typhimuriuni
Salmonella
typhimurium

Strain
TA98
TA100
TA1535
TA1537
TA1538
TA98
TA1535
TA1537
TA1535
TA1537
Cadmium
Compound
Cadmium
chloride
aqueous
solution
Cadmium
red in OMSO
Cadmium
chloride
(solvent
not specified)
nose
0.05
0.5
5.0
50.0
500 ug/plate
1 ug/ml
10, 20, 30,
45, 90 mM
S9 Activation
System
Phenobarbital-
induced rat
liver
Aroclor 1254-
induced mouse
1 iver
Uninduced mouse
liver
Results
Reported as
negative
Reported as
negative
Reported as
negative
Comments
1. Data are not presented
clearly as revertants/
plate for each strain.
2. Purity of compound not
discussed.
1. Data provided only for
the preincubation or
suspension assay. No data
on the spot test given.
2. Only a single dose was
employed; no dose-response data.
1. Spontaneous reversion data
and the experimental
reversion data have not
been given in terms of numbers.
2. Used uninduced mouse liver
S9 activation system.
3. No positive controls.
Reference
Bruce and Meddle
(1979)
Mi Ivy and Kay
(1978)
Polukhina et al .
(1977)

-------
                                                                 TABLE  1.   (continued)
oo

Test System
Salmonella
typhimurium









Bacillus
subtilis
Rec-assay






Strain
TA98
TA100
TA1535
TA1537
TA1538






H17
Rec+
M45
Rec-




Cadmium S9 Activation
Compound nose System
Cadmium 1 Aroclor-
diethyl- 5 induced rat
thiocar- 10 liver
hamate in 50
OMSO 100 ug/plate






Cadmium 0.05 M/plate None
chloride
aqueous
solution
Cadmium
nitrate
aqueous
solution

Results Comments Reference
Reported 1. Lowest effective dose was Hedenstedt et al.
positive for 10 ug/plate. (1979)
TA1538 and 2. Reported positive only for
TA98 in the one dose.
absence of 3. No dose-response rela.tonship.
S9 activation.
Reported weakly
positive -both
in the presence
and absence of
S9 activation.
Reported as Nishioka
weakly (1975)
(+) positive

Reported as
negative


                                                                                                                  (continued on  the following page)

-------
TABLE 1.  (continued)
Test System
Bacillus
subtil is
Rec-assay
Strain
H17
Rec+
M45
Rec"
Cadmi urn
Compound
Aqueous
solutions
of cadmium
chloride,
nitrite, and
sulfite .
nose
n.nns
M/plate
S9 Activation
System
None
Results Comments
Reported as 1. Compounds were pure.
weakly (+)
positive
Reference
Kanematsu et al .
(1980)

-------
    In an abstract published by Kalinina and Polukhina (1977), cadmium
chloride was reported to be nonmutagenic in the Salmonella assay.  However,
important variables such as the number of strains used, the dosage employed,
and the number of revertants per plate were not reported.  Polukhina et al.
(1977) also reported negative results with cadmium chloride on Salmonella
typhimurium strains TA1535 and TA1537 both in the presence and absence of an
S9 activation system derived from uninduced mouse liver homogenate.   In this
report a suspension assay with cadmium chloride concentrations of 10, 20, 30,
45, and 90 mM was employed.  Positive and negative control data were not
presented in this paper, so it is not possible to know whether or not the
assay system was functioning properly.  Toxicity of the test compound was not
reported by these investigators.
    Milvy and Kay (1978) studied the mutagenic effects of cadmium red (cadmium
sulfide and selenium), a dye used in the printing industry, using the
Salmonella spot test (Ames et al. 1973) and the preincubation assay  (Ames et
al. 1975)i  Salmonella typhimurium strains TA1538, TA98,  and TA1535  were
employed in these studies.  The test compound (10 ug)  was dissolved  in 0.01 ml
dimethyl  sulfoxide (DMSO) and added to 0.9 ml of incubation mixture  for 30
minutes at 37°C with shaking before plating 0.1 ml onto minimal  plates.
Experiments were carried out both in the presence and  absence of an  S9
activation system derived from Aroclor 1254-induced mouse liver homogenate.
Cadmium red was reported to be nonmutagenic in both tests.  However, data are
presented only for the suspension assay.  These investigators used only one
concentration, and hence, no dose-response relationship was demonstrated.  The
toxicity of the compound for each strain was not reported.  Consequently, this
study may be regarded as inconclusive.
                                      10

-------
     Hedenstedt et al.  (1979) studied the mutagenic effects of cadmium
diethyldithiocarbamate (used in rubber and plastic industries) in Salmonella
typhiumurium strains TA1535, TA1537, TA1538, TA98, and TA100.  The
concentrations used were 1, 5, 10, 50, and 100 ug/plate.  The compound was
dissolved in DMSO.  Concentrations of 50 and 100 ug/plate were toxic in many
of these strains.  The concentration of 10 ug/plate exhibited mutagenic
activity in strains TA1538 and TA98 in the absence of a metabolic activation
(S9) system obtained from Aroclor 1254-induced rat liver homogenate (Ames et
al.  1975).  In TA 1538 the revertant frequency increased more than twofold at
10 ug/plate, i.e., 26.3 _+ 3.7 revertants/plate compared to a control value of
11.8 +_ 2.6 revertants/plate in the absence of metabolic activation.  In the
presence of metabolic activation the revertant frequencies in treated and
controls were the same.  In TA98, the revertant frequency was 58.8 _+ 2.3 at 10
ug/plate (almost a twofold increase) compared to the control frequency of 31.5
_+ 4.2 revertants/plate in the absence of metabolic activation.  No data were
given for studies in the presence of metabolic activation.  Positive control
data were not presented even though the authors indicate that positive
controls were employed in the experiment.  These results may be regarded as an
inconclusive because the twofold increase was observed for only one isolated
data point, and there was no evidence of dose-response.

Escherichia coli WP2 Assay
    Venitt and Levy (1974), in a report on the mutagenicity of chromates in
the Escherichia coli WP2 forward mutation system, mentioned that they also
tested cadmium compounds for mutagenicity and found them to be negative.
These authors do not mention what types of cadmium compounds they employed or
present data to support their negative conclusions.
                                       11

-------
Bacillus subtil is Rec-Assay
    Nishioka (1975) investigated the mutagenicity of cadmium chloride and
cadmium nitrate using the rec-assay of Kada et al. (1972).   In the rec-assay,
which measures reparable ONA damage, differences in growth  sensitivities of
Bacillus subtil is strains H17 (recombination-competent wild type rec+)  and
M45 (recombination-deficient rec") to mutagenic chemicals are measured.
When a chemical is more inhibitory to rec~ than to rec+ cells, then it  is
suspected of being mutagenic.  Concentrations of 2.5 x 10~7 cells/0.1 ml
were streaked on agar plates from the center of the plate in different
directions.  Aqueous solutions of cadmium chloride and cadmium nitrate
solutions (0.05 M) were applied in 0.05 ml  aliquots to filter paper disks
(diameter of 10 mm) and placed in the center of the agar plate, i.e., the
starting point of the streaks of rec+ and rec~ cells.   All  plates were
incubated at 37°C for 24 hours.  The inhibition of growth was indicated by the
distance (mm) between the paper disk and growth of the bacterial  streaks.
This magnitude of inhibition is called "rec-effect" and is  expressed as:
no difference between rec+ and rec~ plates (-), less than 5 mm difference
(+), 5-10 mm difference (++), and more than 10 mm difference (+++).   Cadmium
nitrate showed no difference in growth inhibition (-), whereas cadmium
chloride exhibited a weak positive response (+).  No positive or negative
controls were employed in this experiment.   Each experiment was repeated  three
times.  These experiments did not use a metabolic activation system. The
cadmium compounds used were of reagent grade but their precise purity was not
reported.
    Similar results were obtained by Kanematsu et al.  (1980) in the  rec-assay
of Kada et al. (1972).  Cadmium chloride, cadmium nitrite,  and cadmium  sulfate
were employed at a concentration of 0.005 M in 0.05 ml aqueous solution.   All
                                      12

-------
these compounds exhibited a weak mutagenic response (+)  (of 4-5 mm growth
inhibition zone).  According to these authors, test compounds  used were  of  the
highest purity commercially available in Japan.

GENE MUTATIONS IN YEAST
    Cadmium chloride has been investigated for the induction of gene  mutations
in the yeast Saccharomyces cerevisiae (Table 2) (Takahashi  1972; Putrament  et
al. 1977).  Takahashi (1972) studied the induction of petite mutations
(p-mutation) and auxotrophs in the Saccharomyces cerevisiae heterozygous
diploid strain C3116.  He treated 104 cells with 10 (5.5 x  1Q-5M), 12 (6.6
x 10'5M), and 20 ppm (1.1 x 10~4M) for 2 days (48 hours) at 25°C.  After 2
days of growth, the cell number was determined and the cell  suspension was
diluted to give a concentration of 2.8 x 103 cells per ml.   One-tenth of the
diluted suspension was spread on the YEPD-agar plate and incubated at 28°C.
When small colonies appeared on the plate, they were replica plated onto
YEP-glycerol-agar medium and minimal medium.  After overnight  incubation at
28°C, induced p-mutants and auxotrophs were scored.  At  the dose of 12 ppm
(1.1 x 10'4M), no p-mutants or auxotrophs were found in  the 786 colonies
counted; at the dose of 10 ppm, 10 p-mutants and three auxotrophs were
detected in the 871 colonies counted; and at the dose of 20 ppm, there were 12
p-mutants and 9 auxotrophs in 1,182 colonies indicating  that cadmium  chloride
may be mutagenic.  In the controls there were five p-mutants and two
auxotrophs in 2,875 colonies counted.  According to this paper, however,
mutants were induced at the dosage of 10 ppm and 20 ppm  but not at the dosage
of 12 ppm.  Such erratic fluctuations in mutation frequency and the low  number
of mutants suggest that the positive values may similarly be questionable.
Since p-mutants occur by damage involving the mitochondrial  DNA rather than
                                      13

-------
TABLE 2.   MUTAGENICITY  EVALUATION OF CADMIUM:  GENE MUTATIONS  IN  YEAST  AND MAMMALIAN  CELL CULTURES
Test System
Saccharomyces
cerevisiae
(Yeast)
P-mutants and
auxotrophs
Saccharomyces
cerevisiae
P-mutants
Mouse lymphoma
Chinese hamster
cells
Chinese hamster
cells
Cadmium
Strain Compound
C3116 Cadmium
chloride
197/2d Cadmium
chloride
L5178Y TK+/~ Cadmium
chloride
Lung (Don) cells; Cadmium
resistance to 8- acetate
azoguanine Cadmium
chloride
Ovary cells Cadmium
(CHO) chloride
Activation
Dose System
10 None
12
20 ppm
8 ppm None
0.05 None
0.06
0.08
0.11
0.15 ug/ml
2.5 None
5
10 ug/ml
2.5 None
5
5
10 ug/ml
Results
Reported as
positive
Reported as
negative
Reported as
weakly
positive
Reported as
positive
Reported as
weakly
positive
Comments Reference
1. P-mutants may not represent Takahashi (1972)
true gene mutations because
they arise by damage in
mitochondrial DNA.
2. Vague protocol .
1. Only one concentration of Putrament et al .
test compound was used. (1977)
2. This concentrations was too
toxic for the cells.
3. No mutants observed in those
few suvivors.
1. Application of t-test Amacher and
to determine the Paillet (1980)
significance has been
challenged by Clive et
al. (1981).
1. Very low survival due Casto (1976)
to high toxicity.
2. Observations not repeated
and confirmed.
1. Data not presented. Hsie et al . (1978)

-------
 nuclear  DNA,  caution  should  be exercised  in assessing the mutagenic potential
 of chemicals  with this system.
     Putrament et al.  (1977)  also  reported a negative result in a test for
 induction of  p-mutation by cadmium chloride in Saccharomyces cerevisiae.  The
 concentration of cadmium chloride tested  (8 mM) was very toxic, however, and
 less than 1%  of the cells survived a 6-hour incubation in YEP-glucose medium.
 No increase of p-mutants was observed and no data were presented.  This study
 is regarded as inconclusive.

 GENE MUTATIONS IN MAMMALIAN CELL CULTURES
    Gene mutation studies in cultured mammalian cells have also been
 summarized in Table 2.  A discussion of each study follows.

 Mouse Lymphoma Assay
    Amacher and Pail let (1980) reported that cadmium choloride (ICN
 Phamaceuticals) was mutagenic in the mouse lymphoma L5178Y TK+/~ assay.
 When cadmium  chloride, dissolved in normal saline, was tested at
 concentrations of 2.35 x 10~7M (cell  survival  100 +_ 11%), 3.57 x 10~7M
 (cell survival 78^24%), 4.5 x 10-7*1 (cell survival 0.62 ±4%), 6.00 x
 10-7M (cell  survival 38 _+!!%), and 8.00 x 10~7M (cell  survival 12^1%),
there was a dose-related increase in the mutation frequency.  The mutation
frequencies per 104 survivors for the above doses were 0.48 +_ 0.01, 0.58 _+
0.06, 0.56 +_ 0.05,  0.63 _+ 0.16, and 0.68 _+ 0.04, respectively.  The mutation
frequency at the highest nontoxic dosage of 6.00 x 10"7M was approximately
1.5-fold higher than the control  frequency of 0.40 _+ 0.03 (survival 100% _+ 5).
The dose-response curve obtained by Amacher and Pail let has been criticized by
Clive et al.  (1981), who claim that the application of a t-test for low
                                      15

-------
numbers of samples to determine significance is misleading.

Chinese Hamster Cell  Assay
    Casto (1976), in  a report submitted to Dr. Richard Troast o'f the Office of
Pesticide Program's, U.S. Environmental Proctetid'ii Agency, stated that cadiriiLim
acetate and cadmium chloride are mutagenic in Chinese Hamster-liihg cells (Don)
as determined by induction of mutations that confer resistance to
8-azdgUanihe.  Cells  were treated with 2.5 (1.36 x 10~8M), 5 (2.72 x
10~^M)i and 10 ug/ml  (5.45 x 10"8M) of cadmium acetate and cadniium
chloride, respectively, for 18 hours followed by 48 hours expression time.
Cadmium acetate induced mutation frequencies of 2.8, 50, and 10 per 10^
survivors, respectively, for above dosages.  The survival rate was 0.70, 0.92,
arid 0.43 percent, respectively.  Cadmium chloride induced mutation frequencies
of 6, 7, 14, arid 37 per 106 survivors.  The negative control  rate was 2 per
10^ survivors.  According to this investigation both cadmium acetate arid
cadmium chloride are  weakly mutagenic, but the results are questionable'
because of low survival rates at these high concentrations.  Hsie et al.
(1978) also reported  cadmium chloride to be weakly mutagenic at the HGP'RT
locus in the Chinese  hamster ovary cells, but no data were presented.
Therefore, although three independent studies have reported cadmium salts to
be weakly mutagenic in cultured mammalian cells, however, in none of the
studies was the demonstration of mutagenicity unequivocal.

STUDIES IN DROSOPHILA AND OTHER INSECTS
    Genetic effects of cadmium in Drosophila are summarized in Table 3.   A
discussion of each study follows.
                                      16

-------
TABLE 3.   MUTAGENICITY EVALUATION OF CADMIUM:  GENE MUTATIONS AND CHROMOSOMAL ABERRATIONS IN OROSOPHILA AND OTHER INSECTS
Test System
Drospphila melanoqaster
sex-linked recessive
lethal test
Drosophila melanogaster
larval development ..
sex chromosome loss
sex-linked recessive
lethal test
Drosophila melanogaster
dominant lethal mutations
Drosophila melanogaster
sex-linked recessive
lethal test
Cadmium
Compound
Cadmium
chloride
Cadmium
chloride
Cadmium
chloride
Cadmi urn
stearate
Dosage
50.0 mq/1
(2.72xlO-4M)
65 mg/1
62 mg/1
+ 3,000 R
x-rays
5
10
20 ppm
10-20 mg/1
50-100 mg/1
100 mg/1 .
3 mg/m3
Treatment
Period
Larvae
feeding
Larvae
feeding
Larvae
feeding
5-10 days
(feeding
larvae)
10-12 days
(feeding
adults)
(feeding
larvae)
(inhalation
adult)
Results
Reported as
negative
Reported as
negative
Reported as
positive
Reported as
negative
Comments
1. Data not presented.
2. Only one dose was used.
1. Treatment was done in larvae
only.
1. Dose-response reported.
2. Confirmation of these results
in an independent laboratory
would be of interest for
comparative purposes.
1. Rationale for selecting
the dosage not given.
Reference
Sorsa and Pfeifer
(1973)
Ramel and Friberg
(1974)
Vasudev and
Krishnamurthy (1979)
Sabalina (1968)
                                                                                                  (continued on the following page)

-------
                                                                  TABLE 3.   (continued-)
CD
Test 'System
Drosophila nelanogaster
chromosomal nondisjuncti on
sex chromosome loss
Cadmium
Compound
Cadmium
chloride
Treatment-
Dosage Period Results
62 ppm- Reported as
negative
Comments
1. No data have
been presented.
Reference'
Ramel and
(1979)

Magnusson
Drosophila melanogaster
sex-linked, recessive
lethal test
Cadmium
. aqueous
chloride 50 ppm
solution
Larvae.
feeding
Reported as
negative
1.,
2.
Toxicity was determined:
Development and survival
was. affected by- cadmium.
Inoue/ and
(1978)
Watanabe
     Poekilpcerus  pictus
     (grasshopper)
     testis  (meiotic
     chromosomal)
Cadmium chloride   0.001%
aqueous solution   0.01%'
                   0.05%
                   per animal
Reported as   1.
positive':
              2.
The-effect nay be  cytotoxic
rather than genetic.
No controls.
Kumaraswamy and
Rajasekarasetty •
(1977)

-------
    Sorsa and Pfeifer (1973) reported that cadmium chloride at concentrations
of 1.25 (6.81 x 10-6M), 2.5 (1.36 x 1Q-5M), 5.0 (2.72 x 10-5M), 10.0
(5.45 x 10-5M), 20.0 (1.09 x 10-4M), and 50 mg/1  (2.27 x 10"4M) of media
caused significant delay in the development of larvae compared to controls.
In the sex-linked recessive lethal mutation test  (Muller-5 test), only one
concentration of 50 mg/1 (2.72 x 10~4M) was used  with no indication of
mutagenic response.  The number of chromosomes tested and the criteria set  for
scoring the lethals were not reported, however, and no data were presented.
    Ramel and Friberg (1974), using a dose of 62  mg'(3.32 x 10~4M) of
cadmium chloride/1 of media, which was the maximum nonlethal  dose in the
toxicity test, found a delay in larval development.  They also studied the
induction of sex chromosome loss and sex-linked recessive lethal  mutations.
In the sex chromosome loss test, a total of 23,360 chromosomes from the
treated group and 28,143 chromosomes from the control group were tested.  The
frequencies of sex chromosome losses were 0.3% and 0.2% for the treated and
the control groups, respectively.  In the sex-linked recessive lethal  test,  a
total  of 1,045 chromosomes from the treated group and 742 chromosomes from  the
control group were tested.  The frequencies of sex-linked recessive lethals
were 8.1% and 6.1% for the treated and control groups, respectively.  Although
these experiments did not indicate a significant  effect of cadmium chloride  in
the sex chromosome loss or sex-linked recessive lethal tests, the number  of
chromosomes tested in the sex-linked recessive lethal test was too low to
preclude all but a strong mutagenic response.
    The mutagenic activity of cadmium sterate was studied by Yu.  A. Revazova
quoted by Sabalina (1968) in Drosophila melanogaster using the sex-linked
recessive lethal  test.   Flies were fed a medium containing
10-20 mg (5.45 x 10~5M to 1.09 x 1Q-4M) and 50-100 mg (2.72 x 10'4 to
                                      19

-------
5.45 x 10~4M) of cadmium stearate/1  substrate for 5-10 and 10-12 days,
respectively.  The number of sex-linked recessive lethal  mutations in 805
chromosomes analyzed was 1 (0.12%) for the 5-10 day treatment, and the number
of sex-linked recessive lethal  mutations in 2,192 chromosomes examined was 8
(0.36%) for the 10-12 day treatment.  When larvae were treated with cadmium
sterate concentration of 100 mg/1  substrate for 12 days and scored for
sex-linked recessive lethal  mutants in 380 chromosomes, no mutants were
discovered.  Cadmium sterate was also administered by inhalation to adult
flies for 32 hours (4 holirs daily for 8 days).  The mean  cadmium concentration
was 3 mg/m^.  The percent sex-linked recessive lethal  mutations among the
498 chromosomes was reported to be 0.2%.  The control  frequency of sex-linked
recessive lethal mutations was  not provided in the paper.  The number of
chromosomes tested was not adequate in this study.  This  study provides no
evidence of mutagenicity of cadmium in Drosophila, but the scale of the study
was too small to be considered  an adequate test even if appropriate controls
were presented.
    Induction of dominant lethal mutations in Drosophila  melanogaster with
cadmium chloride has been reported by Vasudev and Krishnamurthy (1979).  The
doses used were 5 (2.72 x 10~5M), 10 (5.5 x 10"5M), and 20 ppm
(1.1 x 10~^M).  The frequencies of dominant lethals were  11.8, 14.3, and
14.3%, respectively, in 1,244,  1,375, and 1,390 eggs counted.  The control
frequency was 4.83% in 1,076 eggs counted.  These investigators performed the
experiment according to the procedure described by Shankaranarayanan (1967)
and determined the statistical  significance to be at the  5% level, although
they did not mention the type of statistical  test employed.  Based on these
observations, this study is evaluated as an indicator of  a positive response.
A comparable study in an independent laboratory would be  of interest for
comparative purposes.
                                      20

-------
    Inoue and Watanabe (1978) studied the effects of cadmium chloride in the
sex-linked recessive lethal test (attached-X method) in Drosophila
melanogaster, Oregon-R flies.  In this test, the induction of mutations is
measured by the reduction in the proportion of males.  The sex ratio (0.528)
in the experimental group treated with 50 ppm (2.72 x 10'^M) was not
statistically different from the sex ratio of controls (0.54), indicating
cadmium chloride is nonmutagenic.  The dosage selected was a maximally
tolerated dose.  Roth positive (AF-2) and negative controls were used in the
experiment.
    Ramel and Magnusson (1979) failed to detect nondisjunction and sex
chromosome loss in Drosophila following treatment of larvae with 62 ppm (3.32
x 10~4M) of cadmium chloride.  No data were presented; therefore, this study
cannot be evaluated.
    Chromosomal aberrations were observed in the testes of the grasshopper,
Poekilocerus pictus, injected abdominally with 0.001 (5.45 x lO'^M), 0.01
(5.45 x 10-9M), and 0.05% (2.27 x 1Q-7M) cadmium chloride in 0.05 ml
volumes (Kumaraswamy and Rajasekarasetty 1978).  Stickiness of chromosomes,
bridge formation at anaphase-I, and tetraploidy at metaphase were noted.  The
test cannot be considered adequate because no controls were used and no
tabulated data were presented.  The possibility of technical artifacts must
also be considered, particularly because chromosomal preparations were made  by
a squash technique, and there were no controls.

CHROMOSOMAL ABERRATIONS IN HUMAN AND OTHER MAMMALIAN SYSTEMS
    Chromosomal damage studies of cadmium, both in vitro and in vivo, are
summarized in Tables 4 and 5.  A discussion of each study follows.
                                      21

-------
TABLE 4.   MUTAGENICITY EVALUATION OF CADMIUM:   IN VITRO CHROMOSOMAL ABERRATIONS
Duration of
Test System Cultures
Human 72 hrs
blood
lymphocytes




Human 48 hrs
blood 72 hrs
lymphocytes

Human 48 hrs
blood
lymphocytes

Cell line 24 hrs
WI38 and MCR5
Human 48 hrs
blood
lymphocytes
G0 stage
Cadmium Duration of
Compound Dosage Treatment
Cadmium 6.2 x in~2 4 hrs
sulfide ug/ml 8 hrs
(solvent not
specified)



Cadmium 5 x 10"5M 24 hrs
chloride 5 x 10"6M 48 hrs
aqueous 72 hrs
solution
Cadmium
chloride 48 hrs
aqueous
solution
24 hrs

Cadmium 10"8 3 hrs
acetate 10"7
aqueous 10""
solution 1G"5
Activation
System Results
None Reported as
positive





None Reported as
negative


None Reported as
negative


Reported as
negative
None Reported as
weakly
positive


Comments
1. Blood lymphocytes were
derived from only one
indi vidual .
2. Only 50 metaphases for
each end point were scored.
3. Only one concentration of
the test compound was used.
1. Toxicity was determined.
2. Appropriate dosages used.
3. 100 metaphases scored for
each point.
1. Data are not provided.
2. Concentrations of the
test compound not
speci fied.


1. No dose-response.
2. Experiments were not
repeated to confirm
the positive finding.

Refei ence
Shi raishi
et al. (1972)





Dekundt and
Deminatti
(1978)

Paton and
Al 1 ison
(1972)



Gasiorek and
Bauchinger
(1981)

                                                                              (continued on the  following  page)

-------
TABLE 4.  (continued)
Duration of
Test System Cultures
Chinese hamster
"Hy" cell line




Chinese hamster
CHO cell line









Mouse Mammary
Carcinoma
FM3A

Cadmium
Compound
Cadmium
sulfate
aqueous
solution


Cadmium
chloride
'in 0.1 MHC1








Cadmium
chloride
aqueous
solution
Duration of Activation
Dosage Treatment System
in-4M 1 hr and None
harvested
at 2,4,6,8,
10,12,15,18,
21,24,27,30,
days
2xlO-6M 12, 24, 36, None
and 48 hrs









6.4xlO-5M 24 and 48 hrs None
3.2xlO-5M 24 and 48 hrs
1.0xlO-5M 24 and 48 hrs


Results
Reported as 1.
positive 2.

3.


Reported as 1.
positive only
in the presence
of newborn calf
(bovine) or
human serum.
Negative in 2.
the presence
of fetal calf
serum. 3.

Reported as 1.
negative 2.



Comments
Colchicine-1 ike effect
The type of sera has
not been specified.
Protocol for chromosome
preparation has not been
specified.
Threshold dosage was
established as 1 xlO^M
for chromosomal
aberration with
newborn calf and
human sera.
Classif icaton of
aberration types not
given.
Active only in the presence
of fetal calf serum.
6.4 x 10"5 too toxic.
Experiments were
repeated to confirm
the results.

Reference
Rohr and
Bauchinger
(1976)



Deaven and
Campbell
(1980)








Umeda and
Nishimura
(1979)


-------
TABLE 5.   MUTAGENICITY  EVALUATION  OF CADMIUM:  .IN  VIVO  CHROMOSOMAL  ABERRATIONS IN HUMANS

Number of
Exposed Number .of
Species Workers Controls
Human 14 5
blood
lymphocytes

Human 40 13
blood
lymphocytes

Human 7, 12 6, 9
blood
lymphocytes


Human 5 3
blood
lymphocytes from
cadmium-exposed
workers
Itai-Itai patients 4 4
blood
lymphocytes
Human 24 15
blood
lymphocytes


Duration of Number of
Duration of Culture Metaphases
Exposure (hrs) Analyzed
3 months- 48 2800
26 years (exp)
900
(control )
6 weeks- 48 3740
34 years (exp)
1243
(control )
Not given 72 IBS/
person



5-24 years 48-72 100/
person



72 lOO/
person

3-6.5 years '48 4800
(exp)
1650
(control )



Results .
Reported as
negative


Reported as
negative


Reported as
positive



Reported as
negative



Reported as
negative

•Reported as
posvtive





Comment
.1. Sample size
too small .


1. Study has been
conducted following
good cytogenetic
procedure.
1. The history of the
patients including
exposure to other
drugs was not given
in this paper.
1. Sample size
too sma'l 1 .






.1. The po.ssi bitty of
synergistix
action of various
metals -cannot be
ignored.


Reference
Dekundt et -al .
(1973)


O'Riordan et al .
(1978)


Shiraishi and
Yoshida 1972
Shiraishi 1975


Bui et al.
(1975)






Bauchtnger et al .
(1976)




-------
TABLE 5.  (continued)
Species
Mouse bone marrow
Mouse Micronucleus
Rat Embryos
Mouse Dominant
lethal s
Source of
Cells
Bone marrow
cells
Bone marrow
cells
Embryonic
cells
Score dead
proportion of
implants in
the uterus
Cadmium duration of
Compound Dosage Treatment
Cadmium 0.06*, 30 days
chloride in diet
Cadmium 4 mg/kg/ 5 days
chloride day
Kilhman virus
Cadmium
chloride
Cadmium
chloride only
Cadmium 1.35, 2.7, 1 day
chloride 5.4,
mg/kg
Results
Reported as
negative
Reported as
negative
Reported as
positive
Reported as
negative
Reported as
negative
Comments
1. Good technical
procedure.
2. Data were analyzed
statistically.
1. Number of mice per group
was 3.
2. Number of polychromatic
erythrocytes scored was
300 from each mouse.
1. Cadmium enhanced the
effects of vi rus.
2. Cadmium alone was
ineffective.
1. Standard dominant
lethal assay was
performed.
2. The entire spermatogenic
cycle was covered.
Reference
Dekundt and
Gerber
(1979)
Heddle and
Bruce
(1977)
Zasukhina et al .
(1977)
Epstein et al .
(1972)

-------
TABLE 5.  (continued)
Species
Mice Dominant
lethal s
Mice (female)
Dominant lethal s
Mice Heritable
Translocation
Mice (female)
Svrian hamsters
(female)
Mice (male)
Source of
Cells
Score dead
proportional
implants in
the uterus
Score dead
and live
implants in
the uterus
Testicular
cells from
Fj males
Oocytes
Oocytes
Spermatocytes
Cadmi urn
Compound
Cadmium
chloride
Cadmi urn
chloride
Cadmium
chloride
Cadmium
chloride
Cadmium
chloride
Cadmium
chloride
Duration of
Dosage Treatment
1.75 mg/kg 1 day
2 mg/kg^ 0.5 to 4.5
days
1.75 mg/kg
1 day
3 mg/kg 12 hrs
6 mg/kg

0.5, 1.75 90 days
3.0 mg/kg
Results
Reported as
negative
Reported as
negative
Reported as
negative
Reported as
positive
Reported as
positive
Reported as
negative
Comments
1. All cell stages
not sampled.
1. Experiments repeated
three times.

1. Technical difficulties
in processing oocytes
not discussed.
1. Technical problems in
processing oocytes not
not discussed.

Reference
Gilliavod and
Leonard (1975)
Suter (1975)
Gilliavod and
Leonard (1975)
Shimada et al.
(1976)
Watanabe et al .
(1979)
Gill iavod and
Leonard (1975)

-------
Studies on Human Chromosomes In Vitro
    Shiraishi et al. (1972) tested cadmium sulfide for the induction  of
chromosomal aberrations in cultured human blood lymphocytes.  Lymphocytes from
a normal human female were cultured for 72 hours at 37°C.  Eight and  four
hours prior to harvesting, the cultures were treated with cadmium sulfide at a
concentration of 6.2 x 10"2 ug/ml of culture fluid.  Control cultures were
incubated similarly without the addition of cadmium sulfide.  Three hours
prior to harvesting, cells were treated with 0.02 ug/ml of colcemid to obtain
cells in the metaphase stage of mitosis.  Chromosome preparations were made
with the standard procedure (air-drying technique) and stained with Giemsa
stain.  Fifty metaphase cells were scored from each treatment group for
chromosome aberrations.  The types of aberrations described include chromatid
and isochromatid breaks, and symmetrical and asymmetrical translocations.
Increased incidences of chromosomal aberrations, 52% in the 4-hour treatment
group and 60% in the 8-hour treatment group, were noted over the control  value
of 0%.  This study utilized a blood sample from only one donor; the history of
the donor was not discussed.  Only one concentration of the compound was  used,
and hence, no dose-response relationship is available.  No information was
given on the solvent used to dissolve the test compound, and the number of
cells scored was small.  No indication as to the reproducibility of results
was given, and therefore, this study cannot be regarded as strong evidence for
a cytogenetic effect of cadmium.
    Dekundt and Deminatti (1978) investigated the mutagenic effects of cadmium
chloride in cultured human lymphocytes.  They treated two batches of cell
cultures and analyzed chromosomes as follows:  One batch of cultures was
treated at 0 hour and at 24 hours after the initiation of cell cultures with 5
x 10~5 and 5 x 10~6M cadmium chloride.  Chromosome preparations were made
                                      27

-------
48 hours after the initiation of the culture using the standard air-drying
technique.  In cultures treated 0 hour after the initiation, one hundred
metaphases were scored for each dosage.  There were 3% aberrations (1%
aneuploidy, 2% gaps) at 5 x 10~5M and 7% aberrations (5% aneuploidy, 2%
gaps) at 5 x 10-^M.  In cultures treated 24 hours after the initiation of
cultures, there were 5% aberrations (1% aneuploidy, 4% gaps) at 5 x  ICT^M
and 2% aberrations (1% gaps and 1% fragments) at 5 x 10~6M.  The control
aberration frequency was 5% (3% aneuploidy, 2% gaps).   The other batch of
cultures was treated at 0 hour and 24 hours, and chromosome preparations  were •
made 72 hours after the initiation of cell  cultures.  One hundred metaphases
were analyzed for aberration frequencies from each group.  In cultures treated
at 0 hour, there were 4% aberrations (3% aneuploidy, 1% gaps) at 5 x 10~5M
and 3% aberrations (3% aneuploidy) at 5 x 10~6M.  Cultures treated after  24
hours of initiation exhibited 6% aberrations (2% aneuploidy, 1% fragment, 3%
gaps) at 5 x 10~5M and 4% aberrations (1% aneuploidy,  2% gaps) at 5  x
IQ-fyl.  The control frequencies were 1% aneuploidy and 1% gaps.  The first
batch of cultures exhibited aberration frequencies similar to the control
level.  The second batch of cultures, treated only 24  hours after the
initiation, exhibited aberration frequencies two to three times above the
control level.  These aberrations were mostly in the form of aneuploidy and
gaps.  The significance of chromosomal gaps is not yet understood and they may
not represent true chromosomal aberrations because they have a tendency of
restitution.  Furthermore, the slight increase in the  incidence of aneuploidy
may be due to technical difficulties, such as scattering of chromosomes while
preparing slides, leaving uneven numbers.  These results may be treated as a
negative response of cadmium in inducing chromosomal aberrations.
                                      28

-------
     Paton and Allison  (1972) exposed human lymphocyte cultures and cultures of
the  established human  cell lines WI38 and MRC5 to at least two concentrations
(not specified) of cadmium chloride.  The duration of treatment was 48 hours
for  lymphocytes and 24 hours for WI38 and MRC5.  Chromosomal  preparations from
100-200 cells were analyzed for aberrations.  No aberrations were recorded 1n
treated cells, but data are not given.  In the absence of details of
concentrations and data, this study cannot be critically evaluated.
     Gasiorek and Bauchinger (1981) exposed unstimulated human blood
lymphocytes  (GQ) in 1 ml quantities to 10~4, 10~5, 10'6, 10'7, and
10~^M of cadmium acetate for 3 hours.  The cells were washed free of cadmium
acetate and  grown in medium containing fetal calf serum and PHA for 48 hours
at 37°C; chromosome preparations were made with the standard air-drying
technique.   Chromosome analysis from 200 cells per treatment indicated a
dose-related increase in the incidence of chromosome gaps.  The frequencies of
gaps were 0.160, 0.115, 0.135, 0.085, and 0.055 per cell, respectively, for
the above doses compared to the control  frequency of 0.058 per cell.  Data
were analyzed by the Mann-Whitney rank U-test to compare the incidence of
chromosome changes in different samples (significance taken as P < 0.05).  The
frequencies  of structural aberrations (chromatid deletions and acentric
fragments) were 0.025, 0.010, 0.005, 0.020, and 0.010 per cell, respectively,
for the same doses, whereas in controls the frequency of structural
aberrations was 0.005 _+ 0.005 per cell.   Analysis by Mann-Whitney rank U-test
indicated that structural chromosome aberrations were significantly higher
than the controls, although no dose-response relationship was evident.  No
metabolic activation system was used.  Sufficient number of metaphases (200
per dose) was scored and a standard protocol was employed.  Although these
suggest a mutagenic response, the lack of a dose-dependent response makes it
                                      29

-------
important that the experiments be confirmed in another study.

Studies on Rodent Chromosomes In Vitro
    Rohr and Bauchinger (1976) studied the effects of cadmium  sulfate in  the
Chinese hamster cell line "Hy" using three types of experiments.   In  a
long-term experiment without recovery, cells were exposed to cadmium  sulfate
at concentrations from of 10'8 to 10~5M.  Chromosome preparations  were
made following treatment of cells for 16 hours with 0.2 ug/ml  of colcemid and
hypotohic solution.  The 16-hour time period was chosen in order to analyze
the cells after exposure during a whole cell  cycle.  Concentrations of 10~^M
were toxic to cells after 16 hours of exposure, and hence, chromosome analysis
could not be made.  In a short-term experiment without recovery, cells were
treated only for 3 hours at a concentration range of 10~8 to 10"^  mol  and
chromosome preparations were made without the addition of colcemid and
hypotonic solution.  This experiment indicated a typical  stathmokinetic effect
(spindle inhibition) similar to that caused by colcemid.   The  mitotic index
increased with higher concentrations of cadmium sulfate.   In a short-term
experiment with recovery, a concentration of 10~4M was chosen  and  cells
grown on coverslips were exposed for 1 hour.   Cells with  coverslips were
washed free of cadmium sulfate and transfered to fresh medium  and  grown for  2
to 33 hours.  Chromosome preparations were made at 2, 4,  6, 8, 10, 12, 15, 18,
21, 24, 27, 30, and 33 hours after the cells  were transferred  to the  test
medium.  A total  of 500 cells were scored for each recovery period.   The
incidences of aberrations (0.2 to 0.6% structural  and 2.4 to 3.7%  numerical)
after 2 to 12 hours of recovery were similar to those of  control levels (0.1%
structural and 2.4% numerical).  Between 15 and 21 hours, the  structural
aberrations ranged from 10.2% to 22.8% and the numerical  aberrations  ranged
                                      30

-------
from 3.0% to 4.9%.  From 24 to 33 hours, the aberration  frequencies  were  lower
compared to the interval of 15 and 21 hours.  During  this  period,  the
structural aberrations ranged from 1.2 to 4.4% and the numerical  aberrations
ranged from 7.8% to 10.8%.  Only one dose was used in this study.   A common
control was maintained for all these intervals, which may  not be  satisfactory.
The exact type of serum used in this study was not indicated  and  the type of
serum can influence the yield of aberrations.
    Deaven and Campbell (1980) studied the effects of cadmium chloride  on
chromosomes in CHO cells grown in the presence of bovine serum and fetal  calf
serum.  A concentration of 2 x 10~6M cadmium produced 17,  26, 62,  and  74%
damaged cells, respectively, at 12-, 24-, 36-, and 48-hour analyses  of
metaphase chromosomes.  However, the presence of fetal  calf serum and  2 x
10-6M cadmium chloride did not induce growth inhibition  or chromosome
aberrations.  According to these investigators, fetal calf serum  appears  to
protect the cells from the damaging effects of cadmium,  whereas newborn calf
serum and human serum actively transport cadmium ions into the nucleus, thus
damaging the chromosomes.  These authors also examined the frequencies  of
sister chromatid exchanges (SCEs) in cells grown in F-10 containing  15%
newborn calf serum at a concentration of 4 x 10"^M cadmium chloride  (low,
marginal toxicity).  The SCE rate was not elevated above control  levels (10
SCEs/cell).  The range was 2 to 18 for cadmium-treated cells  and  the range for
controls was 4 to 19 per cell.  This report is assessed  as inconclusive
because more information on the exact role of serum in causing chromosome
aberrations is still not known.
    Umeda and Nishimura (1979) investigated the clastogenic effects  of  cadmium
chloride in FM3A cells derived from C3H mouse carcinoma.  Cells were grown in
Eagles minimal essential medium supplemented with 10% bovine  serum.  Cells
                                      31

-------
were exposed to 6.4 x 10-5, 3.2 x 1Q-5j anc) Io0 x 1Q-5M of aqueous
cadmium chloride.  After 24 and 45 hours of exposure, chromosome preparations
were made and analyzed.  One hundred metaphases were scored for each dose.  No
significant increase in the aberration frequency was noted in treated cultures
as compared to control cultures.  There were no metaphases in cells treated
with 6.4 x 10'^M either at 24 hours or at 48 hours,  indicating toxicity.  At
3.2 x 10~5M the aberration frequencies were 2% and 3%, at 24 and 48 hours
respectively.  At the lowest concentration of 1.0 x  10~5, the aberration
frequencies were 1%, each for the 24- and 48-hour treatment.  The control
cultures exhibited 2% aberrations at 24 hours and 1% aberrations at 48 hours.
Experiments were performed using the accepted procedures.  Three
concentrations of the test compound were used and 100 metaphases were scored
for evaluation.  This report is assessed as a negative response of cadmium in
inducing chromsomal  aberrations.
    Zasukhina et al. (1977) reported increased aberration yields in rat
embryos exposed to virus and cadmium chloride.  They infected rat embryo
cultures with Kilhman virus and then introduced cadmium chloride (3.5 x
10"6M) into the cell cultures.   Chromosome preparations were performed 24
hours after the infection.  Examination of metaphase cells revealed 10%
aberrations as compared to the  control  value of 2% aberrations.  In control
cultures infected with virus only, the aberration frequency was 6%, and in
cultures treated with cadmium chloride only, there were 3% aberrations.  These
results indicate that cadmium chloride may enhance virus-induced chromosomal
aberrations but does not induce chromosomal  aberrations by itself.

Studies on Human Chromosomes In Vivo
    Shiraishi and Yoshida (1972) and Shiraishi (1975) obtained markedly
                                      32

-------
positive results from Japanese Itai-Itai patients.  The Itai-Itai disease is
believed to have been induced by cadmium contamination.  Analysis of blood
lymphocytes from 72-hour cultures derived from these patients exhibited a high
degree of chromosomal aberrations (26.7%) compared to the control aberration
rate (2.6%).  Blood cadmium level was not given in this paper.  See Table 5
for exposure parameters.
    The results of Shiraishi and Yoshida (1972) and Shiraishi (1975), were
contradictory to the results of Bui  et al. (1975) who performed chromosomal
analysis in four Itai-Itai patients (blood cadmium level 15.5-28.8 ng/g), five
Swedish workers exposed to cadmium (blood cadmium level 24.7-61 ng/g), four
Japanese controls (blood cadmium level 4.4-6.1 ng/g), and three Swedish
controls (blood cadmium level 1.4-3.2 ng/g).  The incidences of aberrations
after 72 hours of culture were 2.3% numerical aberrations and 6.6% structural
aberrations in the Itai-Itai patients as compared to the Japanese controls
having frequencies of 4.5% numerical  and 6.0% structural aberrations,
indicating there was no difference between the controls and Itai-Itai
patients.  In the five Swedish workers exposed to cadmium, the incidences of
chromosomal aberrations were 1.0% numerical  and 2.0% structural  aberrations,
while in the three Swedish controls the frequencies were 0.7% numerical and
4.7% structural aberrations indicating nonmutagenic response.
    The reason for the discrepancy between the results of Shirashi and Yoshida
(1972) and Bui et al. (1975) in Itai-Itai patients could possibly be due to
some other factor, such as the time of initiation of cultures after the blood
was drawn.  In the experiment of Bui  et al., the subjects were not exposed to
drugs and x-rays, nor did they suffer from viral  infections at the time of
venepuncture.  These factors were not controlled for in the study by Shirashi
and Yoshida.
                                      33

-------
    Dekundt et al. (1973) Investigated the incidence of chromosome aberrations
1n 14 workers exposed to zinc, lead, and cadmium 1n a zinc smelting plant.
The workers were classified Into three groups based on the degree of exposure.
Group 1 consisted of five workers exposed to high levels of zinc
(concentration not specified), low levels of lead (1% w/w of the mineral), and
cadmium (concentration negligible).  Group 2 consisted of five workers exposed
to dust with high levels of all three metals:  zinc (concentration not
specified), lead (4% .w/w), and cadmium (1% w/w).  Group 3 consisted of four
workers exposed to mud and dust containing high levels of lead (60% w/w)  and
cadmium (1% w/w).  The control group consisted of three normal  (control)
Individuals.  Chromosomal analysis from blood lymphocytes cultured for 72
hours indicated 3.87%, 1.6%, and 2.76% aberrant cells, respectively, in groups
1, 2, and 3, while the control frequency was 1.55%.  Since the incidence  of
aberrations in group 3 was less than the incidence of aberrations 1n group 1,
it does not appear that cadlmum enhanced the aberration frequency in this
study.  In addition, analysis of their data using the t-test indicated that
cadmium exposure did not induce a significant increase in the aberration
frequency.  The blood cadmium level was not determined in this experiment.
    Bauchinger et al. (1976) studied 24 workers (25-53 years of age) exposed
to lead (mean blood lead level 1 _+ 7 ug/100 ml) and cadmium (mean blood
cadmium level 0.40 _+ 0.27 ug/ml).  The workers were exposed to these metals
for approximately 3 to 6.5 years at a smelting plant.  Of the 4,800 metaphases
scored from lymphocytes cultured for 48 hours, an increase in both chromosome
and chromatid type aberrations (1.354 +_ 0.994%) was noted compared to a
frequency of 0.467 _+ 0.916% aberrations in 1,650 metaphases derived from  15
controls (mean blood cadmium level 0.15 ug/ml).  The authors point out that
"the observed chromosome aberrations cannot be causally related to cadmium
                                      31*

-------
because the workers were also exposed to lead and zinc."   Dekundt  and  Leonard
(1975) reported a significant (P < 0.02) increase in the  incidence of  "complex
chromosomal aberrations" in a group of 23 men exposed to  high levels of
cadmium and lead (23.5 to 75.9 ug/100 ml) compared to controls.
    O'Riordan et al. (1978) studied chromosomal  aberrations in blood
lymphocytes from 40 workers exposed to cadmium salts, (name not  specified,
mean blood cadmium level 1.95 ug/100 ml  range < 0.2 to 14.0 ug/100 ml) for  a
period of 6 weeks to 34 years.  In 3,740 cells examined from these workers,
there were four chromatid interchanges.   In the control population of  1,243
cells derived from 13 normal  subjects (mean blood cadmium level  less than 0.2
ug/100 ml in 8 donors and 0.6 to 2.9 ug/ml  in 5 donors),  there were no
aberrations at all.  Data were pooled from all 40 workers.  It is  not  clear
whether the four chromatid interchanges  came from one exposed worker or  more
than one worker.  The occurrence of chromatid exchanges though small in
number, 4/3,740 cells, does not necessarily indicate a negative  response but
does indicate that the study may be considered inconclusive.
    Most of these studies on workers reflect mixed exposure to other metals
such as zinc and lead.  Since smelters also commonly process relatively  crude
materials, exposure to other metals such as chromium, nickel, etc., cannot  be
eliminated as possible contributors to the observed effect.  Synergistic
effects may also confuse the results.

Studies on Rodent Chromosomes In Vivo
    Dekundt and Gerber (1979) investigated the in vivo cytogenetic effects  of
cadmium chloride (3.27 x 10~7M, 0.06%) in mice.   Mice were maintained  on a
standard diet (1.1% calcium)  or on a low calcium diet (0.03%) for  one  month.
In both cases the diet was supplemented  with cadmium chloride.  Cadmium  is
                                      35

-------
known to be absorbed through the gut.  Cadmium chloride did not induce
chromosome aberrations in bone marrow cells significantly above the control
level either in normal or low calcium diet groups.  The frequency of
aberrations in animals treated with cadmium chloride in the standard diet
(1.1% calcium) was 2.20%, and the frequency of aberrations in animals treated
with cadmium chloride in the low calcium diet (0.03%) was 1.60%.  The control
frequencies were 1.8% and 2.0%, respectively.  These results indicate that
cadmium chloride does not induce chromosomal  aberrations in mice. -

Micronucleus Assay
    The micronucleus assay is based on the observation that chromosome
fragments are produced by mutagenic chemicals.  Those fragments lacking
centromeres are unable to segregate normally  but lag behind during the cell
division to form small nuclei or micronuclei  in  daughter cells.  Heddle and
Bruce (1977) studied the ability of cadmium chloride to induce micronuclei in
the mouse.  Three groups of mice (Fj of C57B1/6X C3H/He), each group
containing three animals, were given daily intraperitoneal  injections of
cadmium chloride for 5 days with total doses  of  1, 6, and 16 mg/kg,
respectively.  Mice were sacrificed, bone marrow smears were prepared, and 333
polychromatic erythrocytes from each mice were scored for the presence of
micronuclei.  No increase in the incidence of micronuclei was observed.  This
report deals with three mice in each group and only 333 cells from each mouse;
thus, a total of 1,000 cells were analyzed for each dose group.  The
spontaneous frequency of micronuclei was 0.5%.  An observation of 1% over the
control value was considered a positive response.  According to these
authors, the frequency of micronuclei in the  experimental groups was not
different from the control level.  Confirmation  of the results are required
                                      36

-------
using a larger sample of animals (10) per dose group and an analysis  of  at
least 2,000 polychromatic erythrocytes per dose group.   Consequently,  this
report is regarded as inconclusive.

Dominant Lethal Assay
    The ability of cadmium chloride to induce dominant  lethal  mutations, which
result in the death of fetuses during various stages of development,  has been
investigated (Epstein et al. 1972; Gilliavod and Leonard 1975; Ramaiya and
Pomerantseva 1977; Suter 1975; Sutou et al. 1980 a, b).
    Epstein et al. (1972) evaluated the dominant lethal effects of cadmium
chloride in ICR/Ha mice.  Groups of seven or nine male  mice, 8 to 10  weeks  of
age, were injected intraperitoneally with 1.35, 2.7, 5.4, and 7.0 mg/kg  of
cadmium chloride in distilled water.  Treated males were bred with virgin
females, 8 to 10 weeks of age.  Each male was allowed to mate with three
virgin females per week for 8 weeks.  Mated females were sacrificed on the
13th day and analyzed for dead (dominant lethals) and live implants.
According to these authors, cadmium chloride did not induce a statistically
significant increase in dominant lethal mutations over  the control  value.
    Gilliavod and Leonard (1975) investigated the dominant lethal effects of
cadmium chloride in another strain of mice BALB/c.  Only one dose of  1.75
mg/kg cadmium chloride was injected into male mice (11-13 weeks of age)
through the intraperitoneal route.  The treated males were bred with  three
virgin females every week for 3 weeks.  The mated females were sacrificed on
the 10th day and the number of corpora lutea, and dead  and live implants were
counted and compared with the controls.  No dominant lethal effect was
observed in treated and control  groups.  These investigators treated  the
parental  male mice with only one acute dose of the test compound.
                                      37

-------
Furthermore, they bred the treated mates with normal  females for only 3 weeks,
which is too short to sample stages of spermatogenesis.   The standard way  of
doing an dominant lethal  test is to breed the treated males for 8 weeks.
Consequently, this report is treated as inconclusive.
    Suter (1975) studied the mutagenic effects of cadmium chloride using the
dominant lethal  assay in female mice (Fl progeny  of C3H  and C57BLA).
According to this investigator cadmium chloride had no dominant lethal  effects
in female mice.   Female mice, Fj (10 x C3H)  stock,  were  injected
intraperitoneally with 2 mg/kg cadmium chloride and bred with untreated males
0.5 to 4.5 days  postinjection.  Mated females, as evidenced by the vaginal
plug, were sacrificed 12-15 days later and the number of corpora lutea, number
of total implants, number of living implants, and the percent of dead implants
per female were  counted.   No differences were noted between the treated and
control  groups.   In the treated group, the frequency  of  corpora lutea,  total
implants, living implants, and dead implants per  female  were 8.2, 7.8,  6.9,
and 6.9, respectively, as compared to control  frequencies of 7.6, 6.8,  6.4,
and 6.1% per female.
    Ramaiya and  Pomerantseva (1977) investigated  the  mutagenic effect of
cadmium chloride using the dominant lethal test.   Fj  hybrid mice (CBA x
C57BL) aged 2.5  to 3 months, were selected for these  studies.   Males  were
given a single intraperitoneal injection of  aqueous cadmium chloride  solution.
Three doses, 1.0, 2.0, and 4 mg/kg, were employed.   105^ was determined to
be 6.9 mg/kg.  Treated males were mated with untreated females for a  period  of
6 weeks covering the entire spermatogenic cycle.   Dominant lethals, as  noted
by preimplantation and postimplantation losses and  the ratio between  the dead
and live implants, were recorded.  No significant (P  > 0.01) increase in the
dominant lethal  frequency was recorded.  These results are regarded as
                                      38

-------
 negative  since the  authors  followed  appropriate protocols, the dosage
 selection  was  based on  1059,  and data were analyzed statistically.
     From  these studies  it appears that  cadmium chloride has no mutagenic
 potential  in the mammalian  dominant  lethal test.  The dominant lethal test has
 been extensively used in screening for  the mutgenic activity of chemicals.

 Heritable  Translocation Assay
     Gilliavod  and Leonard (1975) evaluated the mutagenic effects of cadmium
 chloride  in BALB/c  mice using the Fj heritable translocation assay.  Male
 mice (number not specified) were treated with 1.75 mg/kg of cadmium chloride
 intraperitoneally and each  treated male was bred with three nontreated virgin
 females each week for 3 weeks.  The  spermatocytes of the resulting 120 Fj
 male progeny were analyzed  for the presence of heritable chromosomal
 translocation  by standard cytogenetic methods.  No evidence of heritable
 translocation  was noted in the spermatocytes of Fj males.  Only a single
 concentration  was used in this experiment.  Mating of treated males was done
 only for 3 weeks instead of 8 weeks.  No controls were maintained.
 Consequently,  this  report is  treated as inconclusive.

 Direct Effects on Germ Cells
      Effects of cadmium chloride on oocytes of mice (Shimada et al. 1976),
 Syrian hamsters (Watanabe et  al. 1979), and on spermatocytes of mice
 (Gilliavod and Leonard 1975)  have been investigated.
      Shimada et al.  (1976)  induced superovulation by injecting female mice,
ddy  strain, with 5 international  units (I.U.)  of pregnant mare's serum (PMS)
followed 48 hours later  by  5  I.U.  of human chorionic gonadotrophin (HCG).
Mice were given 3 mg/kg  or  6 mg/kg of cadmium  chloride 3 hours after the
                                      39

-------
administration of HCG and were dissected 12 hours after the cadmium chloride
administration.  Chromosome preparations were made from unfertilized obcytes
at the second meiotic metaphase as described by Tarkow'ski (1966).  No
structural chromosome abnormalities were found.  However, numerical
aberrations (aneuploidy) were found to be statistically significant (P =
0.015) in the dose group of 3 nuj/kg group as compared to controls.  The
authors postulated that this hondisjunction may be due to the spindle
inhibiting effects of cadmium.
     Watanabe et al. (1979), using Syrian hamster oocytes and cadmium
chloride, reported even more pronounced incidence of aneuploidy.  Cadmium
chloride at concentrations of 1.0, 2.0, and 4 iiig/kg were subctitarieolisly
injected to groups of 20 female Syrian hamsters 5 hours before ovulation.
Matched controls were given equal  volume of normal  saline.  Females were
sacrificed 12 hours  after the treatment, and the oocytes were recovered from
the ampulla.  Analyses revealed that 6 females out of 20 from the 1.0 trig/kg
group, ll females out of 20 from the 2.0 mg/kg group, and 12 females out of 20
from the 4.0 mg/kg group had numerical chromosomal  abnormalities, such as
hyperhaploidy and dipldidy, in oocytes as compared to 3 out  of 20 in control
females.  The results were statistically significant (P < 0.05 and P < 0.01)
in the treated groups compared to the control  group.  Cadmium-treated animals
were also analyzed for cadmium accumulation in the ovary usirtg atomic
absorption spectrophotometry.  The results indicated statistically significant
(P < 0.05) increases in the accumulation of cadmium in the ovaries of treated
females as compared  to control females.  Both of these results appear to
indicate a positive  response of cadmium in inducing numerical  chromosomal
abnormalities in mammalian oocytes.

-------
    Gilliavod and Leonard (1975) investigated the mutagenic effects  of cadmium
chloride in BALB/c mice using the spermatocyte assay.   Males in  groups of 10
were treated with 0.5, 1.75, and 3.0 mg/kg of cadmium  chloride
intraperitoneally.  After 3 months, treated males were sacrificed and
spermatocytes (100 cells per animal) in the testes were analyzed for
translocations that may have been passed on from treated spermatogonia.  No
translocations were found, i.e., the frequency of translocations were 0 in
both the treated and control animals.  This is not a very sensitive  test, and
hence, it is not commonly employed in mutagenicity tests.

Sperm Abnormality Assay in Mammals
    Heddle and Bruce (1977) evaluated the effects of cadmium using the sperm
abnormality assay.  The sperm abnormality assay is based on the  observation of
increased incidence of sperm heads with abnormal shapes as a result  of
exposure to chemical mutagens (Wyrobek and Bruce 1975).  Three groups of mice
of the genotype (C57BL/6 x CSH/HejFj, each consisting  of three mice, were
given daily intraperitoneal  injections of cadmium chloride for 5 days with
doses of 1, 4, and 16 mg/kg, respectively.  Sperm suspensions were made from
sperm collected from the cauda epididymis following the sacrifice by cervical
dislocation of mice.  The sperm suspension was stained with 1% eosin-Y in
water, and smears were made, dried, and mounted under  a coverslip with
permount.  One thousand sperm heads were evaluated for morphological
abnormalities.  The background frequency of sperm head abnormality in the
control  population was 1%.  Under the conditions of the test, there  were no
increases in sperm head abnormalities in the treated over the control  groups.

-------
CHROMOSOMAL ABERRATIONS IN PLANTS
    Levan (1945) reported that treatment of Alllum cepa root-tips with cadmium
chloride induced C-mitosis.  This observation was later confirmed by Avanzi
(1950) using cadmium chloride concentrations ranging from 2 x 10"6M to 5 x
lO-fyl.  Qehlkers (1953) reported that cadmium nitrate induced chromosomal
aberrations in Vicia faba.  Van Rosen (1953, 1954) reported the genotoxi'city
of cadmium as evidenced by chromosomal  aberrations in the root-tips of plants
such as Alii urn cepa, Beta vulgaris, Pi sum abysinnicum, and Vicia sativa.
Similar observations were made by Degraeve (1971) in Horedeum sativum and by
Ruposhev and Garina (1977) in Crepis capillaris.  Aberrations reported in
these studies were of both chromatid and chromosome type with a dose-related
response.  Since many of these publications are in foreign languages, the
material presented here is a summary derived from the review article published
by Degraeve (1981).

OTHER INDIRECT EVIDENCE
    Some information is available on the effects of cadmium on animals, and
although it is not strictly mutagenicity test data, it may be useful  in
evaluating the ability of cadmium to reach and damage the gonads.  Dixon et
al. (1976) reported that cadmium chloride at 2.24 mg/kg administered
intraperitoneally caused damage to rat  testes.  A single 10 mg/kg
intraperitoneal  injection causes selective destruction of rat testes.  Cadmium
chloride, when administered intraperitoneally at 1 mg/kg, reduced fertility of
male mice at sperm cell stages, except  spermatozoa (Lee and Dixon 1973).
However, single oral doses up to 25 mg/kg had no effect on the fertility of
male rats (Dixon et al. 1976), and cadmium chloride at 0.1 mg/1 in drinking
water for up to 90 days had no effect on the fertility of rats.

-------
Intraperitoneal injection of 1 mg/kg, cadmium chloride decreased  the
incorporation of thymidine into spermatogonia in  mice (Lee  and  Dixon  1973).
These authors (Lee and Dixon 1973) also found the binding of cadmium  to  late
spermatids in vivo or in vitro.  Friedman and Staub (1976)  studied  the effects
of cadmium chloride on DNA synthesis in Swiss mice.  Cadmium chloride at 10
mg/kg inhibited DNA synthesis significantly.   An  aqueous solution of  cadmium
chloride was injected intraperitoneally at the above dose into  five male mice
and the mice were sacrificed 3.5 hours later.  Thirty minutes prior to
sacrifice, mice were injected with 10 uCi [^H] thymidine.  Controls received
only 10 uCi [^H] thymidine.  Testes were removed  following  cervical
dislocation, DNA was isolated, and the specific activity was determined.
Cadmium chloride induced a statistically significant (P < 0.01) inhibition of
[^H] thymidine uptake (1.90 +_ 0.58) in the testes as compared to controls
(7.45^1.44).
    Mitra and Bernstein (1977, 1978) reported that when E_.  coli cultures were
exposed to 3 x 10'6M cadmium (Cd2+), 82 to 95% of the cells lost  their
ability to form colonies on agar plate.  Analysis of DNA strands  from cells
treated with various doses of Cd2+ indicated  that there was a dose-related
increase in the breakage of single strand DNA.  These investigators believe
that the loss of viability in cadmium-treated cells is due  to the
single-strand DNA breakage.  Cadmium-treated  cells recovered viability when
grown in Cd2+-free liquid medium containing 10 mM hydroxyurea.
    Si rover and Loeb (1976) investigated the  infidelity of  DNA  synthesis
brought about by cadmium chloride and cadmium acetate.  This assay  measures
the perturbations in the fidelity of DNA synthesis in vitro caused  by soluble
metal salts.  Cadmium chloride and cadmium acetate decreased the  fidelity of
DNA synthesis.  Cadmium chloride also induced concentration-dependent

-------
 inhibition of RNA synthesis  (Hoffman and Niyogi 1977).

 SUMMARY
     Cadmium has been investigated for mutagenic activity in both prokaryotic
 and eukaryotic systems.  Gene mutation studies in Salmonella typhimurium and
£. coli have produced inconclusive results.  In yeast, gene mutation studies
 are also inconclusive.  However, gene mutation studies in mammalian cell
 cultures, mouse lymphoma cells, and Chinese hamster lung and ovary cells, have
 resulted in a weak mutagenic response.  Rec-assay, which is a test for DNA
 repair in Bacillus subtilis, also resulted in a weak mutagenic response.  In
 the Drosophila sex-linked recessive lethal  test, cadmium has been found to be
 nonmutagenic.  However, the negative response may be due to inadequate test
 control.  In contrast, the dominant lethal  test in Drosophila resulted in a
 positive response with a dose-response relationship.  Chromosomal aberration
 studies in human lymphocytes and human cell lines treated with cadmium are
 conflicting and contradictory.  In Chinese hamster cells, chromosomal
 aberrations were noted following treatment with cadmium, however, in mouse
 carcinoma cells no aberrations were recorded in response to cadmium treatment.
 In rodents, treatment with cadmium induced no chromosomal aberrations  or
 micronuclei in bone marrow cells.  Similarly, no dominant lethal  mutations
 were noted in mice treated with cadmium.  Induction of chromosomal
 nondisjunction in female germ cells of mice and Syrian hamsters by cadmium has
 been reported.  Chromosomal aberrations and gene mutations in plants exposed
 to cadmium have also been recorded.
    A clear analysis of the mutagenicity of cadmium in the studies discussed
 herein is rather difficult because of the conflicting results and lack of
 adequate test protocols used.  However, the results of gene mutations  studies

-------
in mammalian cell  cultures, rec-assays in bacteria,  chromosomal  nondisjunction
studies in intact  mammals, as well  as  other indirect evidences  suggest that
cadmium is weakly  mutagenic.

-------
                              CARCINOGENICITY

    The topic of the carcinogenicity of cadmium has been reviewed critically
in earlier documents (IARC 1973, 1976; U.S. EPA 1977, 1981; Sunderman 1977,
1978; Hernberg 1977).  This section updates findings mentioned previously and
discusses recent findings not mentioned in earlier reviews.

ANIMAL STUPIES
Injection Studies in Mice and Rats
    Injection of cadmium metal or certain salts of cadmium has been shown to
produce sarcomas at the site of the injection as well  as testicular tumors
(Leydig cell, interstitial  cell) in experimental animals as summarized in
Table 6.
    Tomatis (1977) reviewed the appropriateness of the subcutaneous (s.c.)
injection route as a carci'nogenesis bioassay by comparing s.c. with other
routes of administration.  He surveyed a number of chemicals tested by the
s.c. route in rodents to see if there was a correlation between their
capacities to induce local  and/or distant tumors in one species and their
capacities to induce tumors by another route in another species.  A total  of
102 chemicals, which have been reviewed by the International Agency for
Research on Cancer (IARC) and have been tested by the s.c. route as well  as by
other routes of administration, were surveyed.  Of those, 69 were positive for
carcinogenic activity when administered by s.c. injection and by another
route, and 18 were negative or inconclusive whether given by s.c. injection or
another route.  Nine were positive only when administered by s.c. injection,
and six were negative by s.c. injection and positive by another route.  The
author concludes that "administration of a chemical by the s.c. route produced
what one could call false negative results for six (5.6%) of the 102 chemicals

-------
                                     TARLE fi.   ANIMAL  TUMORIGENESIS  INDUCED RY CADMIUM INJECTION
Author
Haddow
et al. (1961)

Heath (196?)
Species
Rats
Mice
Hooded Rats
Compound
ferritin containing
Cd
-
Cd powder
Route Tumor and Incidence
s.c. Sarcomas 8/20
Interstitial cell tumors
Sarcomas 0/20
i.m. Sarcomas 2/10

10/20


                                           o*J  [.iwmici
                                           0.28 g  in  0.4 ml  fowl  serum
                                           0.014 g  in  0.4  ml  fowl  serum
                                                                          (later  in  the study 10 more
                                                                          rats  developed tumors; which test
                                                                          group they were in is not stated)
                                                                          Sarcomas  3/10
Heath and
Daniel  (1964)
Kazantzis (1963)
Kazantzis and
Hanbury (1966)
Haddow et al.
(1964)
Hooded Rats
Chester-Beatty
Rats

Wistar Rats
Rats
                        Mice
Cd powder
0.014 g in 0.4 ml  fowl  serum             i.m.
0.02R g in 0.4 ml  fowl  serum

25 mg CdS in 0.25 ml                      s.c.
physiological  saline

25 mg CdS in 25 ml                        s.c.
physiological  saline                      i.m.

25 mg CdO in 0.25 mg                      s.c.
physiological  saline

0.25 ml physiological  saline alone

0.5 mg CdS04.H20                         s.c.
in 1.0 ml sterile
distilled water once  weekly
for 10 weeks

0.05 mg CdSO^^O in  0.2 ml
H20 once weekly for 11  weeks
Sarcomas 9/10
Sarcomas 6/8 (2 were killed early)

Sarcomas 6/10
Sarcomas 6/10, 6/26
Sarcomas 5/14

Sarcomas 8/10
Sarcomas 0/10

Sarcomas 14/20
control 0/15
                                                                          0/20 injection site tumors
                                                                          control  0/15
                                                                                                    [continued  on the following page)

-------
                                                           TARLT 6.  (continued)
    Author
Species
        Compound
                                                                                       'Route
                        Tumor and Incidence
    Roe et  al.  (1964)
Rats
    Gunn  et  al.  (1963a)     Albino Mice

                           Wistar Rats


    Gunn  et  al.  (1964)      Wistar Rats
oo
    Gunn et  al.  (1967)


    Knorre  (1970)

    Knorre  (1971)

    Lucis et al.  (1972)
Wistar Rats


Wistar Rats

Wistar Rats

Wistar Rats
0.5 mg CdSn4.H2n in .1.0. ml
    once weekly for 10 weeks
                   0.05  mg  CdS0.4.4H20  in 0.2 ml  H20

                   0.03  mM/kg  CdCl ?

                   0.03  mM/kg  CdCl 2


                   0.03  mM/kg  CdCl 2
1.8 mg CdCl2


0.003 mM CdCl2/100 g b.w.

0.003 mM CdCl2/100 g b.w.

0.02-0.03 mM/kg CdCl2 in
isotonic NaCl  solution
    Reddy  et  al.  (1973)     Fischer 344 Rats   0.03 mM/kg CdCl2
    Furst and
    Cassetta  (1972)
    Favino et  al.  (1968)
Fischer 344 Rats   5 mg  Cd  powder
                   (suspended  in 0.2 ml)
                   synthetic trioctanoin)
          s.c.




          s.c.

          s.c.


          s.c.
Sprague-Dawley
 Rats
1 mg/100 g CdCl 2
simultaneous s.c.
      and i.m.

   single s.c.

   single s.c.

   single s.c.



   single s.c.
                               2 monthly i.m.
                                  injections
   single s.c.
Interstitial cell tumors  11/15
control 0/15

Interstitial cell tumors  0/16

Interstitial cell tumors  20/26
control 0/25
Interstitial cell tumors  17/25
control 0/25

Sarcomas 9/22; control 0/18
Interstitial cell tumors  21/24
control 0/18

Sarcomas 10/23
Sarcomas 3/26

Sarcomas 6/45

Interstitial cell tumors  10/25

Interstitial cell tumors  13/15
Sarcomas 2/15 (two animals died
early)

Interstitial cell tumors  16/20
control 0/10

Sarcomas 26/50
Interstitial cell tumors 6/6
                                                                                                       (continued on the following  page)

-------
                                                       TABLE fi.  (continued)
Author
Species
       Compound
            Route
Tumor and Incidence
Malcolm (1972)
Levy et al.  (1973)
Rats
C.B. Hooded Rats
CdCl2
0.2 mg 3CdS04.H20 in
0.2 ml H20
                                           0.1 mg 3CdS04.8H20 in
                                           0.2 ml H20
                                           0.05 mg 3CdS04.8H20 in
                                           0.2 ml H20
                                           control - 0.2 mg H20 only
                                                           s.c.
weekly s.c.  injection
into alternate flanks
for 2 yrs
Sarcomas (?)
Interstitial cell tumors (?)
Experiment not completed at
time of publication

Sarcomas 4/25
Interstitial cell tumors 17/25
1 lung adenoma

Sarcomas 1/25
Interstitial cell tumors 17/25
1 malignant lymphoma

Sarcomas 1/25
Interstitial cell tumors 16/25
1 adenocarcinoma of pancreas

Sarcomas 0/75
Interstitial cell tumors 48/75
1 squamous carcinoma of tongue
1 benign liver cell tumor

-------
tested arid, if we accept all the criticisms of this route of administration,
false ppsjtive pesults for pine (8.7%) of the 102 chemicals tested.  Even, so,
according to £he author, it appeals that the s.c. rpu.te of administration is
pot top muc|i wppse th,an apy other; rpijte of administration.

Inhalation Study in Rats
Takenaka et al. (J982)--
    A carcinogenicity study of cadmium administered to male Wistar r^ts by
inhalation has been reported, by Takenaka et al. (1982).  The sninials were
placed in a 225 1  inhalation chamber for exposure to cadmium chloride.
(CdCl2) aerosol.  Aerosol was generated by atorr)izif]g a spjijtipn pf CdClz,
and airflow through the atomizer was 0.7 1/min.  Analytical measurements of
cadm,iu.m levels were made by collecting aerosol samples in menibfane filters in
t(ie intake a.n
-------
TABLE 7.  NOMINAL AND MEASURED CADMIUM CONCENTRATIONS OF
           CdCl2-AEROSOLS USED FOR INHALATION
                  (Takenaka et al. 1982)
Nominal concentrations
Measured concentrations
Standard deviation
Number of measurements
ug/m3
ug/m3
ug/m3
• _ •
50.0
50.8
5.9
212
25.0
25.7
3.6
220
12.5
13.4
2.1
210

-------
    Differences in body weights (Table 8) and mean survival  times  (Table  9)
among control and treated animals were not statistically significant  (P >
0.05).
    A dose-related increase in the incidence of primary  lung carcinomas in
treated animals was evident as shown in Table 9.   The  first  epidermoid
carcinoma and the first adenocarcinoma were found  20 and 22  months,
respectively, after treatment commenced.   Several  treated rats  also developed
adenomas and nodular hyperplasia  in the lung.  Metastases to the regional
lymph nodes and the kidneys and invasion  into the  regional lymph nodes and the
heart occurred in some rats with  lung carcinomas.   No  lung tumors were found
in control  animals.
    Nonneoplastic lesions and various tumors in other  organs were found in
both control and treated animals.   None of these additional  tumor types and
nonneoplastic lesions was significantly (P > 0.05)  different among the four
groups.
    The data in Table 10 show that cadmium was retained  in lung, liver, and
kidney of survivors for as long as 13 months after cessation of exposure.
Anaylsis of these tissues indicates that  cadmium was absorbed and circulated
throughout  the body and that, although the lung was the  target  organ  for
carcinogenicity, the kidney retained the  largest amount  of cadmium.   Increases
in cadmium  levels were dose-related in liver in all  treatment groups  and  in
lung and kidney in the mid-dose and high-dose groups.  Pathologic changes
apparently  were not observed in kidney and liver,  thus suggesting that the
cadmium levels found did not have  a toxic effect in these tissues.
    The authors attributed their  success  in demonstrating cadmium
carcinogenicity to:  1) performance of a  long-term study using  CdCl2
aerosols that were retained at a  rather high level  in  the lungs after
                                      52

-------
                 TABLE 8.  AVERAGE BODY WEIGHTS OF RATS EXPOSED TO CADMIUM  CHLORIDE
                                     (Takenaka et al.  1982)
Exposure groups
Control
12.5 ug/m3
25 ug/m3
50 ug/m3

Control
12.5 ug/m3
25 ug/m3
50 ug/m3
Average
0
135.2*
(4.8)
135.1
(6.6)
133.4
(6.7)
133.3
(6.7)
18t
434.9
(32.4)
424.6
(41.0)
437.6
(38.1)
424.3
(40.6)
Body Weights (Months After the Beginning
3
333.3
(27.4)
320.9
(29.4)
326.6
(28.6)
323.5
(29.0)
21
428.2
(31.4)
421.8
(41.2)
441.2
(37.7)
424.9
(43.8)
7
385.2
(30.5)
375.3
(37.1)
382.1
(32.1)
375.1
(32.2)
24
406.2
(41.3) -
409.5
(45.9)
429.2
(45.9)
415.2
(42.6)
10
411.6
(31.2)
405.2
(39.4)
410.0
(32.8)
403.2
(34.8)
27
405.7
~ (31.3)
408.4
(40.9)
423.9
(37.6)
398.4
(35.8)
of the Inhalation)
12
422.9
(31.7)
417.2
(41.4)
425.7
(35.9)
417.0
(36.6)
30
367.3
(39.8)
372.5
(41.8)
375.4
(47.8)
357.8
(41.5)
15
425.1
(31.8)
420.0
(38.8)
428.3
(36.1)
422.0
(38.5)





*Mean value (_+ S.D.).
tEnd of the inhalation.

-------
       TABLE 9.   SURVIVAL TIMES  AND  LUNG  CHANGES  OF  WISTAR  RATS AFTER THE EXPOSURE TO CdCI2 AEROSOLS
                                         (Takenaka et  al.  1982)
Exposure
Groups
                                                         No.  of Rats With Lung
                    Survival     No.  of
                    Time in     Rats       Adeno-   Adenomas  Carcinomas
           Initial   Weeks       Examined    matous
           No. of   Mean Value  Histo-     proli-
           Rats     + S. D.     logically   feration
                                         adeno  epidermoid  combined    muco-  Total
                                                            epidermoid  epider   (%)
                                                            and adeno   moid
Control
            41
12.5 ug/m3  40

25   ug/m3  40

50   ug/m3  40
122+19

119+17

125^15

116+23
38*

39t

38§

3511
1

6

5

3
0

1

0

1
 0

 4

15**

14**
0

2

4

7
0

0

1

1
0    0

0    6(15.4%)#

0   20(52.6%)**

3   25(71.4%)**
    *Two rats died during the first  18 months;  another rat  was not examined because of autoTysis.
    tOne rat was not examined because of autolysis.
    §Two rats were not examined  because of autolysis.
    lIThree rats died during the  first 18 months;  two  other  rats were not examined because of autolysis.
    #P = _< 0.01.
    **P < 1.0 x 10-5.

-------
TABLE 10.   CONCENTRATION OF CADMIUM IN LUNGS,  LIVER,  AND KIDNEYS OF  RATS EXPOSED TO
         CdCl2 FOR 18 MONTHS (13 MONTHS AFTER  THE  END OF THE  INHALATION)
                              (Takenaka et al.  1982)
No. of Cadmium concentration (ug/g wet weight) in
Exposure Groups Rats Lungs Livers Kidneys
Control 9 0.03 0.1 +_ 0.1* 0.3 _+ 0.1
12.
25
50
5 ug/m3 6 5.6 +_ 1.0 2.2 +. 0.6 13.5 _+ 3.2
ug/m3 9 4.7^1.5 5.9^1.5 16.4 +. 3.6
ug/m3 9 10.4 +_ 4.2 13.5 +_ 3.0 33.6 _+ 10.7
*Mean value + S.D.

-------
cessation of exposure, and 2) continuous observation of the animals over an
extended duration (31 months).  Most of the lung carcinomas were detected
after the 27th month of the study.

Intratracheal  Studies in Rats
Sanders and Mahaffey (1982)--
    Sanders and Mahaffey (1982) evaluated the carcinogenicity of cadmium oxide
(CdO) in male Fischer 344 rats.  Four groups of 46 to 50 rats each  were
treated as follows:   Group 1 (untreated controls) received one intratracheal
instillation of 0.9% sodium chloride solution (the dosing vehicle); Group 2
was given an intratracheal  instillation of 25 ug CdO when 70 days old;   Group
3 received intratracheal instillation of 25 ug CdO when 70 and 100  days  old
for a total dose of 50 ug; Group 4 was given intratracheal  instillations of 25
ug CdO when 70, 100, and 130 days old for a total dose of 75 ug.  The authors
stated that the 25 ug dose was 75% of the 1.050 by the route of
administration used.  Instilled CdO had a count median diameter of  0.5 urn.
    The animals were allowed to survive until spontaneous death.  All animals
were necropsied, organs were weighed, and tumors, lesions,  and major tissues
and organs from all  rats except 12 lost due to autolysis or cannibalism  were
examined histopathologically.
    Median survival  times were 793, 824, 785, and 788 days for Groups 1, 2, 3,
and 4, respectively.  Survival times and organ weights (body weights were not
obtained) were similar  (P > 0.05) between control and treated groups.
Statistical analysis of tumor data by life table and contingency table methods
revealed no significant (P > 0.05) differences among the four groups.  Lung
tumor findings consisted of adenocarcinomas in two rats in Group 3.
                                      56

-------
    Cadmium, as CdO, was not carcinogenic under the conditions  of  this  study;
however, the protocol used may not be as sensitive an indicator of the
carcinogenic potential of cadmium as a design with lifetime exposures by
inhalation, particularly in reference to the carcinogenicity study by Takenaka
et al. (1982) discussed herein.  Lung tissue was not analyzed for  cadmium
content in the Sanders and Mahaffey (1982) study.  However, clearance of  80%
of an intratracheally instilled dose of 15 ug 109CdO from the lung in male
Fisher 344 rats, with an elimination half-life of 4 hours and an experimental
duration of 2 weeks, has been observed (Hadley et al. 1980) as  well  as  repair
of lung tissue to Type 1 alveolar epithelium by 2 weeks following  an
intratracheal instillation of 25 ug CdO into male Fischer 344 rats (Hadley  et
al. 1980).  Hence, a lifetime inhalation exposure to CdO also might have
presented a stronger challenge for carcinogenicity by providing a  greater
cumulative dose of cadmium within target (lung) tissue.

Drinking Mater Studies in Rats and Mice
Schroeder et al. (1964, 1965)
    Schroeder et al. (1964) conducted two lifetime exposure studies (survival
up to 21 months) in Swiss mice.  The animals were given drinking water
containing 5 ppm of cadmium acetate.  The exposure level  was purposely  low  to
simulate the human experience, according to the authors.   Only  males
experienced decreased longevity in comparison with the controls.  The exposed
males had fewer "visible" tumors (1/>50) than the controls (11/>50), possibly
related to shortened lifespan (P < 0.005).
    In another lifetime exposure study by Schroeder et al. (1965),  male and
female Long-Evans rats ingested 5 ppm cadmium acetate in  water  as  the sole
source of fluid; the treated group developed 28/84 tumors versus 24/70  in
                                      57

-------
controls.  The authors stated that "no significant differences  appeared  among1
the Various groups as to type of tumor."  This study, like th'e  1964  study,  was
compile a'tec! by being performed in a low metal environment with  a diet  low in
many trace metals.  When the essential trace element Cr(III) Was added to the
diet of bhe ^roO|j Of rats devoid of cadmiunii they thrived better than  the
control grbtip arid had 34/71 tumors (SchrOeder et al. 1965).

Halcblni (1972)--
    Halcoim (1972) gave male Chester-Beatty hooded rats up to 0.2 nig of
cadmium1 sulphate subctitarieOusly arid up to 0.8 mg weekly by stomach tube  for 2
years.  In another experiment, Swiss mice were given doses of cadmiiJm  sulfate
in distilled Water Up1 to 0.02 mg/5g of body weight subcu'tanebusly at weekly
intervals for 2 years.  Except for a few sarcomas see'ri iri the rats cjive'ri
sUbcUtaneous injectibris and Leydig cell tumors (both also seen  in the
controls), these studies were negative at the time 'reported.

Levy arid 'Clack (1975)--
    Experitiients with male specified pathogen-free Chester-Beatty h'boded  rats,
usiri(j doses of 0.087, 0.18, arid 0.35 mg/kg of cadm'idm sulphate'  in distilled
water given by gastric iristillatibri brice weekly for 2 years, were carried but
by Levy and Clack (19*75) with rib difference iri tumor iriciderice  in exp'bsed and
coritrbl groups.  It is noted, h'Owever, that this particular strain of  rats  has
a very high lifetime incidence of spbntiarieou's interstitial cell tumbr
formatibri (75% in the untreated coritrol group), such that "if exposure to Cd
had any effect on the incidence of the lesions it was entirely  overshadowed by
their spontaneous occurrence," according to the authors.  Effects on the
prostate were especially scrutinized, with no neoplastic  lesions observed.
                                       '58

-------
Dietary Studies in Rats
Loser (1980) —
    A 2-year oral  carcinogenicity study of Wistar rats given  cadmium  chloride
(CdCl2) was carried out by Loser (1980).  Doses of 1,  3,  10,  and  50 ppm  were
given in food to 50 male and 50 female rats, with 100  controls  of each sex.
Food consumption was similar in all  the test groups.   The mean  body weights of
treated males were significantly reduced (P < 0.01) at the highest dose  level.
Other than reduced weight in the high-dose males, the  male and  female
treatment and control groups were comparable for weight,  mortality, and  tumor
incidence.

U.S. Food and Drug Administration (1977)--
    An unpublished chronic toxicity study of CdCl2 was conducted  at the  U.S.
Food and Drug Administration (U.S. FDA 1977).  The compilation  of animals
examined pathologically shows that six groups of Charles  River  COBS (SD) rats,
each consisting of 26 to 32 males and 26 to 29 females, were  studied. These
groups were given 0 (untreated controls), 0.6, 6, 30,  60, or  90 ppm CdCl£  in
the diet for 103 weeks.  Five males and five females per  group  were sacrificed
at 24 and 52 weeks.  All animals were necropsied, and  tissues,  organs, and
tissue masses were examined histopathologically.  Kidney  tissue from  five  or
fewer males in each sacrificed group was evaluated by  electron  microscopy;
sections of liver and kidney from these animals were stained  to assess
fibrosis, lipid content, liver glycogen, and the basement membrane of tubuli
and Bowman's capsules in kidney.
    No significant (P > 0.05) differences in survival  between control and
treated groups were reported, and, excluding interim sacrificed animals, no
more than two animals per group died before 77 weeks.   Results  of necropsy and
                                      59

-------
histopathologic and histochemical  evaluations did not show treatment-related
effects.  Electron microscopy revealed dose-related changes as small
cytoplasmic lipid droplets in renal  tubular epithelium,  increased numbers of
residual bodies in renal  nephron cells, and swelling and sloughing of cells  in
distal tubular epithelium and the collecting ducts of the kidney.
    Cadmium, as CdCl2» was not carcinogenic in this study, but a  stronger
evaluation of cadmium carcinogenicity possibly could have been made through
the use of larger group sizes and, as suggested by the lack of overt  toxicity
in treated groups, higher doses.
    Cadmium appears to be much less  potent by ingestion  than by inhalation  in
terms of the overall carcinogenicity regardless of the site of cancer
induction.  For example,  the total dose of inhaled cadmium in the Takenaka et
al (1982) study, where the rats developed a 71% incidence of lung cancer, was
about 7 mg (0.25 m3/day x 0.05 mg/m3 x 365 days/year x 1.5 years).  By
contrast, in the Schroeder et al.  (1965) drinking water  study in  rats, which
had one of the smallest total doses  of all the ingestion studies, a total dose
of about 60 mg (5 ppm x 0.05 x 0.35 kg x 730 days) induced no cancer response.
If we assume a 10% upper-limit of detection of tumors in the Schroeder et al.
(1965) study, the highest reasonable cadmium potency via ingestion is about
0.0017 (0.1/60), campared with a potency of about 0.1 (0.7/7) for inhalation.
    While it is possible that cadmium is not carcinogenic by ingestion at all,
the negative animal evidence can only set an upper-limit on the carcinogenic
potency of ingested cadmium, which in the rat appears to be about two orders
of magnitude less than for inhalation.
                                      60

-------
Summary
    Evidence for the carcinogenicity of cadmium and certain cadmium salts in
experimental animals has been obtained from studies of cadmium and cadmium
salts subcutaneously administered into mice and rats, as summarized in Table
9, and the inhalation exposure study of cadmium chloride aerosol  in rats by
Takenaka et al. (1982).  Oral-intake and inhalation studies were termed
inadequate by the IARC in 1976, apparently on the basis of the relatively
small  doses employed and the small percentage of absorption from the gastro
intestinal tract; however, the studies by Takenaka et al. (1982), Sanders and
Mahaffey (1982), and Loser (1980) were not reviewed by the IARC.   Schroeder's
work was specifically designed to simulate human exposure, and, for the most
part,  the doses given seem realistic for this purpose; however, the doses used
were apparently below the maximum tolerable doses usually used today in
attempting to establish the carcinogenic potential of various substances.
                                     61

-------
EPIDEMIOLOGIC STUDIES
    Epidemioiogic information dealing with the relationship between cadmium
exposure a'nd prostate cancer has been obtained from mortality studies.   A
major difficulty with mortality studies involves survivorship.   As
survivorship has been improving since the 1940s through  the present time, . ,
detecting an increased prostate cancer incidence in any  group based on
mortality from this disease has become increasingly difficult;  prostate cancer
victims are living longer and, more often than in the past, dying  from  other
causes (Robbins and Angell  1976).
    One of the lowest death rates  for this site is in Japan where  the
age-adjusted rate in 1974-75 was 2.4/100,000,  although Japan has the highest
per capita intake of cadmium in the world.  On the other hand,  Sweden has the
highest age-adjusted death rate from prostate  cancer (21.9/100,000 in 1974-75)
in the world, but a low daily intake and low body burdens of cadmium
(Kjellstrom et al. 197R).  Per capita rates, of course,  include many people
who are not exposed to the agent in question.
    There are 12 epidemiologic studies reviewed here that deal  specifically
with cancer risks resulting from cadmium exposure.  Although five  of these
were reviewed in the OHEA Health Assessment Document for Cadmium (May 1981),
they are covered here also for the convenience of the reader.

Potts (1965)
    Potts (1965) reported the results of a clinical  study of an unstated
number of current and former employees of a British alkaline battery factory
who were exposed to cadmium oxide  dust beginning in 1920 and ending in  1963.
In 1946 the manufacture of these batteries was moved to  a new location  not far
from the site of the earlier factory.  The first measurements of cadmium dust
                                      62

-------
in the air were made in 1949.  At this time,  cadmium in  the  air  varied  from
0.6 to 2.8 mg/m3 of air in the platemaking and assembly  shops  and  236
mg/m3 in the negative active material  department.   After the installation  of
local exhaust ventilation in 1950, cadmium in the  air was reduced  to less  than
0.5 mg/m3.  Improvements to the exhaust system in  1956 further reduced  the
dust to less than 0.1 mg/m3.  The policy at the time of  the  study's
publication was to take steps to reduce the exposure whenever  the  measurement
of cadmium dust exceeded 0.5 mg/m3.
    Of 70 battery workers for which Potts's clinic had medical  records  and who
were exposed for at least 10 years, proteinuria was observed in  44%.  Although
no comparison group was provided, this number is probably excessive  since
proteinuria is the result of renal tubular dysfunction.   A 200-248 ug/day
cadmium dietary intake over a 50-year exposure period is required  to produce a
critical renal cortex concentration that is associated with  renal  dysfunction.
Only 1% of Americans ingest more than 50 ug/day (U.S. EPA 1981).  However, the
author did note that the earlier studies of the urine protein  of
cadmium-exposed workers in this same plant revealed "similar characteristics"
as those of the present study.  Four individuals with persistent proteinuria
were examined further.  Two of them ultimately died.  Kidney function tests
prior to death revealed no abnormalities nor  were  any gross  abnormalities
observed following microscopic examination of the  kidneys of the deceased.
    In a second phase of this study, Potts claimed that  a "careful search"
produced a total of 74 men who had been exposed to cadmium dust  for more than
10 years and eight of them had died.  The author did not reveal  what was
searched, whether it was his clinic's medical records or employment  records of
the factory, nor did he say how these 74 men  relate to the 70  battery workers
mentioned earlier.  Furthermore, the source of his death information on the
                                      63

-------
eight was not available.  Five of the eight deaths were due to cancer and
three of these were cancer of the prostate.  The death data from Potts's paper
is summarized in Table 11.  Presumably, the death information on these eight
came from Potts' clinical  files, although it is not stated.  That the author
made any attempt to determine the vital status on the remaining 66 is unclear.
Since all of the deaths occurred in the early 60s and nearly all  had  lengthy
exposures, this implies that they were all  at risk to the highest cadmium dust
levels during their earlier years of employment prior to 1950.   No information
is given on workers exposed for fewer than  10 years.
    In the absence of selection bias (a distinct possibility if clinical
records were used), the distribution of the eight deaths is striking  as  was
noted by the author.  But  because of the possibility  of selection bias,  lack
of a comparison group, and an unknown age structure of the population of 74,
it is impossible to say if the observation  of three prostate cancer deaths is
statistically significant.  Therefore, this study provides only the suggestion
of an association of prostate cancer and exposure to  cadmium.

Kipling and Waterhouse (1967)
    Kipling and Waterhouse (1967), in a letter to the Lancet,  reported on  246
workers who had been exposed for a minimum  of one year to cadmium oxide  dust.
The authors compared the number of cancers  observed from several  sites with
the number expected from those sites based  on incidence rates  from the
Birmingham Regional Cancer Registry.  The number of observed cancer deaths of
the prostate was significantly greater than expected  (4 observed  vs.  0.58
expected, P < 0.003).  Three of the four prostate cancer cases  are the same  as
those reported in Pott's paper (personal  communication from Kipling to the
IARC in 1976) indicating that there is some acknowledged overlapping,  and

-------
TABLE 11.  MORTALITY DATA FOR CADMIUM WORKERS EXPOSED FOR MORE THAN 10 YEARS
                                 (Potts 1965)
                                  Length of
                                   Cadmium
Year of Death       Age           Exposure (yrs)           Cause of Death


    1960            65                31               Auricular Fibrillation

    1960            75                14               Carcinoma of Prostate

    1961            65                37               Carcinoma of Prostate

    1962            63                34               Bronchitis and Atheroma

    1962            78                18               Bronchitis

    1963            53                35               Carcinoma of Bronchus

    1964            65                38               Carcinoma of Prostate

    1964            59                24               Carcinomatosis
                                      65

-------
therefore they cannot be said to be independent studies.   No significant
difference between observed and expected was found for cancer of the  bronchus,
bladderj testis, or for cancers of all  sites.
    Latency period, although obliquely  referred to in  the  letter,  is  poorly
addressed.  Furthermore, the letter states that expected cases were calculated
by "computing the number of cases of cancer which  would be expected to  occur
in such a group of men of known age" and by excluding  the  time spent  in other
jobs or retirement.  It is not clear how the latter was to be done; the
discussion is sketchy at best.  The authors mention that "judging  from  work in
similar fields, fairly short exposure may be sufficient to initiate a tumor."
Whether this generalized conclusion by  the authors can be  extended to the
specific case of cadmium exposure and cancer remains uncertain.   Failure to
allow for a sufficient latency period weakens the  significance of  the
findings.  Because of these problems and the lack  of an adequate discussion of
the derivation of expected deaths, the  results, although statistically
significant, cannot be considered definitive with  respect  to the
carcinogenicity of cadmium.

Humperdinck (1968)
    Humperdinck (1968) reported on mortality among 536 people who  worked or
had worked at an alkaline dry cell battery plant during the period 1949-67 and
who had been exposed to cadmium hydroxide and "to  a large  extent nickel
hydroxide."  Seventeen of the 536 had died, five from  cancer.  Of  the five who
died from cancer, two died from lung cancer, one from  liver cancer, one from
prostate cancer, and one from cardiac cancer.  The length  of exposure to
cadmium for these cases was: lung, 2.3  years and 9.3 years; liver, 3.5  years;
prostate, 6.4 years; and cardiac, 3.0 years.
                                      66

-------
    There was no comparison group for the 1949-67 time period.   The  author  did
compare, however, the average of the cancer death rates for the years  1963-66
of the city where the plant is located with the average 1963-66 rate of  the
whole plant and the average 1963-66 rate for the departments of the  plant
where there was exposure to cadmium hydroxide.  The author did  not state
whether these rates were age-adjusted, race-adjusted, or sex-adjusted.   No
difference among the three rates was found.  No difference in the proportion
of lung cancer deaths between the city population and the plant population  was
found either.  The proportion of lung cancer deaths for the department where
cadmium exposure occurred was not reported.
    Previously, Baader (1951) had reported on "20 to 30 males and females"
suffering from chronic cadmium poisoning at the same dry cell plant.  Of this
group, Humperdinck reported that four of eight had died, one of lung cancer;
these four are included in the seventeen deaths described previously.  No
mention is made of any of the other "20 or 30" workers.
    Because Humperdinck found no excess cancer mortality amoung workers
exposed to cadmium when compared to the city population or to the plant
population as a whole, he concluded that there was insufficient information to
establish an association between cadmium and cancer.
    A major weakness of this study is that it did not include an appropriate
comparison group for the years of the study, 1949-67.  Comparison of average
death rates for the years 1963-66 among the city, plant, and cadmium
departments is not appropriate since it is not known whether all  workers in
the cadmium departments in 1963-66 had experienced a latency period  of
sufficient duration to have developed cancer.  Second, there is no indication
that the city population or the population of the rest of the battery plant is
similar to the cadmium-exposed group in terms of race, sex, smoking  habits,
                                      67

-------
a,ge, etc. tP  make an objective comparison.   Third, had a proper comparison
group be.e.q used  and an increase in cancer among workers exposed to  cadmium
been demonstrated, a, possible confounding variable would have been  the  nickel
exposure, since  nickel has previously been  associated with cancer of  the  lung,
nasal sinus,  Urge intestine, mouth, and pharynx (Fraumeni 1975).
    In conclu.sion,, the design and methods of  Humperdi.nck are inadequate to
assess whether an association between cadmium exposure and cancer exists  for
the workers in hi? study.

Hoi den (1969)
    Hqlden. (1969), in 3 letter to the Lancet,  reported on 42 men exposed  to
cad.m.iu.m fumes from 2 to 40 years.  He stated  that  six of the men had  been
exposed, to cpn,ce.n,tratiqn,s of cadmium in excess of  4 mg/m3 and the remainder
had been exposed to a,n average concentration  of Q.I mg/m3.  The author
reported, that there w,a,s one case of carcinoma  of the prostate and one case of
carcinoma, of  the bronchus.
    No evaluation of the cancer risk of cadm/ium can be made from this article
since the Author did not report important variables such as the age,  time
s,in,ce first exposure, and smoking history of  each  worker.

Kolopel (1975)
    Ko^one'l (1976) compared the. cad,m,iiU,m. exposure of 6.4 cases of ren.al cancer
to 197 non,mal,ijga,a,n,t cfigje.s.tlve. disease controls and 72 colon, cancer  controls.
According, to  th.e, author, "a ca.n,ce.r contra!  gro.up w,as included to, address,  the
problem o,f potential! n,oin(com.p.arab,i,liiity" between ca,ses $n,d contraljS when, a
aonjca.nce.r con.trolj g,ro,UP "is usedj.  Ca.se;s an,4 controls w,e,re taken, from, pati,en,ts
adm,i,tted from 19.57 to 1(96,4 to. ljto.swe.llj Pa,rk  M,emo,ri;al; lastitute, 8u,ffalO:, New
                                       6.8

-------
York.  Cadmium exposure was assessed using data on occupational  exposure,
cigarette smoking, and dietary intake.  A person was considered  to have
experienced occupational exposure to cadmium only if he had worked for one or
more years at a high risk job within a high risk industry.  High risk
industries included electroplating, alloy-making, welding, and the manufacture
of storage batteries.  A person was considered to be exposed to  cadmium
through smoking if he had at least 10 "pack-years" of cigarette  use during a
lifetime.  Dietary exposure to cadmium was determined by applying reports of
cadmium content in foods to individual dietary histories based on a frequency
recall for a one-week period.  An individual was considered exposed through
diet if his mean daily intake exceeded the third quartile, determined from the
distribution of intakes for the noncancer control group.
    The author found that the odds* of developing renal cancer in
occupationally-exposed patients who smoked was 4.4 when compared to controls
who also smoked and had nonmalignant diseases of the digestive system.  This
is significant at P < 0.05.  The odds of developing renal  cancer in patients
who were occupationally-exposed was 2.5 (P < 0.05) when compared to colon
cancer controls.  The latter is not significant (0.05 < P < 0.10).  Because of
the finding of a greatly increased risk2 when the effects from smoking and
occupational exposure are added together, the author concluded that the
effects of smoking and occupational exposure must be synergistic.
    The odds of developing renal cancer when consideration is given to cadmium
exposure through cigarette smoking only, and separately through  diet only
    •••Although the author referred to relative risk in his article,  it  is
more correct to use the term odds ratio or estimated relative risk.
    2Risk in this context is an estimated relative risk derived by  use of
the odds ratio.
                                      69

-------
(utilizing colon cancer controls), were 1.2 and 1.6, respectively,  neither of
which were significant (0.05 < P < 0.10, two-tailed).
    A major criticism of this study is the confounding exposures to other
industrial materials in the electroplating, alloy-making,  welding,  and
manufacture of storage batteries industries.  The author had  stated that  renal
cancer, as a result of cadmium exposure, is biologically plausible  because,  of
all the organs, the kidney concentrates cadmium most.   Furthermore, Kolonel
pointed out, from an earlier study by Ellman (1959), that  the kidney contains
the body's highest concentration of sulfhydryl  groups  which are  often found  in
zinc-containing enzymes.   Cadmium inhibition of a variety  of  sulfhydryl-
containing enzymes has been reported, the author notes, and this may be the
mechanism of action.  The kidney concentrates many trace metals, however,  and
a variety of metals are found in the industries mentioned  above, including
nickel, lead, and zinc.  Also, it is interesting to note that the odds ratio
for occupational exposure to cadmium is significant (P < 0.05) only when
compared to noncancer controls, but not significant (0.5 < P  < 0.10) when
compared to colon cancer controls.  This indicates that the renal cancer  cases
may not be comparable to the noncancer cases, and selection bias may have
occurred.
    Smoking has previously been associated with kidney cancer (Wynder et  al.
1974, Schmauz and Cole 1974, Kahn 1966, Hirayama 1977) as  well as cancers  of
other sites.  Although cadmium may be the carcinogen in the tobacco smoke
which causes kidney cancer, the issue is confounded by the presence of many
other carcinogens in the smoke.  Although the smoke may serve only  as a
possible synergist or as a carrier mechanism for cadmium exposure from other
sources, it remains to be demonstrated, as indicated above, that cadmium  is
the agent of concern in smoking.
                                      70

-------
    In conclusion, Kolonel's study provides suggestive,  but not sufficient
evidence that cadmium is a renal carcinogen.  More studies, epidemiologic and
animal, are necessary to adequately address the issue.

Lemen et al. (1976)
    Lemen et al. (1976) conducted an historic prospective study on 292 white
male employees of a cadmium smelter who had worked a minimum of 2 years in  the
smelter sometime during the period from January 1, 1940  to December 31, 1969.
Vital status was determined for this group through January 1, 1974.  Death
certificates listing the cause of death were acquired on 89 of a reported 92
deceased.  Some 20 (6.8%) remained lost to follow-up.  For comparison,
expected deaths by cause were generated through a modified life table
technique based upon the product of person-years times the corresponding age,
calendar time, and cause-specific mortality rates for the total United States
white male population.
    The authors stated that the smelter was engaged in the production of
cadmium metal  and cadmium compounds.  However, they reported that some lead
was also produced.  The plant ceased full-scale lead production in 1918 and
began to produce arsenic instead.  In 1925, arsenic production ceased with  the
start of cadmium production at that time.   The authors cited a previous
industrial hygiene survey in 1947 that reported average  air concentrations  of
cadmium fumes ranging from 0.04 to 6.59 mg/m3 and cadmium dust at 17.23
mg/m3.  The authors state that most operations had concentrations lower than
1.5 mg/m3.  The present study included a 1973 industrial  hygiene evaluation
of cadmium dust levels which stated that 8-hour time-weighted average (TWA)
gross concentrations of cadmium ranged infrequently up to 24 mg/m3, but
generally remained below 1 mg/m3.  The authors reported  that in 1973 a
                                      71

-------
respirator program was in use at the plant which allegedly reduced exposure by
a factor of 10.  However, respirators have not usually been found to be useful
in other settings as workers tend to remove them because of their
inconvenience.  Two air measurements taken in the preweld department showed
that besides air concentrations of 74.8 and 90.3 mg/m^ of cadmium, arsenic
was measured at 0.3 and 1.1 ug/m^.  This is about 1% of the cadmium
measurement.  In the retort department, however, where cadmium concentration
measured 1,105 ug/m^, arsenic measured 1.4 ug/m^, which was about
l/l,000t(l that of cadmium.  On the other hand, analyses of bulk samples
revealed 42.2% to 70% cadmium, 3.53% to 6% zinc, 0% to 4.3% lead, and 0.02% to
0.3% arsenic.  The remaining ingredients were not identified.   The authors
concluded  that the exposures from the remaining ingredients were
insignificant.
    A statistically significant excess of total  malignant neoplasms (27
observed vs. 17.6 expected, P < 0.05) was found, as was also the category of
malignant respiratory disease (12 observed vs. 5.1 expected, P < 0.05).
Without regard to latent effects, an excess of prostatic cancer was reported
by the authors to be not significant (4 observed vs. 1.15 expected).  However,
utilizing a One-tailed Poisson variable, the CAG found this observation to be
statistically 'significant (P < 0.05).  After a lapse of 20 years from initial
exposure, the finding of a statistically significant excess in prostatic
cancer (4 "observed vs. 0.88 expected, P < 0.01)  was even stronger.
    Information concerning exposure and latency  of the four prostatic cancer
cases is given in Table 12.
    Of the 12 malignant respiratory cancer cases, the cell  types of eight were
known.  Three were squamous cell carcinoma, one  was undifferentiated small
cell .carcinoma, three were anaplastic, and one was an oat cell carcinoma,
                                      72

-------
TABLE 12.  PROSTATE CANCER DEATHS AMONG CADMIUM SMELTER WORKERS
                  WITH MORE THAN 2 YEARS EXPOSURE
                      (Lemen et al. 1976)
Case Age
1 71
2 77
3 79
4 64
Exposure
4
13
18 -
17
Latency
32
25
31
26
Date of Death
2/26/72
3/19/68
12/10/60
4/3/51
                               73

-------
according to the authors.  Unfortunately, smoking histories were not available
for any members of the cohort.  Therefore, the confounding of the results due
to smoking cannot be assessed.  Only in one other study by Hoi den (1969) of
cadmium-exposed workers does the author take note of a single case of
bronchogenic -ca.ncer appearing in 42 workers exposed to cadmium on which the
author did not comment.  Furthermore, Lemen et al. reported the presence of
other substances in the smelter either known or suspected of causing cancer.
Arsenic, for example, has been shown to be a human lung carcinogen.   Thus,  any
conclusions that are made regarding the carcinogenic potential  of cadmium
should be tempered with the knowledge that arsenic, lead, and .zinc were also
known to be present in the atmosphere of the smelter.  Other constituents of
the processed ore are left unidentified (the percentage content of the  ore
does not add up to 100%).
    However, when consideration is given to the fact that the vital  status  of
6.8% of the study cohort remains unknown, additional causes of  death found  in
this group of 20 people potentially may add a prostate cancer or two to the
observed deaths.  In contrast, expected deaths are overestimated by  counting
person-years to the cut-off date for these same individuals.  This could bias
downward the finding of an excess risk of prostate cancer and bronchogenic
cancer.
    This study provides evidence that exposure to cadmium leads to a
significant excess risk of prostate cancer.  The other metals known  to  be
present have not been shown to be associated with an elevated risk of prostate
cancer.  On the other hand, the presence of arsenic in the atmosphere of the
smelter and the lack of information on smoking habits of the workers casts
doubt on the significant association of bronchogenic cancer in  these workers
with cadmium exposure.

-------
McMichael et al. (1976)
    McMichael et al. (1976), as part of an historic prospective study  of
cancer mortality among rubber workers, followed 18,903 active and retired  male
workers, aged 40 to 84, for a period of 10 years.   They were divided  into  four
separate cohorts each consisting of workers from the four tire manufacturing
plants of the companies under study.
    The mortality experience during the 10-year observation period was
determined from death claims filed with the companies under the group  life
insurance policy in effect.  In three of the four plants, workers were
included if they were employed on January 1, 1964, whereas in the fourth plant
it was January 1, 1963.  About 1% were lost to follow-up according to  the
authors, and death certificates listing cause of death were obtained  on 98% of
the deceased.  Expected deaths were calculated based on the 1968 U.S.  male
race-age-specific death rates.  The calculation of standard mortality  ratios
(SMRs) utilizing such rates produces an underestimate of the risk.  This bias,
known as the "healthy worker effect," is a consequence of the selection of the
healthiest individuals into a given workforce from the general population  from
which the expected deaths were derived.  Apparently, little turnover  occurred
in these four plants because the former employees who switched to another
place of employment formed the group of 1% lost to follow-up during the
10-year follow-up period.
    The total number of deaths equaled 5,106 for an overall SMR of 94. The
total number of cancer deaths equaled 1,014 for an SMR of 100, while  that  for
prostate cancer was 119 (103 observed, nonsignificant at 0.05 < P < 0.1).   The
author noticed an association of prostate cancer with the compounding  and
mixing areas of the four plants, work areas that entail  contact with metallic
oxides (including cadmium oxides).  McMichael  et al. also noted an association
                                      75

-------
of prostate cancer with three additional  work areas (cement  mixing,
janitoring, and trucking) of one particular plant after  "exploratory
work-history" analyses were completed for stomach, bladder,  and  prostate
cancer; lymphosarcoma; and Hodgkin's disease at  this  plant.
    The object of this study was not to single out the association of  prostate
cancer with cadmium exposure as the main  topic of study,  but rather to examine
site-specific cancer mortality, in general, in rubber workers.   Hence, the
authors found excesses in cancer mortality at a  number of different sites but
did not test the significance of any.  These data (McMichael  et  al. 1976) are
summarized in Table 13.  The tests of significance were calculated by  the
Carcinogen Assessment Group (CAG) using the method of Chiang (1961).
    One major problem with this study is  that rubber  workers are potentially
exposed to numerous organic and inorganic chemicals,  some of them known or
suspected carcinogens, including benzene, which  is a  known human carcinogen.
While cadmium is present in these factories, SMRs may be  confounded by
exposures other than cadmium.  Exposure levels of the many different compounds
found in these plants are not given.
    A second problem with this study is the relatively short observation time
(10 years) from the beginning of the study to its cut-off.  This is an
insufficient period in which to assess latent effects, and in fact, no data is
presented in which latency is considered.  This  cohort should be followed for
several additional years before concluding that  there is  no  effect from
cadmium exposure.
    While this paper is of interest as a  basis for further studies, it is
inadequate evidence for associating or not associating cadmium with prostate
cancer.
                                      76

-------
         TABLE  13.   STANDARD  MORTALITY  RATIOS (SMRs)  BY  SITE
                       (McMichael  et  al.  1976)
Site
Lymphatic leukemia
Stomach
All leukemias
Hodgkin's disease
Prostate
Colon
Pancreas
Bladder
Respiratory
Rectum
Brain,
central nervous system
All cancer
All causes
Observed
Deaths
20
80
46
32
103
103
57
32
252
27
14
1014
5106
SMRs
158
148
130
129
119
116
103
92
85
82
78
100
94
Probability of
Occurrence*
0.039
< 0.001
0.073
0.150
0.077
0.131
0.826
0.638
0.002
0.303
0.352
1.000
< 0.001
Taken from Chiang  (1961).
                                  77

-------
Kjellstrom et al. (1979)
    Kjell strom et al. (1979) reported on an historic prospective cohort  study
of 269 male Swedish cadmium nickel  battery factory workers and 94 Swedish male
cadmium-copper alloy factory workers having more than 5 years  exposure since
the factories began production.  As an internal  reference group, 328 alloy
factory workers not exposed to cadmium were also studied.  They were employed
in the alloy factory for at least 5 years but not exposed to cadmium.   It has
been estimated that the average cadmium levels for one of the  two factorys
were as follows:  exceeded 1 mg Cd/m3 prior to 1947; 200 ug Cd/m3 between
1962 and 1974; 50 ug Cd/m3 in 1974; decreased to below 5 ug Cd/m3 at the
time of the study.  At the other factory, concentrations were  in the range of
100 to 400 ug/m3 in the mid-sixties and 50 ug Cd/m3 in 1971 and after.
The battery study population was also exposed to nickel  hydroxide dust.
    National average age-cause-specific death rates and cancer incidence rates
were used to generate expected deaths and expected new cancer  cases  in the two
separate study groups.  New cases of cancer were found in the  battery factory
by matching the names of the 269 workers with those of the Swedish National
Cancer Register.  This was not done with the alloy factory workers.   With
respect to mortality in the battery factory, 43 deaths occurred between  1949
and 1975 of which eight were due to cancer.  This contrasts with 67  expected
total deaths during the same period.  No further breakdown is  given  of the
cancer deaths and no expected cancer mortality is given.  However, the authors
state that there was no increase in "general" cancer mortality.  Furthermore,
the total number of new cases of cancer equaled 15 during the  period 1959 to
1975, while the expected number of new cases equaled 16.4 based in incidence
data provided by the Swedish National Cancer Register.  A breakdown  by site is
given in Table 14.  Only cancer of the nasopharynx was found to be
                                      78

-------
TABLE 14.  EXPECTED AND OBSERVED NEW CASES OF CANCER BETWEEN 1959-1975
        IN THE WHOLE GROUP OF BATTERY FACTORY WORKERS (N = 228)
                        (KjellStrom et al. 1979)
Site
Prostate
Lung
Kidney
Bladder
Colon-rectum
Pancreas
Nasopharynx
Other
All sites
Cancer
Expected*
1.2
1.35
0.87
1.07
2.25
0.60
0.20
9.81
16.4
Cases
Observed
2
2
0
1
5
0
2
3
15
Risk Ratios
1.67
1.48
0
0.93
2.22
0
10. Ot
0.31
0.91
 *Expected deaths based on Swedish National  Cancer Register.
 tStatistically significantly greater than 1 (P < 0.05).
                                   79

-------
significantly in excess (2 observed vs. 0.2 expected, P < 0.05) possibly due
to exposure to nickel dust.
    In the alloy factory, only "preliminary" calculations of prostatic cancer
mortality were done according to the authors; cause-specific mortality or
incidence was not examined in these workers.  Among 94 exposed workers, four
prostatic cancer deaths were noted versus 2.69 expected (P = 0.29).   In the
reference group of 328 nonexposed workers, four prostatic cancer deaths were
noted versus 6.42 expected (P = 0.23) (Table 15).   A corrected "healthy worker
effect" risk ratio was derived by dividing the risk of developing prostatic
cancer in the exposed by that of the reference group.  This equals 2.4 (P =
0.087), which is still nonsignificant.
    Although the results of these two studies are  nonsignificant with  respect
to prostatic cancer and basically inconclusive due to the small  sizes  of the
study groups, they do suggest a positive association of prostate cancer with
exposure to cadmium.
    Two problems with this work are apparent.  The first is that terminated
employees are apparently not included in any of the study cohorts unless they
are terminated by death.  The resulting cohorts are healthier than the general
population because former employees, who would be  expected to carry  the
greatest burden of potential  disease, are not represented.  These employees
are represented in the general population death rates however.  The  net result
is to overestimate the expected deaths, thus masking a potential  risk  in
battery workers.
    Second, in the incidence studies, cancer cases occurring in the  1950s
would be missed since the Swedish Cancer Registry  didn't begin .until  1959,
thus leading to an underestimate of new cancer cases.
    Another potential source of selection bias involves the exclusion  of all
                                      80

-------
TABLE 15.  CUMULATIVE EXPECTED AND OBSERVED NUMBER OF PROSTATIC CANCER DEATHS
                   IN 1940-1975 AMONG ALLOY FACTORY WORKERS
                           (Kjellstrom et al. 1979)
                        Prostatic Cancer Deaths

                       Expected        Observed       Risk Ratios      P value


Exposed group            2.69             4              1.49           0.29


Reference group
 (N = 328)               6.42             4              0.62           0.23
                                      81

-------
members with incomplete information in the factory files.   However,  since
there is no reason to assume differential  selection of subjects  for  study
through this procedure, it may not be a problem.

Hoi den (1980)
    Holden (1980) reported the results of  a preliminary cohort mortality study
of workers employed sometime from August 1940 to  August 1962 and followed to
December 31, 1979, in a British cadmium factory,  while iron and  brass  foundry
workers in a second factory served as a control.   The cadmium factory  was
further subdivided into two sections for analysis purposes.  One section of
the building contained the copper cadmium  alloy department.  In  that
department, 347 men worked a minimum of 12 months while another  624  men  worked
a minimum of 12 months in the remaining part of the factory.  These  were
dubbed "vicinity" workers by the author because they worked within the
building but not in the copper cadmium alloy department.   Another 537  brass
and iron workers were employed in the second British factory for a minimum of
12 months and were reported by the author  to be similar to workers of  the
first factory with respect to their social  and physical  environment.
    Industrial  hygiene surveys carried out at this factory in 1953 and 1957
indicated a mean level of 70 ug/m3 (S.D. = 62 ug/m3) based on 12-hour
sampling data in the cadmium copper alloy  department, while a mean level  of  6
ug/m3 (S.D. = 8 ug/m3) was measured in the vicinity.  Vicinity workers
were manufacturing arsenical copper, and during refining they were exposed to
silver and nickel.  The author reports that those workers  were exposed to
considerably less cadmium than were the cadmium copper alloy workers.
Follow-up was over 95% complete on all three subcohorts.   Expected deaths were
generated based upon death rates for England and  Wales in  5-year age
intervals.
                                      82

-------
    A statistically significant elevated risk of dying from all  causes
(observed = 158, SMR = 112) was evident in cadmium copper alloy  workers.   This
was repeated even when malignant neoplasms were excluded (observed = 122,  SMR
= 113).  Mortality from neoplasms was not significant in cadmium copper  alloy
workers except for leukemia (observed = 3, SMR = 441).  The author contends
that the excess overall was due to pulmonary disease deaths.  On the other
hand, a statistically significant elevated risk of cancer in general  (observed
= 72, SMR = 120) was apparent in "vicinity" workers due chiefly  to an excess
of cancer of two sites, lung cancer (observed = 36, SMR = 138) and cancer  of
the prostate (observed = 8, SMR = 267).  The author attributes the elevated
risk of lung cancer in these workers to the presence of other metals such  as
arsenic.  With respect to prostate cancer, he correctly noted the absence  of a
dose-effect relationship since five of the eight prostate cancers occurred to
individuals with less than 15 years exposure; three were exposed only one  year
if one assumes that "years of exposure" means years of exposure  throughout the
entire plant.  The author attributes only three of the prostate  cancer deaths
to cadmium exposure.  This last observation is somewhat strong in view of  the
fact that every prostate cancer case occurred more than 15 years after initial
exposure.  Latency as a factor was not considered in calculating expected
deaths so that the actual risk may be greater.
    It should be noted that the workforce of any factory may be  rotated  many
times during its operating life.  The fact that cadmium alloy workers, under
the author's definition, apparently experienced a lower risk of  prostate
cancer than did "vicinity" workers may not be so unusual.  It is possible  that
several of the eight cases worked in the copper cadmium alloy department as
well  as in the remaining part of the plant at some time during their working
career.
                                      83

-------
    The observed  risk of cancer may actually be greater than calculated
because of the presence of the healthy worker effect.  The fact that this
factor is '.operating in these .cohorts is 
-------
                   fjg Cd/m3 air
                10,000
                 1000
                   100
                    10
                      1946   1956    1966   1976 Tear
Figure  1.  Concentration  of cadmium in the air  (ug Cd/m3) from 1949 to 1976,
          Arithmetic mean of stationary and  personal samples.
          (Kjell Strom 1982).
                                    85

-------
    From T94'6 t'O 1967 there appears to ''have -been -a .-1,000-fdTd  drop  i'n  average
exposure levels.  A detailed analysts frf past arvd present  ca'dmi-um expbWres  in
this factory has been published '('Adams's-on 1979%  The author report's t!h;at
nickel exposure levels have been at least the same as that "of 'cd'cUntum  and
oftentimes as much as 10 times hi'gher.
    Problems with this study are as follows;  The records >of te'rfhinatetl
employees prior to 1945, a group in which the greatest ris'k is  likely  to be
found, are nonexistent.  Almost 31% of this group hatl less than '2 years
duration of exposure to cadmium.  Almost 50% of the tohbrt (3'01 wb'rkers)
received their first exposure to cadmium after 1959, which means a  large
proportion of the cohort had not been followed even 20 years, an'd thus, there
was probably not enough time to evaluate a cancer risk.  Furthermore^  smbkifig
information was not available for the older workers, a subgroup of  the cohort
in which the greatest cancer risk is likely to be found*  This  conceivably
could have been the reason why no results evaluating the effects of smbkifig
are presented in the study, although a detailed data-base is reported  By the
author to be in the development stages as an extension of the study for future'
follow-ups.  Additionally, the author reports that for cancer 6f the p'rostatei
the rate ratio increases with increasing latency and increasing dose*  He
reports rate ratios of 1.27, 1.33, and 1.55 corresponding to exposure
categories > 0 years, > 1 year, and > 5 years.  In the > 1 y£eif exposure
duration category, prostate mortality rate ratios of l«33j l«44j arid l:8lj
corresponding to latency periods of 1, 10, and 20 years* respectively,- are
given.  However, since no tabular data is presented* it is7 n~bt pbssiBle to
determine how the four observed prostate cartcer deaths <¥re distributed into
the subcategories alluded to by the author.  The autlfbr did fibCe1 that  the
numbers were too small for the detection of statistically s'igriificant
differences.

-------
    Kjellstrom repeated the above exercise for urogenital  diseases.   For those
with more than 20 years exposure and 20 years latency, 4 observed  urogenital
deaths occurred versus 0.93 expected (P < 0.05).   This type of disease  was
exclusively nephritis of the kidney.  Again it is difficult to conclude
without evaluation that cadmium exposure was related, although the author
himself states that it is "clear that cadmium exposure increases mortality
from kidney diseases" after high exposure intensity and long duration of
exposure.  The author notes a tendency in his data for a slightly  increased
nonsignificant risk of prostate cancer from exposure to cadmium.
    In addition to the main study discussed above, Kjellstrom included
discussions of four Japanese studies [Japanese Public Health Association
(JPHA) 1979; Shigematsu et al. 1981; Nogawa et al. 1978, 1981] and a
description of another planned ecological study by himself and the Department
of Epidemiology at the University of Tokyo for which only preliminary findings
are available.  In this latter ecological study age-standardized death  rates
in cadmium polluted areas for persons 35 to 84 years of age were compared with
the respective rates in non-cadmium polluted areas.  Preliminary data,
according to Kjellstrom, suggested a nonsignificant tendency toward  higher
mortality rates in cadmium-exposed areas compared to control areas (an
age-adjusted mortality rate of 176 per 1,000 in cadmium-exposed areas versus
139 in the control areas).  Prostate cancer and kidney disease mortality rates
were also higher in the cadmium-exposed areas, but most of the prostate cancer
mortality excess occurred in individuals 85 and over.  No significant tests
were done.  This analysis was reported by Kjellstrom as tending to support  the
hypothesis of a cadmium effect, but "definite conclusions have to  be left
until  all the analysis are completed."
                                      87

-------
    The first of these (JPHA 1979, Shigematsu et al. 1981) was an analysis of
cadmium exposure and mortality in the general environment.  According to the
author, people in many areas of Japan endure high cadmium exposure up to
several micrograms/day from consumption of contaminated rice.   In each of four
"prefectures" of Japan, age standardized mortality rates were  calculated in a
cadmium-exposed area and compared to those calculated in a nonexposed
reference area of the same prefecture.  It was found that cancer mortality
rates were generally about the same in the non-polluted areas  compared to the
polluted areas, but no significance tests were run.   Only kidney disease death
rates and diabetes death rates were found to be lower in non-cadmium polluted
areas compared to cadmium polluted areas.  With respect to prostate cancer
mortality, two of the polluted areas had higher death rates than did their
controls, while in two others the reverse was true according to the author.
The author noted that the two prefectures with higher death rates of prostatic
cancer compared to their controls are the areas with the "highest likely
cadmium exposure to the population."  The remaining  two prefectures had lower
cadmium exposures.  The former two prefectures tended to have  higher rates of
mortality from kidney disease and hyperplasia of the prostate  as well.  Again,
this is an ecological study and thus can only be considered as suggestive of
areas for future research.
    The second Japanese study (Nogawa et al. 1978) found that  in 2,689 men and
women over age 50, the village-specific prevalence of low molecular weight
proteinuria (LMWP) increased with an increase in the village-specific average
cadmium concentration in rice.  LMWP was measured by urinary retinol  binding
protein.  Since this ecological study more than likely includes more persons
who have never been exposed to cadmium in rice as well  as prevalence rates
that include persons with prior-existing conditions, possibly  introduced long
                                      88

-------
before they were exposed to relatively high concentrations of cadmium in rice,
the positive association noted by the author should not be construed to
signify a causal association.
    Nogawa et al. (1981) conducted a mortality study of the 81 men and 124
women identified in the earlier study as having LMWP.  They were followed from
1974 to 1979 as well as the remaining men and women not found to have LMWP.
He found a nonsignificant (P < 0.05) twofold excess risk of death to men with
LMWP and a nonsignificant 1.2-fold excess risk of death to women with LMWP.
Mortality rates were based on 27 deaths of males with LMWP and 30 deaths of
females with LMWP.  A positive association of LMWP with heart disease,
cerebrovascular disease, nephritis, and nephrosis was noted.  This association
raises the spector of a possible confounding effect of hypertension with LMWP.
If hypertension is a cause of LMWP, the higher mortality in individuals with
LMWP described above may have been a consequence of the hypertension and not
necessarily that of LMWP brought on by cadmium exposure as suggested by the
author.  Thus, the correlation with LMWP may be spurious.  Hence, conclusions
drawn from this study regarding an association of higher mortality with
cadmium exposure must be characterized as certainly no more than suggestive.

Kazantzis and Armstrong (1982 unpublished)
    In a recently completed but unpublished cohort mortality study of 6,995
male cadmium workers with at least one year service in one of five British
industries (primary producers, copper-cadmium alloy, silver-cadmium alloy,
pigments, and plastic stabilizers), the only finding of any statistical
significance was that of bronchitis deaths occurring to employees classified
as having "high" exposure to cadmium (12 observed versus 2.5 exposed,
P < 0.01).  The excess of bronchitis (chronic respiratory disease) occurred
                                      89

-------
among workers exposed mostly to Cadmium fumes* although the author "reports  a
nonstatistically significant excess among those exposed to Cadmium dust  only-
no data is provided to support this finding.   Workers  wha Were borfi after  1940
or who were exposed after 1970 were excluded  from the  study*   The  remaining
cohort was followed to 1979 when it was reported that  1*902 deaths had
occurred versus 2,056 expected (SMR = 93),  a  shortfall  of deaths attributed
most likely to the "healthy worker effect*"  No information is provided
concerning the degree of success of the follow-up*   Standard  mortality rates
were presumably based upon British death rates*  the entire cohort was divided
into three categories* "high*" "medium/ and  "low*"  according to severity of
exposure*  some 3% fell  into the high category 17%  were in the medium
category* and 80% were in the low category* according  to the  author*  These
categories of exposure were devised by industrial  hygienist§  based upon
knowledge of past and present processes and working  procedures*, and upon
biological and environmental monitoring results where  available*   The number
of persons who fell into the highly exposed category (certainly no more  than
210 persons total) was a rather small number  upon which to estimate a Msk  of
prostate cancer*
    No other excessive risks of death by cause (including cancer)  were present
according to the authors*  Unfortunately, only one table of tabular results of
cause«specific mortality are available in this Study by exposure Category*  and
this table has no breakdown by time Since onset Of exposure*   Hence* no
consideration of latent  effects are given in  the Study*  Therefore* it is
impossible to assess the risk Of cancer in  just that group followed for  more
than 15 or 20 years after Initial exposure  to cadmium*
    It Should be noted that this study 1§ not independent of  an iariier  §tudy
by Sorahan (1981, unpublished).  The author acknowledges the  inclusion of

-------
3,026 male (43% of the cohort) nickel-cadmium workers who are the subject  of a
cohort mortality study by Sorahan.   (The Sorahan study is critically reviewed
following this discussion.)  Sorahan also found essentially negative results.
Kazantzis himself admits that with respect to both kidney disease and
prostatic cancer, the "numbers are too small to exclude an increased risk
emerging with time," although he states that if such a risk existed, it "could
only be small in terms of extra deaths expected."  It is difficult to see  how
he could make the latter statement since he made no effort to sort out latent
effects. This study is, in the opinion of the CAG, inconclusive in terms of
finding no risk of prostate cancer or cancer of other sites.

Sorahan (1981)
    Sorahan (1981), in a letter to Dr. Roy Albert (dated 2/18/81), related the
findings of an historic prospective mortality study of 3,026 nickel-cadmium
battery workers employed during the period from 1946 to June 30, 1980, who had
worked at least one month.  A subset of these same workers was studied earlier
by Kipling and Waterhouse (1967), according to the authors.  The cohort was
derived from workers employed in two separate factories that amalgamated in
1974.  The earliest mention of cadmium in the air was reported in 1949.  In
the platemaking assembly shops, the range was 0.6 to 2.8 mg/m3, but  in the
"negative active material" department where cadmium oxide powder was prepared,
the levels were reported to be "considerably higher."  No numbers are
provided.   Extensive local exhaust ventilation was installed in 1950, and  as a
consequence,  cadmium in the air was reduced to below 0.5 mg/m3 in most parts
of the factories.  By 1967, when a new platemaking department was built, the
cadmium oxide dust was reduced to less than the threshold limit value (TLV) of
0.2 mg/m3.  From 1975 on, the plant levels were within the current TLV of
0.05 mg/m3.
                                      91

-------
    For the purposes of the analysis the author excluded 566 female employees
and divided the remaining all  male cohort into two subcohorts:  1,066 who were
first employed before the amalgamation (1947) and 1,494 who were first
employed after the amalgamation.
    Standard mortality ratios (SMRs) were computed.  Expected deaths were
generated assuming that the general  population rates were operating on the
study cohorts.  Overall, the observed numbers of deaths, all  causes combined,
were slightly less than expected (observed = 591, SMR = 97).   With respect to
all forms of cancer, there was virtually no difference between observed and
expected deaths (observed = 152, SMR = 100).  On the other hand, a deficit of
cancer deaths occurred to the subcohort of employees employed prior to the
amalgamation (observed = 80, SMR = 84).  But, in those employed for the first
time after the amalgamation, a significantly increased risk of total  cancer
deaths was apparent (observed = 72,  SMR = 129, P < 0.05).  This increased risk
was attributed by the author to a survivor population effect  in the latter
subcohort.  However, in both subcohorts, before 1947 and after 1947,  an
excessive but nonsignificant risk of cancer of the bronchus was evident
(observed = 45, SMR = 114; observed  = 32, SMR = 134, respectively).  No
significant excess risk of prostate  cancer occurred in either group (observed
= 4, expected = 4.1; observed = 3, expected = 1.9, respectively).   Even after
consideration was given to the time  since first employed, no  significant
excessive risk was seen in workers 15 years after first employment in any of
the following cause of death categories:  all causes combined, all cancers
combined, cancer of the bronchus, and cancer of the prostate.  Nor was there a
significant risk of cancer to employees having left employment before
completion of one full year of employment and one to 14 years of employment.
In an apparant inconsistency, Sorahan reported no prostate cancer deaths in
                                      92

-------
the length of employment category 1 to 14 years;  yet in  the earlier  study  by
Potts (presumably based upon the same data),  three employees whose deaths  were
attributed to prostate cancer were exposed to cadmium for 14, 37, and  38
years, respectively.  Given the cohort definition of Sorahan, at  least one of
these three cases should be included.
    Furthermore, in Sorahan's analysis of latent  effects, implicit is  the
assumption that only terminated employees were considered in the  analysis.
Presumably, person-years of individuals still employed with the company were
not enumerated.  Only if the individual left  the  employment of the company
(through death or other cause) would his person-years be counted.  This has
the effect of reducing the expected deaths by the non-inclusion of
person-years of individuals who are at risk of death but are still alive and
working.  This would tend to bias the SMRs upward.
    On the other hand, without regard to latent effects, the study suffers
from a type of selection bias known as the "healthy worker effect" brought
about by comparison of the observed deaths with expected deaths based  upon the
mortality rates of England and Wales, thus biasing the result toward the null.
Additionally, some 82 persons remain untraced with respect to their  vital
status.  The non-inclusion of the causes of death of the decreased members of
this subgroup would tend to bias the SMRs downward.
    More seriously, tabular data presented classifies the cohort  into  two
categories of exposure:  "exposed" and "non-exposed," although in the
"population section" the author described all the jobs in the factories in
terms of "high," "slight," and "minimal" exposure to cadmium.  It cannot be
determined from the text how the three latter categories were reconstituted as
"exposed" and "non-exposed" for the purposes  of presenting the findings in
tabular form.  The author's treatment of the  subject suggests that some
                                      93

-------
portion of the study population received little or no exposure to cadmium,  and
if this is so, they should have been excluded from the study group,  or else a
better definition of intensity of exposure should have been utilized to
present the tabular findings.  It would have been more appropriate and
informative to present the tabular findings in terms  of "high, slight, and
minimally" exposed groups.  Otherwise,  the cancer incidence cannot be
adequately related to cadmium exposure.
    Overall, this paper presented no evidence of an increased risk of prostate
cancer in cadmium-exposed workers.  However, since many questions remain
unanswered concerning the structure of  the study, it  cannot be said  to provide
conclusive evidence that cadimum is not carcinogenic.

Summa ry
    Of the 12 epidemiologic studies of  cancer in cadmium-exposed  people
reviewed by the CAG, three (Kipling and Waterhouse 1967,  Lemen et al. 1976,
Holden 1980) provide evidence of a significant association  with prostate
cancer.  Although the numbers are very  small, four cases  in each  study, a
statistically significant positive association was observed (P <  0.05) in all
three studies.
    Three other studies (Potts 1965, McMichael et al.  1976, Kjellstrom 1978)
provide the suggestion of a risk of prostate cancer (although statistically
nonsignificant) with exposure to cadmium.   Potts's clinical  study (which
provided three of the four prostate cancer cases in the Kipling and  Waterhouse
study) suffers from a lack of a comparison group.  The McMichael  et  al. study
of mortality in rubber workers suggested a positive correlation of exposure to
cadmium with prostate cancer.  However, the lack of significance  even with
large numbers (observed = 103, SMR = 119)  and the concomitant exposure to

-------
other known and suspected carcinogens make this study equivocal.   Kjellstrom's
"corrected healthy worker effect" risk ratio of 2.4 is still  nonsignificant
because of small numbers, although it approaches borderline significance  at
P < 0.09, offering the suggestion of a possible association of  prostate cancer
with cadmium exposure.
    Two other studies (Humperdinck 1968, Holden 1969) did not report  evidence
of an association of prostate cancer with cadmium exposure chiefly because the
comparison population was either inadequate to assess a risk (Humperdinck) or
absent entirely (Holden).  The ninth study (Kolonel 1976) did not evaluate the
risk of prostate cancer in cadmium-exposed people but did evaluate the risk  of
renal carcinoma and cadmium exposure.
    An update by Kjellstrom (1982) of his earlier 1979 study again failed to
demonstrate a significant risk of cancer of any site.  One of the failings of
this study was that members of his cohort were not observed long  enough to
evaluate latent effects.  More than half of his cohort received no exposure  to
cadmium prior to 1959 and thus could not have been followed for even  20 years.
    The study by Kazantzis and Armstrong (1982) of 6,994 workers, which
included the entire Sorahan cohort, also failed to demonstrate  an increased
risk of cancer of any site.  This study combines cohorts from several
different plants each with their own unique exposure history, none of which
are necessarily comparable.  The main failing of this study, however,  is  the
lack of consideration of latent effects.
    Kolonel (1976) found a statistically significant elevated risk of renal
cancer in persons occupationally exposed to cadmium, and an even  greater  risk
in occupationally exposed people who smoke, thus raising the possibility  of  a
synergism.  The chance of selection bias and concurrent occupational  exposures
to nickel, lead, zinc, and a variety of metals minimizes the importance of the
                                      95

-------
findings.  None of the other human epidemiologic studies reviewed have
demonstrated an elevated risk for the development of renal  cancer and exposure
to cadmium.
    Lemen et al. (1976), in addition to a significant risk  of prostate cancer,
found a statistically significant elevated risk of bronchogenic carcinoma in
cadmium smelter workers (12 observed vs. 5.1 expected, P <  0.05).  But these
same workers were simultaneously exposed to low levels of arsenic,  lead,  and
zinc.  Arsenic is a known pulmonary carcinogen and thus, confounding  cannot  be
eliminated.  Furthermore, smoking habits among workers in the rather  small
study cohort are not known.
    Holden (1980) also noted a statistically significant elevated risk of
bronchogenic carcinoma in British cadmium workers who worked  in the "vicinity"
of the cadmium copper alloy department.  However, the author  attributed the
excess risk to the presence of other metals, such as arsenic.
    The Sorahan (1981) study, which reportedly includes the same group of
workers evaluated by Kipling and Waterhouse, is negative for  an increased risk
of prostate cancer.  Separation of the study group according  to "exposed"
versus "unexposed" is unclear.  Since the control  group constitutes the
population of England and Wales, the "healthy work effect"  may bias the result
downward.  Furthermore, although it is not clear whether the  first  four tables
include only terminated employees, the rest of the tables explicitly  state
that is the case.  If only terminated employees were analyzed, the  SMRs would
be biased upward.  For these and other reasons, the study is  inconclusive to
determine an association between cadmium exposure and cancer.
    It might be of interest to estimate the likelihood that positive  results
in the five independent study populations discussed above could occur by
chance alone under the hypothesis that cadmium had no effect  on deaths due to
                                      96

-------
prostate cancer.  To test this hypothesis, a formal  statistical  test  of  the
joint information contained in all  the studies can be made  in  the  following
manner.  Often the same hypothesis  is tested in a series of independent
studies where an estimate of the probability of rejecting a true hypothesis,
the "P" value, is calculated for each study.
    The results of any one individual study may be only marginally
significant, but the series of consistent results among the studies  gives  the
appearance of much stronger evidence than any one study.  It would be
desirable to combine the results of the studies in some manner to  obtain some
overall statement about the "P" value of the series of experiments in order to
obtain a joint result.  However, often due to differences in how the data  were
analyzed and obtained in the studies, this approach was not possible.  Sir
R.A. Fisher developed a simple, straightforward method to analyze  data in  this
situation where the only information needed is the "P" value for each of the
studies.  He showed that the statistic
                                     m
                             S = -2  E loge Pj
is distributed as an X2 value with 2m degrees of freedom, where Pj  is  the
"P" value for the jth study, and m is the total  number of studies.   However,
caution must be taken in utilizing this approach to be sure that all
methodologically acceptable studies that have been used to test the required
hypothesis are included in the series and that all of these studies are
independent of each other.  If studies that did  not produce a  "significant"
result are not reported in the literature, or one study uses part or all  of
the actual data from another study, the proposed use of the Fisher  method
could lead to a biased result.
                                      97

-------
    In the present situation a total of five independent studies have been
conducted testing the hypothesis of no association between exposure to
airborne cadmium and prostate cancer.  These studies, along with the resulting
"P" valueSj are as follows:
       Study
Kazantzis and Armstrong (1982)
Lemeh et al. (1976)
McMichael et al. (1976)
Kjellstrom et al. (1978)

Hoi den (1980)
Source of Exposure              "P" Value
Combined-Primary Production
Cadmium Copper Alloy
Cadmium Silver Alloy
Plastics Stabilizer
Pigments                          0.52
Cadmium Smelter                   0.013
Rubber Plants                     0.077
Cadmium Nickel Battery Plant
Cadmium Copper Alloy Plant        0.375
Cadmium-Copper Alloy              0.010
    Applying Fisher's method, we find that the statistic

 S = 2{LN {'0*52) + LN {0.013) + LN (0*077) + LN {0.375) + IN {0.01-0)3 = 26.28

has a X^ distribution with 2 x 5 = 10 degress of freedom and a joint "P"
value of '< 0.01 associated with it.
    Although, this calculation indicates that the association of prostate
cancer and cadmium exposure in these five studies is not likely to be 'due to
chance alone, a definite statement about caus'ation cannot be .made considering
the limitations of these studies.
                                       98

-------
Conclusion
    Although problems exist in the design and methodology  of all  12
epidemiologic studies evaluating the association of cadmium exposure  with
prostate cancer, six of them are nevertheless consistent in their findings  of
an elevated risk of prostate cancer.  But only in three are the data
statistically significant.  The remaining six studies are  inadequate  to
evaluate this risk.  Because of small  sample sizes and a negative
dose-response curve in one of the significant studies, the evidence  is limited
that cadmium is a human carcinogen.
    In two studies (Lemen et al. 1976, Holden 1980) where  a significant  risk
of bronchogenic cancer was apparent, there is a suggestion of an association
of exposure to cadmium with bronchogenic carcinoma.  However, the confounding
effects of exposure to arsenic in both these studies makes it questionable
that the association is real with respect to renal cancer, and there  is
sufficient reason to doubt the finding of a significant positive association
in the only study that demonstrated such a risk (Kolonel 1976).
                                      99

-------
                       QUANTITATIVE ESTIMATION

INTRODUCTION
    This quantitative section deals with the unit risk for cadmium in air and
the potency of cadmium relative to other carcinogens that the CAG has
evaluated.  The unit risk estimate for an air pollutant is defined as the
lifetime cancer risk occurring in a hypothetical  population in which all
individuals are exposed continuously from birth throughout their lifetimes to
a concentration of 1 ug/m3 of the agent in the air that they breathe.  This
calculation is done to estimate in quantitative terms the impact of the agent
as a carcinogen.  Unit risk estimates are used for two purposes:  1) to
compare the carcinogenic potency of several  agents with each other, and 2) to
give a crude indication of the population risk which might be associated  with
air or water exposure to these agents, if the actual exposures are known.
    The data used for the quantitative estimate are taken from one or both of
the following: 1) lifetime animal studies, and 2) human studies where excess
cancer risk has been associated with exposure to the agent.  In animal  studies
it is assumed, unless evidence exists to the contrary, that if a carcinogenic
response occurs at the dose levels used in the study, then responses will also
occur at all lower doses with an incidence determined by the extrapolation
model.
    There is no solid scientific basis for any mathematical extrapolation
model that relates carcinogen exposure to cancer risks at the extremely low
concentrations including the unit concentration defined above.  For practical
reasons such low levels of risk cannot be measured directly either by animal
experiments or by epidemiologic studies.  We must, therefore, depend on our
current understanding of the mechanisms of carcinogenesis for guidance as to
which risk model to use.  At the present time the dominant view of the

                                      100

-------
carcinogenic process involves the concept that most agents that cause cancer
also cause irreversible damage to DNA.  This position is reflected  by the  fact
that a very large proportion of agents that cause cancer are also mutagenic.
There is reason to expect that the quantal type of biological  response,  which
is characteristic of mutagenesis, is associated with a linear nonthreshold
dose-response relationship.  Indeed, there is substantial  evidence  from
mutagenicity studies with both ionizing radiation and a wide variety of
chemicals that this type of dose-response model is the appropriate  one to  use.
This is particularly true at the lower end of the dose-response curve; at
higher doses, there can be an upward curvature probably reflecting  the effects
of multistage processes on the mutagenic response.  The linear nonthreshold
dose-response relationship is also consistent with the relatively few
epidemiologic studies of cancer responses to specific agents that contain
enough information to make the evaluation possible (e.g.,  radiation-induced
breast and thyroid cancer, skin cancer induced by arsenic  in drinking water,
liver cancer induced by aflatoxin in the diet).  Some supporting evidence  also
exists from animal experiments (e.g., the initiation stage of the two-stage
carcinogenesis model in rat liver and mouse skin).  Linearity is also
supported when the mode of action of the carcinogen in question is  similar to
that of the background cancer occurrence in the exposed population.
    Because it has the best, albeit limited, scientific basis of any of  the
current mathematical extrapolation models, a linear nonthreshold model has
been adopted as the primary basis for estimating risk at low levels of
exposure.  The risk estimates made with this model should  be regarded as
conservative, representing the most plausible upper-limit  for the risk,  i.e.,
the true risk is not likely to be higher than the estimate, but it  could be
lower.
                                     101

-------
    For several reasons, the unit risk estimate based on animal bioassays is
only an approximate indication of the absolute risk in populations exposed to
known carcinogen concentrations.  First, there are important species
differences in uptake, metabolism, and organ distribution of carcinogens, as
well as species differences in target site susceptibility, immunological
responses, hormone function, dietary factors, and disease.  Second, the
concept of equivalent doses for humans compared to animals on a mg/surface
area basis is virtually without experimental verification regarding
carcinogenic response.  Finally, human populations are variable with respect
to genetic constitution and diet, living environment, activity patterns,  and
other cultural factors.
    The unit risk estimate can give a rough indication of the relative potency
of a given agent compared with other carcinogens.  The comparative potency of
different agents is more reliable when the comparison is based on studies in
the same test species, strain, and sex, and by the same route of exposure,
preferably by inhalation.
    The quantitative aspect of the carcinogen risk assessment is included here
because it may be of use in the regulatory decision-making process, e.g., in
setting regulatory priorities, evaluating the adequacy of technology-based
controls, etc.  However, it should be recognized that the estimation of cancer
risks to humans at low levels of exposure is uncertain.  At best, the linear
extrapolation model used here provides a rough, but plausible estimate of the
upper-limit of risk; i.e., it is not likely that the true risk would be much
more than the estimated risk, but it could very well  be considerably lower.
The risk estimates presented in subsequent sections should not be regarded as
an accurate representation of the true cancer risks even when the exposures
are accurately defined.  The estimates presented may be factored into
                                     102

-------
regulatory decisions to the extent that the concept of upper  risk  limits  is
found to be useful.
    The mathematical formulation chosen to describe the linear nonthreshold
dose-response relationship at low doses is the linearized  multistage  model.
This model employs enough arbitrary constants to be able to fit almost  any
monotonically increasing dose-response data, and it incorporates a procedure
for estimating the largest possible linear slope (in the 95%  confidence limit
sense) at low extrapolated doses that is consistent with the  data  at  all  dose
levels of the experiment.
    In addition to the multistage model currently used by  the CAG for low-dose
extrapolation (a detailed description of the procedure is  given in Appendix
B), three more models, the probit, Weibull, and one-hit, are  employed for the
purpose of comparison.  These models cover almost the entire  spectrum of risk
estimates that could be generated from the existing mathematical extrapolation
models.  These models are generally statistical in character  and are  not
derived from the biological arguments, except for the multistage model  which
has been used to support the somatic mutation hypothesis of carcinogenesis
(Armitage and Doll  1954, Whittemore 1978, Whittemore and Keller 1978).
    The main differences among these models is the rate at which the  response
function P(d) approaches zero or P(0) as dose d decreases.  For instance, the
probit model would usually predict a smaller risk at low-doses than the
multistage model because of the difference of the decreasing  rate  in  the
low-dose region.  However, it should be noted that one could  always
artificially make the multistage model have the same (or even greater)  rate of
decrease as the probit model, by making some dose transformation and/or by
assuming that some of the parameters in the multistage model  are zero.   This,
of course, is not reasonable without knowing, a priori, what  the carcinogenic
process for the agent is.
                                     103

-------
    Although the multistage model  appears to be the most reasonable or at
least the most general model  to use, the point estimate generated from this
model is of limited value because  the issue remains as to what the shape of
the dose-response curve is beyond  the experimental  exposure levels.
Furthermore, the point estimates at low doses extrapolated beyond the
experimental dose could be extremely unstable and could differ drastically,
depending on where the lowest experimental  dose is.  Since the upper-bound
estimates at low doses from the multistage  model  are relatively more stable
than the point estimates, we suggest that the upper-bound estimate of the risk
(or the lower-bound estimates of the dose)  be used  in evaluating the
carcinogenic potency of a suspect  carcinogen.  The  upper-bound estimate can be
taken as a plausible estimate if the true dose-response curve is actually
linear at low doses.  The upper-bound estimate means that the risks are not
likely to be higher but could be lower if the compound has a concave upward
dose-response curve or a threshold at low doses.   Another reason one can, at
best, obtain an upper-bound estimate of the risk  when animal  data are used is
that the estimated risk is only a  conditional probability under the assumption
that an animal carcinogen is also  a human carcinogen.  Therefore, in reality,
the actual risk could range from a value near zero  to an upper-bound estimate.

PROCEDURES FOR DETERMINING THE CARCINOGENIC POTENCY
Description of the Low-Dose Animal Extrapolation  Model
    Let P(d) represent the lifetime risk (probability) of cancer at dose d.
The multistage model has the form

          P(d) = 1 - exp [-(q0 + qxd + q^2 + ...

where
                           q  >.0, i = 0, 1, 2	k

-------
 Equivalently,

          Pt(d) = 1 - exp

 where
                           Pt(d)   P(d) - P(o)
                            tV   - 1 - P(o)
 is the extra risk over background rate at dose d or the effect of treatment.
    The point estimate of the coefficents q., i = 0, 1, 2, ..., k, and
 consequently the extra risk function Pt(d) at any given dose d, is
 calculated by maximizing the likelihood function of the data.
    The point estimate and the 95% upper confidence limit of the extra risk
 Pt(d) are calculated by using the computer program GLOBAL79 developed by
 Crump and Watson (1979).  At low doses, upper 95% confidence limits on the
 extra risk and lower 95% confidence limits on the dose producing a given risk
 are determined from a 95% upper confidence limit, q*, on parameter
 q...  Whenever q1 > 0, at low doses the extra risk Pt(d) has
 approximately the form Pt(d) = q.^ x d.  Therefore, q^ x d is a 95%
 upper confidence limit on the extra risk, and R/q* is a 95% lower
 confidence limit on the dose producing an extra risk of R.  Let LQ be the
maximum value of the log-likelihood function.  The upper-limit q.  is
 calculated by increasing q-  to a value q* such that when the
 log-likelihood is remaximized subject to this fixed value q* for the
 linear coefficient, the resulting maximum value of the log-likelihood Lj
 satifies the equation
                           2 (L0 - L!) = 2.70554

where 2.70554 is the cumulative 90% point of the chi-square distribution  with

                                     105

-------
one degree of freedom, which corresponds to a 95% upper-limit (one-sided).
This approach of computing the upper confidence limit  for the extra  risk
Pt(d) is an improvement on the Crump et al . (1977) model.  The upper
confidence limit for the extra risk calculated at low  doses  is always  linear.
This is conceptually consistent with the linear nonthreshold concept discussed
earlier.  The slope, q*  is taken as an upper bound of the potency of  the
chemical in inducing cancer at low doses.   (In the section calculating the
risk estimates, P^(d) will be abbreviated  as P).   In fitting the
dose-response model, the number of terms in the polynomial is chosen equal to
(h-1), where h is the number of dose groups in the experiment including the
control  group.
    Whenever the multistage model  does  not  fit the data sufficiently well,
data at the highest dose is deleted and the model is refit to the rest of the
data.  This is continued until an acceptable fit  to the data is obtained.  To
determine whether or not a fit is acceptable, the chi-square statistic
is calculated where N-j  is the number of animals  in  the  in  dose  group,
Xj is the number of animals in the ith  dose  group with  a  tumor response,
P.J is the probability of a response in  the itn dose group estimated  by
fitting the multistage  model  to the data,  and h  is  the  number of remaining
groups.  The fit is determined to be unacceptable whenever  X^ is larger than
the cumulative 99% point of the chi-square distribution with f degrees of
freedom, where f equals the number of dose groups minus the number of non-zero
multistage coefficients.
                                     106

-------
 Selection of Data--
    For some chemicals, several studies in different animal species, strains,
 and sexes, each run at several doses and different routes of exposure, are
 available.  A choice must be made as to which of the data sets from several
 studies to use in the model.   It may also be appropriate to correct for
 metabolism differences between species and absorption factors via different
 routes of administration.  The procedures used in evaluating these data are
 consistent with the approach of making a maximum-likely risk estimate.  They
 are listed as follows.
  .  1.  The tumor incidence data are separated according to organ sites or
 tumor types.  The set of data  (i.e., dose and tumor incidence) used in the
 model is the set where the incidence is statistically significantly higher
 than the control for at least one test dose level and/or where the tumor
 incidence rate shows a statistically significant trend with respect to dose
 level.  The data set that gives the highest estimate of the lifetime
 carcinogenic risk, q*, is selected in most cases.  However, efforts are
 made to exclude data sets that produce spuriously high risk estimates because
 of a small  number of animals.  That is, if two sets of data show a similar
 dose-response relationship, and one has a very small  sample size, the set of
 data that has the larger sample size is selected for calculating the
 carcinogenic potency.
    2.  If there are two or more data sets of comparable size that are
 identical  with respect to species, strain, sex, and tumor sites, the geometric
mean of q*  estimated from each of these data sets, is used for risk
 assessment.   The geometric mean of numbers Aj, A2, ..., A,,, is defined as

                            x A2 x ...  x
                                     107

-------
    3.  If two or more significant tumor sites are observed in the same study,
and if the data are available, the number of animals with at least one of the
specific tumor sites under consideration is used as incidence data in the
model .

Calculation of Human Equivalent Dosages from Animal Data —
    Following the suggestion of Mantel  and Schneiderman (1975), we assume that
mg/surface area/day is an equivalent dose between species.  Since, to a close
approximation, the surface area is proportional  to the 2/3rds power of the
weight as would be the case for a perfect sphere, the exposure in mg/day per
2/3rds power of the weight is also considered to be equivalent exposure.  In
an animal  experiment this equivalent dose is computed in the following manner.
Let
    Le = duration of experiment
    le = duration of exposure
    m  = average dose per day in mg during administration of the agent
         (i .e. , during le), and
    W  = average weight of the experimental  animal
Then, the lifetime average exposure is
                                        ,
                                 Le x W2/3
    Inhalation—When exposure is via inhalation, the calculation  of dose can
be considered for two cases where 1) the carcinogenic agent  is  either a
completely water-soluble gas or an aerosol, and is absorbed  proportionally  to
the amount of air breathed in, and 2) where the carcinogen  is a poorly
water-soluble gas which reaches an equilibrium between the  air  breathed  and
the body compartments.  After equilibrium is reached, the rate  of absorption
of these agents is expected to be proportional to the metabolic rate, which in
turn is proportional to the rate of oxygen consumption, which in  turn is a
function of surface area.
                                     108

-------
    Case I—Agents that are in the form of participate  matter  or  virtually
completely absorbed gases, such as sulfur dioxide,  can  reasonably be expected
to be absorbed proportional to the breathing rate.   In  this  case  the exposure
in mg/day maybe expressed as
                                 m = I  x v x r

where I = inhalation rate per day in m3, v = mg/m3  of the  agent  in  air,
and r = the absorption fraction.
    The inhalation rates, I, for various species can be calculated  from the
observations (FASEB 1974) that 25 g mice breathe 34.5 liters/day  and 113  g
rats breathe 105 liters/day.  For mice  and rats of  other weights, W (in
kilograms), the surface area proportionality can be used to  find  breathing
rates in m3/day as follows:
                  For mice, I = 0.0345  (W/0.025)2/3 m3/day
                  For rats, I = 0.105 (W/0.113)2/3  m3/day

    For humans, the values of 20 m3/day is adopted  as a standard  breathing
rate.  The equivalent exposure in mg/W2/3 for these agents can be derived
from the air intake data in a way analogous to the  food intake data.   The
empirical  factors for the air intake per kg per day, i  = I/W,  based upon  the
previous stated relationships, are tabulated as follows:

                         Species       W        i = I/W
                          Man        70         0.29
                          Rats       0.35      0.64
                          Mice       0.03      1.3
                                     109

-------
Therefore, for participates or completely absorbed gases, the equivalent
exposure in mg/W^/3 is
                     .     m     Ivr    iWvr
                    d =
In the absence of experimental information or a sound theoretical  argument to
the contrary, the fraction absorbed, r, is assumed to be the same  for all
species.

    Case 2--The dose in mg/day of partially soluble vapors  is proportional  to
the 02 consumption, which in turn is proportional  to W2/3 anc| js aiso
proportional to the solubility of the gas in body  fluids, which can  be
expressed as an absorption coefficient, r, for the gas.   Therefore,  expressing
the 62 consumption as 02 = kW2/3, where k is a constant  independent  of
species, it follows that
                                m
= k W2/3vr
or
                             d = -^577 = kvr
                                 w2/3

As with Case 1, in the absence of experimental  information  or a  sound
theoretical  argument to the contrary, the absorption  fraction, r,  is assumed
to be the same for all species.  Therefore, for these substances a certain
concentration in ppm or ug/m3 in experimental  animals is  equivalent to  the
same concentration in humans.  This is supported by the observation that  the
minimum alveolar concentration necessary to produce a given "stage" of
anesthesia is similar in man and animals (Dripps et al. 1977).  When the
animals are exposed via the oral route and human exposure is  via inhalation or
                                     110

-------
vice-versa, the assumption is made, unless there is pharmacokinetic evidence
to the contrary, that absorption is equal  by either exposure route.

Calculation of the Unit Risk from Animal Studies—
    The 95% upper-limit risk associated with d mg/kg2/3/day is obtained from
GLOBAL79 and, for most cases of interest to risk assessment, can be adequately
approximated by P(d) = 1 - exp-(q*d).  A "unit risk" in units X is simply
the risk corresponding to an exposure of X = 1.  To estimate this value we
simply find the number of mg/kg2/3/day corresponding to one unit of X and
substitute this value into the above relationship.  Thus, for example,  if X is
in units of ug/m3 in the air, we have that for case 1, d = 0.29 x 701/3 x
10~3 mg/kg2/3/day, and for case 2, d = 1,  when ug/m3 is the unit used to
compute parameters in animal experiments.
    If exposures are given in terms of ppm in air, we may simply use the fact
that
                1 ppm = 1.2 x molecular weight (gas) mg/m3
                              molecular weight (air)

Note, an equivalent method of calculating  unit risk would be to use mg/kg for
the animal  exposures and then increase the jth polynomial coefficient by an
amount
                    (Wn/wa)J/3     j = 1,  2, .... k
and use mg/kg equivalents for the unit risk values.
                                      11

-------
Model for Estimation of Unit Risk Based on Human Data
    If human epidemiologic studies and sufficiently valid exposure information
are available for the compound, they are always used in some way.   If they
show a carcinogenic effect, the data are analyzed to give an estimate of the
linear dependence of cancer rates on lifetime average dose.   If they show no
carcinogenic effect when positive animal  evidence is available, then it  is
assumed that a risk does exist, but it is smaller than could have  been
observed in the epidemiologic study, and an upper-limit to the cancer
incidence is calculated assuming hypothetically that the true incidence  is
just below the level of detection in the cohort studied, which is  determined
largely by the cohort size.  Whenever possible, human data are used in
preference to animal bioassay data.
    Very little information exists that can be utilized to extrapolate from
high exposure occupational studies to low environmental  levels. However,  if a
number of simplifying assumptions are made, it is possible to construct  a
crude dose-response model  whose parameters can be estimated  using  vital
statistics, epidemiologic studies, and estimates of worker exposures.
    In human studies, the response is measured in terms of the relative  risk
of the exposed cohort of individuals compared to the control  group.  The
mathematical model  employed assumes that for low exposures the lifetime
probability of death from lung cancer (or any cancer), Pg, may be
represented by the linear equation

                                P0 = A + BHx

where A is the lifetime probability in the absence of the agent, and x is  the
average lifetime exposure to environmental levels in some units, say ppm.  The
factor BH is the increased probability of cancer associated  with each unit
                                     112

-------
increase of r, the agent in air.
    If we make the assumption that R, the relative risk of cancer for exposed
workers, compared to the general population, is independent of the length or
age of exposure but depends only upon the average lifetime exposure,  it
follows that
                       R - JL - A + BH (XT  + x?)
                         - P0 - A + BH Xl
or
                         RP0 = A + BH (Xl + x2)

where xj = lifetime average daily exposure to the agent for the general
population, X2 = lifetime average daily exposure to the agent in the
occupational setting, and PQ = lifetime probability of dying of cancer with
no or negligible exposure.
Substituting P0 = A + BH Xj and rearranging gives

                           BH = P0 (R - l)/x2
To use this model, estimates of R and X2 must be obtained from the
epidemiologic studies.  The value PQ is derived from the age-cause-specific
death rates for combined males found in 1973 U.S. Vital Statistics tables
using the life table methodology.

CADMIUM RISK ESTIMATES
Unit Risk Estimate Based on an Animal Study
    The bioassay by Takenaka et al. (1982) using male Wistar rats and cadmium
chloride aerosol was chosen for estimating the quantitative unit risk of
cadmium.  This was the only positive animal  inhalation study with cadmium
                                     113

-------
and/or cadmium compounds, with a dose-response trend of primary lung
carcinomas to animals continuously exposed to CdCl2 aerosols  for 18 months.
The primary lung carcinomas were histologically differentiated as
adenocarcinomas, epidermoid carcinomas,  combined epidermoid and
adenocarcinomas, and mucoepidermoid carcinomas, but were combined for this
unit risk analysis.  The incidences of total  primary lung carcinomas was  15%
(6/39), 53% (20/38), and 71% (25/35) for the  low (12.5  ug/m3), medium (25
ug/m3), and high (50 ug/m3) exposure groups,  respectively.  No tumors were
found among 38 controls.
    In calculating an upper-limit unit risk estimate for humans, dose is
calculated on a lifetime continuous basis with 2 years  considered a full
lifetime exposure for rats.  Thus, by multiplying by 0.75, the measured
concentrations of 13.4 ug/m3, 25.7 ug/m3, and 50.8 ug/m3 for  the three
dose groups, the lifetime continuous exposure can be estimated as 10.05
ug/m3, 19.3 ug/m3, and 38.1 ug/m3, respectively.  The corrections for
animal to human weight differences are given  below.
    In transforming from animal  exposure to human equivalence, the method for
treating inhalation of an aerosol [presented  earlier in the section for
calculation of human equivalent  dosages  from  animal data (Case 1)], assumes
aerosols to be absorbed proportional to  the breathing rate.   This breathing
rate is also given there for 113 g rats  as 0.105 m3/day.  For the Wistar
rats used in the Takenaka et al. bioassay the average weights at 18 months
were 424.6 g (for the 13.4 ug/m3 group), 437.6 g (for the 25.7 ug/m3
group), and 424.3 g (for the 50.8 ug/m3  group).  Adjusting for these weights
we use the formula

                      I = 0.105  (W/0.113)2/3  m3/day

-------
where I = the daily inhalation rate of a rat weighing  W kilograms.   For the
three groups the I values are 0.254 m3/day, 0.259 m3/day,  and  0.254
m3/day, respectively.  Combining these with the lifetime continuous  exposure
estimates above, we estimate daily exposure as 2.55 ug/day,  5.00  ug/day, and
9.68 ug/day, respectively.  Equivalently, we can estimate  dose on a  ug/kg/day
basis as 6.0 ug/kg/day, 11.4 ug/kg/day, and 22.8 ug/kg/day.
    Based on the above data, the 95% upper-limit unit  risk for induced cancer
based on cadmium chloride exposure is q* = 6.3 x IQ-^ug/kg/day)"1
based on animals using the linearized multistage model.  When  transformed to
equivalent human dose, the CAG method requires multiplying q*  by  the
weight ratio factor (Wn/Wa)l/3, where W^ = weight of a human,  which  we
assume to be 70 kg.  Thus,
         = 
-------
that a 70 kg human breathes 20 m3 air/day.   Thus,

        q* = 3.4 x lO-^ug/kg/day)-1 x   1    x 20 m3 =
         "                             ToTg"   liy
                                                       9.7xlO-2(ug/m3)-1

for cadmium chloride exposure, and

         q* = 9.7 x 10-2(ug/m3)-l/0.613 =  1.6 x  lO-
 based on inhalation exposure to the cadmium  ion.   Therefore,  the  unit  risk
 from the inhalation of one microgram of elemental  cadmium  per cubic  meter of
 air is approximately

                         R = 1 - exp -(0.16 x 1) =  0.15

     This is an upper-bound estimate of  risk  based  upon  the best direct
 experimental  evidence presently available.   Using  other dose-response models
 to estimate risk (as shown in Appendix  A) can  give considerably lower
 estimates than obtained using the upper-bound  multistage model.   However,
 there is no direct evidence suggesting  that  these  alternative models provide  a
 more rational  basis for estimating risk than the upper-bound  multistage model.
 It must be kept in mind that the alternative models have the  potential for
 seriously underestimating the true risk at low levels of environmental
 exposure to cadmium.

 Unit Risk Estimate Based on a Human Study
 Data Base--
     At the present time the strongest evidence in  humans suggesting  a cadmium-
 induced carcinogenic response is found  in the  Lemen et  al. (1976)  study.  It
 was observed in a cohort of cadmium smelter  workers with more than 2 years of
                                      116

-------
exposure that 12 deaths were due to a malignant neoplasm of the respiratory
system, while only 5.11 would have been expected based upon the age,
calendar-time, and cause-specific mortality rates for the total United States
white male population (relative risk = 2.34).  Assuming that the U.S.  white
male population is a valid control population for the cohort of cadmium
smelter workers, we find that the probability of obtaining 12 or more
respiratory cancer deaths is P = 0.0063 using the exact Poisson Test.
    A number of problems exist in using these data to obtain a quantitative
estimate of human respiratory cancer risk due to cadmium exposure.   Among them
are the following:
    1.  The smoking histories of the cadmium workers are unknown, and  if their
        smoking rates were substantially higher then white U.S. male  averages,
        this fact alone could explain the twofold increase in relative risk.
    2.  The exposure to cadmium is confounded with exposure to arsenic, a
        known respiratory carcinogen.
    3.  Very limited evidence exists concerning the exposure rate and  the
        duration of exposure for the members of the cohort.
    4.  No exposure estimates exist for individuals.
    5.  To obtain an estimate of risk, a mathematical model must be assumed
        that cannot be evaluated for goodness of fit in any reasonable manner
        using the available data.

    In spite of these considerable shortcomings, it was still felt  that a risk
estimate based on the limited and potentially biased data base could be of use
for the following reasons:
    1.  The observed human respiratory cancer response corresponded to the
        animal response in regard to site, which increases the likelihood that
        the response is real.
    2.  All the factors that are potentially biasing would work to  increase
        the apparent cadmium-induced cancer risk.  Thus, such a risk estimate
        should be considered an upper-bound estimate.  If this upper-bound
        estimate is lower than the one obtained in the animal experiment, it
        should be used in preference to the animal  estimate.
                                     117

-------
    Another approach for obtaining a risk estimate would be to use the
information on prostate cancer.   In the cohort studied by Lemen et al. (1976),
after a lapse of 20 years from initial  exposures,  4 cases were observed  versus
0.88 expected (SMR = 4.55), which is a  statistically significant finding
(P < 0.01).  Prostate cancer was also marginally significant in the Holden
(1980) study, but exposure information  for individuals in his  cohort was not
well-enough defined to permit a  risk estimate.  Use of the prostate cancer
data avoids the difficulties in  numbers 1 and 2 relating to respiratory  system
cancers, but the other difficulties remain.

Risk Model--
    We make the following assumption to obtain a simple risk model.  It  is
assumed that an average daily lifetime  exposure increases the  relative risk
over the entire lifespan by an amount that is linearly related to the
exposure.  This model would tend to overestimate a risk unless early exposures
in one's lifetime are much more  critical  than exposure during  the working
lifetime.  Under this model, the unit risk estimate has the following simple
form:
                                p = PQ  (R-D
                                        X

where P0 is the U.S. lifetime respiratory or  prostate cancer risk in the
absence of cadmium exposure, R is the relative risk for cadmium smelter
workers, and X is the lifetime average  exposure received by the cadmium
workers.
    For respiratory cancer the value for P0 is 0.036 based upon 1973 U.S.
Vital Statistics data.  The term R, the relative risk, is 2.34 with a 95%
confidence interval of 1.35 to 4.10 based upon the exact Poisson distribution.
                                     118

-------
For prostate cancer the value of P0 is 0.0187 using  the  same data source and
the relative risk is 4/0.88 = 4.55.
    To estimate the average lifetime exposure on  a continuous  basis, a number
of factors need to be estimated.  They are:

    The average age, t, of the cohort at the end  of  the  observation period.
    The average duration, d, of exposure for the  cohort  in  years.
    The average exposure rate, e, on the job.
    The fraction, f, of time per year exposed on  the job.

    Given these factors, an estimate of the  average  exposure rate over the
cohort's lifetime is

                                  X = def/t

    Unfortunately, information is very limited concerning  any  of the above
factors.  The approach to be taken is to use the  limited information to make
an educated guess as to what is the best estimate of each  quantity.   In
addition, an upper- and lower-bound is given for  each quantity, which most
likely brackets the true value.  We shall  discuss each of  the  following terms
in order.

    Average age, t, at the end of the observation period—The  only hints on
the age of the cohort is that the people were employed for  at  least 2 years
during the period 1940 to 1970 and were followed  up  to the  start of 1974.
    As an upper-bound it is assumed that the average age at the end of the
observation period is 70.  The lower-bound is set at 40  and the best guess is
taken be the midpoint, or 55.
                                     119

-------
    Average duration, d, of exposure—We know that each individual  was exposed

for at least 2 years and that the four individuals that died of prostatic

cancer were exposed for 4, 13, 17, and 18 years.   If we make the potentially

biased assumption that these four observations are a random sample  from the

total  smelter worker population and obtain a 95%  confidence interval  around

the mean of the sample, we have that best estimate is 13 with bounds  from 3 to

23.



    Average exposure rate, e, on the job--Information concerning exposure

rates  during the appropriate tie frame that we have at our  disposal is taken

from the Lemen et al. (1976) study.

    In 1947, a previously cited study by Princi reported an average air
    concentration of cadmium fumes that ranged from 0.04 to 6.59 mg/m3 and
    of cadmium dust at 17.23 mg/m3, with one man  exposed 2-3 hr/day to an
    average cadmium dust concentration of 31.3 mg/m3.  Most operations,
    however, had concentrations lower than 1.5 mg/m3.

    As a low estimate we take the geometric mean  of the range of the  cadmium

fumes, (0.04 x 6.59)1/2 = 0.5 mg/m3.  As a best estimate we use  the 1.5

mg/m3  and as a upper-bound, 6 mg/m3.



    Fraction, f, of time per year exposed on the  job--We will  assume  that as a

low, "best," and high estimate the number of hours  per day  exposed on  the job

is 2,  4, and 8 hours, respectively, and the number of days  exposed per year  as

120, 180, and 240 days, respectively.



Estimation of Unit Risk—

    The values assumed for each parameter used in the estimation  of the unit

risk based on respiratory cancer data is shown in Table 16.   For  prostate

cancer the "best" estimate of risk can be obtained,  using the  equation
                                     120

-------
  TABLE 16.  "BEST" ESTIMATE AND BOUNDS ON TERMS USED TO ESTIMATE UNIT RISK
            FROM RESPIRATORY CANCER DATA IN THE LEMEN et al.  STUDY
Factor
t = age at end of
observation period
d = years duration
exposure
e = exposure rate
on job (mg/m3)
h = hours in exposure
per day
n = days worked
per year
f = hn*(24 x 365)
fraction of time per
years exposed
x = 1,000 def/t
average lifetime
exposure x ug/m3
R = relative risk
Value
Maximizing
Risk
70
3
0.5
2
120
0.0274
0.587
4.10
Value Giving
"Best" Estimate
of Risk
55
13
1.5
4
180
0.0822
29.1
2.34
Value
Minimizing
Risk
40
23
6
8
240
0.2192
756
1.35
                       Unit Risk Based on Above Values
P = 0.036(R-1)/I =
1.9 x 10-1
1.6 x 10
                                                     -3
1.7 x 10-5
                                     121

-------
given previously for estimating the unit risk, by multiplying the risk for
respiratory cancer by the factor

    (Rp - l)P0,p/(Rr - l)P0,r = (4-55 - l)0.0187/(2.34 - 1)0.036

where the subscripts p and r refer to prostate and respiratory cancer data,
respectively.  Therefore, the "best estimate of the unit risk based on
prostate cancer data is 1.6 x 10'3 x 1.38 = 2.2 x 10~3.  The effect of
compounding, especially multiplying together, a series of assumptions that
consistently overestimates or underestimates the true values of parameters
used to estimate risk leads to very different estimates.  This is true even
though the mathematical model  itself, a major source of uncertainty,  remains
the same.  However, it is highly unlikely that either extreme is close to the
true value.  We take as our estimate the value obtained by compounding the
series of best guesses.  Even though a single term may not be conservative,
the overall result is probably more reasonable than either of the extremes.
    One final correction is needed.  We assume that human exposure was to
cadmium oxide, CdO; thus, the risk to elemental  cadmium is increased  by the
ratio
                       (CdO/Cd) = (128.4/112.4) = 1.14

with a corresponding unit risk estimate of

                P = 1.65 x 10~3 x 1.14 = 1.88 x 10~3 from respiratory data
and
                P = 2.2 x 20-3 x 1.14 = 2.51 x 10~3 from prostate data.

    This estimate is two orders of magnitude lower than the estimate  based on
the rat inhalation sutdy, which was 0.156.  Some of this difference might be
explained as a generalized difference in the biological activity between
                                     122

-------
      in rats and Cd dust and fumes in humans.   In any event,  our  final

estimate is based on the human study, which is  used for the relative  potency

calculation.



RELATIVE POTENCY

    One of the uses of unit risk is to compare  the potency of  carcinogens.   In

this section the potency of cadmium is compared with that of other chemicals

that the CAG has evaluated as suspect carcinogens.  To estimate the relative

potency on a per mole basis, the unit risk slope factor is multiplied by the

molecular weight and the resulting number expressed in terms of

(mMol/kg/day)-l.  This is called the relative potency index.

    Figure 2 is a histogram representing the frequency distribution of potency

indicies of 53 chemicals evaluated by the CAG as suspect carcinogens.  The

actual  data summarized by the histogram are presented in Table 17.  When human

data were available for a compound, they have been used to calculate the

index.   When no human data are available, animal oral studies  and  animal

inhalation studies have been used in that order.  Animal oral  studies are

selected over animal inhalation studies because most of the chemicals have

animal  oral studies; this furnishes a more consistent basis for potency

comparisons.

    The potency index for cadmium based on the  Lemen et al . (1976) study of

cadmium smelter workers is 7.4 x 10+2 (mMol /kg/day)-1.  This is derived as

follows:  the slope estimate from the human study, 1.88 x 10~3

(ug/m3)~l, is first converted to units of (mg/kg/day)~^, assuming  a

breathing rate of 20 m3 of air per day and a 70 kg person.


     1.88 x 10-3(ug/m3)-1 x l daV x   1 "9  x 70 kg
                            20 m3   10"3 mg
                                      23

-------
                                    4th
                                 quartile
   3rd
quartile
   2nd
quartile
                                             1st
                                          quartile
                                        1x10
                                            +i
                           4x10
                  2x10
                            CM
                                                  CM
                                                       °n
               -2
i
0
    2        4

Log of Potency Index
      i
      6
    8
Figure 2.  Histogram representing the frequency distribution of the potency
           indices of 53 suspect carcinogens evaluated by the Carcinogen
           Assessment Group.

-------
TABLE 17.  RELATIVE CARCINOGENIC POTENCIES AMONG 53 CHEMICALS EVALUATE!!
BY THE CARCINOGEN ASSESSMENT GROUP AS SUSPECT HUMAN CARCINOGENS1'2'3
Compounds
Acrylonitrile
Aflatoxin B^
Aldrin
Ally! Chloride
Arsenic
B[a]P
Benzene
Benzidine
Beryl 1 ium
Cadmium
Carbon Tetrachloride
Chlordane
Chlorinated Ethanes
1,2-dichl oroethane
hexachl oroethane
1 ,1 ,2,2-tetrachl oroethane
1,1,1-trichl oroethane
1,1,2-trichl oroethane
Chloroform
Chromium
DDT
Dichlorobenzidine
1 ,1-dichloroethylene
Dieldrin
Slope
(mg/kg/day)'1
0.24(W)
2924
11.4
1.19x10-2
15(H)
11.5
5.2xlO-2(W)
234(W)
4.86
6.65(W)
1. SOxlO-1
1.61
6.90x10-2
1.42x10-2
0.20
1.6xlO-3
5.73x10-2
7x10-2
41
8.42
1.69
1.47x10-1(1)
30.4
Molecular
Weight
53.1
312.3
369.4
76.5
149.8
252.3
78
184.2
9
112.4
153.8
409.8
98.9
236.7
167.9
133.4
133.4
119.4
104
354.5
253.1
97
380.9
Potency
Index
1X10*1
9xlO+5
4xlO+3
9x10-1
2xlO+3
3xlO+3
4x10°
4xlO+4
4x10+1
7xlO+2
2xlO+1
7x10+2
7x10°
3x10°
3x10+1
2x10"!
8x10°
8x10°
4xlO+3
3xlO+3
4x10+2
1x10+1
1x10+4
Order of
Magnitude
(1 og^Q
Index)
+1
+6
+4
0
+ 3
+3
+ 1
+ 5
+ 2
+ 3
+ 1
+ 3
+1
0
+1
-1
+1
+1
+4
+3
+3
+ 1
+4
                                  125
                                          (continued  on the  following  page)

-------
TABLE 17.  (continued)
Compounds
Dinitrotoluene
Diphenylhydrazine
Epichlorohydrin
Bis(2-chloroethyl )ether
Bis(chloromethyl )ether
Ethylene Dibromide (EDB)
Ethylene Oxide
Formaldehyde
Heptachlor
Hexachlorobenzene
Hexachlorobutadiene
Hexachl orocycl ohexane
technical grade
alpha isomer
beta isomer
gamma isomer
Nickel
Nitros amines
Dimethyl nitrosamine
Diethylnitrosamine
Dibutylnitrosamine
N-nitrosopyrrol idine
N-nitroso-N-ethylurea
N-nitroso-N-methylurea
N-nitroso-diphenylamine
PCBs
Slope
(mg/kg/day)-1
0.31
0.77
9.9x10-3
1.14
9300(1)
8.51
0.63(1)
2. 14xlO-2(I)
3.37
1.67
7.75X10-2
4.75
11.12
1.84
1.33
1.15(W)

25.9(not by q*)
43.5(not by q})
5.43
2.13
32.9
302.6
4.92xlO-3
4.34
Molecular
Weight
182
180
92.5
143
115
187.9
44.0
30
373.3
284.4
261
290.9
290.9
290.9
290.9
58.7

74.1
102.1
158.2
100.2
117.1
103.1
198
324
Potency
Index
6x10+1
1x10+2
9X10-1
2xlO+2
1x10+6
2xlO+3
3xlO+1
6X10-1
lxlO+3
5x10+2
2xlO+1
1x10+3
3x10+3
. f\
5x10+2
4x10+2
7xlO+1

2xlO+3
4x10+3
9x10+2
2x10+2
4xlO+3
3x10+4
JLAiU
lxlO+3
	 r-l 	 f 1 q —
Order of
Magnitude
(Io9l0
Index)
+2
+2
0
+2
+6
+3
+1
0
+3
+3
+1
+ 3
+3
+ 3
+3
+2

+ 3
+4
+ 3
+2
+4
+4
0
+ 3
— T— 	 i
         126

-------
                           TABLE 17.   (continued)
Compounds
Phenols
2,4,6-trichlorophenol
Tetrachlorodioxin
Tetrachl oroethylene
Toxaphene
Trichloroethylene
Vinyl Chloride
Remarks:
Slope
(mg/kg/day)'1
1.99xlO'2
4.25xl05
5.31x10-2
1.13
1.26xlO-2
1.75xlO-2(I)

1. Animal slopes are 95% upper-limit
Molecular
Weight
197.4
322
165.8
414
131.4
62.5

slopes based
Potency
Index
4x10°
lxlO+8
9x10°
5xlO+2
2xlOn
1x10°

on the lineari
Order of
Magnitude
Oogm
Index)
+ 1
+8
+1
+3
0
0

zed multistage
2.


3.
They are calculated based on animal oral  studies, except for those indicated by
I (animal inhalation), W (human occupational exposure), and H (human drinking water
exposure).  Human slopes are point estimates based on the linear non-threshold
model.

The potency index is a rounded-off slope in (mMol/kg/day)"l and is calculated by
multiplying the slopes in (mg/kg/day)"^ by the molecular weight of the compound.

Not all of the carcinogenic potencies presented in this table represent the same
degree of certainty.  All are subject to change as new evidence becomes available.
                                     127

-------
    Multiplying by the molecular weight of 112.4 gives a potency index  of
7.4 x 10+2.  Rounding off to the nearest order of magnitude gives a  value of
10+3, which is the scale presented on the horizontal  axis of Figure  2.   The
index of 7.4 x 10+^ lies in the second quartile of the 53 suspect
carcinogens.
    Ranking of the relative potency indicies is subject to the  uncertainty of
comparing estimates of potency of different chemicals based on  different
routes of exposure to different species using studies of different quality.
Furthermore, all  the indicies are based on estimates  of low-dose risk using
the linearized multistage extrapolation model fitted  to the data at  relatively
high doses.  Thus, relative potencies could be different at high exposures,
where non-linearities in the dose-response curves could exist.
                                     128

-------
                                   APPENDIX A
             COMPARISON OF RESULTS BY VARIOUS EXTRAPOLATION MODELS

     The estimate of unit risk from animals presented in the body of this
document is calculated by the use of the linearized multistage model.
The reasons for its use have been detailed therein.  Essentially, it is part
of a methodology that estimates a conservative linear slope at low extrapolation
doses and is consistent with the data at all dose levels of the experiment.
It is a nonthreshold model holding that the upper-limit of risk predicted by
a linear extrapolation to low levels of the dose-response relationship is the
most plausible upper-limit for the risk.
    Other models have also been used for risk extrapolation.  Three
nonthreshold models are presented here:  the one-hit, the log-Probit, and the
Weibull.  The one-hit model  is characterized by a continuous downward curvature
but is linear at low doses.   It can be considered the linear form or first
stage of the multistage model because of its functional form.  Because of this
and its downward curvature,  it will always yield estimates of low level risk
which are at least as large  as those of the multistage model.  Further,
whenever the data can be fit adequately by the one-hit model, estimates from
the two procedures will be comparable.
    The other two models presented below are the log-Probit and the Weibull.
They are often used to fit toxicological data in the observable range, because
of the general "S" curvature.  The low-dose upward curvatures of these two
models usually yield lower low-dose risk estimates than those of the one-hit
or multistage models.
    The log-Probit model  was originally used in problems of biological assay
such as the assessment of potency of toxicants and drugs and has usually
                                      129

-------
been used to estimate such values as percentile lethal dose or percentile
effective dose.  Its development was strictly empirical, i.e., it was
observed that several log dose-response relationships followed the
cumulative normal probability distribution function,   .   In fitting the cancer
bioassay data, assuming on independent background, this becomes

             P(D;a,b,c) = c + (1-c) $ (a+blogjo D)   a,b > 0 _< c < 1

where P is the proportion responding at dose D, c is an estimate of the
background rate, a is an estimate of the standarized mean  of individual
tolerances, and b is an estimate of the log dose-Probit response slope.
     The one-hit model  arises from the theory that a single molecule of a
carcinogen has a probability of transforming a single normal cell
into a cancer cell.  It has the probability distribution function

                       P(D;a,b) = l-exp-(a+bd)   a,b > 0

where a and b are the parameter estimates.  The estimate a represents the
background or zero dose rate, and the parameter estimated  by b represents
the linear component or slope of the dose-response model.   In discussing the
added risk over background, incorporation of Abbott's correction leads to

                          P(D;b) = l-exp-(bd)   b > 0

Finally, a model from the theory of carcinogenesis arises  from the multihit
model applied to multiple target cells.  This model has been termed here the
Wei bull model.   It is of the form

                        P(D;b,k) = l-exp-(bdk)   b,k > 0
                                      130

-------
For the power of dose only, the restriction k > 0 has been placed on  this  model.
When k > 1, this model yields low-dose estimates of risks usually significantly
lower than either the multistage or one-hit models, which are linear  at  low
doses.  All three of these models usually project risk estimates significantly
higher at low exposure levels than those from the log-Probit.
    The estimates of added risk for low doses for these models are given in
Table 18 for the cadmium chloride rat inhalation studies by Takenaka  et  al.
Both maximum likelihood estimates and 95% upper confidence limits are presented.
The results show that the maximum likelihood estimates of risk for the
log-Probit model are all less than those for the other models and this
difference increases sharply at low doses.  The one-hit model yields  maximum
likelihood estimates slightly higher than from the multistage model,  with  the
Wei bull  somewhat lower.
                                      131

-------
              TABLE 18.   ESTIMATES  OF  LOW-DOSE  RISK  TO  HUMANS EXPOSED TO CADMIUM CHLORIDE BASED ON MALE WISTAR
                     RATS FROM THE  TAKENAKA  ET  AL. INHALATION STUDY  DERIVED FROM FOUR DIFFERENT MODELS
                  Maximum Likelihood  Estimates  of                     95% Upper Confidence Limit of
                          Additional  Risks                                   Additional Risks

Dose       Multistage  One-Hit      Weibull     Log-Probit       Multistage  One-Hit     Weibull    Log-Probit
ug/m3        Model       Model       Model        Model           Model*     Model        Model        Model


10-4       5.5xlO-6    S.lxlO-6    1.9xlO-7        0           9.7xlQ-6    l.OxlO-5    1.3xlO'6   1.2xlQ-38


10-3       5.5xlO-5    8.1xlO-5    4.1xlO-6        0           9.7xlQ-5    l.OxlO-4    2.6xlO~5   8.9x10-25


lO-2       5.5xlO'4    S.lxlO'4    8.8xlO-5    2.0xlO-15        9.7xlQ-4    l.OxlO-3    S.SxlO'4   4.4X10'14


ID'1       5.5x10-3    S.lxlO-3    1.9x10-3    l.SxlO'7         9.7xlQ-3    LOxlO"2    5.9xlQ-3   1.5X10'6


1          5.5x10-2    7.8x10-2    3.9x10-2    7.0x10-3         9.2xlQ-2    9.5xlQ-2    S.lxlO"2   2.3X10'2


    *  q* = 9.7 x 10-2(ug/m3)-1 for the  multistage model;  P(d)  = 1  -  e -

-------
                                  APPENDIX B



    INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC) CLASSIFICATION FOR



       UEIGHT-OF-EVIDENCE FOR CARCINOGENICITY OF A SUSPECTED CARCINOGEN








ASSESSMENT OF EVIDENCE FOR CARCINOGENICITY FROM STUDIES IN HUMANS



    The degrees of evidence for carcinogenicity from studies in humans are



categorized as:



    1.  Sufficient evidence of carcinogenicity, which indicates that there is



a causal relationship between the agent and human cancer.



    2.  Limited evidence of carcinogenicity, which indicates that a causal



interpretation is credible, but that alternative explanations, such as chance,



bias, or confounding, could not adequately be excluded.



    3.  Inadequate evidence, which indicates that one of three conditions



prevailed:  (a) there were few pertinent data; (b) the available studies,



while showing evidence of association, did not exclude chance, bias, or



confounding; (c) studies were available which do not show evidence of



carcinogenicity.







ASSESSMENT OF EVIDENCE OF CARCINOGENICITY FROM STUDIES IN EXPERIMENTAL ANIMALS



These assessments are classified into four groups:



    1.  Sufficient evidence of carcinogenicity, which indicates that there is



an increased incidence of malignant tumors:  (a) in multiple species or



strains; or (b) in multiple experiments (preferably with different routes of



administration or using different dose levels; or (c) to an unusual  degree



with regard to incidence, site, or type of tumor, or age at onset.  Additional



evidence may be provided by data on dose-response effects,  as well  as



information from short-term tests or on chemical  structure.
                                     133

-------
    2.  Limited evidence of carcinogenicity, which means that the data suggest


a carcinogenic effect but are limited because:   (a) the studies involve a


single species, strain, or experiment; or (b) the experiments are restricted


by inadequate dosage levels, inadequate duration to exposure to the agent,


inadequate period of follow-up, poor survival,  too few animals, or inadequate


reporting, or (c) the neoplasms produced often  occur spontaneously and, in  the


past, have been difficult to classify as malignant by histological  criteria


alone (e.g., lung and liver tumors in mice).


    3.  Inadequate evidence, which indicates that because of major qualitative
                                                         i

or quantitative limitations, the studies cannot be interpreted as showing


either the presence or absence of a carcinogenic effect; or that within the


limits of the tests used, the chemical is not carcinogenic.  The number of


negative studies is small, since, in general, studies that show no effect are


less likely to be published than those suggesting carcinogenicity.


    4.  No data indicates that data were not available to the working  group.





EVALUATION OF CARCINOGENIC RISK TO HUMANS


    At present, no objective criteria exist  to  interpret data from studies  in


experimental animals or from short-term tests directly in terms of human risk.


Thus, in the absence of sufficient evidence  from human studies, evaluation  of


the carcinogenic risk to humans was based on consideration of both the


epidemiological and experimental evidence.  The breadth of the categories of


evidence defined above allows substantial variation within each.  The


decisions reached by the group regarding overall risk incorporated these


differences, even though they could not always  be reflected adequately in the


placement of an exposure into a particular category.
                                     13*

-------
    The chemicals, groups of chemicals, industrial  processes,  or occupational
exposures were thus put into one of three groups:

Group 1
    The chemical, group of chemicals, industrial  process, or occupational
exposure is carcinogenic to humans.  This category  was used only when  there
was sufficient evidence from epidemiological  studies to support a causal
association between the exposure and cancer.

Group 2
    The chemical, group of chemicals, industrial  process, or occupational
exposure is probably carcinogenic to humans.   This  category includes exposures
for which, at one extreme, the evidence of human carcinogenicity is almost
"sufficent," as well as exposures for which,  at the other extreme, it  is
inadequate.  To reflect this range, the category was divided into higher
(Group A) and lower (Group B) degrees of evidence.   Usually, category  2A  was
reserved for exposures for which there was at least limited evidence of
carcinogenicity to humans.  The data from studies in experimental animals
played an important role in assigning studies to category 2, and particularly
those in Group B; thus, the combination of sufficient evidence in animals and
inadequate data in humans usually resulted in a classification of 2B.
    In some cases, the working group considered that the known chemical
properties of a compound and the results from short-term tests allowed its
transfer from Group 3 to 2B or from 2B to 2A.

Group 3
    The chemical, group of chemicals, industrial  process, or occupational
exposure cannot be classified as to its carcinogenicity to humans.
                                     135

-------
                                 REFERENCES
Adamsson, E.  1979.  Long-term sampling of airborne cadmium duct in an
    alkaline battery factory.  Scand. J. Work Environ. Health 5:178.

Amacher, D.E., and S.C. Paillet.  1980.  Induction of
    trifluorothymidine-resistant mutants by metal  ions in L5178Y/TK"1"/"
    cells.  Mutat. Res. 78:279-288.

Ames, 8.N., W.E. Durston, E. Yamasaki, and F.D. Lee.  1973.  Carcinogens are
    mutagens:  A simple test system combining liver homogenates for activation
    and bacteria for dectection.  Proc. Natl. Acad. Sci. (USA) 70:2281-2285.

Ames, B.N., J. McCann, and E. Yamasaki.  1975.  Methods for detecting
    carcinogens and mutagens with the Salmonel1 a/mammalian microsome
    mutagenicity test.  Mutat. Res. 31:347-364.

Avanzi, M.  1950.  Osservazioni sull'attivita citologica di alcuni composti
    chimici.  Caryologia 3:234-248.

Baader, E.W.  1951.  Dtsch. Med. Wochenschr.  484.

Bauchinger, M.E., et al.  1976.  Chromosome aberrations in lymphocytes after
    occupational lead and cadmium exposure.  Mutat. Res. 40:57.

Bui, T.H., et al.  1975.  Chromosome analysis of lymphocytes from cadmium
    workers and Itai-Itai patients.  Environ. Res. 9:187.

Casto, B.  1976.  Letter and mutagenicity data to  Richard Troast, EPA, on
    cadmium chloride and cadmium acetate.

Chiang, C.L.  1961.  Vital Statistics.  Division of Vital Statistics,
    Department of Commerce, Special Report No. 47, p. 275.

Clive, n., G. Hajian, and M.M. Moore.  1981.   Letter to the Editor.  Mutat.
    Res. 89:241-244.

Cox, C.R.  1972.  Regression model and life tables.  J. Roy. Stat. Soc. B
    34:187-220.

Crump, K.S.  1979.  Hose-response problems in carcinogenesis.  Biometrics
    35:157-167.

Crump, K.S., and W.W. Watson.  1979.  GLOBAL  79.  A fortran program to
    extrapolate dichotomous animal carcinogenicity data to low dose.
    Natl. Inst. Environ. Health Science, Contract  No. l-ES-2123.

Crump, K.S., H.A. Guess, and L.L. Deal.  1977.  Confidence intervals and test
    of hypotheses concerning dose-response relations inferred from animal
    carcinogenicity data.  Biometrics 33:437-451.
                                     136

-------
Heaven, L.L., and E.W. Campbell.  1980.  Factors affecting the induction of
    chromosomal aberrations by cadmium in Chinese hamsters cells.  Cytogenet.
    Cell Genet. 26:251-260.

Degraeve, N.  1971.  Modification des effets du methane sulfonate d'ethyl  au
    niveau chromosomique.  I. Lesions metalliques.  Rev. Cytol. Biol. Veget.
    34:233-244.

Oegraeve, N.  1981.  Carcinogenic, teratogenic, and mutagenic effects of
    cadimum.  Mutat. Res. 86:115-135.

Dekundt, Gh., and G.B. Gerber.  1979.  Chromosomal aberrations in bone marrow
    cells of mice given a normal or a calcium-deficient diet supplemented with
    various heavy metals.  Mutat. Res. 68:163-168.

nekundt, Gh., and M. Deminatti.  1978.  Chromosome studies in human
    lymphocytes after in vitro exposure to metal salts.  Toxicology 10:67-75.

Dekundt, Gh., and A. Leonard.  1976.  Cytogenic investigations on leukocytes
    of workers occupationally exposed to cadmium.  Mutat. Res. 38:112.

Dekundt, Gh., et al.  1973.  Chromosome aberrations observed in male workers
    occupationally exposed to lead.  Environ. Physiol. Biochem. 3:132.

Oixon, R.L., et al.  1976.  Methods to assess reproductive effects of
    environmental chemicals: Studies of cadmium and boron administered orally.
    Environ. Health Perspect. 13:59.

Doll, R.  1971.  Weibull distribution of cancer.  Implications for models of
    carcinogenesis.  J. Roy. Stat. Soc. A 13:133-166.

Dripps, R.D., J.E. Eckenhoff, and L.D. Vandam.  1977.  Introduction to
    anethesia, the principles of safe practice.  5th ed.  Philadelphia, PA:
    W.B. Saunders Company, pp. 121-123.

Ell man, G.L.  1959.  Tissue sulfhydryl groups.  Arch. Biochem. Biophys.
    82:70.

Epstein, S., et al.  1972.  Detection of chemical mutagens by the dominant
    lethal assay in the mouse.  Toxicol. Appl. Pharmacol. 23:288.

Federation of American Societies for Experimental Biology (FASEB).  1974.
    Biological Data Books.  2nd ed., Vol. III.  Philip L. Altman and
    Dorothy S. Dittmen, eds.  FASEB, Bethesda, MD.  Library of Congress No.
    72-87738.

Favino, A., et al.  1968.  Testosterone excretion in cadmium chloride induced
    testicular tumors in rats.  Med. Lav. 59:36.

Fraumeni, J.F.  1975.  Persons at high risk of cancer.  Academic Press, Inc.

Furst, A., and D.  Cassetta.  1972.   Failure of zinc to negate cadmium
    carcinogenesis.  Proc. Am. Assoc. Cancer Res. 13:62.
                                     137

-------
Gasiorek, K., and M. Bauchinger.  1981.  Chromosome changes in human
    lymphocytes after separate and combined treatment with divalent salts of
    lead, cadmium and zinc.  Environ. Mutagen. 3:513-518.

Gilliavod, N., and A. Leonard.  1975.  Mutagenicity tests with cadmium in the
    mouse.  Toxicology 5:43.

Gunn, S.A., et al.  1963.  Cadmium-induced interstitial  cell  tumors in rats
    and mice and their prevention by zinc.  J. Natl.  Cancer Inst.  31:745.

Gunn, S.A., et al.  1964.  Effect of zinc on carcinogenesis by cadmium.   Proc.
    Soc. Exp. Biol. Med. 115:653.

Gunn, S.A., et al.  1967.  Specific response of mesenchymal tissue to
    carcinogenesis by cadmium.  Arch. Path. 83:493.

Haddow, A., et al.  1961.  Carcinogenicity of iron preparations and
    metal-carbonate complexes.  Rep. Br.  Emp. Cancer  Campaign. 39:74.

Haddow, A., et al.  1964.  Cadmium neoplasia: Sarcomata  at the site of
    injection of cadmium sulphate in rats and mice.  Br.  J. Cancer 18:667.

Hadley, J.G., A.W. Conklin, and C.L. Sanders.  1980.   Rapid solubilization  and
    translocation of 109CdO following pulmonary deposition.  Toxicol. Appl.
    Pharmacol. 54:156-160.

Heath, J.C.  1962.  Cadmium as a carcinogen.  Nature  (London) 193:592.

Heath, J.C., and M.R. Daniel.   1964.  The production  of  malignant  tumors by
    cadmium in the rat.  Br. J. Cancer 18:124.

Heddle, J.A., and W.R. Bruce.   1977.  Comparison of tests for mutagenicity  or
    carcinogenicity using assays for sperm abnormalities, formation of
    micronuclei, and mutations in Salmonella.  In: H.H.  Hiatt et al., eds.,
    Origins of human cancer.  Cold Spring Harbor N.Y.:  Cold Spring Harbor
    Laboratory, pp. 265-274.

Hedenstedt, A., U. Rannug,  C.  Ramel, and  V. Wachtmeister.  1979.  Mutagenicity
    and metabolism studies  on  12 thiuram  and dithiocarbamate compounds used as
    accelerators in the Swedish rubber industry.  Mutat.  Res. 68:313-325.

Hernberg, S.  1977.  Cadmium.   In:  H.H.  Hiatt, ed.  Origins of human cancer.
    Book A. Incidence of cancer in humans.  Cold Spring  Harbor Conferences  on
    Cell Proliferation. 4:149.

Hirayama, T.  1977.  Changing  patterns of cancer in Japan with special
    reference to the decrease  in stomach  cancer mortality.  In: Hiatt,  H.H.,
    Watson, J.D., Winsten,  J.A., eds.  Origins of human  cancer. Book A:
    Incidence of Cancer in  Humans.  Cold  Spring Harbor,  New York:   Cold
    Springs Harbor Laboratory, pp. 55-75.
                                     138

-------
Hoffman, D.J., and S.K. Niyogi.  1977.  Metal  mutagens and carcinogens affect
    RNA synthesis rates in a distinct manner.   Science 198:513-514.

Holden, H.  1969.  Cadmium toxicology.  Lancet 2:57.

Holden, H.  1980.  Further mortality studies on workers exposed to cadmium
    fumes.  Report on seminar:  London, March  20:  Occupational exposure to
    cadmium.  Cadmium Association, London, England.

Hsie, A.M., J.P. O'Neil, J.R.S. Sabastian, D.B. Couch, et al.  1978.
    Quantitative mammalian cell genetic toxicology.  EPA-600/9-78-027:293-315.

Humperdinck, K.  1968.  Kadmium and lungenkrebs.  Med. Klin. 63:948.

Inoue, Y., and T.K. Watanabe.  1978.  Toxicity and mutagenicity of cadmium and
    furylfuramide in Drosophila melanogaster.   Jpn. J. Genet. 53:183-189.

International  Agency for Research on Cancer (IARC).  1973.  Monographs on the
    evaluation of carcinogenic risk of chemicals to man.  Cadmium and
    inorganic cadmium compounds. 2:74.

International  Agency for Research on Cancer (IARC).  1976.  Monographs on the
    evaluation of carcinogenic risk of chemicals to man.  Cadmium and cadmium
    compounds.  11:3974.

Japanese Public Health Association.  1979.  Epidemiological  research on the
    causes of death in cadmium polluted areas.  Japan Public Health
    Association, Tokyo, Japan (in Japanese).

Kada, T., K. Tutikawa, and Y. Sodaie.  1972.  In vitro and host-mediated assay
    "rec-assay" procedures for screening chemical mutagens and phloxine, a
    mutagenic red dye detected.  Mutat. Res. 16:165-174.

Kahn, H.A.  January 1966.  The Dorn study of smoking and mortality among U.S.
    Veterans:   Report on eight and one-half years of observation.  In:
    Heanzel, W., ed.  Epidemiological approaches to the study of cancer and
    other chronic diseases.  National Cancer Institute Monograph No. 19, U.S.
    Department of Health, Education, and Welfare, Public Health Service,
    National Cancer Institute, pp. 1-125.

Kalinina, L.M., and G.N. Polukhina.  1977.  Mutagenic effects of heavy metal
    salts on Salmonella in activation systems  in vivo and in vitro.   Mutat.
    Res. 46:223-224 (abstract).

Kanematsu N., M. Hara, and T. Kada.  1980.  Rec-assay and mutagenicity studies
    on metal compounds.  Mutat. Res. 77:109-116.

Kazantzis, G.   1963.  Induction of sarcoma in  the rat by cadmium sulphide
    pigment.  Nature 198:1213.

Kazantzis, G., and W.J. Hanbury.  1966.  The induction of sarcoma in the rat
    by cadmium sulphide and cadmium oxide.  Br. J. Cancer 20:190.
                                     139

-------
Kazantzis, G., and B.G. Armstrong.  1982.  A mortality study of cadmium
    workers in England:  A synopsis of the study.  TVC Centenary Institute of
    Occupational  Health, London School of Hygiene and Tropical  Medicine,
    London, England.

Kipling, M.O., and J.A.H. Waterhouse.  1967.  Cadmium and prostatic carcinoma.
    Lancet 1:730.

Kjellstrom, T.  1982.  Mortality and cancer morbidity in people exposed to
    cadmium.  Report  prepared for Health Effects Research Laboratory, U.S.
    Environmental Protection Agency, Research Triangle Park, N.C.  Grant No.
    R806036101.

Kjellstrom, T., et al.   1978.  Cadmium in feces as an estimator of
    daily cadmium intake in Sweden.  Environ. Res. 15:242.

Kjellstrom, T., L. Friberg, and B. Rahnster.  1979.  Mortality  and cancer
    morbidity among cadmium-exposed workers.  Environ. Health Perspect.
    28:199-204.

Knorre, D. von.  1970.   Zur Induktion von Hautsarkomen bei der  Albinoratte ,
    durch Kadmiumchlorid.  Arch. Geschwulstforsch. 36:119.

Knorre, D. von.  1971.   Zur Induktion von Hodenzwischenzelltumoren an der
    Albinoratte durch Kadmiumchlorid.  Arch. Geschwulstforsch.  38:257.

Kolonel, L.N.  1976.   Association of cadmium with renal  cancer.  Cancer
    37:1782-1787.

Kumaraswamy, K.R., and  M.R. Rajasekarasetty.  1977.  Preliminary studies on
    the effects of cadmuim chloride on the meiotic chromosomes.  Curr. Science
    46:475-478.

Lee, I., and R. Dixon.   1973.  Effects of cadmium on spermatogenesis studied
    by velocity sedimentation cell separation serial mating.  J. Pharmacol.
    Exp. Then. 187:641-652.

Lemen, R.A., J.S. Lee., J.K. Wagoner, and H.P. Blejer.,  1976.  Cancer
    mortality among cadmium production workers.  Ann. N.Y. Acad. Sci. 271:273.

Levan, A.  1945.   Cytological reactions induced by inorganic salt solutions.
    Nature (London) 156:751-752.

Levy, L.S., and J. Clack.  1975.  Further studies on the effect of cadmium on
    the prostate gland.  I. Absence of prostatic changes in rats given oral
    cadmium sulphate  for two years.  Ann. Occup. Hyg. 17:205.

Levy, L.S., et al.  1973.  Absence of prostatic changes  in rats exposed to
    cadmium.  Ann. Occup. Hyg. 16:111.

-------
Loser, E.  1980.  A two year oral carcinogenicity study with cadmium on rats.
    Cancer Letters 9:191.

Lucis, O.J., et al.  1972.  Tumorigenesis by cadmium.  Oncology 26:53.

Malcolm, D.  1972.  Potential  carcinogenic effect of cadmium in animals and
    man.  Ann. Occup. Hyg. 15:33.

Mantel, N., and M.A. Schneiderman.  1975.  Estimating "safe" levels, a
    hazardous undertaking.  Cancer Res. 35:1379-1386.

McMichael, A.J., D.A. Andjelkovic, H.A. Tyroler.  1976.  Cancer mortality
    among rubber workers: an epidemiologic study.  Ann. N.Y. Acad. Sci.
    271:124.

Milvy, P., and K. Kay.  1978.   Mutagenicity of 19 major graphic arts and
    printing dyes.  J. Toxicol. Environ. Health 4:31-36.

Mitra, R., and I. Bernstein.  1977.  Nature of the repair process associated
    with the recovery of E_. coli after exposure to Cd^+.  Biochem. Biophys.
    Res. Commun. 74:1450-1455.

Mitra, R., and I. Bernstein.  1978.  Single-strand breakage in DNA of £. coli
    exposed to Cd2+.  J. Bacteriol. 133:75-80.

Nishioka, H.  1975.  Mutagenic activities of metal compounds in bacteria.
    Mutat. Res. 311:185.

Nogawa, K., et al.  1978.  Statistical observations of the dose-response
    relationships of cadmium based on epidemiological studies in the Kakehashi
    River basin.  Environ. Res. 15:185.

Nogawa, K., S. Kawano, and M.  Nishi.  1981.  Mortality study of inhabitants in
    a cadmium-polluted area with special reference to low molecular weight
    proteinuria.  In: Proceedings of the 3rd Cadmium conference, Miami
    Florida.  In press.

Oehlkers, F.  1953.  Chromosome breaks influenced by chemicals.  Heredity
    Suppl. 6:95-105.

O'Riordan, M.L., E.G. Hughes,  and H.J. Evans.  1978.  Chromosomal studies on
    blood lymphocytes of men occupationally exposed to cadmium.  Mutat. Res.
    58:305-311.

Paton, G.R., and A.C. Allison.  1972.  Chromosome damage in human cell culture
    induced by metal salts.  Mutat. Res. 16:332.

Polukhina, G.N., L.M. Kalinina, and L.I. Lukasheva.  1977.  A test system for
    the detection of the mutagenic activity of environmental pollutants.  II.
    Detection of mutagenic effect of heavy metal salts using in vitro assay
    with metabolic activation.  Genetika 13:1492-1494.

Potts, C.L.  1965.  Cadmium proteinuria - The health of battery workers
    exposed to cadmium oxide dust.  Ann. Occup. Hyg.  8:55.

-------
Putrament, A.M., H. Baranowska, A. Ejchart, and W. Jachymczyk.  1977.
    Manganese mutagenesis in yeast.  VI. Mn2+ uptake, mit DNA replication
    and ER induction, comparison with other divalent cations.  Mol.  Gen.
    Genet. 151:69-76.

Ramaiya L.K., and M. [). Pomerantseva.  1977.  Investigation of cadmium
    mutagenic effect on germ cells of male mice.   Genetika 13:59-63.

Ramel, C., and M. Friberg.  1974.  Carcinogenic and genetic effects.   In:
    M. Friberg, et al., eds.  2nd Ed.  Cadmium in the environment.   Cleveland,
    Ohio: CRC Press, p. 133.

Ramel, C., and J. Magnusson.  1979.  Chemical induction of nondisjunction in
    Drosophila.  Environ. Health Perspect. 31:59-66.

Reddy, J., et al.  1973.  Cadmium-induced Leydig  cell tumors of rat  testis:
    Morphologic and cytochemical study.  J. Natl. Cancer Inst. 51:891.

Robbins, S.L., and M. Angell.  1976.   Pathology.   2nd ed.  Philadelphia,  PA.:
    W.B. Saunders Company, p. 555.

Roe, F.J.C.,  C.E. Dukes, K.M. Cameron, R.C.B. Pugh, and B.C.V. Mitchley.
    1964.  Cadmium neoplasia:  Testicular atrophy and Leydig cell  hyperplasia
    and neoplasia in rats and mice following the  subcutaneous injection of
    cadmium salts.  Br. J. Cancer 18:674.

Rohr,  G., and M. Bauchinger.  1976.  Chromosome analysis in cell  cultures of
    the Chinese hamster after application of cadmium sulfate.  Mutat.  Res.
    40:125.

Ruposhev, A., and K. Garina.  1977.  Modification of mutagenic effects of
    ethyleneimine by cadmium in Crepis capillaris.  Genetika 13:32-36.

Sabalina, L.P.  1968.  Industrial hygiene in the  production and use  of cadmium
    stearate.  Hyg. Sanit. (U.S.S.R.) 33(7/9):187-191.

Sanders, C.L., and J.A. Mahaffey.  1982.  Carcinogenicity of single  and
    multiple  intratracheal instillation of cadmium oxide in the rat.   In
    press.

Schmauz, R.,  and P. Cole.  May 1974.   Epidemiology of cancer of the  renal
    pelvis and ureter.  J. Natl. Cancer Inst. 52(5):1431-1434.

Schroeder, H.A., et al.  1964.  Chromium, lead, cadmium, nickel,  and titanium
    in mice:  effect on mortality, tumors, and tissue levels.  J.  Nutr. 83:239.

Schroeder, H.A., et al.  1965.  Chromium, cadmium, and lead in rats:   Effects
    on life span, tumors, and tissue  levels.  J.  Nutr. 86:51.

Shankaranarayanan, K.  1967.  The effects of nitrogen and oxygen  treatments  on
    the frequency of X-ray induced dominant lethals and on the physiology of
    the sperm in Drosophila melanogaster.  Mutat. Res. 4:641-661.

-------
Shigematsu, I., S. Kitamura, and J. Takeuchi.  1981.  A retrospective
    mortality study on cadmium polluted population in Japan.  In:  Proceedings
    of 3rd cadmium conference, Miami, Florida.

Shimada, T., et al.  1976.  Potential mutageriicity of cadmium in mammalian
    cocytes.  Mutat. Res. 40:389.

Shiraishi, Y., and T.H. Yoshida.  1972.  Chromosomal abnormalities in cultured
    leucocyte cells from Itai-Itai disease patients.  Proc. Jap. Acad. 48:248.

Shiraishi, Y., et al.  1972.  Chromosomal aberrations in cultured human
    leucocytes induced by cadmium sulfide.  Proc. Jap. Acad. 48:133.

Shiraishi, Y.  1975.  Cytogenetic studies in 12 patients with Itai-Itai
    disease.  Humanogenetik 27:31-44.

Si rover, M.A., and L.A. Loeb.  1976.  Infidelity of DNA synthesis in vitro:
    Screening for potential metal mutagens or carcinogens.  Science 194:1434.

Sorahan, T.  1981.  A mortality study of nickel-cadmium battery workers.
    Cancer Epidemiology Research Unit, University of Birmingham, Birmingham
    B15 2th, England.

Sorsa, M., and S. Pfeifer.  1973.  Effects of cadmium on development time and
    prepupal puffing pattern of Prosophila melanogaster.  Hereditas
    71:273-277.

Sunderman, F.W., Jr.  1977.  Cadmium.  In:  R.A. Goyer and M.A. Mehlman,
    eds.  Advances in modern toxicology.  Volume 2.  New York:Hemisphere
    Publishing Corp., John Wiley and Sons.

Sunderman, F.W., Jr.  1978.  Carcinogenic effects of metals.  Fed. Proc.
    37:40.

Suter, K.E.  1975.  Studies on the dominant-lethal and fertility effects of
    the heavy metal compounds methylmercuric hydroxide, mercuric chloride, and
    cadmium chloride in male and female in mice.  Mutat. Res. 30:365.

Sutou, S., K. Yamamoto, H. Sendota, K. Tomomatsu, Y. Shimizu, and M. Sugiyama.
    1980a.  Toxicity, fertility, teratogenicity, and dominant lethal tests in
    rats administered cadmium subchronically.  I. Toxicity studies.
    Ecotoxicology and Environmental Safety 4:39-50.

Sutou, S., K. Yamamoto, H. Sendota, and M. Sugiyama.  1980b.  Toxicity,
    fertility, teratogenicity, and dominant lethal tests in rats administered
    cadmium subchronically.  II. Fertility, teratogenicity, and dominant
    lethal tests.  Ecotoxicology and Environmental Safety 4:51-56.

Takahashi, T.  1972.  Abnormal mitosis by some mutagens in S. cerevisiae.
    Bull. Brewery Sci. 18:37.

Takenaka, S., H. Oldiges, H. Konig, D. Hochrainer, and G. Oberdorster.  1982.
    Carcinogenicity of cadmium aerosols in Wistar rats.  In press.

-------
Tarkowski, A.K.  1966.  An air-drying method for chromosome preparations
    from mouse eggs.  Cytogenetics 5:394-400.

Tomatis, L.  1977.  The value of long-term testing for the implementation of
    primary prevention in organs of human cancer.  Hiatt, H. H., Watson, J.D.,
    and Winsten, J.A., eds.  NY: Cold Spring Harbor Laboratory, pp. 1339-1357.

Umeda, M., and M. Nishimura.  1979.  Inducibility of chromosomal aberrations
    by metal  compounds in cultured mammalian cells.  Mutat. Res. 67:221-229.

U.S. Environmental Protection Agency.  1977.  The Carcinogen Assessment
    Group's assessment of cadmium.  Washington, D.C.  Unpublished.

U.S. Environmetnal Protection Agency.  May 1981.  Health Assessment Document
    for Cadmium.  Office of Health and Environmental Assessment.  Washington,
    D.C.

U.S. Food and Drug Administration.  1977.  Total diet studies (7320.08).
    Compliance Program Evaluation, Bureau of Foods.

Vasudev, V.,  and N.B. Krishnamurthy.  1979.  Dominant lethals induced by
    cadmium in Drosophila melanogaster.  Curr. Science 48:1007-1008.

Van Rosen, G.  1953.  Radiomimetic activity and the periodical  system of the
    element.   Bot. Notis. 140-141.

Van Rosen, G.  1954.  Breaking of chromosomes by the action of elements of the
    periodical system and some other participants.  Hereditas 40:258-263.

Venitt, S., and L. Levy.  1974.  Mutagenicity of chromates in bacteria and its
    relevance to chromate carcinogenesis.  Nature (London) 250:493-495.

Watanabe, T., T. Shimada, and A. Endo.  1979.  Mutagenic effects of cadmium on
    mammalian oocyte chromosomes.  Mutat. Res. 67:349-356.

Wynder, E.L., K. Mabuchi, and W.F. Whitmore, Jr.  December 1974.  Epidemiology
    of adenocarcinoma of the kidney.  J. Natl. Cancer Inst. 53(6):1619-1634.

Wyrobek, A.J., and W.R. Bruce.  1975.  Chemical induction of sperm
    abnormalities.  Natl. Acad. Sci. U.S.A. 72:4425-4429.

Zasukhina, G.D., T.A. Sinelschikova, G.N. Lvova, and Z.A. Kirkova.  1977.
    Molecular mutagenic effects of cadmium chloride.  Mutat. Res. 45:169-174.
                                     144

-------