-------
 In the present table, dosage is expressed in mg and is based on 10 m3 inhaled
 air.)  There is no tendency towards a dose-response relationship at total  expo-
 sures from 105 to  3890  mg, but a sharp increase is noted at higher dosages.
 Furthermore,  Blejer and  Wagner  (1976) reported  that of the  173  deaths,  138 had
 occurred  among workers with less  than one  year  of  exposure.   In that group, 16
 deaths  were due to respiratory  cancer.   As seen in Table  7-2, 15 of those  deaths
 occurred  in the six  groups with  average total  exposures  estimated  to be from
 105  to  3890 mg.  There are no quantitative data on other compounds that these
 workers might have been exposed  to during their total  time with the company;
 however,  it is known that in addition to the arsenic-containing insecticides,
 the  plant processed and  packaged  several other  products,  the most important of
 which were powdered sulfur  and  dry  lime  sulfur.
      The  retrospective cohort analysis was  based  on a roster of 603 workers
 who  had worked for at least  one  month in the  actual  unit  between  1940 and
 1973.   Workers leaving the company  before  retirement  as  well as workers who
 had  been  exposed to asbestos were not included  in  the  analysis.    It was stated
 that virtually all  men with at  least one year of exposure had been  identified.
 Person-years,  by 10-year age groups,  for five calendar year groups were calcu-
 lated and expected number of deaths  were  calculated by using  United States
 white male  mortality  data.  Table 7-3 shows  that by this analysis a  significant
 increase  (p <  0.01)  in  deaths due  to respiratory cancer was  found  among
 exposed workers.   There  was also a  significant increase  (p <  0.01)  in the
 number  of deaths  due to malignant  neoplasms  in lymphatic and  hematopoietic
 tissues,  except leukemia.
     A  chemical plant in Baltimore was  the  subject of two  studies concern-
 ing  the working environment and one  study  concerning  the outer environment.
This  plant  started producing arsenic acid  in the early 1900s and in  the early
                                    7-16

-------
           TABLE 7-3.   OBSERVED AND  EXPECTED  DEATHS  FOR SELECTED  CAUSES  IN
                         RETROSPECTIVE COHORT ANALYSIS  (1940-1973)




AT 1 causes
Malignant neoplasms, total
Respiratory system
Digestive organs & peritoneum
Lymphatic & hematopoietic
tissues except leukemia
All other sites


Observed
deaths
95
35
20
7

5
3
Expected
deaths
(U.S. white
males)
113.5
19.4
5.8
6.3

1.3
6.0
Ratio of
observed
to expected
deaths
.84
1.80
3.45t
1.11

3.85t
.50
Diseases of cardiovascular system    41
Emphysema, chronic bronchitis,
  & asthma

All external causes

All other causes
4

9

6
58.5


 2.8

12.7

20.1
 .70


1.43

 .71

 .30
t  p < 0.01

Source:  Ott et al.  (1974).
                                    7-17

-------
 1950s  production of  arsenical  pesticides  started.   Lead arsenate,  calcium
 arsenate  and sodium arsenate were  among the compounds produced.  Production
 terminated  in 1974 (Matanoski et al., 1976).  The plant also produced chlori-
 nated  hydrocarbons and  organic phosphates.  Data on  air concentrations of
 arsenic are  lacking.
     In the  first study, Baetjer et  al.  (1975) reported on both the propor-
 tionate mortality and age-specific  death rates of workers who  had  retired
 between 1960  and  1972 and generally had  at  least  15 years'  employment.   Seven-
 teen of 22 deaths among  male white workers  and  two of  five  among female  workers
 were due  to  malignant  neoplasms!   The female  deaths  did not appear to be
 related to  occupational  exposure,  whereas,  the male  deaths did.  Of the 17
 malignant neoplasms in males, 10 were  in  the respiratory  tract,  3 were lympho-
 sarcomas, and the remaining 4 were of other tissues.  The  proportionate mor-
 tality analysis,  based  on mortality data for the  city of Baltimore,  showed  an
 observed/expected ratio  of  6.58 for respiratory cancer (p <.05)  and  15.79 for
 cancer of the lymphatic and hematopoietic  system (0.1 <  p  < 0.05).  The age-
 specific death  rate  analysis  showed that deaths from respiratory cancer were
 16.67 times  the expected,  and,  that  for  lymphosarcomas,  the  observed  number
was 50 times  that of the expected number.   The  authors  reported  that observed
 and expected  rates  for  non-cancer deaths did not differ significantly  (p <
 0.05).
     In a second study (Mabuchi  et al., 1979),  a follow-up was made of workers
 employed from 1946 to 1974.   Since  exposure data  were  lacking, an attempt was
made to classify workers according to exposure  to arsenicals and non-arsenicals.
A roster of 3,141 workers was obtained.  Since  2,189 workers had been employed
for less than 4 months,  a 20 percent random sample was drawn from that  popula-
tion and together with the remaining 952 workers constituted the study  popula-
                                    7-18

-------
tion:   1,050 males  and  343 females, mainly white.   Exposure assessments were
made,  and the workers were categorized into one of six exposure groups.
     Of the  study  population,  240 had died:  197 males and 43 females.  Ex-
pected deaths  were calculated from  the  city of Baltimore statistics.  The
observed/expected  ratios  for  lung cancer were analyzed by exposure, year of
first  employment,  and duration of  employment.   A  statistically significant
increase  in  lung cancer  mortality (SMR = 336, p  < 0.05) was found among "pre-
dominantly arsenical  production"  workers.   There was  a clear  lung cancer mor-
tality dose  response among  these workers  by duration  of exposure.   Those
employed  15-24 years and 25+ years  both had statistically  significant (p <
0.05)  lung  cancer  SMRs  (1365  and  2750,  respectively).   Interestingly,  statis-
tically significantly elevated SMRs were only found in "predominantly arsenical
production"  workers,  but not  in workers  engaged entirely  in  arsenical  produc-
tion.  Further analysis  revealed that the  proportion of  workers engaged en-
tirely in arsenical  production  for 5 years or  more  was  relatively low (1
percent),  while the  proportion  of  workers exposed predominantly,  but not
entirely,  to arsenic for  5 years or more was much  higher (29 percent).  This
difference  in  duration  of exposure may have  accounted  for the absence of
excess lung cancer mortality among  the  workers  engaged entirely in arsenical
production.   Data  on smoking  were not obtained.
      Occupational  exposure to  arsenical pesticides  has  been common  among
vintners  and agricultural  workers.   Exposure has mainly been  to  lead  arsenate.
In a study  of orchard workers who used  lead arsenate in  the  Wenatchee Valley
in the state  of  Washington,  Nelson  et  al.  (1973)  found  a deficit of  lung
cancer mortality for the  period  1938-68 when compared with  mortality for  the
 state of  Washington.
      Because the  results of this  study  were at variance with previous evidence
 on the long-term  effects  of  arsenic exposure,  NIOSH  reviewed data  from other

                                     7-19

-------
 sources  and used alternative procedures in an attempt to verify the findings.
 Analysis  of respiratory cancer mortality data by occupational  category for  the
 state  of  Washington for the  period 1950-71 found that orchardists  had  19 percent
 more  deaths from  respiratory cancer than expected.  An  analysis  of cancer
 mortality rates for 1950-69 for the counties comprising the locale from which
 the orchardist  sample was  drawn by Nelson  et  al.  found  that males  had  a 7 per-
 cent  higher respiratory cancer  rate  than  expected among males.  The  county
 from  which most  of the orchardists were  drawn  had a 31 percent excess (P
 <0.01) of lung cancer mortality.  Thus, it was NIOSH's (1975)  conclusion that
 the Nelson et al.   report  did not accurately depict the cancer experience  of
 persons exposed to  lead arsenate spray  in  the Wenatchee Valley.
     Several  studies in Germany indicate that  workers exposed to  arsenic
 trioxide  when  spraying  vineyards had a  high  mortality in cancer,  especially
 lung cancer.   In  one report  (Roth, 1958), it was  stated  that of 47  autopsies
 among  vintners  with chronic  arsenic intoxication,  30 (64 percent)  were due  to
 cancer, and 18  to  lung  cancer  (60 percent of all  cancer  deaths).  The author
 did not  state how  the  cases were selected, nor were  controls  used in the
 study.
     Gilbert et al.  (1983) studied a group  of  182 workers  in  Hawaii exposed
 for at least 3  months  during the  period 1960-1981 to  the wood-treating che-
micals chromated-copper-arsenate (CCA), pentachlorophenol  (penta),  tributyl
tin oxide  (TBTO)  and lindane.  The study  was divided  into  two  parts:  (1)  a
cohort comparison  study of 88 workers and  61  controls and (2) a historic pro-
spective study which consisted  of a morbidity and  a mortality analysis of the
entire cohort.
     In the  cohort  comparison study,  61  controls were matched on the basis  of
age, sex,  race, level  of physical activity, and weight to  88 wood  treatment
                                    7-20

-------
workers who were "qualified and agreed to participate in the study."  Controls
were recruited from among the membership of carpenters', ironworkers', masons',
plumbers', and  stevedores'  unions  and  from  the  names  of friends  and relatives
referred  to  the study by participating members  of  the  occupational cohort.
Fourteen  of the controls were reported to be carpenters and 13 of them had had
exposure  to either CCA or penta or both.  Their  urine arsenic levels, however,
were  reported  not to differ significantly  from  those of other  controls.  The
exposed group  and the control group were each  given a comprehensive health
examination  consisting  of a questionnaire  and  clinical  and laboratory  tests
including analysis of a  urine sample for penta,  arsenic, copper, chromium, and
tin levels.  Only penta  was found  to be significantly elevated in the urine  of
the exposed  group over that of the controls.   The authors reported that there
were  no "clinically  significant" differences between  the exposed group and the
controls.  No  significant differences  were  found between the wood treaters and
controls  with  respect to educational level, smoking history, or alcohol  consump-
tion.
      The  historical  prospective  study  identified three  cases of cancer,  one  by
means of the questionnaire in the cross-sectional  study and the other  two  by
means of the Hawaii Tumor  Registry.   One  of the cancer  cases was  a bladder
cancer case; the other  two were colorectal cancer cases.  For the mortality
analysis, the vital status of 125 of the 182 workers (69 percent) was able to
be determined.  Of  these 125,  six deaths  occurred,  five from cardiovascular
 disease,  the other  from an  unknown cause.   The authors calculated that eight
 deaths would have been  expected in this group,  three of which would have been
 from cancer.   The authors did not state the basis of the expected number of can-
 cer deaths.
      There are  limitations in  the study design and sample size of this study
 which make the results  with  regard to the evaluation  of  a cancer  risk among

                                     7-21

-------
  the  wood treaters inconclusive.  A  cohort comparison study on the basis of a
  single  medical  examination  is  an inappropriate approach to determine  if  a
  cancer  risk exists in the wood  treater group, particularly when the group con-
  sists primarily  of persons  who  are  currently employed as  wood treaters  (60 of
  88 were current  employees).   Most  persons  who  have  developed  cancer will
  either  have died  or  left the work  force.   In addition, the sample size (88
  current  or former wood  treaters and 61 controls) would  have been too  small to
 detect an excess  cancer risk.   Also,  it  should  be noted that the inclusion in
 the controls of  13 carpenters who had been  exposed  to  arsenic-treated wood
 certainly presents a  potential bias, despite the fact that the  authors claimed
 that the levels of arsenic  in the urine of  these  13 were not  significantly
 different from those in other controls.   In this regard  it  should be noted
 that  no  effort had apparently been  made  to  restrict  the intake of seafood  or
 other foods  or liquids, which might have elevated urine arsenic levels, prior
 to collection  of  the  urine samples.   The  mean level of urinary arsenic in both
 the study group and in  the controls  was significantly  (P <0.01) higher than what
 the authors reported  would be a  normal level.
      The historic  prospective study also had limitations.  Again, the sample
 size  was too  small to adequately determine  whether an excess risk of  cancer
 existed  in the study  cohort.  Only  125 were  included  in the mortality study
 and only 182 were included   in  the  morbidity study.   Arsenic  exposure via
 inhalation and  ingestion is   known to be associated with lung and nonmelanoma
 skin  cancer, respectively.   Nonmelanoma skin  cancer is rarely fatal and thus,
 an  excess risk  of skin  cancer would  not  have been  detected in  such a small
mortality study.   The Hawaii Tumor Registry, which was  one of  the primary
 sources of information used  to identify cancer cases for the morbidity study,
does not even report nonmelanoma skin cancer.
                                    7-22

-------
     There is also a question of whether the authors had sufficiently allowed



for a cancer latency period in their study.  The authors did not indicate the



length of  follow-up  of  the  cohort members.  Some  indication  is  provided,



however,  in the  data  reported by the authors on the length of employment of



the 88 workers  in  the study cohort of  the  cohort  comparison study.   These



workers were part of the population in the incidence and mortality studies.   Of



the 88 workers,  60 had worked 10 years or less and 80 had worked  15  years  or



less at the time of the physical  examination in 1981.   Twenty-two  of  the study



cohort in  the cohort comparison study were  former employees.  At the minimum,



assuming that the workers had had no breaks in employment, 58 of 125  (46 percent)



in the mortality study would have been followed for only 15 years  or  less and 38



of 125 (30 percent) would have been followed for only 10 years  or  less.  In  the



morbidity study, if,  again, no breaks in employment of the workers are assumed,



58 of  182  (32 percent) would  have been  followed for only 15 years or less and



38 of 182 (21 percent) would have been followed for only 10 years  or  less.   In



conclusion, this study is inadequate to conclude whether an excess risk of cancer



exists in the wood treater population in Hawaii.



     Occupational exposure to  arsenic  also occurs  in smelters where  exposure



is  predominantly to  arsenic  trioxide.   Several  studies  have  been done on



mortality  among  workers  at the copper  smelter in Tacoma,  Washington.   Pinto



and Bennett (1963) reported on the proportionate mortality of 229  workers from



1946  to  1960.   Workers  leaving the  plant  before retirement were not  included.



The proportionate  mortality  of the smelter workers was compared with that of



males  in the state of Washington in 1958.  Of all  cancer deaths,  the respec-



tive  proportions of lung cancer were 41.9 and 23.7  percent, respectively.  The



authors  then classified  the smelter workers according  to  arsenic  exposure and



did not  find any difference between exposed  and  non-exposed.   However, the



"non-exposed"  had elevated arsenic  levels in urine, suggesting possible  exposure.





                                    7-23

-------
More extensive studies  have since been published, showing an increase in lung



cancer among arsenic-exposed  workers  (Milham and Strong, 1974;  Pinto et al.,



1977; Pinto  et al., 1978; Enterline and  Marsh,  1980;  Enterline and Marsh,



1982).



     Milham and Strong  (1974) examined county records from  1950  to 1971 to



find the  number of deaths due to respiratory  cancer among  county residents



employed  at  the smelter.   Expected  number of  cases  were calculated for the



smelter population  by  using  the 1960 age-cause specific mortality statistics



for white males in  the  United States.  Forty deaths  were observed and 18 were



expected.



     The  two papers by Pinto et al. (1977, 1978) refer to the same study and



the following is based on the 1978 paper.   The cohort studied consisted of 527



men who were living pensioners on January 1, 1949 or who became pensioners before



January 1, 1973.  Complete job histories were obtained for 525 men.   The average



duration  of employment was 28 years, ranging from 7 to 54 years and beginning  in



1910.  Death certificates were obtained for all 324 men who had died during the



observation period  (1949-1973).  Expected numbers of deaths were calculated from



statistics of the state of Washington.



     An exposure  index was constructed by  using  data on urinary levels of



arsenic obtained  in 1973.  Mean  urinary concentrations were  calculated  for 32



departments, and  the individual exposure  index was  obtained by multiplying



urinary arsenic level  with years of work in a department.   If  an individual



had  worked in more than  one department,  the  index  values  were added.   By



dividing  the exposure  index by the total  number  of  years in the  smelter, an



index of  intensity  of exposure was obtained, i.e., the average urinary arsenic



level.  These  indices were created to enable interdepartmental comparisons and



did not reflect past exposure  since air analysis  in  the  1930s to 1940s indicated
                                    7-24

-------
that exposure might  have  been 5-10 times higher at  that time.   It may have



been still higher in 1910.



     Data were also  obtained  on smoking habits from all men still alive and



from relatives of  men  who had died since January,  1961.   Table 7-4 (Pinto et



al., 1978) shows that  there was a significant increase  (p < 0.05) in deaths



from all  causes, in  cancer deaths in general, and,  specifically,  in deaths



from respiratory cancer.  Almost all  of the excess  mortality could be explained



by  the  increase  in  lung  cancers  which  could not be explained  by smoking.



     Table 7-5 shows respiratory  cancer deaths in  relation to exposure index



(a  value  which  reflects both the  duration  and  intensity of exposure).  An



increase in SMR with exposure is seen.  Table 7-6 shows that both duration and



intensity of  exposure  contributed to the excess in  respiratory cancer.   As



stated above, the  urine values are relative and do  not reflect the actual



exposure.



     It was also shown that the main excess occurred  in ages 65-74 years,



whereas at higher  ages the lung cancer  rate was closer to expected rates.



     More  recently,  Enter!ine  and Marsh (1980,  1982)  conducted additional



studies on workers at  this same copper  smelter in Tacoma.  A cohort of 2802



males who  worked a year or more  during  the period 1940-1964 was  identified.



Since a one-year work  exposure was required for eligibility into the cohort,



actual follow-up did not start until 1941 and extended through 1976.  In the



cohort, the vital  status of 51  could not be verified, leaving 2751 persons.   In



that group, 1061 deaths had occurred.   There was  a  significant increase in



total cancer  mortality which wholly depended on an  increase  in deaths from



lung cancer.  Arsenic  exposure was estimated for  each man  on the  basis of  a



representative average  urinary  arsenic level for workers in a given department.



Using this  representative value,  an  individual value was calculated for each
                                    7-25

-------
    TABLE 7-4.  OBSERVED AND EXPECTED DEATHS AND STANDARDIZED MORTALITY RATIOS
         FOR SELECTED CAUSES OF DEATH OF 527 MALES OF COHORT UNDER STUDY*
Cause of death
All Causes
Cancer
Digestive
Respiratory
Lymph, etc.
Urinary
All Other Cancers
Stroke
Heart Disease
Coronary Heart Disease
All Other Heart Disease
Respiratory Disease
All Other Causes
Disease
Classificationt

140-205
150-159
160-164
200-203, 205
180,181
330-334
400-443
420
480-493, 500-502

Observed
324
69
20
32
2
3
12
43
144
120
24
11
57
Expected
288.7
46.5
16.4
10.5
2.1
3.3
14.2
38.0
132.3
110.2
22.1
10.8
61.8
SMR
112.2+
148.4+
122.0
304.8+
95.2
90.9
84.5
113.2
108.8
108.9
108.6
101.8
92.2
*Cohort consisted of living male pensioners from a copper-smelting plant who were
 living January 1, 1949, and whose causes of death were noted through December 31,
 1973.

tNumbers from rubrics of 7th Revision of International Classification of Diseases.

+P<.05

Source:  Pinto et al. (1978).
                                         7-26

-------
           TABLE 7-5.   OBSERVED AND EXPECTED RESPIRATORY CANCER DEATHS AND
               STANDARDIZED MORTALITY RATIOS BY ARSENIC EXPOSURE INDEX
Respiratory Cancer Deaths
Exposure index
Under 2,000
2,000-2,999
3,000-5,999
6,000-8,999
9,000-11,999
12,000 and over
Mean index
1,514
2,513
4,317
7,473
10,135
14,712
No. of men
36
109
205
109
38
29
Observed
1
4
11
7
4
5
Expected
0.9
2.1
3.9
2.3
0.7
0.6
SMR
111.1
190.5
282.0*
304. 3*
571.4*
833.3*
*p <.05

Source:  Pinto et al.  (1978).
    TABLE 7-6.   OBSERVED AND EXPECTED RESPIRATORY CANCER DEATHS AND STANDARDIZED
          MORTALITY RATIOS BY INTENSITY AND DURATION OF EXPOSURE TO ARSENIC


  Intensity of                           Duration of exposure
    exposure             less than 25 years                25 years and more
([jg/liter urine)  Observed     Expected     SMR     Observed     Expected     SMR
  50-199             2

  200-349            4

  350 and over       3
2.1        95.2       10

1.5       266.7        8

0.5       600.0*       5
3.6        277.8*

2.2        363.6*

0.6        833.3*
*P<.05

Source:  Pinto et al.  (1978).
                                       7-27

-------
man for  each  year of employment in a given department,  and  a  total  exposure
estimate per  individual  was made by summing values  across all jobs and all
years of employment.
     This method of estimating exposure differed from earlier methods employed
by Enterline  and  co-workers (Pinto et al., 1977, 1978)  in that estimates of
historic exposure,  based upon  simple linear interpolations and extrapolations
of actual data from 1948-52 and 1973-75, were used to characterize exposure by
department, rather than by 1973 urinary measurements.  Use of this new method of
analysis partially  helped  to  eliminate exposure underestimates of  workers
employed in the  early years of the smelter operation.   However, the present
method did not totally eliminate this bias because urinary arsenic levels were
only determined  for workers starting in 1948.   For  workers  exposed  prior to
1948 (approximately  80  percent of the present study cohort), urinary arsenic
values for  1948-52  were assumed to apply.  Furthermore,  the average urinary
arsenic  level  for some  departments may  have been based on only a few samples,
thereby  limiting the usefulness of the departmental average as a representative
measure  of  any given worker's urinary arsenic  level.  The authors  did note
that in  areas of the smelter  where arsenic levels were  reported to  be  high,
workers  tended to be measured  more often  for  urinary arsenic.  Thus, averages
for those areas were, in fact, more representative.
     Using  this  time-weighted  measure  of exposure,  a  life table  method for
accumulative  dose, a 10-year lag period and a standard population of mortality
rates in the  state  of Washington,  the authors  reported that  SMRs  for respira-
tory cancer ranged  from 155 in the  lowest  exposure  category  to 246 in the
highest  category.  Table 7-7   shows the relationship between the time-weighted
estimates of  arsenic exposure lagged 0 and 10 years and respiratory cancer.
This relationship was much weaker than that  previously  reported  by  Pinto  et
al. (1977,  1978).   Enterline and Marsh noted that results from the two  sets of

                                    7-28

-------
 TABLE 7-7.   RESPIRATORY CANCER DEATHS AND SMRs BY CUMULATIVE ARSENIC EXPOSURE
                 LAGGED 0 AND 10 YEARS, TACOMA SMELTER WORKERS

Cumulative
Exposure (ug/As/1)
(urine-years)
<500 ( 302)
500-1500 ( 866)
1500-3000 ( 2173)
3000-5000 ( 4543)
7000+ (13457)

0 Lag
Observed
Deaths
8
18
21
26
31
Lag


SMR
202.0
158.4
203.2**
184. 1**
243.4**

10-Year
Observed
Deaths
10
22
26
22
24

Lag

SMR
155.4
176.6*
226.4**
177.6*
246.2**
* p<0.05          «
**p<0.01
QMean of class interval
Source:  Enter!ine and Marsh (1982).
studies were  not  totally comparable due not only  to  the differences in the
exposure estimates  noted above, but also to differences  in follow-up periods.
In earlier reports, the  follow-up started after exposure stopped at retirement,
whereas in the  present  study,  follow-up started at various points  in the work
experience of the workers, allowing for follow-up and dose accumulation to pro-
ceed concurrently.  When the authors analyzed a subsample of the present study
cohort, consisting  of 582 workers  retired  at age 65 and  over  (paralleling the
experimental design of the earlier studies), a stronger dose-response relation-
ship was, in fact, observed.
     Table 7-8  shows  the respiratory cancer deaths and  SMRs  by duration of
exposure  and time  since first exposure.   From Table  7-8,  it appears that
neither duration  of exposure nor long latent periods made strong contributions
to  excess  respiratory  cancer.   The authors suggested that this may  have been
due  to the high SMRs  observed  shortly after termination of employment but not
noted  thereafter; thus, for workers with  less  than 10 years  of exposure the
                                     7-29

-------
CO
oi
LU
a

&.
LU
i
LU
s:
co
«=c
•z.
0
O
—
ca
to
CO

a
z.
«ac

CO
a:
r-
^

UJ
1
ca
>
U
c
OJ
•P
CO
1








































cc
CO



-a
i — CD t/)
re > -c
-p J- +i
O CO (O
r— to 0)
.a o
o



Cv*
s:
CO
o
co -a
CD to
> -C
i- -P
0) CO
w cu
-Q O
O




§r
CO
cn
CM

o -o
CM CU W
;> r^
5- -P
d) CO
v> cu
jQ Q
o

s:
cn co
rH
|
o -a
rH O) to
> r—
i- -P
O) (0
to cu
-Q O
O



«*v*
y
CO
0
v -a

f^ iT^
0




C •'"^
O W)
•1- t-
-P 
^ ^*>
3 •>— •
O




^/ , ^ .
^C • ^C • 3C • K •
en CM rH co
co r^ ic r-5
r~. co CM CM
rH rH CM CM


3( >
^ •
0 00 CO 0
•3- rH CM CM



X ' $C •
9C • ^C *
cn cn co co

r*^ r^ to rH
CO CO Ii3 CM
rH rH CM CM




CM «>t- CO O
rH rH CM




*•
* *• *•
rH O O
o oo r-~ i
rH r-~ cn
CM CM rH





^ CO CO 1
rH rH



<•'«*•
LO l£> 1 1
CO LO
CM rH




0 CO 1 1
rH





CO
• 1 1 1
LO
LO





rH 1 1 1











o cn cn
rH rH CM +
V 1 1 CZ>
o o co
rH CM




JC •
*•
.
CO
cn
rH



•51"
O
rH


^ '
^ .
CO

rH
cn
rH




cn





^ •
X •
CO
co
rH
CM





CO
CO



K •
O
rH
CM




CO
rH





CD
,
LO
LO





rH











^~.
fO
-P
O
I—


























































.
^-^
CM
CO
cn
rH


.e
to
i-
CO
-a
c
(0

-------
SMR is highest  10-19  years after date of  hire;  for workers employed 10-19



years, the SMR is highest 20-29 years after date of hire, etc.   This high rela-



tive risk of lung cancer mortality in the decade following termination of employ-



ment, which essentially disappears in the following 20 years, prompted the authors



to suggest that  arsenic may be  a  promoter  rather than an  initiator.  However,



when the authors reanalyzed the data according to a method which followed workers



only from the point of termination or retirement, both duration of exposure and



intensity of exposure contributed more strongly to respiratory cancer mortality



(Table 7-9).   The difference in the two analyses led the authors to suggest that



the weak  dose-response  relationship  observed in the first  analysis may  have



resulted from a tendency for workers in high-exposure jobs to leave employment



more quickly than those in low-exposure jobs.



     In studying the  interactive  effects of sulfur  dioxide, the authors  did



not find significant differences in respiratory cancer incidence in two depart-


                                                       3
ments which  both had  high arsenic exposures (7500  ug/m  )  but  differing SOp



exposures, one having low to moderate exposures (520 ppm) and the other having



essentially  none.   Because  the  respiratory cancer SMRs were quite similar in



the two departments,  the  authors  suggested that SOp exposure did not play an



important role in respiratory cancer excess at that particular smelter.



     In discussing  their overall  study  results,  Enterline and Marsh noted



that, in  this particular  case,  a  dose  response was  not observed when dose was



measured  in  terms of  cumulative dose.    The  results seen  in Table  7-8 suggest



that short exposures  seemed to have a disproportionately greater effect than



long  exposures  and  that effects  of  early  exposure  tended to diminish  with



time.  Table 7-9 suggests  that short high-intensity exposures  may have a



greater effect  than  longer-term,  more low-level  exposures.  The  fact  that



different results are obtained when different exposure/follow-up methods are
                                    7-31
-------
to
LU

rv*
O
3:

fyf
LU
LU
*»•
to

^C
^r*
O
O

C£

to
Q
z.
•a:

to

F^

LU
a

a:
til
LLJ
o
•z

o

^_
rv*
O
J—
i
»— i
<">
to
LU



,
CD
1


LU

CO

in
 -C
S- P
,p (0
W ID
^^ ^J
O


(0 ^
in

C3 T



Q£
^|
IO


T3
0) in
P .C
0 P

a. cu
x -a
LU




•a
QJ m
> x:
S- P
QJ (0
tn at
J3 T3
o




P
(C -^

• 'r—
O 5-
z.








q-
o
cu
C S- /~N
O 3 in
•r- in S-
•P O fO
m a. cu
t- x >,
3 0} *~s
a

¥.¥.¥.* ¥.
¥. * * * *
CD ID ID 0 <•
• • • • •
CO rH r-> r-- rH
O CM r--. -sf CD
CM ro in co CM



p*s p*s CM
ID CJ r-~ rH tD
CM •=*•••
. . CM CO rH
CM CO CM
rH








LO rH ID H CO
CM rH rH rH 
-------
 employed suggests that choice of experimental design has a possible influence


 on results.   The possibility that  some workers may have simply  been  more


 susceptible than others may also have accounted for the dose-response  relation-


 ships observed in this study.


      Because the authors suggested that the role of arsenic  in the carcinogen-


 esis  process may have been as  a  promoter rather than an initiator, the  signifi-


 cance of cumulative exposure to  arsenic as a measure of dose would be  question-


 able  in regard  to  the induction of respiratory  cancer  as observed in this
t

 study.   The role of  arsenic as  an initiator as well as a promoter cannot be


 ruled out, however.  Until  further research is done, the authors'  conclusions


 from  this  study remain speculative.


      Mortality of workers in a smelter  in Magna, Utah, was studied by Rencher


 et al.  (1977).  Average hourly air concentrations of arsenic in 12 work areas


 were  between  zero  and 22 mg/m   in  1975 (NIOSH, 1975).  However,  exposure


 before  1959 had  likely  been higher,  since a processing of  ore with a  rela-


 tively  low arsenic content  began  in that year.  In the period 1959 to 1969,


 965 deaths  had occurred among  all  current or former employees,  and death


 certificates were  obtained  for  virtually  all  deceased.   The proportionate


 mortality  for smelter workers  was  compared with that for mine workers,  concen-


 trator  workers, refinery workers, and  office  workers,  and for the population


 above 20 years of age for  the  entire  state of  Utah  in 1968.   Among the  smelter


 workers,  7 percent of  the  deaths  were  due to  respiratory cancer,  whereas  the


 percentage for the  other factory employees varied from  0 to  2.2 percent and was


 2.7 percent for the state.   Data on  smoking were obtained for all  smelter  workers


 and for subsamples of the other employee groups and indicated that nonsmoking


 smelter workers had the same percentage of deaths from  lung  cancer as mine and


 concentrator workers who smoked.  By applying lifetable methods, age-adjusted
                                     7-33
-------
death rates were obtained.   For  smelter workers, a death rate for lung cancer
of 10.1 per 10,000 was obtained  compared with 2.1 and 3.3 per 10,000 for mine
workers and the state population, respectively.   Causes of death were compared
with a cumulative  exposure index obtained by multiplying the number of days
spent in each  department  by  the  average exposure level and then summing.  The
average exposure to  arsenic  as well as sulfur  dioxide,  sulfates,  lead, and
copper was found to  be higher  for the  lung cancer cases than for other causes
of death.
     In addition to  these studies,  a  large  study by  Lee and Fraumeni (1969)
involving 8,047 white  males  was conducted at the Anaconda Copper Smelter in
Montana from 1938  to 1956.*   Data were obtained on time, place, and duration
of  employment  for  each individual,  all subjects having  worked at least 12
months in the  smelter  during the indicated study period.   Follow-up was from
1938 to 1964.   Death certificates were obtained, and  the lifetable method was
used to compute the  expected number of deaths, using mortality rates of the
state of Montana.  The smelter workers were classified into 5 cohorts,  based
on total years of  smelter work  and the time period in which the years  were
worked:  (1) 15 or more years  worked before  1938; (2) 15 or more years  worked
between 1938 and 1963; (3) 10-14 years; (4)  5-9 years; and (5)  1-4 years.   An
attempt was also made to classify the workers according to exposure to arsenic
and sulphur dioxide.   Exposure to arsenic for more than 12 months had occurred
among 5,185 men.  These workers were divided into three groups; heavy, medium,
and light exposures.   This division was based on exposure times and place at
work.   Arsenic concentrations in air were primarily determined from a
     *While  the original Lee  and  Fraumeni  study reported on the  mortality
experience of male smelter workers in unidentified states, subsequent analyses
of  these  workers  indicate  that they were employed  solely  at the Anaconda
smelter in Montana.
                                    7-34
-------
1965  survey by  Public  Health Service officials  (NIOSH,  1975).   These air
data are shown in Table 7-10.
     According to a former public health official associated with this survey,
actual measurements were  collected during two different periods—one in 1965
and one  "about  five to six years  earlier"--and  at several locations within
given departments  (Archer,  1983;  personal communication).  It is impossible,
however, to determine how the values listed in Table 7-10 are distributed over
these different time periods and locations.  Further, it is also impossible to
determine the number  of hours sampled at  a  given location.   Therefore, the
arithmetic  means  used  by  Lee and Fraumeni to characterize heavy, medium and
light exposures can only be viewed as very rough estimates of arsenic exposure,
their primary value being in their relative scale of measurement.
     In Table 7-11  it  is  seen that  heavy  and  medium exposures  resulted in
significant increases  in  SMR for  respiratory cancer.   The SMR was  highest
(800) in workers  belonging  to cohort 1 (more than  15 years of smelter work
before 1938),  and then decreased  to 263  among v/orkers with  1-14 years of
smelter work.   Among the  workers  with light exposure, the SMR was 250, 310,
and 214  in  cohorts  1,  2,  and 3-5,  respectively.   Among the smelter workers
with less than 12 months'  exposure to arsenic,  the SMR was 286.
     Lee and Fraumeni  also  looked at factors that  possibly  had  confounding
effects on their study results.   Smoking histories were not recorded; however,
based upon information obtained from other studies,  Lee and Fraumeni  concluded
that smoking alone could not have accounted for the excess respiratory cancer
mortality of the magnitude observed in their study.
     In contrast  to smoking, the authors  did note  a positive relationship
between exposure to sulfur dioxide and respiratory cancer mortality; however,
they found  it difficult to  separate the relationship of arsenic from sulfur
                                    7-35
-------
      TABLE 7-10.   1965 SMELTER SURVEY ATMOSPHERIC ARSENIC CONCENTRATIONS (mg As/m?)
"Heavy exposure area
Arsenic Roaster Area
0.10
0.10
0.10
0.10
0.10
0.10
0.17
"Medium exposure area
Reverberatory Area
0.03
0.22
0.23
0.36
0.56
0.63
0.66
0.76
0.78
0.78
0.80
0.83
Treater Building and Arsenic
0.10
0.10
0.10
0.11
"Light exposure areas
Copper Concentrate Transfer
0.25
0.65
1.20
Samples from Flue Station
0.10
0.24
Reactor Building
0.001
0.002
0.002
0.002
" as classified by Lee
Mean:
0.20 Median:
0.22
0.25
0.35
1.18
5.00
12.66
" as classified by Lee
Mean:
0.93 Median:
1.00
1.27
1.60
1.66
1.84
1.94
2.06
2.76
3.40
4.14
8.20
Loading Mean:
0.48 Median:
0.62
3.26
7.20
11 as classified by Lee
System Mean:
Median:


Mean:
Median:

Mean:
0.003 Median:
0.009
0.010

and Fraumeni
1.47
0.185






and Fraumeni
1.56
0.88











1.50
0.295



and Fraumeni
0.70
0.65


0.17
0.17

0.004
0.002



Source:  National Institute of Occupational Safety and Health (1975).
                                         7-36
-------
TABLE 7-11.   OBSERVED AND EXPECTED DEATHS FROM RESPIRATORY CANCER, WITH STANDARDIZED
      MORTALITY RATIOS (SMR),  BY COHORT AND DEGREE OF ARSENIC EXPOSURE, 1938-63



Cohort
All cohorts
combined

1


2


3-5+



Respiratory
cancer
mortal ity
Observed
Expected
SMR
Observed
Expected
SMR
Observed
Expected
SMR
Observed
Expected
SMR



Heavy
18
2.7
667?
8
1.0
800?
6
0.9
667?
4
0.9
444§
Maximum exposure to
arsenic (12 or more
months)*
Medium
44
9.2
478?
22
3.3
667?
12
2.2
545.?
10
3.8
263§



Light
45
is; 8
239?
14
5.6
250?
9
2.9
310?
22
10.3
214?
Number of persons in
  arsenic category*
402
1,526
3,257
*The remaining 2,862 men in the study worked less than 12 months in their
 category of maximum arsenic exposure and had an SMR of 286?.

?Significant at the 1% level.

+Cohorts 3, 4, and 5 were combined, since observed and expected deaths
 were small for each cohort alone.

§Significant at the 5% level.

Source:   Lee and Fraumeni (1969).
                                         7-37
-------
dioxide.   Most of the work areas having heavy arsenic exposure were also areas



having medium sulfur  dioxide exposure; conversely, all work areas with heavy



sulfur dioxide exposure  were areas of medium arsenic  exposure.  The authors



did  note,  however,  that persons with  the  heaviest exposure to arsenic and



moderate or  heavy exposure to  sulfur dioxide were  those most  likely to  die of



respiratory cancer in this particular  instance.



     Since the Lee and  Fraumeni  study, additional  research  has been conducted



on the employees of the Anaconda smelter.  Lubin et al. (1981) studied 5403 of



these employees.  These  workers had all  been employed for  12 months or more



between January 1, 1938, and December 31, 1956, and were known to be alive as of



December 31, 1963.   Essentially,  this  cohort was equivalent to the surviving



members from the Lee and Fraumeni cohort.



     Exposure was from  date  hired to  December 31,  1963;  follow-up was from



1964 to 1977.   Classification of exposure categories  was similar  to that  of



Lee and Fraumeni.  However,  unlike the study of Lee and Fraumeni, a cumulative



arsenic exposure index was calculated  for each worker.  This  index was derived



by,  first, weighting the three exposure categories  and then  multiplying the



number of  years  an  individual worked  in a given  category by  this  weight and



summing over categories.   The weights were derived  from  mean airborne dust



concentrations taken  during  1943 to 1958, which averaged 11.3, 0.58 and 0.29



mg As/m3  in  the respective  heavy, medium and light categories.  These  values



differed from those used by  Lee and Fraumeni to group  departments  (Table 7-9).



It should  be noted  that the exposure  estimates used by Lee and Fraumeni were



based upon more  recent monitoring data  primarily  collected in 1965.    Lubin



et al. reduced weights  in the  heavy category  by  a factor  of 10 in order to



account for  the  wearing of  respirators  as was  observed "at  least in  recent



years."  SMRs were calculated  by comparisons to U.S. white  males.
                                     7-38
-------
     The mortality experience  of  workers during the years  1964  to 1977 was
similar to that  of  the workers studied by Lee and Fraumeni during the period
1938-1963.   Excess  deaths  from respiratory cancer corresponded  to areas of
highest arsenic  levels.  The authors noted an overall strong gradient  in risk
associated with  the indices of cumulative  arsenic exposure.   However,  the
authors also  noted  that this  gradient was less  clear  when weighting of the
high-exposure category  was  reduced ten-fold to account for respirator usage.
According to  Welch  et  al.  (1982), however, respirator  usage  was not common
prior to 1964.
     Some study  differences were  noted by Lubin et al.  between their respec-
tive study and  that of  Lee and Fraumeni.  Differences  in  excess  respiratory
cancer—65 percent in the more recent period versus a three-fold excess in the
earlier period—were partially attributed to differences in respiratory cancer
rates  observed  between  the two comparison  populations.   Respiratory cancer
rates  in  the general  populace have increased  in  recent years;  therefore,
comparisons  to  recent  populations  will  produce  lower  relative  risks than
comparisons  to  past general populations  in which the rates of  respiratory
cancer were  lower.   The authors  also  noted that the comparison populations
differed in  composition.  Lee and Fraumeni used white  males  in  the state  of
Montana as their standard population, whereas  Lubin et al.  used U.S. white
males.  As noted in this study,  as  well  as elsewhere (Welch  et al.,  1982;
Higgins et al. ,  1982), death  rates for specific causes  of death  (inclusive of
respiratory  cancer)  have been reported to be lower in Montana.   Finally,  the
authors suggested the  possibility that  individuals  most susceptible to lung
cancer  contracted the  disease in  the  earlier period and were, thus, lost  to
the  study follow-up  due to  death.
      In  looking  at SO;,, exposures, the authors were unable to  totally separate
the  possible interactive effects  of S02 with  arsenic.   However,  they did note

                                     7-39
-------
that after controlling for arsenic, no significant increase in mortality could
be associated with heavy or medium S02 exposures, whereas the association with
arsenic exposure persisted after controlling for Sf^.  This finding is consis-
tent with  that of other researchers studying  smelter populations  (Enterline
and Marsh, 1982; Welch et al., 1982; Higgins et a!., 1982).
     In an update of the earlier study co-authored with Fraumeni, Lee-Feldstein
(1983) observed  the  mortality experience of the  same Anaconda  workers  (with
the exception  of two women) from 1938 to 1977.  The workers (8045) were assigned
to one of five cohorts on the  basis  of  total  years  of  employment.  Cohort 1
worked 25+ years;  cohort  2,  15-24  years;  cohort 3,  10-14 years;  cohort  4, 5-9
years, and cohort 5, 1-4 years.   SMRs were calculated by comparison to the
combined  white male  populations in the states  of Idaho, Wyoming and Montana.
     Of  the  thirteen  specific  causes of  death considered,  tuberculosis,
digestive  and  respiratory cancer,  vascular  lesions of the  CMS,  diseases of the
heart, emphysema,  and cirrhosis of the  liver  showed a  significant excess of
observed  deaths over  that  expected;  however,  only  excesses  in respiratory
cancer showed  a positive gradient with  length of employment when  comparing
cohorts  1 through 5.   The  ratio  of observed  to expected mortality from
respiratory  cancer was  approximately  5.1, 4.5  and 2.3 in the  heavy, medium and
light  arsenic-exposure groups,  respectively.   This was in accord with the
earlier  results of  Lee and  Fraumeni,  except  that  the ratio  of  observed
respiratory  cancer deaths to that expected in the  heavy exposure  category in
the  earlier  study was  7.
     Brown and Chu (1983c)  discussed  the  Lee-Feldstein  (1983) data with regard
to its  implications on the  multistage theory of cancer.   They  indicated that
analysis  of the data  suggests  that arsenic acts as a  late-stage  carcinogen
                                     7-40
-------
because:  (1)  there  was an increasing excess lung cancer mortality risk with



increasing age  at  initial  exposure,  and  (2)  the  excess mortality was  indepen-



dent of time after exposure stopped.



     Higgins et al. (1982) and Welch et al.  (1982) reported on a sample of 1800



workers at the Anaconda smelter.  Compared to the 8047 workers studied by Lee and



Fraumeni, this  cohort  of  1800 workers  included all the workers originally de-



signated in Lee and Fraumeni's heavy exposure category (277) and a random sample



(20 percent) of the remaining known workers.   The date of entry into the study



cohort ranged from 1938 to 1956, providing the individual had one year of work



experience.   Unlike other studies on these workers, smoking histories on the 1800



workers were obtained either by direct questioning or by proxy respondent.   SMRs



were based on  comparisons  to white  males  both in the  state  of Montana and the



United States.



     From industrial  hygiene records for the period 1943 to 1965, estimates of



airborne arsenic concentrations  within 35 smelter departments were provided.



A total of 818 samples were collected from  18  departments and departmental



averages were  calculated  from  these measurements.  The  remaining 17  depart-



ments were estimated  by analogy with those that were known.  The departments



were then grouped  into four categories in which arsenic exposure was charac-


                          3                        33
terized as low (<100 ng/m  ), medium (100-499 yg/m ),  high  (500-4999 (jg/m )  or



very high (>5000 pg/m3).



     Workers were  assigned to these categories based  upon estimations of both



time-weighted  average  (TWA) arsenic exposure levels  and ceiling levels. TWA



values were individually calculated based upon the time that a worker spent in



a given department and the average arsenic  concentration estimated for that



department.   This  quantity was  summed across all  departments the individual



worked in and  was  divided by total  time  worked  to yield a  TWA.  TWA  arsenic
                                    7-41
-------
 exposures  were calculated at entry into the study cohort and at the beginning
 of January,  1964,  corresponding  to  the  end  of  the  follow-up  period  used  by  Lee
 and Fraumeni.  TWA differences between  these two periods  could  have increased,
 decreased, or  remained  the  same  depending on the work  history of  an individual
 worker.   In  contrast,  ceiling levels were  defined as the highest  level  to
 which an employee  was exposed for a period  of  30 days  or  more.  Ceiling  levels
 were calculated  at entry into cohort,  at  the  beginning  of  1964  and at the
 beginning  of 1978.  Unlike a TWA value,  a worker's ceiling  level could only
 increase or  remain the  same from point  of entry into the  cohort.
     Data were analyzed according to five different exposure/follow-up methods
 which varied in  the amount of overlap  allowed between exposure and follow-up
 periods.   Method I, the primary method used  by the authors, included  each
 worker's arsenic exposure up to  the date he entered the cohort.    Follow-up  was
 from entry to  1978; thus, there was  no overlap of the two  periods.  Method
 IV~exposure from  date  hired to  1964, follow-up from 1964 to 1978--also  had no
 overlap.  Methods  II  and V had  complete overlap—exposure from date hired  to
 1964, follow-up  from  1938 to 1964  and  exposure from date hired till termina-
 tion, follow-up  from 1938  to 1978, respectively.   Method  III  had partial
 overlap—exposure  from  date hired  to  1964, follow-up from 1938 to  1978.
 Except where specifically noted, results were given  according  to Method I.
     The results of the study supported the thesis that  exposure to arsenic
was  strongly related  to respiratory cancer mortality  in  workers  at the  Ana-
 conda smelter.   SMRs for the total  cohort for  all causes  of  death were identi-
 cal  when compared  to  either the state  of  Montana or  U.S. white males (both
 SMRs = 133, significant at 0.01  level); however, SMRs  for specific causes were
 somewhat higher  when  compared to Montana males  than when compared to U.S.
 males.   Exposure to arsenic appeared to be the principal  factor  in the observed
                                    7-42
-------
increased risk for respiratory cancer, the study cohort having 3 times the ex-
pected death  rate  for white men living  in Montana.   Exposure  to other occu-
pational  contaminants,  such  as  sulfur dioxide  and  asbestos,  did not  appear to
account for  respiratory cancer  excess, while  smoking  explained only a small
fraction of the excess.
     Of particular note were the differing respiratory cancer  results obtained
under the  two categories of arsenic  exposure  (Tables 7-12 and 7-13). The SMR
for men  in the lowest TWA category  was  elevated,  although  not significantly
so, whereas  the SMRs in the  other  TWA categories were significantly  elevated.
Mortality  for respiratory cancer by  ceiling  arsenic exposure showed  that SMRs
were  only  signficantly elevated in the high  and very high categories, whereas
they  were  close to expectation  in  the two lower exposure categories.  Date of
hire  showed  a  definite relationship to mortality  from  respiratory  cancer.
Workers  who  were employed in the  early  years  of  the smelter operation—from
1884  to  1938, but particularly  prior to 1923, when  a  selective  floatation
process  which markedly improved fume and dust recovery was introduced—had
higher SMRs,  indicating that the overall  higher arsenic  exposures  in  the early
years were associated with higher  death  rates  from respiratory cancer.   Age at
hire  did not seem to have a confounding effect, although the authors did note
that  the SMRs for  respiratory cancer showed  more fluctuation with  age than did
all causes of death.
      Tables  7-14  and 7-15  show comparisons of TWA and  ceiling respiratory
cancer mortality as  analyzed by the  different  exposure/follow-up methods.  The
authors  noted that,  while there was some variation in the SMRs derived  from
the different methods, the  same basic pattern  of  increasing respiratory  cancer
SMRs  with increasing TWA and ceiling arsenic exposures could be seen for each
method.   The authors indicated  that  none of  the exposure/follow-up methods found
 a significantly (P <0.05) increased  respiratory cancer risk for ceiling  level

                                     7-43
-------
  TABLE 7-12.  MORTALITY FOR ALL CAUSES AND RESPIRATORY CANCER FROM 1938 TO 1978
    BY TIME-WEIGHTED AVERAGE (TWA) ARSENIC EXPOSURE AS OF ENTRANCE INTO COHORT
TWA
Arsenic
(ug/m?)
<100
100-
500-
5000-
N
547
542
565
146
Person-
Years
13152
14157
13460
3552
All Causes
Obs Exp SMR
219
216
292
89
196.
178.
184.
56.
7
0
7
1
111*
121**
158**
159**
Respiratory
Cancer
Obs Exp SMR
11
22
29
18
7.
7.
7.
2.
9
3.
7
6
138
303**
375**
704**
 * Significant at 0.05 level
** Significant at 0.01 level

Source:  Higgins et al. (1982).
    TABLE 7-13.  MORTALITY FOR ALL CAUSES AND RESPIRATORY CANCER BY CEILING
                  ARSENIC EXPOSURE AS OF ENTRANCE INTO COHORT
Cei 1 i ng
Arsenic
(ug/m?)
<100
100-
500-
>5000
N
445
276
833
246
Person-
Years
10591
7083
20757
5889
All Causes
Obs Exp SMR
165
80
416
155
152.
80.
288.
94.
1
5
5
4
108
99
144**
164**
Respiratory
Cancer
Obs Exp SMR
8
4
41
27
6.
3.
11.
4.
2
4
8
1
129
116
348**
662**
*  Significant at 0.05 level
** Significant at 0.01 level

Source:  Higgins et al. (1982).
                                    7-44
-------





0
CD
£
- Q.
* CO rH

* -X -X
±. ±, +f
CO (T> 00 CO
to LO CD r**»
CO •>!• tD


CD oo i — i r — •
r^« r^^ P^ CM


LO 00 CD 00
CM CM rH
CM O LO CO
CM OO rH OO
LO LO LO i — 1



* * *
•X * -X
00 CO LO «3-
co o r»» o
rH CO CO r>-



CD co r~ to
r*1** r**^ r*** ^^J





rH CM CD 00
rH CM CM rH





r*. CM in to
^J" ^^ tD *^*
LO LO LO I— 1

1 1
O O O O
o o o o
rH rH LO 0
V LO
All














00
0"i
•""*
O'

>}
S—
•p
C
cu
•p
o
0
u

o
Q.
=>
k
' —
O
LJ_

..;
s-
"c
O)
•p
5-
£
0
o

o
"cu "a!
> > T3
O) CO <1)
- - .«:
LO rH J=
O O

O O -P
re
•P +J T3
re re
+J P 0
c c s-
re re M-
^ .^ cu
.t t 3
C C U5
cn en o
•r- T- Q.
to to X
* £ re^


















.
00
i^»
CD
i— 1
O
•P
>> 00
-P CD
C rH
OJ
O
-P +->
o <*•
^= tD
O CD
U rH
£ E
O O
I- i-
*4~« M~"

Q. Q.
3 3
i k
O O
t— r~~
'o o
LL_ U.

• •
ID U5
CD CD
r~H nH

o o
+J -p
-a -a
Q) CU
s- s-
.E £:

-------




-
J2
O
a
5
u

z
UJ
ce
•c
_i
•—•
UJ
o
I
M
S2
5J
Z
u.
o
a
1
>•
CO


»w

i
ce
UJ
u
I
u
1

s
UJ ,'
ae
.
in

r^

«c
*"*










3s
^
UJ
•a
* M

°

z

CK
3?
ex
X
UJ
2 w
vo a
vo

f""1 z

Ur>
O
ee
a x
1 "
UJ
a.
X
UJ

J3

in
S

z


- i


o.
UJ

«

v>
s


z



o» u*-»
•S *c «
^" Of ***
•r* tf» O»
•f ^ a.
<3 < *^?

^^ f^ C3 ^^
1*. 00 O rH
en vo
 «t I-. 
ro o o in
CM rH

00 CM VO cu
r«. in in o
rH rH 10 CM

« «
u^ on o^ ^j
CM VO
o in oo CM

* CM co in
rH


CO CM rH »T
m oo o r^

CM CM on ro

« «
« *
3 rH CO VO


CM * 00 fH
• • * •
VO CO rH *



• * 5! R!


m vo co vo

5 CM 00 CM


1 1
o g § §

Al


















00
en
rH
3
>,

1
t


o
u

1

§

*o
u.
e
01
4J>
O
JS
8

^ ^ S

> > -a
€t 41 41
r— «— «.
in rH f
22*
4* *3 5
"J 10
11 1
u u

e e u>
vo lo S"
« « UJ
« « .






















CO
r»
S
s
>^

^
c

t
o
u
§
u

a.
*o
u.

3
S

S

•o
£
'j=
3


|

£
U>
O
a.
X



























00

en
rH

O
S
s
§
u
H-
Q.
I
*0
U.

s
s

s

TS
£
JC
«
•o

§
V-

£
3
Wl

UJ
u






















00

s
s
J^

^J
e
41
t
8
S
L.
»-
a.
*o
u.

s
a

s

•o
£
JC
3
••S

|

£
3
|










































^

en
rH

IQ
tf
Ul
c
en
o>
•r-
X

c!


i»
7-46
-------
categories less than  500  (jg/m   of arsenic (Table 7-15).   For the respiratory

cancer SMR analysis by TWA exposure, a significantly (P <0.05) elevated SMR was
                                                           3
not found in the lowest dose category.  In the 100-499 (jg/m  TWA exposure category,

there was a significantly (P <0.05) elevated respiratory cancer SMR (Table 7-14).

The respiratory cancer  SMR  in  this category would not have been significant,
                                                            3
however, if those  respiratory  cancer deaths with <500 (jg/m  ceiling arsenic

exposure had been excluded from the analysis.  The authors concluded that if a
                                                     3
worker in the study had not been exposed to <500 (jg/m  ceiling level, he would

have had little, if any, excess cancer risk.

     A number of  criticisms  have been made of this study (48 FR 1864), a few

of which bear mentioning at this point.  The representativeness of the exposure

estimates as they relate to overall exposure conditions in the smelter and spe-

cifically to the  cohort's TWA  and ceiling  exposures  has been  criticized.  The

authors noted that individual worker exposure estimates,  based upon area measure-

ments rather than  personal  samples, would  not  likely  have  great precision.

Furthermore, estimates of analogy—as in 17 of the departments—weaken the re-

liability of overall exposure estimates, although areas that were thought to be

"problem" areas of high exposure were areas that were generally measured.  The

extent to which  respirator  usage was an effective protective measure is also

unknown.  In interviews with former workers, however, indications were that the

use of respirators was not widespread or regular during the period prior to 1964.

Collectively, the authors stated that exposure estimates probably tended to affect

results such that  exposure values were underestimated  from  1884  to  1938, were

reasonable for 1938 to 1964, and were overestimated for the period after 1964.

     The authors found that workers who had not experienced a ceiling level of
        3
500 pg/m  of arsenic  did not have an elevated risk of lung cancer.  However,

the number  of  lung cancer deaths  that could be  predicted  using  a linear dose-
                                    7-47
-------
response model  (see Section 7.3.1) in the  group  of workers with <500 pg/m



arsenic ceiling  exposure would not be  significantly different  than  the number



of  lung cancer deaths actually found.  This  was  determined using the linear



risk model  (Section 7.3.1) and data obtained from Higgins and made available



by  Consultants  in Epidemiology and Occupational  Health  (1982) on the cross-



tabulation  of  the number of workers in cumulative and ceiling exposure cate-



gories.  Higgins  et al.  are presently undertaking a study of the entire Anaconda



cohort; however,  results of this study will not be forthcoming until late 1984.



     In foreign  smelters,  an excess in lung  cancer mortality  has also been



found. The  Rb'nnskar smelter in Sweden, which  has  been  processing arsenic-rich



ores since the 1920s, has been the subject of several studies.   Axel son et al.



(1978) conducted  a case-control  study  of  mortality  from  respiratory cancer  in



relation to employment at that smelter.   In the parish surrounding the smelter,



369 deaths have been recorded  in the registers for men aged 30-74 years during



the years 1960 to 1976.   Causes  of death  were obtained in all  cases.  Smoking



habits were  obtained  from medical  files.   Cases were defined as subjects who



died of malignant tumors of the lung, other cancers, cardiovascular disease,



cerebrovascular disease,  and cirrhosis of the liver.  The  control  group was



made up of  persons from the same  parish  who  died from causes  other than the



above, excluding  44 persons  with  diabetes,  mental  deficiency and  unclear



diagnoses.   Attempts were made to assess exposure  and four exposure groups



were constructed, based  on intensity and  duration of exposure and time between



initiation  of  exposure  and death.   It  was found that occupational exposure  to



arsenic was associated with a  significant increase in deaths from lung cancer.



For the three exposure groups with at least 3 months of exposure occurring 5 years
                                    7-48
-------
prior to death,  the  lung cancer mortality ratio was  4.6.   In these groups,
there was an  increase with exposure  intensity.  For the fourth group in which
either persons were  not exposed or the first exposure was for a period less
than 3 months  and/or death occurred within 5 years of this first exposure, a
risk of respiratory cancer was not found.
     Other agents, including sulfur dioxide,  did not seem to be associated with
lung cancer in this study.  In 83 percent of the exposed lung cancer cases there
was a history of smoking.  A study of smoking habits by Pershagen (1978) showed
that the excess lung cancer mortality could not be explained by smoking habits.
     Because  a  high lung  cancer mortality  rate was  noted among males in
Saganoseki-machi, Japan, for the period 1967-69, Kuratsune et al.  (1974) did a
case-control study of  lung cancer cases in that town.  The nineteen cases of
lung cancer for the period 1967-69 were compared with nineteen controls randomly
selected from  deaths of  diseases  other than cancer of the  lung,  skin,  or
bladder for the  1967-69 period.  Smoking and drinking habits, residential and
occupational histories, and exposure to atomic bomb radiation were the factors
compared.   Fifty-eight  percent  of  lung cancer cases were found to be  former
smelter workers vs. 15.8 percent in the controls (p <0.01).   The relative risk
was reported to be 9.0  (confidence limits not reported).   No difference was found
between the cases and controls for smoking habits,  residential history, drinking
habits, or atomic bomb  radiation.
     Tokudome and  Kuratsune  (1976) did a cohort study of 2765 male workers,
including 839 copper smelter workers at the metal  refinery in Saganoseki-machi,
Japan.   Deaths  which occurred  between 1949  and 1971 were  analyzed  in the
                                    7-49
-------
study.  The  expected number of deaths  was  calculated using  mortality  data  for
Japanese males.  A significantly increased mortality was noted for lung cancer
(SMR =  1189;  observed = 29; expected = 2.44;  p  <0.01)  and colon  cancer  (SMR =
508; observed = 3; expected = 0.59; p <0.05).  A dose response was demonstrated
between lung cancer mortality and the degree of exposure measured by length of
employment and level of exposure.  A  very high excess mortality from  lung
cancer (SMR = 2500; 10 observed; 0.4 expected; p <0.01) was found among smelter
workers who  had  worked in the  heaviest  exposure  category  and who had  been
employed over  15  years before 1949.  The  latent period  in  this  study ranged
from 13 to 50 years.
     Studies have  also been performed  to see  if there  is any  increase in lung
cancer  among  residents in  areas  surrounding  smelters.   Lyon  et al.   (1977)
compared the  incidence of  lung cancer  and lymphoma in Utah residents  from
1970-1975 in  relation to the distance of  these residents from a smelter and
found no association.   It should be noted that distance from the smelter was
based on address  at the time of  diagnosis;  nothing regarding the length of
time lived near the  smelter was factored into the analysis.   Furthermore,  use
of  lymphoma  cases as controls  is questionable  since lymphomas  may also be
associated with arsenic (Ott et al., 1974).  Thus, the conclusion of  the study
that there was no association between lung cancer and distance from the smelter
is questionable.
     In a  more recent  study,  Rom et al.  (1982) compared lung  cancer with
breast and prostate  cancer in residents living near a non-ferrous smelter in
El Paso, Texas.   The study period ranged from 1944-1973.  Similar to methods
used by Lyon et al. (1977), comparisons were made in relation to distance from
the smelter.   Breast and prostate cancer were chosen as control  cancers because
they have no  known association with arsenic exposure.  The authors reported
                                7-50
-------
that the  distribution  of lung cancer cases  (575)  and control  cancer cases
(1490)  was roughly the same for the different distances studied.  No associa-
tion between  lung cancer and  distance from  the  smelter was found,  nor were
there any associations for race,  age or sex.  However, the authors did note
that they were  unable  to control  for such factors as smoking,  occupation and
migration and were  unable to  obtain environmental exposure measurements over
most of the years studied.
     In Montana, the studies  by Newman et al. (1976)  showed that there was  an
increase  in  the incidence of  lung cancer among  men  in cities where copper
smelters  are  located.  In one of  the  cities  there  was also  an  increased inci-
dence of  lung cancer  among women.  It was not stated to what  extent occupa-
tional  exposure had caused the increase, and there was no control for smoking.
     In the Blot and Fraumeni  study (1975), the cancer mortality in 71 counties
with smelters  and refineries  was  studied  from 1950 to 1969.  Comparisons were
made with the remaining  2,985 counties in 48  states.  In 36  counties with
smelters  processing copper,  lead, or zinc  ores, lung cancer  mortality was
significantly  higher both among males (p  < 0.01) and females  (p  < 0.05).   For
all 36 counties with these  industries the median SMR  was 112 for males and  110
for females.  Although occupational cancers were included in the analyses and,
therefore, contributed to some of the excess  risk,  the number of workers  in
the smelter  industry generally comprised a small fraction of the total popula-
tion  (less  than 1  to  3  percent); thus,  it  was  unlikely  that occupational
exposures alone accounted for the observed  excess mortality  in males  or  the
increased risk for females.   Although this study is suggestive of a lung
cancer  effect in populations surrounding smelters,  it is unknown if the lung
cancer  cases were  even  exposed to arsenic.  Thus, the results,  although sug-
gestive,  are inconclusive.
                                 7-51
-------
      In the Baltimore, Maryland area,  Matanoski  et al. (1976, 1981) studied
 cancer mortality in areas  near the earlier mentioned  pesticide facility.  They
 found an excess  of all  cancers and respiratory  cancers  among  males  in  the area
 nearest to the factory but  not among  females.  Soil  arsenic  levels  generally
 corresponded  to  lung cancer  and all  cancer incidences,  with the highest average
 of  arsenic in the soil (63 ppm) reported in the area with the greatest cancer
 risk.   The authors were unable to account for the differences noted between
 males  and females, but suggested  that  smoking  may have contributed to these
 differences.   Other  environmental  factors and/or occupational exposure may
 have  also influenced cancer mortality  excess.   Further environmental sampling
 of  arsenic  and conducting  of community  surveys were recommended to address these
 unknown factors.
      Pershagen et  al.  (1977) studied cancer deaths in the area  surrounding the
 Rb'nnskar smelter in Sweden.   From  1961 to 1975  there  was a  significant excess
 of  respiratory  cancers in the male  population  (SMR = 250)  compared to resi-
 dents  in an area without  known emissions of arsenic but in  which the same age
 distribution  and  occupational  profiles applied.   When the occupationally
 exposed  cases were excluded, significant increases in respiratory cancers were
 no  longer detected (SMR 173);  however,  the male  population still showed a
 tendency  to excess lung cancers.   No similar occupational group was excluded
 from the comparison population, however.  There was no tendency to an  increase
 in  lung cancer among women.
 7.1.2.2   Cancer of the Skin and Precancerous Skin Lesions—An  elevation  in
the proportionate  mortality  of skin cancer was reported by Hill and  Faning
 (1948)  for  factory workers manufacturing sodium arsenite and  by  Roth  (1957,
 1958) for German vintners.   In addition, an increased incidence of skin cancer
has been reported after long-term oral  exposure to arsenic.
                                7-52
-------
     In Taiwan a large population had long-term exposures to inorganic arsenic
in drinking water.   Exposure  started in 1910-20 when water was obtained from
deep wells,  100-200  m.  below the surface.   Already  in the 1920s, vascular
changes began to  appear,  and in the 1950s, the first epidemiological studies
were conducted.   The  arsenic  content of the water varied from 0.01 to 1.82
mg/1 (Ch'i  and Blackwell,  1968;  Astrup, 1968; Tseng  et  al.,  1968;  Tseng,
1977),  generally  being  0.4 -  0.6  mg/1,  whereas water  from  shallow wells or
other  surface  waters generally  contained from near zero to  0.15 mg As/1.
     Tseng et al.  (1968)  and Tseng  (1977)  have  reported the results from a
large-scale epidemiologic  survey  of arsenic-related diseases in an area with
high arsenic concentrations  in  drinking water.  The population  at risk was
103,514 persons.   Thirty-seven villages  with  a population  of 40,421 were
surveyed  house-to-house.   Examinations were made with special  attention to
pigmentation, hyperkeratosis, and cancer of the skin.   Four  hundred twenty-
eight  cases  of arsenical  skin  cancer were  found,  resulting in a rate  of
10.6/1000.  No cases were under 20 years of age.  The prevalence rate increased
markedly  with  age except for females >70.  Over 10 percent of the people >59
were affected  by  skin cancer.  The  overall male/female ratio was 2.9:1, with
males  having a higher rate in all age groups >29.
     The  villages were  divided into 4 exposure levels:  <0.3, 0.3-0.6, >0.6,
and undetermined,  based on their water arsenic content.  There was a clear cut
ascending gradient of prevalence from low  to  high  in each of 3  age groups
(Table 7-16).
     Hyperpigmentation  (melanosis)  was  found  in 18.4 percent of  the  total
population:  19.2 percent  for males  and  17.6 percent for  females.  Usually the
prevalence was  higher for males  than  females.   The  rates increased steadily
with age  for males and did  likewise for females until a peak was reached at
50-59, followed by a  gradual  decline.

                                7-53
-------
                TABLE 7-16.  PREVALENCE OF SKIN CANCER (per 1000)
                       BY AGE AND ARSENIC EXPOSURE (ppm)
Arsenic content
of drinking water
(ppm)
<.3
0.3 -
>0.6

0.6

20-39
1.3
2.2
11.5
Age
40-59
4.9
32.6
72.0
>60
27.1
106.2
192.0
Source:  Adapted from Tseng et al. (1968).

     The overall prevalence  rate for keratosis was 7.1 percent -7.5 percent
for males  and 6.8  percent for females.   Males  had  higher  rates  in  the greater
than-49-year group.  The prevalence increased for both males and females up to
age 70 and then declined.
     As was  the  case  for skin cancer, the prevalence  rates  for  hyperpigmenta-
tion and keratosis suggested  that positive correlations existed between  these
conditions and the arsenic content of the  water  in the artesian wells; the
greater the arsenic content, the higher the prevalence.
     In the  total  survey of 40,421 people, 7418 cases of hyperpigmentation,
2868 of  keratosis, 428  of skin  cancer,  and 360 of Blackfoot  disease (see
Section 5.2.2)  were found.   Many of these occurred  in combination  in  the
same individual.
     The data were examined by comparing expected (based on overall rates) and
observed rates for various combinations  of the 4  end points.   The obtained
ratios indicated quite strongly that a common underlying cause existed for the
4 conditions, presumably chronic arsenicism.
     A control population of  7500 persons from nonendemic areas was examined
in the same  way  as the arsenic-exposed persons.   Four thousand, nine hundred
                                7-54
-------
and seventy-eight  people  lived on Matsu Island;  its  water supply was  from
shallow wells and  no  arsenic was detectable by the analytic methods  used  in
the main  series.   The remaining portion of control population  members came
from 5 villages  on Taiwan whose water  source was shallow wells with  arsenic
levels between 0.001-0.017  mg/£ (ppm).   No cases of melanosis,  keratosis, or
skin cancer were observed in the entire control  population.
     Although age  at  onset  of the conditions was difficult to  assess, some
information on  latency  was  obtained.  "We  know from the  study that the  young-
est cancer  patient was  24,  the youngest with hyperpigmentation was 3,  and  the
youngest  with  keratoses was  4."    This meant that hyperpigmentation  could
occur  in  patients  who were  exposed  from birth for at  least  3 years,  keratosis
for 4 years, and cancer for 24 years (Tseng et al.,  1968).
     While the Taiwanese data collected in the late 1960's strongly implicated
arsenic  as  the  etiological  source of the  observed  diseases, recent  findings
have  called  to  question the hypothesis  of  arsenicism  as  sole causative source
in the induction of cardiovascular effects.  The discovery by Lu et al. (1977b,
1978)  of fluorescent  compounds identified  as  alkaloids—either  lysergic acid,
dihydrolysergic  acid  or a derivative of ergotamine tartfate—has  opened the
possibility  that other  toxic mechanisms may have been involved.  Ergotamine-
 like  compounds  in  combination with high alkalinity,  characteristic  of these
waters,  have been  shown to  cause gangrene  (Lu et al. , 1977a,b; 1978).   Whether
these compounds have  a  confounding effect  on skin cancer is presently  unknown.
.  .    In  addition,  recent  analyses by Irgolic (1982)  on  a limited number  of
 samples   have shown the water samples  to  contain predominantly  pentavalent
 arsenic  and no  organic  arsenicals.  Samples taken from two wells in  the Yenshei
 Province of Taiwan were collected in plastic cubitainers.   Two samples  each of
 unpreserved water  and water preserved either by addition of 0.1 weight percent
                                 7-55
-------
of ascorbic  acid or by acidification of 0.1 M HN03 were sent to the U.S. for



analysis a few days after collection.  Treatment of samples by HNO- or ascorbic
                                                                  »5


acid was  done immediately  upon collection  (Irgolic, 1983;  personal  communica-



tion).  Samples  were  analyzed during a two-week period  after arrival  in the



U.S., the total time lapse between collection and analyses ranging from one to



three weeks.   The results  of the analyses  can  be  seen in  Table  7-17.  Addi-



tional  water samples collected  from other parts  of  Taiwan  also contained



pentavalent  arsenic.  However,  these samples were less  reliable in that the



collection period was  unknown, and, upon arrival  in  the U.S., these samples



had  a  yellowish  hue with  some flocculated matter present (Irgolic, 1982).



     Several  questions  still  remain to be answered in regard to water-usage



patterns of  the  Taiwanese.   For instance, it is  not  clear how  quickly  the



water drawn  from  the wells was consumed; nor is it clear how much of the water



was consumed in  tea or other  beverages where cooking preparations,  such as



boiling, would have altered the chemical form of the arsenic.   These questions



need to be  answered in light of the possible effect these answers  might have



on interpreting the observed skin cancer incidence.



     Ch'i  and Blackwell  (1968) conducted a case-control  study in the  area of



Taiwan  studied by Tseng et al.  The authors  compared a  variety of factors



between 353  cases of  Blackfoot disease and 353 controls matched for sex and



age.   Socioeconomic status, occupation, cigarette  smoking, diet, and consump-



tion of deep well  water  which was  arsenic-contaminated were the factors



compared.   Two factors  were  found to be significantly different between the



cases and controls.   Significantly  (P  < 0.01) more cases  than controls were



found to consume  deep  well  water and to  have  a lower socioeconomic status.



Though  the author concluded  that the primary contributing cause of Blackfoot



disease was  consumption of deep  well  water,  the socioeconomic  differences



cannot be completely discounted.





                                    7-56
-------


to
o
1— 1
1
r^
Z
t— I
S
UJ
1 __t
UJ
o
UJ
I—
-
Z LU
I I 1 -T-
CO 1—
ae.
 :r

•r- >>
C. &
0)
t/)
i-
^£




E .Q
a. a-
a o

0
HH
Z
UJ
to
a:
•=c


0)
s-
Q.




5T* 1
•M
iH O
. Z
0 3=





^A
U H

-Q O
S-
O T3
U •<-
O) U
^C ^C




Q^
r—
Q.

re
to



(H 00
o o
o o
1 +1 +1
LT> CO
00 00
0 CD


rH pH
O O
0 0

O CD
1 +1 +1
•3- CM
CM CM
0 0
« ,
O O




o o o
o o o
+1 +1 +1
CO CM CO
r>» r«- r~.
0 O 0





CM H rH
O O O
O 0 O
+1 +1 +1
«^- 00 LO
00 00 00
o o o







1 1 +









+ 1 1












1 + 1






HH

r—
^~
0)
3:



CM
o
o
1 1 +1
00
o
iH



rH
0
*
o
1 1 +1

CM
0
,
O




o o o
o o o
+1 +1 +1
m to oo
CT> CO CD
0 O O





SCO CM
O O
O O O
+1 +1 +1
in to r-t
0 rH rH
iH iH iH







1 1 +









+ 1 1












1 + 1





t—*
I— (

I—-
r—
0)
3:




















,
CD
^~
a.
E
re
U)
s_
a.

in 5*>
o •P
•r- (U
-P E
re o
C S-
•1- -p
E U
S- 0)
a> a.
4^ in
01
•a c
0
at T- . '
fli i/j
U/ Vf
t- in
JZ «r—
•P E
0)
•p
in re
re E
a) in
r- re

•P 0) Q.
re 3
cr c
M— *r— O
O C CD
J= S-
c u re
O 0)
•r- +J T3
•P 0)
re c •—
•r— O CX "'^
> -r- 3 CM
CD +-> o oo
•o re o en
4- rH
, ^
C35 C ^
re ai a) u
S- O) > -r-
a» T- r—
> OU -P O
re TU > l— )
D) ± ••
re II  O O 3
< 3C *-i. O
* re JQ to
7-57
-------
     Other  studies  have also explored  the  relationship of arsenic to  skin



cancer  and  various  skin lesions.  Similar chronic  effects  as  seen  in Taiwan



have been reported in other countries.



      In Antofagasta, Chile, a new water supply was put into service in 1958.   In



the 1960s, physicians began noticing dermatological manifestations and even deaths,



especially among children.  In an investigation of skin pigmentation in 27,088



school  children  from  the province of Antofagasta, an overall incidence of 12



percent was  reported  (Borgono and Greiber, 1972).   It was discovered that the



drinking water contained 0.8 mg As/1.  Borgono and Greiber compared 180 inhabi-



tants of Antofagasta with a community without exposure to arsenic via drinking



water.  Abnormal skin pigmentation was reported to be present in 80 percent and



hyperkeratosis in 36  percent of  the  Antofagasta  inhabitants, whereas none was



found in the control group (p <0.05).



     In 1970, a  water treatment  plant was installed and there was a consider-



able drop in arsenic.   According to  Borgono et al.  (1977),  there were no skin



lesions in  children born  since the water treatment  began.   However, it  should



be noted that in this more recent study the sample size of children born since



the water treatment plant began  was small  (306) and no comparison population



was studied.  Therefore, any conclusions associating the lack of dermatological



manifestations with the decrease in arsenic must take into account the small



sample  size.  In regard to skin cancer, the follow-up may not have been long



enough to detect a difference.



     In addition to the work of  Borgono and co-workers, Zaldivar and Guillier



published a series  of papers  on the Antofagasta  situation  (Zaldivar,  1974;



Zaldivar,  1977;  Zaldivar  and  Guillier,  1977).   The first of these (Zaldivar,



1974) describes a study on a total  of 457 patients (208 males and 249 females)



bearing cutaneous lesions  (leukoderma, melanoderma, hyperkeratosis, and squa-



mous cell carcinoma).   The cases were collected  both  by  the author and the





                                    7-58
-------
local  hospital  during the period 1968 to  1971.   Children 0 to  15  years  of age
accounted  for  69.2 percent of the male cases  and 77.5 percent of the female
cases.  The average incidence/100,000 for cases with cutaneous  lesions  in 1968
to 1969 were 145.5 and 168.0 for males and females, respectively.   By 1971 the
incidence  rates  had  dropped to 9.1 and 10.0  for males and females, respec^
tively.  The decline  in  morbidity was so  rapid that a  conclusion  that arsenic
was the cause of the skin lesions would have to be considered suspect.
     The existence of arsenical  waters in an eastern area of the province of
Cordoba, Argentina, has  been known for many decades (Arguello et al. ,  1938;
Bergoglio,  1964).   Effects noted on  the  population  from this  area  include
hyperpigmentation, keratosis,  and  skin and respiratory cancer.  A large area
of the  province,  mainly  in the east  and  somewhat to the  south, is the  focal
point for chronic endemic regional arsenical intoxication (CERAI)  (Arguello et
al., 1938).  This  CERAI  is  due to the ingestion  of well water  coming from  the
uppermost  sheet  of underground watei—the  principal  source of arsenic—as
well as to  the ingestion of arsenic from the wells,  which varies  among wells
throughout  the  region.   The  concentration  also  varies with  rainfall.
Vanadium is also elevated in areas with high arsenic content.   A  later  report
(Bergoglio, 1964)  indicates that progress  had been made  in  improving  the
hygienic condition of the drinking water.
     Arguello  et  al.  (1938) summarized the early history of investigations
(1913 to mid-1938) into  the health outcomes associated with the ingestion of
arsenic-contaminated water.  They also  reported the  results of an investiga-
tion of a  large  series of epitheliomas collected from mid-1932 to September
1938 at the dermatosyphilology clinic of a medical  school  in the arsenic-
contaminated area.  The  series consisted  of 323 cases of epithelioma of which
39 or 12.1 percent were cases with clinical  evidence  of CERAI.
                                    7-59
-------
     Patients exhibiting  CERAI  had characteristic cutaneous lesions,  repre-



sented by  the symmetrical  palmo-plantar  keratoderma  and  melanoderma,  although



the latter symptom was seen less frequently.  Foot and hand keratoderma were



seen in 100 percent of the CERAI patients.  Appearance usually occurred 2 to 3



years after  the  onset of intoxication.  A prodrome  of the keratosis is the



appearance  of an erythema in the  same location,  and a  subjective crawling



sensation and local fever.  Otherwise, it is a state of dryness which precedes



the keratosis.



     Argue!lo et  al.  (1938)  also reported that most patients also had hyper-



hidrosis and abnormalities of pigmentation.   The melanoderma appeared early in



the process  and  was variable among patients.  It was described as small dark



spots ranging in  diameter from 1 to  10  mm.  They had  a  tendency  to coalesce



and appeared predominantly  on  the trunk in the areas not exposed to the sun.



Atropy was  associated with telangiectasia and loss  of color,  or  leukoderma,



between the hyperpigmented areas (the "raindrop" appearance cited by Reynolds,



1901).



     Geographically,  the  largest  proportion  of cases in the clinic came from



the areas  with the highest incidence  of  CERAI.  Because  the authors'  data are



not population-based,  however,  it cannot be  stated that  the incidence of skin



cancer is  significantly  increased in these areas.   The authors reported that



the hands and feet were the locations of choice for the arsenical  epitheliomas



(38.5 percent vs  3.9  percent)  compared to the nonarsenical epitheliomas.   An



example of this  was seen in the  head where  81.6 percent of the nonarsenical



epitheliomas occurred versus 15.4 percent for arsenical epitheliomas.



     Bergoglio (1964)  did a proportionate mortality study of residents of



certain departments (counties)  in the Province of Cordoba, Argentina, where



endemic arsenic  levels in the water supply are reported  to be  very high.  The
                                7-60
-------
proportion of  cancer deaths  was  higher in those departments  than  for the



province as a  whole  (23.84 percent versus 15.3 percent;  P <  0.05).   Of the



cancer  deaths,  respiratory cancer  constituted 35 percent and  skin cancer



2.3 percent.   From the description  by the  author, it can  probably be  inferred



that these data are not age-adjusted.  No comparison is made of the percentage



of respiratory  and skin  cancer deaths in  the  affected departments  with the



respective proportion for Cordoba Province.



     Morton et  al. (1976)  investigated the relationship of skin cancer  morbi-



dity and the ingestion of  arsenic-contaminated drinking water  in Lane County,



Oregon.  The southcentral  region  of Lane County  is underlaid by an arsenic-



rich stratum called the Fisher formation, which is known to produce high arsenic



levels in waters from wells drilled into the land.  An extensive search of the



pathology records of  medical  providers  in Lane County, Oregon, was conducted



in 1972  for the occurrence of  skin cancer during the years  1958  to  1971.



Cases were thoroughly screened  to eliminate .duplications and were then coded



to 1970 census  tract  numbers  according to  the  residential address at  the time



of diagnosis.   Water  samples  were obtained in all census tracts at selected



points in all  municipalities  and  water districts in the  county,  as well  as



from single-family water sources.   The single-family water sources were neither



randomly nor  uniformly distributed  throughout the  county but instead  were



heavily concentrated  in  the regions believed to have water arsenic problems.



Water samples  collected  during  1968 to 1974 were compiled into mean  concen-



trations for each  census  tract and census tract  region.  The  authors  stated



that it seemed  reasonable  to  assume that  the  samples  were representative  of



the earlier time periods as well.



     The skin  cancer  data  were  expressed  in  four  sets  of  rates because  of  the



availability of 4 sets of population estimates.  Overall mean annual incidence
                                7-61
-------
rates for the entire 1958 to 1971 period used mean population estimates based
on all four sets of denominators.   Census tract regions were devised to simplify
analysis and presentation.   Table 7-18 presents the water arsenic level obtained.
As can be readily seen, the South rural area had a much greater arsenic exposure
than any of the  other  sections of the  county.  Figure  7-1 contrasts the parts
of the county underlaid by the Fisher formation,  and subsequent higher arsenic
levels, with the parts of the county  experiencing higher squamous cell skin
cancer.  Relatively little concordance is noted.   A multiple regression analysis
performed by the authors demonstrated essentially no  relation  between skin
cancer and water arsenic.   It should  be noted that  water arsenic levels in
this study varied from 0 to 2150 ppb.   The authors  point out that the Lane
County water arsenic levels were much  lower than those reported for Taiwan and
Antofagasta.   In  particular,  only  5 percent of the  Lane County  samples con-
tained 100 or more  ppb of arsenic in contrast to  48  percent  of the samples  in
that range in the Taiwan data.
     Similar findings  were  reported by Southwick et al.  (1981)  in a  study
conducted on residents of West Mil lard County, Utah.   As  an area in  which
naturally occurring arsenic  in public drinking water  had been reported, West
Millard provided an "excellent opportunity to study the  effects of arsenic
exposure on  a homogeneous,  stable  population  with  a  predominantly  'Mormon
lifestyle"1 (Southwick et al., 1981).  The exposed communities of Hinckley  and
Deseret had average arsenic concentrations of 0.18 mg/1 and 0.27 mg/1, respec-
tively  (based upon  monthly water samples taken between May  1976  to May 1977).
The  control  community of Delta  had average arsenic concentrations of 0.02
mg/1.  All drinking water in the study  communities  came from wells and the
predominant species of arsenic was reported as the  pentavalent  form (85 per-
cent).
                                7-62
-------
00
CO
_i
UJ


LU
O
LU
CO
Di
LU

-
Q.
(0
CO
O
CU
CD






*^ CT> CT> CO CD ^*
r^- CO <^ in rH in










O «* CO CO O O
I^«- <*J" 00 rH ID f*^
«3" ID CO f** CM «^
CD CO 1*- CM rH •=!•
CM iH





rH cn 0 CD Cn l-»
CD CM 00 OO in CO
<*• rH rH




O O
in in r*»
rH <* C5 rH O CO
CM CM 1^ CM rH rH
1 1 1 1 1 1
0 0 0 O 0 rH
V



^.
in ID oo o !"••• r*»
ID f^- rH CO CO <^~
rH rH CO






4J
in
^ x: eu -P
+J -P -P 3 w
t— S- in 3 73 <0
(0 O (0 O T- O
S- Z LU CO S O
3
O£.






i-HCMCOOOlDinrHO
CnCflCTlOOCriCOOOO
rH









omooooom
co co CD co r^- co cn ^"
CMCM^CM^^-CO«d"
LOCOOOCncDOrHCM
COrHrHCMrHCMrHCM
rH




r**- "^d" ^f" CD CD CD 00 LO
CD rH rH CM






0 00
co o m co
OOOOCOI^-OOCOOOOO
1 1 1 1 1 1 1 1
OOOOOrHOrH
V V



1
cocnocoinoocoo
«^-cocoincQcoco«3-

73 73
CU CU
•1- T- O)
4-  cn
•r- -r- C C CU C 3
S- S- CD CD O1 CU LU
Q. Q. cn cn 3 cn
CO CO 3 3 LU 3 CU
C LU LU LU t-
(0 LU 3 SB O
f*t "%£ CO CO rHE «Z 21 CJ
i-
o
                                                                      co
                                                                      o
                                                                      co
                                                                      r^-
                                                                      CD
                                                                      00
                                                                      in
                                                                      in
o
in
rH
CM

o
                                                                      ID


                                                                      CO
                                                                       -p
                                                                       c
                                                                       3
                                                                       o
                                                                       o

                                                                       CD
                                                                       c
                                                                       (0
                                                                              in

                                                                              CD
                                                                              m

                                                                              o
                                                                               cn
        73
        CD
        73

        3
        O



        .c
        u
                                                                               o
                                                                              •p
                                                                               u
                                                                               S-
                                                                              «f—

                                                                              73
                                                                               01
                                                                               -p
                                                                               u
                                                                               73
                                                                               C
                                                                               LO
                                                                               4-
                                                                               O
         cn

        73
                                                                               ro
                                                                               c
                                                                               j=   .
                                                                               -P in
                                                                               in i
                                                                               cu
                                                                               3 O>

                                                                              'm  •
                                                                              > o
                                                                              *c
                                                                                        in
                                                                                        cn
                                                                                        rH
                                                                                         CD
                                                                                         C
                                                                                         O
                                                                                        i-
                                                                                        o
                  cu
                  (J

                  3
                  o
                  CO
                                       7-63
-------
                           -a

                           CO
                            (0


                            o
                            Q)
                            o

                            3
                            (A
                            O

                            2
                            0)

                            D)

                           "5
                            c
                           .2
                           'C
                            0)
                            Q.
                            X
                            0)

                           # o5
                            O U

                           s§

                            (0 h

                            l§

                            0) O
                            U i_



                           li
                            o o

                           •S£
                            (8 ^
                            Q. (A



                           If
                           a> .c

                           3. -9
s
0)

c
o
0)
u

3
O
(O
7-64
-------
     All study participants  from the exposed communities were required to be

five years  of  age or older and to have been residents of Hinckley or Deseret

for at  least the previous five years.  Control participants were selected at
                                                                        i
random  from age   and  sex categories  matched  to the exposed participants.

Control participants  were required to have  lived their entire  lives  either in

Delta or communities where arsenic in drinking water did not exceed the national

standard of 0.05 mg/1.
                                                                         ]
     Physical examinations were conducted to detect signs and symptoms associ-

ated with chronic arsenic poisoning.   A total  of 250 people participated in
                                                                         1
these examinations  (145  exposed,  105 control); not all  of these participated

in each part of  the examination,  however.   No  explanation was  provided by the

authors as to the differences in participation rate for different parts of the

examination.  Urine  and  hair samples  were collected from 94 and 74 percent of

the participants, respectively.   Dermatological examinations were conducted on

249  individuals.    Neurological  and  hematological  examinations  were  also

conducted and are discussed  in  Sections 5.2.1  and  5.2.4,  respectively.   In

addition,  the incidence  of  cancer and vascular diseases  in  the study popu-

lation was compared to other counties in the state of Utah.

     The study  results showed  a clear relationship between the  amount  of

arsenic consumed  in drinking water and the  amount measured in  hair and urine.

Differences between exposed and control  populations were statistically signif-

icant.

     Of the 249   participants examined  for  dermatological  signs of  arsenic

toxicity (palmar and plantar keratosis and hyperkeratosis,  tumors,  diffuse

pigmentation, arterial insufficiency), only 12 showed such signs and no parti-

cipant had  more  than one sign.   The  12 individuals were not clustered among

the more heavily exposed,  and  when the dermatological signs were regressed
                                7-65
-------
against annual arsenic  dose and the log of the dose, no significant associa-



tions were found.



     Age-adjusted cancer  incidence  rates showed Hinckley to  have  a  somewhat



lower cancer  incidence  than Delta.   Cancer death rates, 1956 to 1976, showed



Hinckley to have the  highest rate (138 per  100,000)  when compared  to 42  other



Utah communities.  Table 7-19 shows age-specific death rates for Utah and three



Millard County communities.  (Fillmore  is the County seat;  comparisons to Deseret



were not made due to the community's small population).   Between 1956 and 1976,



14 cancer  deaths were reported  for  Hinckley.  All of these  deaths  occurred  in



individuals 45 years or older and the cancers were types most frequently reported



for Utah:  lung, breast, large intestine, prostate, stomach, leukemia, kidney,



uterus, bone and connective tissue.   Hinckley had generally lower death rates for



cardiac and vascular diseases than did  the control community of Delta.  The authors



noted that no unusual death patterns likely to be associated with arsenic exposure



were seen in Hinckley.



     Certain weaknesses  exist with  this study, most notably, that of a small



study population from which to derive meaningful statistical analyses.  Further-



more, children  and  adults were  not treated  separately.  In regard to sample



size, the  authors felt  that the  small  sample  size was somewhat  compensated  by



the  homogeneity  and  stability  of the  predominantly nonsmoking population.



Data on food consumption were also missing which might have influenced urinary



arsenic  levels.   This,  in turn,  may have resulted in overstating the strength



of the  dose-response  relationship for  urinary arsenic and arsenic  in drinking



water.   It  should also  be  noted that some participants were reported to have



an average daily water consumption greater than 8£, which seems very high, even
                                    7-66
-------





^
1 —
— '
o
o
Q
	 1
_J
I— 1
s:
1 i 1
LLJ
LU
Qi
Qi
0 -
\,
5 CD
^» ~J
5 •>
O
£^ O
0 rH
-
CO
E
^^
•a
c
rO
to
Q.
O
S-
C3

CO
cn














LO
CO




sf
CD
LO
Sf




sf
sf
LO
CM





sf
CM
LO
rH






rH
1
LO




Sf
1
O







to
u
-P
Community
Character!' s

0
CO /-N/-S
co LO cn /— •.
r^ CM CM oo

00 CD CD OO
LO O 00 LO
LO i*** cn o
rH

cn
r** OO r-H ^*^
sf rH rH CO
cn CD cn o
sf- cn oo cn
rH rH i-H OO



LO
CO r-H sf
V^V^V^
cn r*» rH o
rH rH CO



OO
CM x-~s
CM rH
CO 0 CM 0
CM


s~\
sf
CM rH
CO sf CD C5
rH


x*^s
00
LO
rH
Vw*1
r~- o o o









S CO*
O i—
C£ +> r- 4-> 0
LU ro i— i— c
O 4J •!- CO -i-
z co LU Q 3:

CO f~« O O
rH rH OO



/^)
oo
rH O O O


^>^
CO
rH
t
O O O O



LO
CM
S*-l'
r-\ O O O







Qi
g
0 CO >,
to s- co
 co E ro ^:
O 4-> •— 4-> U
i— t ro i— i — c
Q 4-5 «i- CO •!-
oi co LU ca a:

rH OO OO CD

!**• r^s 00 Cn
Cn rH 00 CO
r^ G) ^D r^
i-H rH

rH
LO
i— 1 sf CM
CM 0 sf 0
LO CO OO




rH
O
CM
s~- '
sf O O O




/-^
CO
oo
rH O O O


^r^
OO
CM
LO
O O O O



^
CM
^*-/
rH 0 0 0






o:
*3i
— i
o
co co >,
 0 r—
o co E ro ^i
ce: 4-> i — 4-> u
03 ro i— i— c
LU 4-> -r- CO v-
Q; co LU Q or
LU
0

rH
CO ''"**
CM rH 00 00

CM rH CM LO
O rH r**. Cn
CM 00 
rH
CM
LO
O O O O




c^>
CM
O O O O



OO

o
C3 O O O






0000





CO
1— 1
CO
o
LU
— 1 CO >,
O $~ CO
CO O r—
o co E ro -v
t-t 4-5 r— 4-> O
a: ro i — r— c
LU 4-> T- CO •!-
}— co LU Q n:
. sf OO
I^. rH rH
oo"
CO

OO CM r-H f*-
i — i oo cn r^N
CO OO CM
cn"
LO
rH




CM oo o cn
o cn oo r^
LO CM OO
^
CM
CM



r*- CM sf r-.
O O CM LO
CD CM CM
LO

rH



r^** r^"» *~^ co
i-H sf O O
sf 00 sf r-H
sf
CM



sf r~- CM o
O LO CD sf
O rH CM
CD
rH
r-H




LU
i — i
CO
CO >,
Q- i- CO
:D o •—
o co E ro -^
D£ 4-5 i— 4-5 U
o ro i — i — c
4-> T- CO •!-
LU co LU o. o:
C3


-—
c
ro

CO
to
ro
CO
to •
•i- Q.
•O 3
i. i-
ro en
*3 CO
u cn
to ro
ro
> CO
0 -C
i- 4->
-Q
CO <4-
S- 0
CD
O CO
N
f\ i i r\
VL' VI
to
ro co
co cn
to ro
T3 O)
i- ro
ro
*3 JC
U 4J
in
> c
O ro
**"*
W CO
ro 4->
u ro
CO
-TJ
U O
ro to
U i-
O)
S- J3
£ 1
c
LO
r-. cn
cn c
rH T-
i ^
co o
LO -C
cn to
rH
E to
O T-
i- to
4- 0
Death rates
arteriosclel
































.
rH
00
cn
, — |


^
ro

+j
CO
^
o
'5
.e

o
CO
QJ
O
3
O
CO
7-67
-------
taking into account  the  hot summers.  In reporting  on the lack of arsenic-
related effects  in this  study population compared to  others  (Tseng et al.,
1968; Borgono and Greiber, 1972; Borgono et al. ,  1977; Zalclivar, 1977; Zaldivav
and Guiller,  1977),  the  authors did  note that populations  in Taiwan and Anto-
fagasta were exposed to considerably higher concentrations of arsenic in their
drinking water.
     Interestingly, the importance attached to the fact that no adverse effects
were  seen  in this group of individuals exposed to  drinking  water arsenic
levels four times  that allowed by the Interim Primary Drinking Water Regula-
tions (0.05 mg  As/1) is  somewhat compromised by the very  characteristics  of
the population  that  make it useful for epidemiological study.   The fact that
this  study  population is  so homogeneous and stable  and,  therefore,  lends
itself to a  relatively well controlled statistical  analysis,  also makes it
less  useful in  terms of  being  representative  of the  overall population of  the
U.S.  For this reason, any generalizations that might be drawn from this study
population are subject to limitations.
     In addition to  exposures  via drinking water, food exposures  and thera-
peutic exposures have also caused skin lesions and cancers.  Hyperpigmentation
and depigmentation of the  skin were  found to  be common among the survivors of
the  Morinaga  milk poisoning  in 1955.  A follow-up  study  conducted on the
exposed children when they were 17-20 years  of age  showed the  prevalence  of
lesions to be 15 percent (Yamashita  et  al. ,  1972).   It is not known if uny
skin cancers have developed.
     Hutchinson (1888) first reported on a possible association of skin cancer
with  the  use  of arsenical medicines.  Prior to that, the association between
these arsenical  agents and  keratotic  lesions had been recognized.   In  his
classic paper,  Hutchinson  reported  on 6 patients with  case histories who
                                7-68
-------
exhibited the keratotic lesions associated with arsenical  poisoning.   He  felt

that the clinical series supported two principal  conclusions:
          Prolonged internal use of arsenic may seriously affect the
          nutrition of the  skin  and that use may produce warty  or
          corn-like indurations.

          Continued use  of arsenic may  result  in  a tendency  for
          these "arsenic  corns"  to grow downward and  pass into
          epithelial cancer.
     Neubauer (1947)  later compiled a series of reports on 143 cases of medi-

cinal arsenical epitheliomas.   He  excluded five categories of cases reported

in the  literature  to  keep the series consistent with regard to diagnosis of

case and  history of arsenical use.  Seventy-one percent of the patients being

treated with arsenical  medicine  were  patients suffering from skin  diseases,

especially psoriasis  (54  percent).   In  contrast,  only a small percentage of

the cases reported came from patients treated with arsenic for various internal

disorders.  In  nearly all  cases the drugs used were inorganic and almost all

were trivalent.  The  most commonly used arsenical was potassium arsenite.  No

externally applied arsenic-related skin cancer case was reported in the series.

     The  elapsed time from  the beginning of administration of the  arsenical

drug to the beginning of the epitheliomatous growth was variable,  but averaged

18 years  regardless  of the  type of lesion.  In cases with  keratosis, the

latent  period to  the  onset of keratosis  was about  half the latent  period to

the  onset of  the  epithelioma, i.e., about  9 years.  Thirty-three percent of

the  patients were  40  or younger.  Of the 143  patients,  13 had or developed

miscellaneous cancers at  other  sites, but such  cases were  not reported syste-

matically.

     Fierz (1965)  reported a follow-up study of patients treated with  arsenic

by a private practitioner.  An accurate assessment of the total arsenic intake
                                7-69
-------
 in  terms of the amount of Fowler's Solution administered was available to the



 investigator from patient records.  The  follow-up  examination was conducted



 under  the auspices of a  local  polyclinic.   Fourteen hundred  fifty  patients



 were  identified as having  received  arsenic treatment,  and  invitations were



 mailed  for them to come for a free follow-up medical examination.   During the



 period  March 1963 to April  1964, 262 patients presented themselves for exami-



 nation.   Two hundred  eighty patients were not located while 100 patients



 actively refused to participate.  Only patients  under  65 years of  age were



 invited to participate in the  study.   The author admits that the patients



 reporting  for examination were not a  representative  sample.   In fact,  he



 categorizes  them into three groups which  range the  spectrum of likely biases.



 There were patients satisfied with the results of  the  arsenic treatment  and



 wishing to express thanks,  patients in  whom disturbing side effects were



 occurring,  and  finally patients who were still suffering  from the  initial



 disease and who were eager to get  a consultation.



     Arsenic treatment was prescribed for individuals suffering primarily from



 three main skin  diseases:   psoriasis  (64  patients),  neurodermatitis  (62),  and



 chronic  eczema (72).  In addition, treatment was also prescribed for 64 patients



 suffering  from assorted skin diseases other than those listed  above.



     Fierz noted  that the arsenic treatment  showed  good success.  Of the 64



 cases of psoriasis,  55 reported a favorable  effect  while taking the drops.



 Forty-eight of 62  patients  with neurodermatitis reported a favorable effect.



This effectiveness  was the  cause  for the  patients' reliance  on  the drug.



     Upon  examination,  106   of  262 patients (40.4 percent)  reported  hyper-



 keratoses, although frequently  a detailed examination was  necessary to find



the changes.   Hyperkeratoses were round, superficially verrucose papules,  1 to



3 mm in diameter.  The number and the exact presentation of the hyperkeratoses



varied from case to case.





                                7-70
-------
     In the series, 21 cases of skin cancer were found comprising 8 percent of



the total  subjects.   As  was in the case of hyperkeratosis, some variability



was observed  in the  expression  of the skin cancers.   Multiple  basal  cell



carcinoma  was  the most frequent  histologic type  observed,  occurring in 48



percent of the  cancer patients.   The basal cell carcinomas appeared morpho-



logically  as polycyclic,  sharply  bounded erythemas with slight infiltration.



In 13 of the 21 cancer cases, multiple carcinomas were observed,  a much higher



proportion than had been observed with other causes of skin cancer.  Of the 21



patients with carcinomas,  16 showed distinctly developed arsenic warts on  the



palms and soles simultaneously with the skin tumors.



     Both  hyperkeratosis  and skin cancer were found to vary with  increasing



arsenic intake.   Above 400 ml  of Fowler's solution, more than 50  percent  of



the patients studied showed hyperkeratosis.  As the dose increased there was a



corresponding increase in the simultaneous occurrence of hyperkeratosis of the



palms and  soles.   Only 2/15 cases of  hyperkeratosis with intakes of up to  250



ml had a simultaneous presentation, while 16/27 had a simultaneous presentation



with intake of 250 to 500 ml.             '



     As with the hyperkeratosis,   skin cancer increased with increasing arsenic



doses.  Below  500 ml  only basal  cell carcinomas were  found;  spindle cell



carcinomas were only found above  that dose.  The latency period for hyperkera-



tosis was  at a  minimum 2.5 years; skin cancer, however, had a minimum  latency



period of 6 years with a mean latency period of 14 years.



     Cuzick et  al.  (1982) conducted a mortality analysis of a cohort  of 478



patients given  Fowler's  solution  to determine if persons taking the solution



were  at an increased risk of internal malignancies.  The cohort consisted of



213 males  (45 percent) and 265 females (55 percent) who took Fowler's solution



for various  lenghts  of time ranging from 2 weeks to 12 years (mean duration,
                                7-71
-------
 8.92  months) during the period  of 1945-1969.   Follow-up was  until January 1,
 1980;  the mean  follow-up  time for the  entire  cohort was  20.3 years.  The
 median  cumulative dose of arsenic  was  reported as  448  mg, with most  patients
 consuming  arsenic at a rate  of about  250 mg/month.
     Within  the  defined risk period,  139 patients died, 265 patients  were known
 to be alive  and  at risk, 29  had  passed  85 years of  age, 19 had emigrated, and 26
 could not  be traced.   Because of the  unreliability  of the expected values in the
 open-ended group beyond 85 years of age,  deaths and person-years at risk after
 age 85 were  ignored.   Persons who  emigrated were censored as  of the date of emi-
 gration.
     A  subset  of 142 of the cohort were physically examined  during  1969-70,
 and the  presence of arsenical keratoses, hyperpigmentation,  and  skin cancer
was recorded.  No  signs of internal malignancy  were clinically apparent at the
time  of examination.    This  subset was  chosen  because the individuals were
 known to be  alive  and were able to  be traced.
     The  results showed that risk ratios were similar for  both  males and
females.  No  signs  of  internal malignancy were clinically apparent  in the
patients given physical examinations, and mortality was not found to be signi-
ficantly increased for any internal malignancies in  the cohort, although it
was somewhat elevated  for  death  from  bladder  cancer (observed = 3, expected =
1.19,  P =  0.12).  The ratio of observed-to-expected  deaths from all neoplasms,
from cancer  of  the digestive  organs,  from cancer of  the respiratory  organs,
and from all  causes was analyzed by cumulative  dose level.   No trend by cumula-
tive dose  level  was found;  however, the risk ratios were low for deaths from
all neoplasms, digestive cancers,  and respiratory cancers  in the lowest dose
group, with  the  risks  for deaths  from  all  neoplasms being  significantly
(p <0.05)  lower  than expected.  A significant positive trend in the ratio of
                                7-72
-------
observed-to-expected mortality for  all  neoplasms was found for  dose  in the
period 5-9 years  from  first exposure, but this  trend was not found in  other
time intervals since first exposure.
     Of the  subset  of 142 patients  examined  for  skin manifestation of arseni-
cism, their overall  mortality was consistent with the remainder of the cohort.
Within this group, 45 percent had keratosis,  14 percent had hyperpigmentation,
and  11 percent  were found to have skin cancer.  The authors did not indicate
what the  expected proportions  of patients with  these  signs  would be.   Some
patients showed 2 or all 3 signs of arsenicism, and 49 percent showed at least
one of the signs.  The appearance of signs was dose- and age-related.   Patients
with signs  had  higher median doses  (672 mg)  than those without any signs (448
mg), and this was significant (P <0.001) by the Wilcoxon rank-sum test.   Forty
percent of  patients  with signs had cumulative doses  above  1000  mg, whereas
only 22 percent of  the patients without arsenical signs had cumulative doses
above 1000 mg.  On the average, patients with signs were 7 years older and had
received  their  first exposure 6  calendar  years  earlier than those without
signs.  Within  the  group with exposures > 1000  mg,  those with  signs were,
again, an  average of 7  years older, and they had their  first exposure 3 years
earlier.  It  is  interesting  to note that all  seven of the subsequent deaths
from internal malignancy in this subgroup had been identified as showing signs
of  arsenicism.   No  deaths from cancer occurred  in the group without physical
signs.
     In summary,  this study provided little evidence that ingestion of arsenic
is  related to a risk of mortality from internal  malignancies.  Non-fatal skin
cancer was found, however, to be related as a function of dose to ingestion of
Fowler's  solution.   The  authors indicated that perhaps persons who demonstrate
signs of  arsenicism following arsenic exposure retain arsenic longer than those
                                7-73
-------
who do not show signs, which may suggest that persons who demonstrate signs of



arsenicism are  at a greater risk of internal  malignancies.  This  was  suggested



by the fact that  all of the deaths  from internal malignancies in the subset of



individuals that  was examined for signs  of arsenicism occurred  in  individuals



who demonstrated  such signs.



     Knoth (1966) also  reported on two patients treated with arsenical medi-



cinals and who developed skin cancer.  Roth (1958) reported that of 47 autopsy



cases  of vintners  with  chronic arsenic intoxication,  13  had skin tumors.



     Gilbert  et  al.  (1983) conducted a study on wood treaters  in  Hawaii who



were occupationally exposed to  arsenic  (see Section  7.1.2.1 for  complete dis-



cussion  of study).   The  authors found no evidence  of risk  of skin  cancer,  but



inadequacies  in  the study design and the  small sample size limit the ability



to draw  valid conclusions from  the  data.



7.1.2.3   Other  Cancers—Whereas there are  many studies suggesting that there



is an  association between  inhalation  exposure and  respiratory cancer and oral



exposure and skin cancer, respectively, there are no consistent data with regai



to cancer in internal organs.



     Reymann  et  al.  (1978)  have investigated the relationship  between the



intake of arsenic for medicinal purposes  and subsequent internal  neoplasms.



Study  subjects  were identified  by examining the files  of a dermatology clinic



in Denmark for  the years 1930  to  1939.  Two  rosters of study subjects were



generated:  1) persons treated  with arsenic for multiple basal cell carcinoma,



Bowen's  disease,  psoriasis,  verruca planus, and lichen planus,  and 2) persons



with  keratoses  from  the  first roster plus 30 other persons identified as



having keratoses.   The  first roster of 413 persons comprised the basic study



population; 24 persons were excluded for various reasons, resulting in a final



study population  of  389 persons.  The malignancy histories of both populations
                                7-74
-------
were traced through  the  years 1943-1974 in the Danish Cancer  Registry.  For
each person  a length-of-observation period was  determined and an expected
number of  internal  malignant neoplasms was determined using the former data.
     In the main  study population, 41 cases were  observed versus 44.6 cases
expected.   Therefore,  no increased  incidence  of  internal  malignancies was
noted in the  arsenic-treated patients.  However,  examination  of the  data  by
individual skin disease  categories showed  that women  with multiple basal cell
carcinoma  had  a  significantly higher  incidence of internal cancer.   The same
trend, although  not significant, was  observed in  female  patients exhibiting
verruca planus.   Next, an analysis was made of a possible dose-response rela-
tionship using duration  of treatment as the exposure  variable.  The categories
were low,  medium  and high, based on <1 month, 1-3  month,  and >3 month duration
of  standard  dose  administration (6-8 mg As203).  No  relationship between dose
and internal  organ cancer was  noted for  the total population or by  sex.   In
addition,  the form  of  arsenic administered did not seem to affect the incidence
of  internal  cancer, nor was  any effect noted  due  to period of observation.
Reymann et al. did  not provide  a definition of the term "internal cancer."  It
is  presumed that  the authors  included  lung cancer  in  their definition.  If so,
the authors  did  not provide any  adjustment  for smoking.  Regardless, the
sample  size  of 389 persons  is  probably too small  to detect an excess in the
incidence  of  internal  cancer, even if  lung cancers were excluded.
     The  keratosis group was not  analyzed in  the  same way.  Eight of the  19
men with   keratosis died before 1974.  Four of these  were due to cancer of  the
 internal  organs.   Expected deaths  were 1.9.  Similarly, 5 deaths due  to cancer
of the internal   organs  were  observed  in  the 34 women with keratosis  compared
with 3.3  expected.   Together there were 9  deaths due  to cancer of the internal
 organs  compared  to 5.2 expected.   Although the figure is  almost doubled, it is
                                 7-75
-------
still  not  significant.   On the average, the patients with keratosis received



higher doses of arsenic than the other patients.



     Hemangioendothelioma  and  reticulosarcoma  of the liver has been reported



by Roth (1958) to occur among German vintners exposed occupationally to arsenic.



The  same  type of malignancy occurrence  has  also been reported in  isolated
                                                                      s~

cases  by  other authors (Pershagen and Vahter,  1979).   Higgins  et al.  (1982)



reported increased  mortality for  cirrhosis of  the  liver and  urinary cancer in


workers at the Anaconda Montana smelter.



     More recently,  Falk  and  coworkers (1981a) described a nationwide (U.S.)



review of deaths from hepatic angiosarcoma during the years 1964-1974.  Of 168


confirmed cases  included  for survey, a group of seven cases with a  history of



prolonged use  of  Fowler's solution was found.   Taken with the  data of Roth,


chronic arsenic exposure,  mainly  via ingestion,  appears  to be associated with


this very rare liver cancer.



     In a review  of four cases of hepatic angiosarcoma,  Falk et al.  (1981b)


determined that  one case, that of a  child,  involved environmental arsenic



exposure in  a  copper-smelting  community  in Arizona.  Since this child had a



history of pica,  the usual exposure routes were augmented by further arsenic


intake.



     Roat et al.  (1982)  reported a case  of hepatic angiosarcoma in a subject


who, 33 years earlier, had ingested Fowler's solution for the relatively short



period of 6  months.   The  pre-existence of skin  cancer  in this  subject, from



age 25, of the type associated with arsenic exposure, was further support of an



arsenic association.



     Knoth (1966) reported on two patients who had been treated with arsenical



medicinals and developed  tumors at sites other  than  the  skin or lung.  One of



the patients was  a  61-year old woman who had been treated with an  arsenical
                                7-76
-------
medicinal and  developed mammary  carcinoma.   The other case was  that  of a
53-year old male who had been treated with an arsenical medicinal  for cirrhosis
vulgaris and developed a reticulosarcoma of the glans penis.
     In the studies by Ott et al.  (1974), an increased proportionate mortality
due to  malignant  neoplasms of lymphatic and hematopoietic tissues was found.
Axel son  et  al.  (1978)  found an increased  risk of leukemia and myeloma in a
case-control study of workers exposed at a smelter.
7.2  ANIMAL STUDIES
    .While  arsenic  carcinogenicity  in test animals  has  not been  observed  in
most studies, some recent  reports have noted positive  results.  The literature
on  experimental  inorganic  arsenic carcinogenesis, as summarized in  Table 7-20
and as  reviewed by scientific bodies (MAS, 1977;  NIOSH, 1975;  IARC, 1973 and
1980) and individuals (Sunderman, 1976; Wildenberg,  1978; Pershagen and Vahter,
1979) supports  this conclusion.
     In light of the presently recognized anomalous metabolizing  of inorganic
arsenic by  rats,  studies  which used  rats  as  the experimental  subjects should
be  viewed  cautiously.   Studies on other animal models have generally resulted
in  negative findings.   A  few  of  these  studies are discussed below, but most
have been summarized  in Table  7-20.
     In a  study by Baroni  et al.  (1963), Swiss mice were given either arsenic
trioxide dissolved in  drinking water  (concentration of 100 mg/£) ad libitum
for the duration of the 70-week  experiment;  or, sodium arsenate  in a concen-
tration of 15.8 gm/£ in a 2.5 percent  solution of Tween 60 in water,  applied
twice  weekly for the duration of the experiment.   Each  compound was tested
 alone,  in  combination with skin applications of croton oil  (to test for initi-
 ating   action), and after  initiation  with a single  skin application of 7,12-
 dimethylbenz(a)anthracene  or with  administration of urethan by  stomach tube
                                 7-77
-------









I_

os



o
u
s
t.
in ai
l-H «t-
CO 01
uj o:
UJ
0
CC
g
K
U in
l-H +»
W *3
co in
£
w^
UJ
H- t
Q

V> #O
| 1
UJ
£
H-l
UI
X
UJ
u.
0 ^
UJ g
CO O
1 I

in
?
c
LU *r-

1 I
v^

.
01 01

•"SloP
f— Ol O
u c
•O +J CO
oi x: •
e t- +J r—
Ol O 0) O
£ c . +3
o ro o- c
-c o. ro o
co ui>^ o



tM
CO c-H
I- -
Ol O)
Q. C
3= OL.






t

IA

o
0>
3
t~ or
CO >* O

|2 o a







i1)*
o

<





us
|H
; ca
to
u
u

'£


15

o














~
'ra
+>
ai

c
o m
ra ai






i

t.
01

S

Ol
c
c
T
a







pj»
o

in






in

CO
S

s:


ra
i.
o
+> 01
o i ro is u
c c c c c
C3T- O r—. 10 Ol -a •
• co+* ina£^^ coo) -Ora^*
O)o>u*>•!— oi a.
I*-U> T3O r-HHOO -r-O  i— 3
•r- O 01 Ol Q. CO CT1 >T3 -r-CT3o
i— T3OCO) > • >> r-l i- i— T-OICS-


o)ouCa)o &.+J*oro oioi £-(no-oo>
a> "~ 'C -p a) <-> to a) s « 01 o I=o3">ro
i-e"'c-a'e.ro !_+>•!- 1-^ 3'^ jrs^oiS
coincojauxi o. -i- fc. ^_- aero— I £csicoro
ai
r* c ai
u> >> 10
T3rH CI*.-O O- r^OrH.
CCO rOOiCrH r— Cr-
to - to rHro^f ~O a ro - 10
t- en s. oi c s-
roairH oi * r- rH ra +J raai-t->
S T3 N i— r— O 3 13 0)
roai* 3Olrot- t- ro 0)
U1OJ= rorOfOI Ol C inoin
•i- fc. 4J S- E S- i— D.OJ or*. -i- 1- 01 «i-
roue u ui ro -^ 3O1 >>a> ro u s. o>
^£:i£ COI-4U.£ aZrH COrH^dCO^tH





**™N
1 Ci II 1 1 1 1 1 I 1
*
U 1 t. S-
a> 3 ai -o ^-> 0)

ro I O ro *r- (Sj ro
3: s- t- ui 3 M- u>rH X
OO -i- t-3v_/
ai ui u) 01 u aio +> 01 <-> j-> 4j 4J
•^ xb'i?"!- T- Q. CO 3 "o •!- -r- .^-r-.J!.?!
j<£ s-a> -^ in 3 cr c a -^ aaaa
c aiaii — cc roro c
•r- 3 +J » O -r- +J O. f -r-
5- s- c o T- t. o ro s- *J i.
a a 1-1 u- +J a z I— oai a

2
O
LU

N-tt* N cj-

W • M*OO CO OlOd'O^O^*
O TO' O Vt V> O O U) O OO(AO

ro IA to to • to •  *^- CO  i- to
V> 3E "N, £ O O) -P
CO O 0) W O *'-
a>« ^»S S Si2-

3C Z 0>(A ui ui ui ui
"r" fQ (T) <0 1) fl
Z OS C£ OC Ctf OS

*nj IB 'ro IQ <0 ro to to
t_ t_ £.£. S- £_&.£.
OO OO O OOO
7-78
-------















^
•o
o>

•g
o
u
o
C>J
1
UJ
m
































Remarks


0)
u
e
Referei
in
3
in
oe




01
u
1





•o

^
o

o




c
"re
£«




> ••-
*fc *> 01
IB &. a.
Weight loss, e
mortality, sho
duration of ex|
ment.

J"
re
o5
c
25
£3

1






+>
0*
5




^
o

NO
o w

11"











o








2
o

.C O
O) 1-
s. c
o
Reported in an
abstract form,
incidence in c
-
re
*J
3
10
t- r-.
O 01
CO i-H

i



^
o
in
2
re
3
o
I







34*
O
VI
1


^J
0)

•f—
u
O)
CL
U)

-p
£
0)
I

c
0
"I—

re
"re
c
>->
Ol

* *>
§re c
o> re
*> t- o
0) t-
C C •!-
O'r- C
"re
0)
•r—
f
C
U) 01
»— I r- 1

1 1 1




•c
0)
£
Q.
in
I


0)
O u in
t-S« c -a
n«i B a>

N O. • ft* 3
V) O CO < i —


^^ ««?




ai
^
>M»
0)



in
5
in
re
ee.
i— C
re o
15
u re
re •—
(. r—

t. U)
+> c
C 'P*
fe)
ai ui
t- *J
One tumor obse
in treated, no
nificant.

01
•p- UJ
in 0t
•i- rH
.C
in i-^
t-i re

i






c
in o
3 ••-
o •*>
Ol 3
3 r—
O" O

in
5
«»/
06
r- C
re o
O) *i—
u re
re i—
t- >—

*> c
c ••-
H-l
C
01
at in
A C-
re
Results cannot
attributed to
alone.

01
U 01
> Ol
5-"
c -
re •
> i—
i-i re

•f-






in -a
X C C/-1
z> a> -i- to N •
re t- re ^-N
ai 3 *> vo:
•o+j coo
i. X O CO *~>
o -i- u 3 re
GO E *-*O O




N •
»
O
in

«-_.
co-
re
o











i«^
a
CO
^
5
t— C
re o
QJ 'r*
f *>
u re
re t—
t- i—
re *J

C -r-
1— 1
C/-i
re r-i in
1 .CO >
••- •*> •
**<• o ro in
•i- t- i— .
e ai v c re u>
cn+J •>- E i—
•f- 10 CL ••- O
in 01 N^ u> c t.
t. re re *>
in o) in E c
re f— o *o o
e >, o c ai u
§•— t- •.- *J
*> *» o re c
ai c c t- ai -f-
T3 re o re s-
«K U U O 
U Q
re i—
t. r—
+J ••-
re *»
t. in
•u c
c ••-
i— <

i
•r- | C
t- HI O
3 •D in
4j re
re 
t
3
«
5
CL
in
0
f








T5--
0
*





u>
t-
0)

w

-------




























TO
J3
C
•r-
.p
C
O
u



o
CM
1
tu



t—


































•if
re
e
ce






0)
S
2


O

•K
in
*2
VI



01



.c
Q)










1
3
0


C
re
i.
v>
\MI*

in
01

u
J-





S
3
O
oe



i
1 CO O 1 -P .
"a (-2 c c Hp
c a. i to   exit—
r— *^- Q) 3 C "i~ 4«> CJ
JO I*- 0) •.- >
•r- 01 fi- « fi- in S 0»
U) 0) -P re O) r-
in our c a. u (-
oc4->aia.ca>u>
a.>i-otnrea>u«£





1
re
O)


•C Ol
O rH


+

T3
.*
1^.
•r—
U
a)
o.
in

jJ
o
z



c re o
0
U in in
 -p
0( -I- O C
t- re 3 o
o +J -P o












a

r- C

us
u re


ro *»


C -r-
l-l




fCiH.2
m o re
rH O >

O 01
s- -p in
0) J3
3 T3 O
re o>
^2 i — in
O) re £
Z 11- -P
CM


t-H

re >;
U 10
•P c
•r- C
01 01



,








C
re


LU






T3
c
re
>N -PJ
0 0
in -M
T3
0)
•1—
1^
0)
Q.
in

4J
O
c;

8

,^

4J 4J
c re
•r- U
re -r-


C Q*


IO
O
03
Ol
rH

^ CO
TJ
01 •
J-5
> **^
0)
S- O
yi 
J C OJ
•N O O U
0 r- +J (TJ
in r— o
Si ° *~ c












0}
U
*
^

'•P *>
c re

Q.r-
cx
C 0.
•r- re

to
o



>. -
CM

0) •
O) O
o>
t C_3
in ^c

m
to
en
rH

~"
i
^
3
O
CQ


1



o
C
m


03



00







o -a
U) (I)
S|
55




i
re **"i
r- 0)
^ in
U O
O a.
& sT

«_?•
'£ 15
01 'e
c o


•>- o
re v-


c a.
•r- re

10
                                        c
                                        O ro
                                        (. ID
                                        re o>
                                       CO rH
                                                              T3 -P
                                                              i— fi-
                                                              re ai
                         p c
                         o 01
                         z 01
                         S
                         re
i—         .a co
'r-
o      < -a  c


o cj   o  o  -N c
P N      r- O  re
O c   ^M—  in .c
i- a>   CM  o <; -p
u .a   ^>»t- :2  ai
                                re
                                c
                                o


                                O ^>TJ
                                E -P  at


                                Q.T-  a!
                                   •p  in
                                O O  J3
                                z re  o
 c
 o    o
 u  Diva


 0)  C  Cli
•P T-  01
 re  re  *
3 •
 in    »r-
< T3 O     CQ
 ro 3 c z
Z O OQ

/•^r— O TJ
CM O fi. C
^-"f- o re
                                               rere
                                                               o'o*
                                                               ui tn
                                                               
                        c re
                                       ?g
                                      .S  S
                                       re f-
                                                o  c
                                                ai  o

                                                re '-P
                                               •p  u
                                                3  01
                                                o •<->
                                               £>  C
7-80
-------
             o>  •

             o> "o
               ce
             in •<
            <£ HI
                met.
                o  o o

               "5>£ Q.
                O  CL
               .	 •!- in
                O  5- S-
               +J  O O
                ut  in S
                                                
 C       01
•I- Ul    3
1- *>    O
1-0)   r—
                                 c
                               3 O
                               C-
                               0)

                               I
                                                                           re
                                                                                u
                                                                                to
                                                                                              a>
        s.
              •a
               £
 o   -00

£    £.5
A   .a .a

"re    s "re
      o
 Ol   *O  4»
                                                     c
                                                     ai
                                                     t-
                                                    •c
                                       I
                                       I
                                                     •P
                                                     re
                                                     a:
                                                            .g

                                                        -O ••—  3    U      -c: •»- i_
                                                        co       -r-  N'O +J o a)
                                                        re    s.    i—o  v       u

                                                         *£°    -E-So'^iS
                                                        O-P<    res2^--r-+^-
                                                         N-O CO    U)   r— ^t O
                                                        infi-£    &.£-oin£.
                                                        
                                                                           u
                                                                           X
                                                                                                      3  3=
                       3
                       O
                       a>
                       c •
                I r—   3 01
                i a.   o 'r-»
                i.g   -§.5
re -P re  c
+j u c  i
3 o) i +J
u '"^ 01  in
jO c t-  o
3 •!- a. o.
CO   >—'
                                     3  C
                                     u  o
                                     Wl •(-
                                     3 •!->
                                     e  u
                                     re  ai
3  c
u  o
in «r-

 >•«    "re

        o
                                                                 -
                                                              E U
                                                              re oi
                                                                    2
                                                                    o
                                                                                     +

                                                                                     x^
                                                                                     *
                                                7-81
-------
 (to test for promoting  action).   All  tests failed to  show  any carcinogenic
 activity of the two compounds under the given  experimental conditions.
      Kanisawa and  Schroeder  (1969)  treated Swiss mice with  sodium  arsenite
 (equivalent to 5 (jg/ml  As)  in drinking water for the  30-month duration of
 their study.   In the  treated animals,  the authors noted 6  malignant  and  11
 combined malignant and  benign tumors  out  of 103 animals versus 15 malignant
 and 50 combined malignant and benign tumors out of 170 control animals; thus,
 the results were  negative  for showing  any  carcinogenic  effect- of sodium arsenite
 in  this  study.
      In  a study of Leitch and Kennaway (1922;  as reported in IARC, 1980),  100
 mice were  given  skin  applications of  a  solution of potassium  arsenite in
 ethanol  containing  1.8 percent arsenic  trioxide  (later  reduced to  0.12  percent
 due to a high death rate),  thrice weekly for  3 months.  Of  the 33 mice that
 lived  for 3 months, only one  developed  a metastasizing  squamous  cell carcinoma.
      In  certain animal  systems,  positive carcinogenic responses  have  been
 reported, however.  For  example, Osswald and Goerttler  (1971)  exposed pregnant
 Swiss  mice  to daily parenteral dosing  of sodium arsenate  (0.5 mg/kg,  solution
 of  0.005 percent  arsenate salt) for a  total  of 20 injections.  Part of the
 offspring groups  received  20 injections of the  same  level subcutaneously  at
weekly intervals.   Leukemia  or lymphoma was  seen in 46  percent of the  mothers
 (11/24)  at  the  end of 2 years versus  none in the controls.    Of the treated
 offspring,  41 percent (17/41) of  the  males and  about  half  of the females
 (24/50)  developed  leukemia versus  only 3  of 55  male  and  female control  off-
spring (approximately  6  percent).   IARC (1973) has criticized this study for
the  absence of exposure  of  controls to the appropriate  vehicle solution.
     Knoth  (1966)  noted  a  significant  frequency of tumors in  30 mice exposed
to  Fowler's solution orally  (one drop/week, 20 weeks, approximately 5.3  mg As
                                7-82
-------
total),including adenocarcinomas of the skin, lung, and lymph nodes.  The ab-



sence of experimental details makes critical assessment of this study difficult.



     Some recent animal  studies  have  employed different exposure conditions



than have been  employed in the past.   Berteau  et  al.  (1978) have exposed a



tumor-susceptible strain of  female mice  to a respirable aerosol of inorganic



arsenic  (1  percent  aqueous solution of sodium meta-arsenite, 20 to  40 minutes



daily, 5 days/week, 55 weeks total).  The 30 exposed mice showed neither gross



nor histological evidence of neoplasia.



     In  a two-part  study,  Ishinishi et al.  (1977)  examined  the carcinogenic



and co-carcinogenic  effects  of various  arsenic compounds on male Wistar-King



rats.  In the first part of the study, arsenic trioxide, an arsenic-containing



copper  ore  (containing  3.95  percent  arsenic)  or  metal  refinery  flue dust



(containing 10.5  percent arsenic)  were administered to 51 rats via 15 weekly



intratracheal  instillations.   The  rats were observed  over  their lifespan.   Of



the 25 surviving rats,  one adenocarcinoma was seen  in the group receiving flue



dust.   One  lung metastasis from osteosarcoma of the femur  and  one adenoma was



reported in the group  exposed to  the  copper ore.   No malignant  tumors were



reported in the group  receiving arsenic  trioxide;  however,  one adenoma was



reported for this group.   All  groups displayed  squamous cell metaplasia in  the



airway and  osteometaplasia in  the  alveolus  of the  lung.



      In  the second part of the study,  87 rats were instilled with each of the



above  compounds suspended  in a saline  solution containing benzo[a]pyrene



(B[a]P)  or with  B[a]P, alone.  Control  rats  (23)  were instilled  with the



saline solution.   Of the  34  surviving  exposed rats,  one adenocarcinoma  was



seen in the group  receiving  B[a]P plus  copper ore.  All  exposed groups  had



squamous cell  carcinomas of the lung.   Of particular interest to the authors,



was the noted  co-carcinogenic  effect  of arsenic trioxide  with B[a]P;  rats  in
                                 7-83
-------
 this group  exhibited  a 43  percent  incidence rate (3/7) for  squamous  cell
 carcinomas.   This compared with  a  14 percent (1/7)  incidence  rate  in  animals
 given B[a]P, alone.  No benign or malignant tumors were seen in the 7 surviv-
 ing control rats.  Again,  squamous  cell  metaplasia or  osteometaplasia  were
 seen in  all  groups.
      The results  indicated  a positive interaction  between arsenic trioxide and
 B[a]P;  however,  the authors  noted that the  numbers of  surviving animals were
 too small to permit drawing any firm  conclusions from the study.
      In  another  study,  Ishinishi  et al.  (1980) gave 30  male adult Wistar rats
 intratracheal  instillations of arsenic trioxide in  suspension for  15  weeks.
 Of  the  19 rats that survived, only one malignant  squamous  cell carcinoma was
 observed over  lifetime.  No  tumors were found  in the controls.
      Ivankovic and  co-workers  (1979)  exposed  rats, via  intratracheal instilla-
 tion, to a  pesticide mixture corresponding to that used in  the past for vine-
yard  treatment and consisting of a mixture  of copper (II) sulfate, calcium
 hydroxide and  calcium  arsenate.   Of  25 rats exposed to  approximately 0.07 mg
 arsenic, 10  died from lung necrosis or pneumonia.  In the survivors, 9 animals
 (60  percent) showed multi-focal bronchogenic adenocarcinomas and bronchiolar/
 alveolar cell carcinomas.
     This study  appears to  offer experimental  evidence that  the  vineyard
pesticide mixture,  employed as such,  could  have been carcinogenic to  vine
dressers  working  with  the material.   One  difficulty  with  this study is an
ambiguity regarding its  full  significance for the  general  issue of arsenic-
inducing carcinogenic effects by itself.   Clearly,  the high mortality rate,  40
percent, and the known toxicity of Bordeau mixture (copper sulfate  plus calcium
hydroxide)  to  animals  (Pimentel  and  Marques,  1969)  and man (Villar, 1974,
Pimentel and Marques, 1969)  suggest carcinogenesis; however, it is  impossible
                                7-84
-------
to clearly ascribe such activity to arsenic alone given the presence of other
compounds within the  mixture.   In  their studies, Ivankovic et al.  (1979)  did
not  include  animal  groups exposed to calcium  arsenate  alone or to Bordeau
mixture alone.
     Pershagen et al.  (1983, in press) have studied the pulmonary carcinogeni-
city of  arsenic  trioxide  alone  and in combination with  benzo(a)pyrene  in male
golden  hamsters  given 15 weekly  intratracheal instillations of 3 mg  As/kg
and/or 6 mg/kg B[a]P.   In this  study, a carbon dust carrier with dilute sulfuric
acid  was employed to enhance  retention of arsenic.  Carcinomas of  larynx,
trachea, bronchi and  lungs were found in 3 animals given just arsenic  versus a
zero response in controls.   The incidence of adenomas,  papillomas and  adenoma-
toid  lesions was higher  in  arsenic  versus  control  groups  (p < 0.01).   While
the  carcinomas in the arsenic  group  did not reach statistical  significance,
the  significantly  higher  incidence of the adenomas, papillomas,  and  adenomatoid
 lesions support the authors' conclusion that the carcinomas were not a chance
 finding.   The use of the carrier to enhance lung retention of arsenic appears
 to be the key  to  the results  in  this  study.   By itself, the oxide  of  arsenic
 is rather rapidly cleared from the lung (see below).
      Rudnai  and Borzsony (1981) reported that a combination of pre- and post-
 natal  exposure of  CFLP  mice to injected arsenic trioxide (1.2 pg As/g, days
 15-18 of gestation; 5 ug As/animal for 3 days post-birth)  resulted in a 63 per-
 cent lung tumor incidence  at 1 year of  age versus approximately 18 percent  in
 controls.   In  this study,  a detailed histological description  of the tumors
 was not  given,  but it can  be  assumed  that  both adenomas  and adenocarcinomas
 were observed.
      Chung  and Liu (1982) reported that the repeated intratracheal instillation
 of  ore dust containing  arsenic in  rats  resulted  in a lung cancer incidence
                                  7-85
-------
 rate of approxiamtely 15 percent (6/41) over a period of 3 years, while controls
 (N=18) showed  no lung tumors.  There was extensive metastasis in one animal.
 In this study,  two compositions of metal content in ores were used, with the
 arsenic varying by 8-fold.  The lung cancer rate for the two forms of ore dust
 were comparable,  despite  the  8-fold difference in arsenic,  i.e.,  there was  no
 simple dose  response.  On  the other hand, the ores contained variable amounts
 of iron and lead, which may have affected both the dose response and, possibly,
 the actual incidence of the lung cancers.
      Ishinishi  and Yamamoto (1983)  exposed  female golden hamsters to arsenic
 trioxide by intratracheal  instillation of a  suspension in phosphate buffer,  15
 times weekly, for 4 months, at a total dose of 5.3 mg As.  A second group was
 given a total dose of 3.8 mg As.  All  animals were followed over their entire
 life  span.  The incidence of lung tumors (adenomas) for the combined exposure
 groups  was approximately 17 percent versus  no  tumors  in surviving controls.
 While the  sample size was  small, there was  the suggestion  of a dose  response
 across  the two  exposure levels  (3/10  for the higher dose;  1/10 for the lower
 dose).  These data suggest that arsenic trioxide  is  tumorigenic  in the golden
 hamster and may be carcinogenic.
      Recently,  both  Inamasu et al.  (1982) and  Pershagen et al.  (1982) have
 studied the effects of intratracheal  instillation of  calcium arsenate (see
 Section 4.1.1 for complete discussion of studies).  Inamasu et al.  gave single
 intratracheal instillations  of arsenic trioxide or calcium  arsenate  to male
Wistar rats.  Pershagen et al.  instilled male Syrian golden hamsters with four
weekly suspensions of arsenic trioxide, arsenic trisulfide and calcium arsenate.
The results of  both  studies showed that arsenic trioxide was rapidly cleared
from the lungs, whereas  calcium arsenate was slowly eliminated.   The differ-
ences in clearance appeared to  be related to  solubility,  with the less soluble
calcium arsenate exhibiting the slowest clearance.

                                7-86
-------
     These recent findings  might  help to explain the differences noted above
in the earlier  studies  of Ishinishi et al.  (1977; 1980) and Ivankovic et al.
(1979).   In regard to the Ivankovic study, it may be that the calcium arsenate,
of itself,  contributed  to the high mortality  rate  observed in the exposed
rats.
     Schrauzer and co-workers  (Schrauzer  and Ishmael, 1974;  Schrauzer  et al.,
1977;  Schrauzer et al.,  1978) have reported experimental results using oral
arsenic and the tumorigenie effect  of the agent  on  spontaneous mammary adeno-
carcinomas in an  inbred strain of mice (CgH/St  Mice).   These workers noted
that while  arsenic retards  the overall  incidence of tumor formation (10 or 80
ppm As),  it stimulates  the  growth of  tumors that otherwise  occur.  When these
animals were exposed  to only 2 ppm As (arsenite) in drinking water,  compared
to levels of 10 or 80 ppm, there was  no effect on frequency of tumors, although
the same  enhanced tumor growth was  seen as before with  levels of 10 or 80 ppm
As  (Schrauzer  et al.,  1978).  Furthermore,  a  higher incidence of multiple
tumors of  the  mammary gland was observed, as was the abolishing of the anti-
carcinogenic effect  of  selenium  in this  system when both elements were given
together.
     In support of the  positive responses noted  in the above animal  studies,
Dipaolo and Casto (1979) reported the transformation of cultured Syrian hamster
embryo  cells  by  direct exposure  to  sodium  arsenate (NA2HAsO^).   In  their
assay, 300  cells  from secondary cultures  of  Syrian  hamster  embryo cells  (HEC)
were plated  in  complete medium with  20 percent  serum  in  50-mm plastic Petri
dishes along with 6  x  10   HEC cells  which  had  been irradiated as confluent
monolayer  cultures.   Sodium arsenate at  concentrations of  0 ug/ml (control),
2.5 ug/ml and 5.0 ug/ml  in complete medium (with  a  final concentration of ace-
tone less than 0.02 percent) were added 24 hours  later.  Dishes were fixed and
stained nine days after  seeding.  Colonies were  scored  blind by two observers.
                                7-87
-------
      For  the  two  sodium  arsenate  concentrations,  the  authors  reported  that  the
 number  of transformations were 1.16 and 2.08/dish, respectively, or 1.74 and
 4.13  percent, respectively,  on the basis of transformed colonies relative  to
 total colonies  scored.   The authors also reported that typical colonies scored
 as  transformed  after direct treatment with a metal carcinogen were identical
 to  those  obtained with  other chemical carcinogens.   In addition, the  authors
 claimed  (data not shown) that when 2.5 pg  benz(a)pyrene  per ml  of medium were
 used, the transformation frequency, on a colony  basis,  was 4 to 6 percent.
 7.3   QUANTITATIVE CARCINOGEN RISK  ESTIMATES
 7.3.1 Introduction
      This quantitative section deals with the unit risk for arsenic in air  and
 water and the potency of arsenic relative to other carcinogens that the Carcino-
 gen Assessment  Group (CAG) of the  U.S.  Environmental  Protection Agency has
 evaluated.  The  unit risk estimate for an  air  pollutant is defined as the
 lifetime  cancer  risk occurring in a population  in which all  individuals are
 exposed continuously from  birth throughout  their  lifetimes  to a  concentration
          3
 of  1  jjg/m  of the agent in the air they breathe.  The unit risk estimate for
water is  defined  similarly,  but with  a water concentration of 1 |jg/£.  Unit
risk  estimates  are  used  for two purposes:   (1)  to compare several agents with
each  other  in terms of  carcinogenic potency, and (2) to give a crude  indica-
tion of the human health risks that might be associated with exposure to these
agents,  if the actual exposures are known.
     The data used  for quantitative estimates can be  of two types:  (1) life-
time  animal studies,  and (2)  human studies where cancer risk has been asso-
ciated with exposure to  the  agent.  It is  assumed, unless evidence exists to
the contrary» that  if a  carcinogenic response occurs at the dose levels used
in a  study, then  responses at all  lower doses  will  occur with an incidence
that can be determined by an appropriate extrapolation model.

                                7-88
-------
     There is  no  solid scientific basis for  any  mathematical  extrapolation



model that relates  carcinogen  exposure to cancer risks at the extremely  low
                                                           f
                                                           \

concentrations which must  be  dealt with in evaluating environmental hazards.

                                                           1

For practical  reasons,  such  low levels of risk cannot  be measured directly



either by animal  experiments  or by epidemiologic studies.   It is  necessary,



therefore, to  depend on current  knowledge  of  the mechanisms  of carcinogenesis




for guidance as to the correct risk model to use.



     At the present time,  the dominant view is that most cancer-causing agents



also cause irreversible damage to DNA—a position supported by the fact that a



large proportion  of agents that cause cancer  are  also  mutagenic.   There is



reason to expect  that  the  quantal  type of  biological  response, which  is char-




acteristic of mutagenesis, is associated  with a  linear non-threshold dose-



response relationship.  Indeed, there  is substantial evidence from mutagenesis



studies with both ionizing radiation and a wide variety of chemicals that this



type of dose-response model is the appropriate one to use in estimating cancer



risks from environmental  exposures.   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 non-threshold dose-response relationship is



also  consistent with the   relatively few  epidemiologic  studies of cancer re-



sponses  to  specific agents that contain  enough information to  make the evalu-



ation possible (e.g.,  radiation-induced leukemia, breast and thyroid  cancer,



skin  cancer  induced by arsenic  in drinking water, liver cancer induced  by



aflatoxins  in the diet).   There is also  some evidence from animal  experiments



that is  consistent with the  linear  non-threshold model (e.g., the initiation



stage  of the  two-stage carcinogenesis model  in rat  liver and  mouse  skin).



      Because of these  factors,  the linear  non-threshold model  is considered to



be a viable possibility for  the true  dose-response relationship and, unless





                                 7-89
-------
direct evidence  to  the  contrary  is presented,  it  is used as the primary basis
for risk extrapolation at low levels of exposure.
     The quantitative aspect  of  carcinogen risk assessment is included here
because  it  may be of use in  setting regulatory priorities, evaluating the
adequacy of technology-based  controls,  and other aspects  of  the  regulatory
decision-making  process.  However, the  imprecision  of  presently  available
technology  for estimating cancer risks  to humans at  low  levels  of exposure
should be  recognized.   At  best, the linear extrapolation  model  used here
provides a  rough  but plausible estimate of the upper limit of risk—that is,
with this model  it  is not likely that the true risk would  be much more than
the estimated  risk,  but it  could be considerably lower.  The risk estimates
presented below should not be regarded,  therefore, as accurate representations
of true  cancer risks even when the exposures involved are accurately defined.
The estimates presented may, however, be factored into regulatory  decisions to
the extent that the concept of upper-risk limits is  found to be useful.
7.3.2  Unit Risk for Air
7.3.2.1  Methodology for Quantitative Risk Estimates—The  methodologies  used
to arrive at quantitative estimates  of risk must be  capable of being implemented
using the data available in existing epidemiologic studies of exposure to air-
borne arsenic.   In  order to extrapolate from the exposure levels and temporal
exposure patterns  in these studies  to  those for which risk  estimates  are
required, it will be assumed that the age-specific mortality rate  of respiratory
cancer per  year  per 100,000 persons for a particular 5-year age  interval,  i,
can be represented using either of two models:
a.(D) = a.|
                                        a'Dk]
(1)
(a relative or multiplicative risk model), or
                         a..(D) = a. + 100,OOOa'D
                                7-90
(2)
-------
(an absolute  or additive risk model).   With either model, a.  is  the  age-


specific mortality rate per year of respiratory cancer in a control population


not exposed to arsenic, a1  is a parameter representing the potential of airborne


arsenic to cause respiratory cancer, and D is some measure of the  exposure to


arsenic up to  the  ith age interval.  For example, D might be the  cumulative


dose in |jg/m3  years,  the cumulative dose neglecting exposure during the last


10 years prior to the ith age interval,  or the average dose in ug/m3 over some


time period prior to the ith age interval.   The forms to be used for D  will be


constrained by  the  manner  in which dose was treated in  each  individual epi-


demiologic  study.    The  parameter  k  determines  the   shape  of  the


dose-response curve.  Attention will be given particularly to the values k = 1


and k = 2.  If k =  1,  the age-specific  incidence  rates vary linearly with  the


dose level (a  linear  model), and if k  = 2  they vary quadratically. At low


exposures the  extra lifetime probability of  respiratory  cancer mortality will


vary correspondingly  (e.g.,  linearly  for k = 1 and quadratically for k = 2).


     The dose-response  data  available  in the epidemiclogic studies for esti-


mating the parameters  in these models consists primarily of a dose measure D.
                                                                             J

for the jth  exposure  group,  the person-years of observation Y., the observed
                                                              0

number of  respiratory cancer deaths 0., and the  number  E. of these deaths
                                      J                   J

expected in a control population with the same sex and age distribution as the
exposure group.  The expected number E. is calculated as
                                      J
                           . = \ Y..a./100,000,
(3)
where Y.. is the number of person-years of observation in the ith age category


and  the  jth exposure  group  (Y.  =  .Y..).   This  is  actually a  simplified  repre-


sentation,  because  the  calculation also takes account of the change in the
                                7-91
-------
age-specific  incidence rates  with  absolute time.   The expected number of


respiratory cancer deaths for  the ith exposure group is
E(0.) =   Y..a.
   J    I   J I   1
                                             a'D.)/100,000
                                                J
                                                                      (4)
under the relative risk model, and is
                         E(0.) = '7 Y..(a. +100,OOOa'D. )/100,000
                            J     '  J '   '            J


                                      k
                         = E. + a'Y.D.
                            o      J J
                                             (5)
under the  absolute  risk model.   Consequently,  with  either  model,  E(0.)  can be
                                                                     J

expressed  in  terms  of quantities typically available from the published epi-


demiologic studies.   Note that person-years of observation are not required if


the multiplicative model is used.


     Making the  reasonable  assumption  that 0.  has a Poisson distribution, the
                                            J

parameters a1  and k can be estimated from the above equations using the method


of maximum  likelihood.   Once  these  parameters  are estimated,  the  age-specific


mortality  rates  for respiratory cancer can be estimated for  any  desired ex-


posure pattern.


     To estimate the corresponding additional  lifetime probability of respira-


tory cancer mortality,  let  b..,...,  b-,8 be the  mortality rates, in the absence


of exposure,  for all  cases  per year per 100,000 persons for the age  intervals


0-4, 5-9,...,  80-84, and 85+,  respectively;   let a.^...^.^  represent the


corresponding rates  for malignant neoplasms  of the  respiratory system.  The


probability of survival  to  the beginning of the  ith 5-year age interval  is


estimated as
                                7-92
-------
                         H  [1 - 5b./100,000].
                                   J
(6)
Given survival to the beginning of age interval i, the probability of dying of

respiratory cancer during this. 5-year interval is estimated as
                         5ai7100,000.
(7)
     The probability of  dying of respiratory cancer given survival to age 85

is estimated  as  a18/b18-   Therefore, the probability of dying of respiratory

cancer in the absence of exposure to arsenic is estimated as
                         17             i-1
                    Pn = I [5a.7100,000) n (1 - 5b.7100,000)]
                     U  1=1   1         j=l       J                   (8)
                               17
                    +(a18/b18) n (1 - 5bj./100,000)
Here  the  mortality  rates  a. apply to  the  target population for which  risk

estimates  are desired, and  consequently will  be  different from  those  in

(l)-(5),  which  applied to the epidemiologic  study  cohort.   If the 1976  U.S.

mortality  rates  (male, female, white, and non-white combined)  are  used in this

expression, then PQ = 0.0451.

      To  estimate the  probability  PFp  of respiratory cancer mortality  when

exposed  to a particular exposure pattern EP,  the  formula (8)  is  again  used,

but  a.  and b. are replaced  by  a.(D.)  and  b.(D.),  where  D.  is the exposure

measure  calculated for the ith age interval from the exposure  pattern EP.   For

example,  if the dose measure used in (1) is cumulative dose to the beginning

of the  ith  agl  interval  in  |jg/m3-years, and the  exposure  pattern EP is a
                                7-93
-------
lifetime exposure  to  a constant level of  10  ug/m3,  then D. = (i-l)(5)(10),
where the  5  accounts  for the fact  that  each  age inverval has a width of  5
years.  The  additional  risk of respiratory cancer mortality  is estimated  as

                              PEp - P0.                               (9)
If the  exposure  pattern EP is constant  exposure to  ug/m3, then P£p - PQ  is
called the "unit risk."
     This approach can easily be modified to estimate the extra probability of
respiratory cancer mortality by a particular age due to any specified exposure
pattern.   It is also  clear that the applications of  the approach are not
limited to respiratory cancer.
7.3.2.2  Risk Estimates from Epidemiologic Studies—Prospective studies of the
relationship between mortality and  exposure to airborne arsenic have been  con-
ducted for the Anaconda, Montana smelter (Lee and Fraumeni, 1969;  Lee-Feldstein,
1983; Higgins  et al., 1982; Brown  and  Chu,  1983 a,  b,  c); and the  Tacoma,
Washington smelter (Pinto et al., 1977;  Enterline and Marsh,  1982).
     The study  of  Lee-Feldstein (1983)  reported on an additional  14 years of
follow-up  of the cohort of 8047 studied by Lee  and Fraumeni  (1969), and used
essentially  the  same  methods  of analysis as  the earlier study.   Therefore, it
will  not  be necessary to consider the Lee and Fraumeni  study  in  any  detail in
this  report.   Higgins et al. (1982)  followed for an additional 14  years  a
sample  of 1800  men from the cohort  studied  by  Lee  and Fraumeni,  but used
different  exposure classifications  and different methods  of analyses.
      Brown  and  Chu (1983 a, b,  c), in a series of papers, arranged  the Ana-
conda smelter data in  such  a manner that a mathematical  model could  be derived
from  it to account for  the effect of the timing of  exposure  as  predicted by
the multistage model.
                                 7-94
-------
     Whereas Pinto et  al.  (1977) studied a cohort  from the Tacoma smelter



consisting of 527 men  followed only after retirement, the study by Enter!ine



and Marsh (1982) involved 2802 men and included follow-up prior to retirement.



Consequently, the  Enter!ine and Marsh study  appears  to provide a stronger



basis for quantitative risk assessment than the Pinto et al. study.



     In addition to  studies of copper smelter workers,  there have also been



studies of workers exposed to arsenicals in  the production or  use of pesti-



cides.   Ott  et  al. (1974)  studied the mortality experience  of workers exposed



to lead arsenate and calcium arsenate.



7.3.2.2.1  The  Lee-Feldstein  (1983)  study.   This  study  included 8047  white



males who were  employed as  smelter workers for 12 months  or more before 1957,



and whose mortality  experience was  observed  from 1938  through  1977.   Alto-



gether, the  study involved  192,476 person-years of  follow-up and 3550 deaths,



including 302 from respiratory cancer (Table 7-21).   Expected numbers of cancer



deaths were  calculated on an age-adjusted basis using the combined mortality



experience of the white  male population of Idaho,  Wyoming, and Montana.  As



Table  7-21  indicates,  malignant neoplasms of the digestive and respiratory



tracts had  SMRs of  125 and 285, respectively, both of which were  significant



at the 1 percent level (SMR = [observed/expected][100]).



     Workers were  categorized both  by duration of  employment and level of



exposure to  airborne arsenic  in  order to determine  the  effect of these param-



eters  upon  mortality.   For each year of the  study  period, workers were as-



signed to one of five  groups on the  basis of  total years  of smelter employment



completed (Table 7-22).   Work areas in  the smelter v/ere divided into heavy,



medium,  or  light exposure  areas.    Based upon  this division,  workers were



categorized  into  heavy,  medium,  or  light exposure  groups,  as  determined by



their  maximum exposure for 12 or more months.  The results for respiratory



cancer based upon these categorizations  are given in Table  7-23.





                                7-95
-------
    TABLE 7-21.   OBSERVED AND EXPECTED DEATHS DUE TO SELECTED CAUSES, WITH
      STANDARDIZED MORTALITY RATIOS (SMRs) AMONG SMELTER WORKERS,  1938-77
Cause of Death
Tuberculosis
Respiratory
Other
Malignant neoplasms
Digestive
Respiratory
Other
List No.a
001-019
001-008
010-019
140-199
150-159 .
160-164°
140-148, 165-170
Number of deaths
Observed
53
47
6
609
167
302
140
Expected
27.93
25.51
2.42
370.74
133.58
105.81
131.35
SMRb
190C
184°
248
164C
125C
285C
107
Vascular lesions of
  central nervous
  system
177-181, 190-199

         330-334
262
211.56
124L
Diseases of heart 400-443
Influenza and pneumonia 480-483,
490-493
Emphysema (1963-77 only) 527
Cirrhosis of liver 581
Accidents 800-962
Motor vehicle 810-825, 830-835
Other 800-802, 840-862
Suicide and homicide 963-964, 970-979
980-985
All other causes Residual
Total
Seventh revision of International Lists
1366
88
90
76
288
106
182
83
606e
3522
of Diseases
1056.55
76.05
34.58
36.53
280.27
115.55
164. 72
86.08
548. 50
2728.79
and Causes
129
116
260C
208°
103
92
110
98
129°
of Death.
 SMR = (observed/expected) x 100.
Significant at 1% level.  Bailar and Ederer (1964)
 Among the 302 deaths from respiratory cancer, the site was lung and
 bronchus (162,163) in 289 cases,  larynx (163) in 9, mediastinum (164)
 in 3 and (160) in 1.
 Includes 19 emphysema deaths occurring in the years preceding 1963, for which
 emphysema death rates are not available from individual states.

Source:  Lee-Feldstein (1983).
                                7-96
-------
 TABLE 7-22.  DESCRIPTION OF LENGTH OF EMPLOYMENT GROUPS, WITH NUMBERS OF SMELTER
     WORKERS, NUMBERS OF DEATHS, PERSON-YEARS AT RISK, AND DURATION OF SMELTER
        EMPLOYMENT (BASED ON TOTAL WORK EXPERIENCE THROUGH SEPT. 30, 1977)
Length of
employment group3
1 (25 or more years)
2 (15 to 24 years)
3 (10 to 14 years)
4 (5 to 9 years)
5 (1 to 4 years)
TOTAL
Number of
persons
1899
1138
678
1082
3248
8045
Number of
deaths
1169
586
328
433
1006
3522
Number of
person-years
of follow-up
27,053
26,556
19,734
30,854
88,279
192,476
 Employees in all cohorts were living on Jan. 1, 1938.

 Group assignment of each person here was based on his status at the
termination of employment or on September 30, 1977 (whichever date was
earlier).

 Represents cumulative follow-up experience over the study period, 1938-77,
with a total  of 67,569 person-years of follow-up in the period 1964-77.
Individuals were initially counted at risk upon completing 1 year of employ-
ment or on Jan.  1, 1938, if employed at least a full year before that date.
In each calendar year of the study period, employees were counted in the group
reflecting their cumulative work experience to date.

Source:  Lee-Feldstein (1983).


       Exposure  to  airborne  arsenic was estimated  from  702  samples collected

  at 56  sampling stations  during the years  1943-1958  (Morris,  1975).   Morris

  estimated that airborne  levels  averaged 11.27,  0.58,  and  0.27  mg/m3  in the

  heavy, medium, and  light exposure areas, respectively.  Respirators were used

  with varying degrees of faithfulness in the high exposure areas; consequently,

  average individual exposures  in these areas were probably much less than 11.27

  mg/m .  A  rough estimate  is  that use of  respirators  reduces  the exposure

  levels by a factor of 10 (OSHA, 1978).
                                  7-97
-------
    TABLE 7-23.   OBSERVED AND EXPECTED DEATHS FROM RESPIRATORY CANCER,  WITH
      PERSON-YEARS OF FOLLOW-UP, BY COHORT AND DEGREE OF ARSENIC EXPOSURE
Maximum Exposure to Arsenic (12 or more months)

Years of Exposure
25 years*
15-24
Less than 15 years
Heavy
Obs/Expb P-YC
13/2.5 2400
9/1.3 2629
11/2.4 6520
Medium
Obs/Exp P-Y
49/7 6837
13/4.0 6509
31/9.3 24594
Light
Obs/Exp
51/16.3
16/8. 6
69/31
P-Y
14573
12520
78245

arsenic exposure were not included in this table.
 Observed/Expected.
cPerson-years of follow-up furnished by Dr. Lee-Feldstein (personal communi-
cation).

Source:  Adapted from Lee-Feldstein (1983).


     The Lee-Feldstein (1983) study has a number of features which support its

use in making quantitative estimates of  respiratory cancer  risk from airborne

arsenic.   It was  a  large study that involved  observations  of a considerable

number of respiratory  cancer deaths.   A substantial amount of follow-up was

conducted of persons who  had been  exposed  for  15 years or more.   Estimates  of

exposure  levels  and work histories are  available  for  estimating  individual

exposures and for determining dose response.

     It would have  been more appropriate for making quantitative  estimates  of

risk to  have  categorized workers by their  individual  cumulative  or average

exposures, rather than  by their maximum exposures for  1 year or more.   In

developing the  quantitative estimates,  it  will  be assumed that  a  worker's

average exposure  during work hours was equal to the exposure for  the category

to which  he was  assigned.  However,  because these  assignments were  based  upon
                                7-98
-------
maximum exposures  for at  least  a 12-month period, this  approach  tends  to
overestimate exposures, and  consequently,  to underestimate the carcinogenic
potency of arsenic.
     Because smoking  is also an  important  risk  factor for respiratory cancer,
it would  have  been very useful to  have  smoking histories for the workers.
Higgins et  al.  (1982) collected  some limited smoking data for this cohort.
Higgins and  co-workers  suggest that the smelter workers smoked somewhat more
than the average U.S. white male population, but the difference was not enough
to have a major effect upon the outcome of the study.
     Development of risk estimates:  The  data from the Lee-Feldstein (1983)
study used in the  risk assessment are listed in Table 7-24.  The relative risk
(observed/expected) from this table are graphed in Figure 7-2, and the absolute
risks  ([observed-expected/person-years)  in  Figure  7-3.   It  is  clear from
these graphs that  the risk for  the  high-exposure  group exposed for greater
than 25 years  is not commensurate with  the risks  for the other groups.   Be-
cause of  this,  and also because  the exposures in the high exposure groups are
much more uncertain than those of the other groups, it was decided to estimate
risk  using  only the  low  and medium exposure groups.   Results of applying
chi-square goodness-of-fit tests of the relative and absolute  risk models with
k  =  1 and k =  2 are recorded in  Table  7-25.   The maximum  likelihood estimates
of the  carcinogenic potency parameter a1  are also listed  in this table.  The
maximum likelihood fits  of these models  are graphed  in Figures  7-2 and  7-3.
All  of  the  fits are  poor  (p less than 0.0001) with the exception of that for
the  absolute-risk  model; this  latter fit  is  marginally acceptable  (p = 0.025).
     The  unit  risk (additional risk of respiratory cancer death  from lifetime
exposure  to 1  ug/m3 airborne arsenic) obtained from  the  absolute-risk  model
with  k =  1 is  also listed in Table 7-25.   This  risk was estimated by applying
(2),  (8), and  (9)  with D.  based  upon a constant exposure  of 1  pg/m3.

                                 7-99
-------
TABLE 7-24.   DOSE-RESPONSE DATA FROM LEE-FELDSTEIN (1983) USED FOR RISK ASSESSMENT
Cohort
1
(25 + years
(of exposure)
2
(15-25
years of
exposure)
3
(less than
15 years of
exposure)
Maximum
Exposure
to Arsenic
Heavy
Medium
Light

Heavy
Medium
Light

Heavy
Medium
Light
Cumulative
Exposure3
(ug/m3-years)
36064
18560
9280

22250
11600
5800

5973
3074
Person-Years
of .
Observation
2400
6837
14573

2629
6509
12520

6520
24594
Observed
Deaths
13
49
51

9
13
16

11
31
Expected
Deaths
2.5
7.0
16.3

1.3
4.0
8.6

2.4
9.3
 Exposures are in ug/m3-years estimated as (air concentration) (duration).   For
 light,  medium, and heavy exposures, air concentration was estimated as 290,  580,
 and 1127 ug/m3, respectively (OSHA, 1978).   Duration was estimated as follows
 (cf.  Table 7-22):

     Cohort 1:  Persons in this cohort had at least 25 years'  exposure.   If
               all  had worked continuously throughout follow-up,  average
               duration would have been 25 + 27053/1899 = 39  years.   There-
               fore the midpoint (39 + 25)/2 = 32 years was used.

     Cohort 2;  The midpoint of the employed interval, i.e., (15 + 25)/2 =  20
               years was used.

     Cohort 3:  A weighted average of the midpoints of the employment intervals,
               i.e. ,
(3) (88279) + (7.5)(30854) + (12. 5)(19734)   K ,
- 88279 + 30854 + 19734 -  = 5'3
                                                                      WaS USed'
 Furnished by Dr.  Lee-Feldstein.
                                          7-100
-------
                                                            _c
                                                            '3
                                                            +•>
                                                            en
                                                            2

                                                            •?
                                                            6
                                                            o>
                                                             o

                                                             2
                                                             03
                                                            TJ


                                                            I
                                                             V>
>|S|y
    7-101
-------
                                                                I
                                                                 e
                                                                "5
                                                                to
                                                                .3
                                                                H-
                                                                 O

                                                                 «2
                                                                 03
                                                                 0)
                                                                 o
                                                                 o

                                                                i
                                                                •o
                                                                 c
                                                                 CO
                                                                 CO
                                                                I
                                                                 o
                                                                 (0
                                                                a
                                                                 0)
                                                                 3
                                                                 en
000 L x >|S!H
           7-102
-------















l/>
LL!
j
z

10
LU
P
r-
HH
J£
*


U.
o


Of
?




in
CM

3
CD




































U)
*~
U.
0
in
IA
Ol
1
O
o
<*-
o
UI

*3
0)




















































.Q
W
•^
i.
*f
c
3
2


01
3
2
^
Q.


^
IK
•d
09
X

(0
o >•>
0) *^
f &
25
0 S.




XX










01

jj







0)


18 3

5
I/) 18
a











e
0
•a-"
a> P
tn at
o i—
a. 3
•o

it- »- if-
CO If- If- t-
1 O O O
»^
§^ It It
u u o
• (O 10 (8



rH «*• rH
O 0 O
in o o o
CM 0 0 0
o o o o
o o o o




in
i^. in in m
CM o oo cn
rH ID CM f*



^
P^ rH ^ CO
1 1 1 1
co cn o cn
•d- 0 0 O
CM CM CO CM




rH CM rH CM





y y
in in
*£ 'C
5 2
3 *•*"
1- -P
O IB
U) r-
fl 2



^^>
ro ai

.|E£,*i
at S
•P ^ O
in CM
2,1.2
0) 3
U- 01 UI
01 S Q.
— 1 r- Ol













•a i. M
C 01 k
5 « §

**
If-
1 1 1 O
Sir) co Ji
O 0 U
... 18



in rH ID O
I-. O «*• O
O O O O




^^ co co co
CM * in *
• o -in
rH rH CM i-H



^
:•> rH •* oo
i i i i
SrtrnS
CM rH CO CM




l-t CM rH CM





w y
tfl V)
'k *k
O) 0)
•P >
3 *r-
"3 18
in f—
A 0)
10 S-






"io

•P ID
01 CM
s-
•r- Ot

S ™















P -p
'C .U> «t
o in « > «
. O rH CO
r- in CM rH CM



r*, U /-»
rH f^* rH ^™N ^»
1 1 1 «S- 1
in ^ co ^«-> rH
«J- O * 10 0
On >D rH «»• rH




rH rH CM rH CM





*j> *t \*
in in in
'£ 'si C
0) 01 01
p *» >
3 3 i-
1- r= 4J
o o re
in tn r-
^3 ft 0)
(8 10 t.



in
t.

^ 3E

xi i co cn
tj r- a> at i n)
CM c r* ^
HI r- r^ 01 O

f3 -o at !a o>
10 C *> (3 N
09 UJ













s- in
O 0) t.
U P 0)
a: i— ji
< 01 fc.
Sli

B-P P **
t>- it- *-
CO If- If- --i-i in cn
o • • •
r^SScS



^^^
i i i i
in rH CO ID
00 O rH 0
00 CM in rH




rH CM rH CM





\* j^
in in
'u 'C
a> o>
•P >
3 *r-
r- +>
O 10
in r-
* 2



in
t.

Z **^
cn
eg 10
CO r-
at co
C 1 I-
•^ p^ fO
r- 0)
t- 01 >,
ai i— i
P J3 O
C 18 rH
UJ t-*~f

















CM1?
tD 00
CO ID
• •



ID CO
ID CM
O O




^-./-^
O -
•r- p


. ^r o

P 3
at 01 in
r— O
O H- 0)



in
c-


i.


•r* 0)
U S-
••- 3
in u
IS









ai
!a
IB
(0
o.
§
u
o
0}
s-
18

rj
C
ai
3
cr
at
UI
c
Q
U
c
10
.
+?
X
at

at
at
in
•a
at
tn
3
i—
01
.tJ

C
o
Q.
3
O!
C

•a
c
at
a.
--S

M

'E
3
P
2
ai

if-
•r-
•o

c


2
10

in
at
3Potenci







u
c
0)
ut
t-
a>
05
^
cn
3.
rH
5
2
3
in
O
Q.
2
I—
5
i
o
s.
0)

I

+J
<*-

r—
s
s
<*-

>J
•tf
r— •
t

s

0)
u
c
R)
U

^
o
p
10

•r-
o.
• tn
2

o
tn

T
9)
I
•r-

01

•r1


10
C
bAdditio






















































•
*;
S
X

s

tD

in
e



y*>

iL
S
ID
U
































rH
CD

o'

C
10
j:
-p
UI
U)
0)

vi

•p
to
5


M-

if—

in
tn
at
c
•a
o
o
cn

ai
i.
10
3
cr
in
i
«r-
U


O

0)
3
>
C,
•a
7-103
-------
Specifically,
                                           2.5]
(10)
was  used,  which represents the cumulative exposure in ug/m3 resulting from a
constant exposure  to ug/m3 from birth  to  the midpoint of the  ith 5-year age
interval.  The  factor 4.56 is needed to account for the fact that the workers
in the occupational  study  used to estimate the  carcinogenic potency of arsenic
were only  exposed  during work hours.  Assuming that workers, were exposed for
an average of 8 hours per  day, 240 days per year, an environmental exposure to
1 |jg/m3 for 1 year is equivalent to  an occupational exposure to

      (1 ug/m3)(24 hours/8 hours)(365 days/240  days) = 4.56 ug/m3     (11)

for 1 year.

7.3.2.2.2  The Higgins et  al. (1982) study.  Higgins et al. conducted additional
independent follow-up  through 1977  of a sample of 1800 men from the Anaconda
smelter cohort studied by  Lee and Fraumeni (1969).  The sample included all of
the men classified in  the  heavy exposure  category by  Lee  and Fraumeni  (1969),
as well as a  random  sample of 20 percent  of  the remaining cohort.  There were
80 deaths  from  respiratory cancer in the  sample.   Expected numbers were  based
upon the mortality rates of Montana white males, except for "all respiratory
diseases";  U.S.  white males were the  referenced population for the  latter
category.
     Higgins et al.  also reviewed the industrial hygiene data and calculated
average concentrations for the period 1943-1965 for 18 departments.  No measure-
ments were  available for  17 departments, and average concentrations were
estimated for these.   These estimates were coupled with work histories
                                7-104
-------
updated through 1978  to  obtain exposure measures  for  each  individual  in the


study.  Three types of individual exposure measures were considered:  ceiling,


time-weighted average  (TWA), and cumulative.   Ceiling exposures were esti-
              3
mates, in ug/m  ,  of the  highest exposure  a  man experienced for 30  days or
                                                      3
more.  TWA exposures  were  estimates, in units of ug/m  , of the time-weighted

average exposures  during the period  of  employment.   Cumulative  exposures were

                                            3
estimates of total exposure in units of ug/m -years.


     Higgins et al.  investigated 5 combinations of follow-up and time period


during which exposure was assessed:  I.  Exposure was assessed up to the date a


worker entered the study  and follow-up was from entry  into the study through

1978;  II.   Exposure  was  assessed up to 1964  and  follow-up was  also  through


this date;  III.   Exposure was assessed through 1964 and follow-up was through


1978;  IV.   Exposure  was  assessed through 1964 and follow-up  was from 1964


through 1978;  V.   Exposure  was assessed through 1978  and follow-up was also


through 1978.  These  different  methods  were  considered principally  because of

the  perceived  difficulties  of  overlapping exposure and follow-up  periods.


Thus, with methods I  and IV, exposure  and follow-up periods  were disjoint,


whereas with methods II and V they coincided.

     The analyses based upon ceiling exposures are not  considered suitable for


quantitative risk assessment because it seems extremely unlikely that respira-


tory  cancer  risk  would be  a  function of peak exposures for  any  30-day period,


regardless of the other  exposures that might have been experienced.  This is


also  the  case with the TWA analyses, because the exposures  were  averaged only

over  the  period  of employment,  without, regard to the duration of employment.


If either of these dose measures were appropriate, it would mean, for example,
                         3
that exposure to 500 ug/m  for 1 month would produce the same risk as exposure

           3
to 500 ug/m  for 30 years—which seems highly unlikely.
                                7-105
-------
     Thus, the analyses of Higgins et al. which appear to be most appropriate

for developing quantitative estimates are those based upon cumulative exposure.

This particular type of analysis was applied only to method V and applied only

in Higgins (1982).   The results of this analysis for lung cancer are listed in

Table 7-26.
 TABLE 7-26.  RESPIRATORY CANCER MORTALITY 1938-1978 FROM CUMULATIVE EXPOSURE
        TO ARSENIC FOR 1800 MEN WORKING AT THE ANACONDA COPPER SMELTER3
Cumulative
Exposure
ug/m3-years
0-500.
(250)D
500-2000
(1250)
2000-12000
(7000)
6 12000
(16000)

Person-Years
of Observation
13845.9
10713.0

11117.8

9015.5


Observed
Deaths
4
9

27**

40**


Expected
Deaths
5.8
5.7

6.8

7.3

aFrom Higgins (1982), Table 6.
 Numbers in parenthesis indicate assumed average exposures.

  Significant at 0.01 level.


     Information on  the  smoking habits of 80.6 percent of  the  1800 men was

obtained from questionnaires  administered directely to those still  living and

to close friends or  relatives of those who were deceased.   Sixteen  percent  of

the smelter  workers  were "non-smokers" compared with  24-36 percent of U.S.

males from  1955 through  1978.   Thus,  it appears that the  smelter workers

smoked somewhat more than the average U.S. male.  However,  no confounding was

detected between arsenic  exposure  and  smoking; 15.1 percent of those  in the

"heavy"  exposure  group were  non-smokers, versus  16.3 percent  in the  other
                                7-106
-------
exposure groups.  Significant  increases  in respiratory cancer were observed
even among non-smokers exposed to high levels of arsenic.
     Development of risk estimates:   The  relative risks  (observed/expected)
from Table 7-26 are graphed  in  Figure 7-4, and  the absolute risks  ([observed-
expected]/person-years) in Figure 7-5.   Results of applying chi-square good-
ness-of-fit tests of  the  relative-  and absolute-risk  models  with  k = 1 and
k = 2 are  recorded  in Table 7-25.   The maximum likelihood estimates of the
carcinogenic potency  parameter a1 are also listed in Table 7-25.   The maximum
likelihood fits of  those  models are graphed  in  Figures 7-4 and 7-5.   The  fits
for k = 1 are both excellent, with  the absolute-risk model providing  a slightly
better fit than the relative-risk model (p = 0.75 vs.  p =  0.46).   The fits for
k = 2  are much less  adequate  (p = 0.017  for  the absolute-risk model  and
p = 0.0015 for the relative-risk model).
     The  unit risks  (defined  as the  additional risk  of  respiratory  cancer
death  from lifetime exposure to 1  ug/m3  airborne arsenic) obtained from  the
absolute-  and relative-risk models with  k = 1  are also listed in  Table 7-25.
     These  risks  were estimated by applying (1)  or  (2),  (8),  and  (9) with  D..
 based upon  a constant exposure  of  1 ug/m3.   Specifically,
         = (4.56)(72)
(12)
 was used, which represents  the  average lifetime cumulative exposure  in  pg/m3
 resulting from a constant  exposure  to 1 pg/m3.   Average lifetime exposure  is
 used because  it  seems  most commensurate with the  treatment of exposure by
 Higgins et a!.;  in  their analysis all of the person-years attributable to a
 single worker were  placed  into  a single exposure  category based upon total
 lifetime exposure,  and  consequently  person-years of observation were placed
 into exposure categories according to exposures which had not yet occurred.  It
                                 7-107
-------
7-
                                                      Dose-response data
                                                      is from Table 7-26.
                                                        Fit is by relative
                                                         risk model.
                 4000
     8000

Cumulative Dose
 (yug/m3 - years)
12000
16000
Figure 7-4. Relative risks and 90% confidence limits for data of Higgins (1982).
                                7-108
-------
    6-
§
 0)
 <->

 "5
 CO
 .n
2-
                                                          Oose-response data
                                                           is from Table 7-26.
                                                           Fit is by absolute
                                                              risk model.
                      4000            8000            12000           16000

                                 Cumulative Dose
                                   (//g/m3-years)

      Figure 7-5. Absolute risks and 90% confidence limits for data of Higgins (1982).
                                    7-109
-------
would  have  been more appropriate for purposes of quantitative risk assessment
had  exposures  been related to each  5-year age interval (as was done in the
analysis  of Enter!ine and Marsh, 1982)  rather than to the total  observation
period  of an individual.   The factor 4.56 converts from occupational  to envi-
ronmental exposures and is explained at  equation  (11).
7.3.2.2.3   The  Brown  and Chu  estimates from the Anaconda data.
Development of  Risk Estimates.   As  noted  by  Whittemore (1977) and Day and
Brown  (1980),  the multistage theory for the  carcinogenic  process  predicts
that the carcinogenic response is a function of the following factors:

     (1)  exposure rate
     (2)  duration of exposure
     (3)  age at initial exposure
     (4)  time since cessation of exposure.

     Brown  and  Chu (1983a) discuss in detail  the ways in which these factors
influence the  age-specific carcinogenic  rate  at  various stages  of the car-
cinogenic process.
     Using  the  updated Anaconda  copper  smelter  workers  cohort originally
studied by  Lee and  Fraumeni  (1969) and recently  extended through 1977 by
Lee-Feldstein (1983), Brown  and  Chu (1983b) concluded  that airborne  arsenic
most probably acted on a late stage of the carcinogenic process.  As a result,
they hypothesized  that  the carcinogenic  risk  from  arsenic exposure could be
quantified  by assuming a multistage model  in which  only the penultimate stage
is affected by  exposure.   Under this assumption,  the  risk  may be expressed  in
the form
r(d, to) = C[(d + to)
                                    k"1
                                         -t
(13)
                                7-110
-------
where d is the duration of exposure, t  is the age at initial exposure, and C

and k are unknown parameters.   The parameter C depends upon the exposure rate,

and the parameter k upon the time effect of exposure.

     Brown and Chu  (1983b)  noted a deviation from this model on the part of

workers who  left  employment at  the  copper  smelter before  the  age of 55.  As a

result, the  mortality  experience of that  group  after leaving employment was

not included in the analysis.

     In order  to  estimate the  unknown parameters C and  k, the basic mortality

data were arranged in the three-way table  reproduced as Table 7-27.  The three

classifications used in this table are as  follows:


     (1)  Level  of exposure,  corresponding  to  Lee  and  Fraumeni •. (1969)--
          classified into heavy, medium, and light exposure groups;

     (2)  Duration  of  employment,  classified  into the following  five  sub-
          groups; 0-9, 10-19, 20-29, 30-39, and  40+ years;

     (3)  Age  at  initial  employment, classified into the following five sub-
          groups; 20,  20-29, 30-39, 40-49, and 50+ years.


     For  each  of the 3 x 5 x 5 = 75 cells in  the  table,  the following three

 variables were given:
      Obs  = observed number of respiratory cancer deaths;

      Exp
expected number of respiratory cancer deaths (based upon the U.S.
white male age-specific calendar-time-specific respiratory cancer
mortality rates); and
      Pyr = person-years of observation


      A single individual  could  supply information for more than one cell as

 his duration of  employment  increased over the follow-up period.   The person-

 year weighted average  duration  of employment, and age at initial  employment,

 were calculated for each cell.
                                 7-111
-------
      Assuming that age  at  initial  exposure is equivalent  to  age at initial
 employment, and that duration of employment and exposure are equivalent,  Brown
 and Chu (1983c)  fitted  equation 13 to the data in Table 7-27.   They used the
 maximum likelihood method, assuming  a binomial distribution where Obs is the
 number of positive responses, Pyr is the sample size,  and the rate of response
 is p = Exp/Pyr  + r(d,to),  where d,tQ are the averages for each cell.  Using
 this approach, the value for k is estimated to  be  6.8,  and c  = .603, 1.42,
          -13
 1.74 x 10     for  the  light,  medium, and  heavy exposure  categories,  respec-
 tively.
      Brown and Chu (1983c) did  not attempt to  give  an exposure rate  estimate
 to the heavy, medium, and  light exposure groups  of  Lee and Fraumeni,  (1969).
 One reason for this was that "heavy" and "medium"  were  defined as "having
 worked at  least  one year in a heavy or medium exposure area."   The "total  time
 worked" was not necessarily an indication of the total time worked in a heavy
 or medium  exposure area.   As  a  result,  the use  of  the exposure rate  in  the
 areas  defined as medium or heavy would  tend to overestimate the true  average
 exposure over an individual's working  history.   This  bias did not exist for
 those  in  the light exposure group,  since almost all  of their working time was
 spent  in light exposure  areas.   In  addition, these low environmental exposures
 are  of greater utility in estimating  risks.
     As a  result of these  factors, only the light exposure group  was  used to
 obtain  a  dose-response model.   In this group, Brown  and Chu (1983c) estimated
 that the respiratory cancer rate for  an  individual first  exposed at age t  for
 a  duration  of  d years would be
          r(d,tQ) = .603 x 10~13 [(d + t )5'8  - t 5'8].
(14)
     Only limited  information  exists  concerning  the  time-weighted  exposure of
workers in  the  light  exposure  areas.   Arsenic  concentrations  in  several  light
                                7-112
-------
  TABLE 7-27.   OBSERVED AND EXPECTED LUNG CANCER DEATHS AND PERSON-YEARS
BY LEVEL OF EXPOSURE, DURATION OF EMPLOYMENT, AND AGE AT INITIAL EMPLOYMENT
Age at
C3
Initial
Employment


Duration of
0-9
10-19



Employment (years)
20-29
30-39
40+
High Exposure Level Group
<20 Obs
Exp
pyr
20-29 Obs
Exp
Pyr
30-39 Obs
Exp
Pyr
40-49 Obs
Exp
Pyr
50+ Obs
Exp
Pyr
Medium Exposure
<20 Obs
Exp
Pyr
20-29 Obs
Exp
Pyr
30-39 Obs
Exp
Pyr
40-49 Obs
Exp
Pyr
50+ Obs
Exp
Pyr
0
0.001
206
0
0.008
624
0
0.030
398
0
0.083
210
0
0.066
78.0
Level Group
0
0.010
1801
0
0.035
2636
0
0.167
1939
0
0.167
1190
1
0.262
295
0
0.009
408
0
0.051
637
0
0.077
207
0
0.054
80.0
0
0.027
23.2

0
0.039
1763
0
0.118
1622
0
0.473
1137
0
0.414
448
0
0.076
71.2
0
0.065
588
2
0.164
495
3
0.106
155
0
0.034
49.1
0
0.0
0.0

1
0.171
1500
2
0.331
1099
1
0.329
438
1
0.098
98.9
0
0.011
14.5
3
0.249
499
0
0.277
308
0
0.053
59.1
0
0.007
6.88
0
0.0
0.0

4
0.591
1206
4
0.717
951
3
0.161
194
3
0.010
12.1
0
0.0
0.0
0
0.193
172
2
0.082
64.4
0
0.001
0.86
0
0.0
0.0
0
0.0
0.0

I
0.597
579
7
0.514
654
0
0.045
68.2
0
0.0
0.0
0
0.0
0.0
                               7-113
-------
                            TABLE 7-27.  (continued)
Age at
Initial
Employment
Low Exposure
<20


20-29


30-39


40-49


50+


Obs
Exp
Pyr
Obs
Exp
Pyr
Obs
Exp
Pyr
Obs
Exp
Pyr
Obs
Exp
Pyr
Duration of Employment ( years)
0-9
Level Group
0
0.056
8524
0
0.115
9951
0
0.390
5218
2
1.29
3703
3
1.62
1945
10-19

0
0.117
5249
0
0.334
4724
3
0.802
2218
1
1.18
1319
2
0.385
371
20-29

1
0.478
4038
2
0.892
2965
1
0.937
1364
1
0.344
386
0
0.041
65.4
30-39

1
1.59
3175
5
1.74
2117
0
0.662
715
1
0.035
52.7
0
0.0
0.0
40+

3
1.57
1376
6
0.796
834
1
0.062
74.6
0
0.001
2.00
0
0.0
0.0
 Source:   Brov/n  and  Chu  (1983a).

 exposure  areas, as  given in a  NIOSH criteria document (1975), are shown in
 Table 7-28.
     In the  absence of information  to  the  contrary,  it is assumed that the
 person-hours  spent  in each area are  equal.   Thus  an estimate of the  time-
weighted average for workers in the  light exposure category is

     1/3 x .7 + 1/3 x .17 + 1/3 x .004 = .291 mg As/m3.

     Under the  linear assumption, equation  14 may  be  expressed in terms of mg
As/ms working exposure by dividing by .291,  which gives the result
          Kd,to) = 2.07 x 10"13 [(d + to)5'8 - t 5'8].
(15)
                                7-114
-------
  TABLE  7-28.
ARSENIC EXPOSURES:  1965 SMELTER SURVEY ATMOSPHERIC ARSENIC
         CONCENTRATIONS (mg/As/m3)
          "Heavy  exposure  area"  as  classified by Lee and Fraumeni
Arsenic







Roaster Area
0.10
0.10
0.10
0.10
0.10
0.10
0.17
0.20
0.22
0.25
0.35
1.18
5.00
12.66
Mean: 1.4/
Median: 0.185






          "Medium exposure areas" as classified by Lee and Fraumeni
Reverberatory Area
          0.03
          0.22
          0.23
          0.36
          0.56
          0.63
          0.66
          0.76
          0.78
          0.78
          0.80
          0.83
               0.93
               1.00
                .27
                ,60
                ,66
                .84
                ,94
                .06
                .76
                .40
               4.14
               8.20
                                                  Mean:
                                                Median:
                                          1.56
                                          0.88
1.
1.
1.
1.
1.
2.
2.
3.
Treater Building and Arsenic  Loading
          0.10                0.48
          0.10                0.62
          0.10                3.26
          0.11                7.20
                                                  Mean:
                                                Median:
                                          1.50
                                          0.295
                "Light  exposure  areas"  as  classified by Lee  and Fraumeni
 Copper  Concentrate  Transfer System
           0.25
           0.65
           1.20

 Samples from Flue Station
           0.10
           0.24
 Reactor Building
           0.001
           0.002
           0.002
           0.002
          0.003
          0.009
          0.010
                                                  Mean:
                                                Median:
                                                  Mean:
                                                Median:
                                                  Mean:
                                                Median:
                                           0.70
                                           0.65
                                           0.17
                                           0.17
                                           0.004
                                           0.002
 Source:  Table X-3, NIOSH Criteria Document (1975).

                                 7-115
-------
      In this case,  exposure is expressed in mg As/m3 per 8-hr working day.   To
 change the relationship so  that  it expresses the risk  due  to a lifetime of
 continuous exposure  to  1 ug  As/m3,  we  assume' 240 days worked  per  year,
 one mg As/m3 on the job  gives  the same  cumulative exposure as 103 x
 1/3 x 240 = 219 |jg  As/m3 continuous exposure.
       365
      The age-specific rate due to a continuous 1 (jg  As/m3 exposure is obtained
 by  substituting tQ = 0 and d = t = age into equation (15) and dividing by 219
 to  arrive at the correct  number  of exposure  units.   This gives  the  result

                                                                       (16)
     r(t) = 9.45 x 10"16 t5'8.
     The  unit  risk  is  approximately  equivalent  to  the  risk  of  induced  respira-
tory cancer in the median  life  span.  Based  upon  1976 U.S. vital statistics,
the median  life span is 76.2 years,  so that the unit risk is expressed approxi-
mately as
                                                                           (17)
     An additional  approximation consistent with the  previous  unit calcula-
tions is obtained by assuming that
r(t) = 9.45 x 10
                               ~16
                                    t. + t.-!15.8  t. , < t l    *j ___ J      J J-       J
(18)
where t.  are  the ages at the interval boundaries given in U.S. vital statis-
       J
tics  records.   This  assumes  that the age-specific  death rate due to  the
exposure  is constant  throughout the interval and equal to the true value at
the midpoint  of the  interval.   Under  this approximation, using 1976 vital
statistics, the unit risk is estimated to be P ^ 1.25 x 10~3.
                                7-116
-------
Evaluation of Goodness-of-Flt.   It is desirable to assess whether the data for



the low-exposure group  is  consistent with the model utilized to estimate the



unit risk.  However,  two  factors tend to create  a  situation that would de-



crease this goodness-of-fit and bias the results.   First, the exact values for



each cell  of  d,  t  are presently not available.  Second, the value  of  k was



determined on the  basis of all  three exposure  groups,  and does not give as



good a fit as would be obtained  using the low-exposure group alone.



     Brown states (personal communication, 1983) that the exact values of d,tQ



are very close  to the  midpoint  of  the  interval,  and that  the  values of k



appear to  be  statistically consistent between exposure  groups.   Thus, distor-



tions of the  data because of  the use  of midpoint values and average k, al-




though inevitable, are  not appreciable.



     The expected  number  of cases in each cell  are  calculated  using  the rela-




tionship
          E - Exp +  Pyr x  . 603 x lo"13  [(d + t  )5<8  -tj5'8],
(19)
 in  which Exp and Pyr  are  taken  from Table 7-27  and d,tQ are the midpoints of



 the intervals.   These results are shown in Table 7-29.   A standard chi-square



 goodness-of-fit  test is then run,  resulting  in  a chi-square value  of  X§! =



 13.85  with 23.2 = 21  degrees  of freedom and an  associated  p-value of .88.



 Unfortunately,  due  to the  low expected number of cases  in many of the  cells,



 the X2 approximation is  of questionable validity for  this situation.



     To obtain a more stable approximation, cells with low frequency that are



 as  close as possible  to each other are usually  combined.   It  is  important to



 have some criteria  for  combining the  data that  do not  depend  upon inspection



 of  the data itself.   Two  methods of combining  the data  are used  here.   The



 first   is across  columns (duration exposed),  so  that the maximum  number of
                                 7-117
-------
         TABLE  7-29.   OBSERVED  AND  EXPECTED  NUMBER  OF  RESPIRATORY CANCER
          DEATHS  FOR  EACH  CELL  IN THE  LOW-EXPOSURE  GROUP OF  TABLE 7-27 .
d
*o
18
25
35
45
55
5
0
.088
0
.258
0
.723
2
2.023
3
2.582
15
0
.314
0
.858
3
1.641
1
2.508
2
1.234
25
1
1.20
2
2.145
1
2.556
1
1.428
0
.418
35
1
3.504
5
4.358
0
2.791
1
.371

45
3
3.836
6
3.321
1
.497
0
.026

X|]. = 13.85, p = .88


cells are  obtained,  with the constraints  that  combined cells must have at

least three expected cases, and that all cells combined are consecutive within

a row.  The second approach uses the same technique within a column (age first

exposed).  The  results  are  shown  in Tables  7-30  and  7-31  respectively,  giving

chi-square values of  7.01 and 7.61, with  p-values of  .41 and  .38.  Thus the

assumed model is  shown  to be consistent with the observed low-exposure data.

7.3.2.2.4  The Enter!ine and Marsh (1982) study.  This  study included all men

(2802 in all) employed  at the Tacoma, Washington copper smelter for a year or

more during 1940-1964.  Their mortality experience was  observed through 1976.

The study  involved over 70,000  person-years of  observation, and  104  deaths

from cancer of the respiratory system were recorded (Table 7-32).   Respiratory

cancer deaths had an  SMR of 189.4, which was significantly increased at the

1 percent level.  Expected  deaths  for Table 7-32 were  based upon U.S. white
                                7-118
-------
         TABLE 7-30.  CELLS FROM TABLE 7-29 COMBINED WITHIN ROWS TO
     OBTAIN CELLS WITH THREE OR MORE EXPECTED RESPIRATORY CANCER DEATHS
d
to 5 15

18

25

35

45

55
25 35
2
5.106
2 5
3.261 4.358
4
4.920


5
4.234
45
3
3.836
6
3.321
1
3.288
5
6.356


X2 = 7.01,  p = .41
        TABLE 7-31.   CELLS FROM TABLE 7-29 COMBINED WITHIN COLUMNS TO
        OBTAIN CELLS WITH 3 OR MORE EXPECTED RESPIRATORY CANCER DEATHS
d
to 5 15
18
25
25 35 45
1 3
3.504 3.836
3 5
3.345 4.358
35
45
5 6
55 5.674 6.555
1 7
3.162 3.844
2
4.402
X2 = 7.61, p = .38
                                7-119
-------
        TABLE  7-32.   OBSERVED DEATHS AND SMRs  FOR 2802 SMELTER WORKERS
  WHO WORKED A YEAR  OR MORE 1940-64, FOLLOWED  THROUGH 1976, BY CAUSE OF DEATH
Cause of death (7th revision code) Observed Deaths SMR
All causes of death
Tuberculosis (001-019)
Malignant neoplasms (140-148)
Buccal cavity and pharynx (140-148)
Digestive organs & peritoneum (150-159)
Esophagus (150)
Stomach (151)
Large intestine (153)
Rectum (154)
Biliary passages and liver (155-156)
Pancreas (157)
All other digestive organs (residual)
Respiratory system (160-164)
Larynx (161)
Bronchus, trachea, and lung (162-163)
All other respiratory system (residual)
Prostate (177)
Testes and other genital (178-179)
Kidney (180)
Bladder and other urinary organs (181)
Malignant melanoma of skin (19)
Eye (192)
Central nervous system (193)
Thyroid gland (194)
Bone (196)
Lymphatic & haematopoietic (200-205)
Lymphosarcoma and reticulosarcoma (200)
Hodgkins1 disease (201)
Leukemia and aleukemia (204)
Other lymphopoietic tissue (202, 203, 205)
Other malignant neoplasms (residual)
Benign neoplasms (210-239)
Diabetes mellitus (260)
Stroke (333-334)
Heart disease (400-443)
Hypertension without heart disease (444-447)
Nonmalignant respiratory disease (470-527)
Influenza and pneumonia (480-493)
All other respiratory diseases (residual)
Ulcer of stomach and duodenum (540-541)
Cirrhosis of liver (581)
Chronic nephritis (592)
External causes of death (800-998)
Accidents (899-962)
Suicides (963, 970-979)
Other external causes (residual)
Other causes of death (residual)
Unknown causes
1061
4
231
7
65
3
17
21
9
3
11
1
104
2
100
2
11
1
6
4
0
1
3
0
2
17
4
2
6
5
10
2
12
91
412
1
60
24
36
7
22
6
81
61
17
3
85
47
103.2
27.6**
123.6**
110.7
108.9
66.2
122.1
120.4
122.4
64.1
106.0
71.6
189.4**
67.7
194.9**
305.0
79.0
92.6
133.3
63.0
—
492.7
59.8
--
175.0
93.8
93.2
83.9
78.7
130.4
82.3
78.6
84.8
111.4
92.5
18.8
108.6
92.9
122.4
75.5
101.9
87.5
94.2
100.6
84.8
56.2
86.1
—
* p <.05, ** p <.01
Source:  Enter!ine and Marsh (1982).
                                         7-120
-------
male mortality  rates.   The respiratory  cancer  SMR increases to 198.1 when
Washington State mortality rates are applied.
     Enterline and Marsh  estimated individual exposures to  airborne  arsenic
using individual work  histories, urine arsenic measurements,  and an estimated
correlation between  exposure to airborne arsenic  and  resulting  levels of
arsenic in urine.  Average urine arsenic levels were available by department
for the years 1948-52, 1973,  1974,  and 1975.  Linear  interpolations were used
to estimate levels between 1952 and 1973.   Levels during  1949-1952 were as-
sumed to  hold prior  to that time.   By coupling these data with employee work
histories, Enterline and  Marsh  estimated individual cumulative exposures for
various times in units of |jg-years/£ urinary arsenic.
     Pinto et  al.  (1977)  compared  airborne  concentrations of arsenic with
urinary arsenic  levels for 24 workers wearing personal  air samplers for 5
successive days.   A regression analysis of these data showed a highly signifi-
cant linear  correlation  between airborne and urinary  arsenic  (p  < 0.01).
Average airborne arsenic  in  units  of |jg/m3  was  estimated  to be 0.304 times
average urinary arsenic levels in units of [nq/SL.
     To investigate  dose-response, Enterline and Marsh divided  the  total
person-years  of  observation  into. 5 groups by cumulative arsenic exposure (0
lag), and also by cumulative arsenic exposure up to 10 years prior to the year
of observation (10-year lag).   In this type of analysis, as a worker continues
to be exposed to arsenic,  he or she will contribute person-years to progres-
sively higher exposure categories.   The  numbers of respiratory cancer deaths
and corresponding expected numbers  for each of these groups are given in Table
7-33.  In this table,  urinary arsenic levels provided by Enterline and Marsh
have been converted  to airborne exposures in ug/ms-years,  using  the  factor
0.304 estimated  by Pinto  et  al. (1977).   The observed numbers of cancers are
all significantly increased at the higher exposure levels.

                                7-121
-------
          TABLE 7-33.  DATA FROM TABLE 8 OF ENTERLINE AND MARSH (1982)
                         WITH PERSON-YEARS OF OBSERVATION ADDED
Cumulative Exposure9
(jg/m3-years

91.8
263
661
1381
4091

91.8
263
661
1381
4091
Person-Years .
of Observation

10902
21642
14623
13898
9398

27802
16453
11213
9571
5423
Observed
Deaths
0 Lag
8
18
21
26
31
10-Year Lag
10
22
26
22
24
Expected
Deaths

4.0
11.0
10.3
14.1
12.7

6.4
12.5
11.5
12.4
9.7
 Exposures are  in |jg/ms-years  estimated  by  the  formula  (I ug/1-years)  (0.304)
 where I  is mean urinary  exposure  index  from  Enter!ine  and Marsh  (1982) Table
 8 and 0.304  is the relation  between  urinary and  airborne arsenic estimated
 by Pinto et al. (1977).

 Furnished by Dr.  Enter!ine (personal communication).


     Although the Enter!ine and  Marsh  (1982)  study is not as  large as  that of

Lee-Feldstein (1983),  it  does  involve a sizable number of respiratory cancer

deaths (104).   Workers  were followed for an extended period—an average of 25

years per  individual.   Several features of the analysis render it more amen-

able to  quantitative  risk estimation than the analysis  used by Lee-Feldstein.

Enter!ine  and Marsh made  estimates  of individual  exposure histories, whereas

Lee-Feldstein did not.  The  type of dose-response analysis used by Enter!ine

and Marsh  is  also  more suitable for quantitative  risk estimation.  The expo-

sure estimates  based  on a 10-year  lag  probably yield  a more realistic dose

response than those that do not utilize a lag.  Because the latency period for

respiratory cancer  is generally greater than 10 years  (cf.  Doll  and  Peto,
                                7-122
-------
1978), exposure during  the  last 10 years prior  to  observation would not be
expected to affect respiratory cancer mortality.
     By applying urinary arsenic measurements made during the years 1948-52 to
earlier years,  Enter!ine and Marsh probably  underestimated exposures prior  to
194&.  This would  result  in an overestimate  of  the carcinogenic potency of
arsenic.   A calculation by  Enterline and Marsh  of  SMRs  by both year of  hire
and  cumulative  exposure indicates that workers in a given cumulative exposure
category tend to  have at  least roughly  comparable  SMRs  irrespective of the
year of hire.   This suggests that exposure estimates for earlier years are not
greatly in error.   However,  further investigation  of this problem would be
useful.
     Because smoking  is also an important risk factor for respiratory cancer,
it would have been helpful  if data  on smoking habits had been available for
analysis.   Pinto and  Enterline  (undated) report  on  smoking histories obtained
in 1975 from 550 active employees at the Tacoma  smelter.  Of  these  employees,
59.6 percent were  active  smokers,  compared  to 45.4 percent in 1970  for  U.S.
males aged 21-64.   If this excess of smokers  holds  in general  for  the smelter
workers, then a small fraction  of the excess  in  respiratory cancer  could have
been due to smoking.
     Development of risk estimates:   The data from  Table 7-33 were  used for
quantitative risk  assessment.   The  relative  risks  (observed/expected)  from
this  table  are  graphed in  Figures  7-6  and  7-7, and  the absolute  risks
([observed- expected]/person-years) in Figures 7-8 and 7-9.   Although there is
no clear trend  of  increasing SMRs with increasing exposures,  such a trend is
present for  absolute risk.   Results of  applying chi-square  goodness-of-fit
tests of the  relative-   and absolute-risk models with  k =  1 and  k = 2  are
recorded in Table  7-25.  The maximum likelihood  estimates of  the carcinogenic
                                7-123
-------
    3-
cc

§
1
Q)
OC
                                                        Dose-response data
                                                        is from Table 7-33.
                                                          Fit is by relative
                                                           risk model.
    o-
1000           2000            3000

    Cumulative Dose (//g/m3- years)
                                                                       4000
   Rgure 7-6. Relative risks and 90% confidence limits for zero-lag data of Enterline
   and Marsh (1982).
                                   7-124
-------
0)

•i-j
JO
-------
o

X

(/>
E
o
 O
 (0
.Q
                                                         Dose-response data
                                                          is from Table 7-33.
                                                          Fit is by absolute
                                                             risk model.
                     1000            2000            3000

                         Cumulative Dose (//g/m3- years)
4000
   Rgure 7-8. Absolute risks and 90% confidence limits for zero-lag data of Enterline
   and Marsh (1982).
                                   7-126
-------
o
o
o
to
ir
0
•*-"

"o
CO
                                                          Dose-response data
                                                           is from Table 7-33.
                                                           Fit is by absolute
                                                              risk model.
                      1000            2000            3000
                          Cumulative Dose Ot/g/m3- years)
4000
  Rgure 7-9. Absolute risks and 90% confidence limits for 10 year lag data of Enterline
  and Marsh (1932).
                                    7-127
-------
potency parameter  a1  are also listed in Table  7-25.  The maximum-likelihood
fits of these  models  are graphed in  Figures  7-6 through 7-9.  The  quadratic
fit  (k = 2) is  poor  for both the absolute-  and  relative-risk models with
either 0 lag or 10-year lag data (p less than 0.001 in each case).  The linear
fits to the  relative-risk models are  also  relatively poor (p = 0.02 for the 0
lag  data and 0.006 for the 10-year lag data).  On the other hand, the linear
fits of the  absolute-risk  model  are  all acceptable (p = 0.24  for 0 lag data
and 0.14 for 10-year lag data).
     The unit  risks  (additional  risks of respiratory cancer death from life-
                       3
time exposure  to 1 ug/m  airborne arsenic)  obtained from each  of  the  fits  for
which the chi-square p-value is 0.01 or higher, are also listed in Table 7-25.
These  risks  were estimated  by  applying  (1)  or (2), (8), and (9) with  D. based
upon a constant exposure of 1 (jg/m .   Specifically,
                            = 4.56[5(i-l) + 2.5]
was used for the 0 lag data, and
                            = 4.56[5(i-l) - 7.5]
(17)
(18)
was used for the 10-year lag data.  These D. represent the cumulative exposure
        3                                               3
in ug/m  resulting from a  constant  exposure to 1 ug/m   from  birth to the
midpoint of  the ith  5-year age  interval.   The factor 4.56 converts  from
occupational  to  environmental  exposures,  and is explained at  equation (11).
7.3.2.2.5  The Ott et al. (1974)  study.  Ott et al.  (1974) compared the age-
specific death patterns of  174 decedents exposed to  arsenic  in the  production
of pesticides to those  of 1809 decedents who were not exposed to arsenicals.
By fitting the  death  patterns of the  unexposed decedents to a mathematical
function, an estimate was obtained of the probability that a death at a parti-
                                7-128
-------
cular  age  and during  a particular  epoch  was due  to respiratory cancer.



This  function  was used  to estimate expected  respiratory cancer deaths in



various exposure  categories  for the exposed decedents.  Cumulative exposures



were estimated for  exposed decendents, using work histories and estimates of



average exposures in various jobs.  The exposure estimates were made by indus-



trial  hygienists  familiar  with the processes.   Expected  cancer deaths were



compared with  observed to  obtain observed-to-expected ratios.  Table  7-34



shows the  results of Ott et al.'s  dose-response analysis.  The data in this



table are  all  reproduced  directly from Table  4  of Ott et  al.  (1974),  except



for the cumulative  exposures.   Average total exposures in mg provided by Ott



et al.  were converted to cumulative exposures in |jg/m3 years by multiplying by



the factor
                                  1000 ug/mg
                    (4 m3/day)(21 days/mo)(12 mo/year)
(19)
          TABLE 7-34.   DATA FROM TABLE 4 OF OTT ET AL.  (1974)
Cumulative Exposure3
|jg/m3-years
41.8
125
250
417
790
1544
3505
6451
29497
Exposures are in pg/ms-years

Observed
Deaths
1
2
4
3
3
2
3
5
5
estimated by:
d mg x 1000 ng/mg
Expected
Deaths
1.77
1.01
1.38
1.36
1.70
0.97
0.77
0.79
0.72


                    (4 m3/day)(21 days/month)(12 months/year)



     where d is  average total  exposure from Table 4 of Ott et  al.
                                7-129
-------
The values  included  in this  factor  are  not  in doubt because use of the factor
simply negates the calculation of total exposure made by Ott et al.
     Decedent studies such  as this  are more subject to bias then prospective
studies such  as  those of  Lee-Feldstein  (1983) and  Enterline and Marsh (1982).
If, for example, in some age category arsenic exposure increased the mortality
from  some  other disease  in addition to respiratory  cancer,  an  analyis  of
decedents might  show an  artificially low effect of arsenic upon  respiratory
cancer for  this  age group (because  there might be an artificially large number
of total deaths).   It  is  also of some  concern that Ott et  al.  did not clearly
describe how  the study cohort was defined.
     This study  involved primarily  short-term exposures, as less than 25 percent
of the decedents had worked  with arsenicals for more than  one year.   Thus, this
study  is  less appropriate  for  estimating  risks  from lifetime environmental
exposure than a comparable  study involving longer exposures.   The study also
was  quite   small; only 28  respiratory  cancer deaths occurred  among  exposed
decedents.
      Development of  risk estimates:  The dose-response  data in Table 7-34 were
used  in  an assessment of risk.   Because  of the nature  of the study, only  a
relative risk model  could  be  applied  to  these  data.   The dose-response for
relative risk is graphed in Figure 7-10.   The  response in the  most highly
exposed  group falls far below  that predicted by the  lower-dose  data and  is
omitted  from Figure 7-10.   This  is possibly  due to the fact  that some of  the
more  highly exposed workers  wore respirators.   Because of this shortfall  in
response,  and also because  the exposures to this group were the furthest from
the  low-level environmental  exposures  of  interest, the data  from the  highest
exposure group were  omitted from the analysis.
      Results of applying chi-square goodness-of-fit  tests  of  the relative-risk
model with k =  1  and k  = 2 are listed in Table 7-25.   The maximum-likelihood

                                 7-130
-------
                                             CB
                                             CD
                                            T3

                                             O
                                            H-

                                            .±f
                                             CD
                                             O
                                             C
                                             Q)
                                             O
                                             o
                                             Si
                                             -^ a
                                             .2 3
                                             c o
                                             +3 0)
                                             05 0)
                                             iT 2
7-131
-------
estimates of the  carcinogenic  potency parameter a' are  also  listed in this
table.   The maximum-likelihood fits are graphed in Figure 7-10.   Both of these
fits are acceptable  (p = 0.66  for  k = 1 and p = 0.23 for k = 2), although the
data appear to be more linear  than quadratic.  It  should be kept in mind that
the sample size was  quite  small in this study, and consequently a wide range
of curve shapes would probably provide an acceptable fit.
     The risk estimation method described in the  previous  section  is based
upon the life table  method of  analysis, and does not seem particularly  appro-
priate for a  decedent analysis.  Because the  method employed  by Ott et al.
seems to estimate  a relative probability of respiratory cancer death, it was
decided to estimate the extra lifetime probability of respiratory cancer death
                                  3
from lifetime  exposure to  d pg/m  airborne arsenic,  using the expression

          P0(l + a'[(72)(4.56)d]k) - PQ = PQa'[(72)(4.56)]k.          (20)

Here PQ  is  the lifetime probability of respiratory cancer mortality given by
(8), and is equal to 0.0451 if  1976 U.S. mortality rates are used.  The factor
72  represents  life  expectancy  in the  U.S.  in years.  The factor 4.56  converts
from occupational  to environmental exposures, and is  explained at equation
(11).  Thus,  the  term in the  square  brackets  in  (20)  represents the  average
                                         3
total  exposure  over a life  span in  ng/m  years,   which  is  the  same as the
measure used in estimating the  potency a1.
7.3.2.3  Discussion—Table  7-25  summarizes  the  fits of both absolute-  and
relative-risk models, with either  k = 1 or k = 2,  to dose-response  data from 4
different  studies.   Table  7-25  also displays  the  carcinogenic  potencies  a1.
It  should  be  noted that the potencies estimated from  different models  are  in
different units, and are therefore not comparable.
     In  every  case, a linear  model  (k = 1) fitted the  data better than the
corresponding quadratic model  (k = 2).  In every case  but two,  the  fits of  the

                                 7-132
-------
quadratic model  could be  rejected  at the 0.01 level.  The  two  exceptions


involved the two smallest  data sets (Higgins et al. absolute risk, and Ott et


al.) and in the  former case the fit  was  very  marginal  (p = 0.017).   On the


other hand, for each  data  set a linear model provided an adequate fit.  Also,


in every case, an  absolute-risk linear model fit  the data better than the


corresponding relative-risk linear  model.   The  p-values for the fits of the


absolute-risk linear model  ranged from 0.025 to 0.75.


     The estimated  unit  risk  is presented for each fit for which the chi-


square goodness-of-fit p-value  is  greater than 0.01.   The unit risks derived


from linear models—8 in all — range from 0.0013 to 0.0136.   The largest of


these is from the  Ott et al.  study, which probably is the least reliable for


developing quantitative estimates,  and which also involved exposures to penta-


valent arsenic, whereas  the other studies  involved trivalent  arsenic.   The


unit risks derived from the linear (k = 1) absolute-risk models are considered


to be the most reliable;  although derived from 5 sets of data involving 4 sets


of investigators and 2 distinct exposed populations, these estimates are quite


consistent, ranging from 0.0013 to 0.0076.


     To establish a single point estimate, the geometric mean for data sets is


obtained within distinct exposed populations, and  the final  estimate  is taken


to be the  geometric mean of those  values.   This  process is illustrated in


Table 7-35.

                                    -3                           3
     The final estimate  is 4.29 x 10  , where exposure is in pg/m  of continu-

                                         3
ous exposure.  Based  upon an assumed 20 m  tidal volume of air and a 30 percent

                                                           -3
absorption rate, this amounts  to a unit  risk  of  4.29 x 10   -=-  (.3  x 20 x


.001/70) = 50.1 in units of mg/kg absorbed dose per day.


     Although the estimates derived from the various studies are quite consis-


tent, there are a number of uncertainties associated with them.  The estimates
                                7-133
-------
   TABLE 7-35.  COMBINED UNIT RISK ESTIMATES FOR ABSOLUTE-RISK LINEAR MODELS
Exposure Source
Anaconda smelter
ASARCO smelter
Study
Brown & Chu
Lee-Feldstein
Higgins
Enter! ine &
Marsh
Unit Risk
1.25 x 10~,
2.80 x 10~^
4.90 x 10
6.81 x 10~3
7.60 x 10
Geometric
Mean Unit
Risk
2.56 x 10~3
7.19 x 10"3
Final Estimated
Unit Risk
4.29 x 10"3
were made from occupational studies that involved exposures only after employ-



ment age was reached.  In estimating risks from environmental exposures through-



out life, it  was  assumed, through either the  relative-risk  model  (1)  or the



absolute-risk model  (2), that  the increase in  the age-specific mortality



rates of lung cancer was a function only of cumulative exposures, irrespective



of how  the  exposure  was accumulated.   Although  this  assumption  provides  an



adequate description of  all  of the data, it may be in error when  applied to



exposures that begin very early in life.  Similarly, the  linear models pos-



sibly are  inaccurate at  low exposures,  even  though they provide  excellent



descriptions of the experimental data.



     The risk assessment methods employed were  severely  constrained  by the



fact that they were based only upon the analyses performed and reported by the



original authors—analyses  that had been performed for  purposes  other than



quantitative  risk assessment.   For example,  although other measures of expo-



sure might  be more  appropriate,  the analyses were necessarily  based upon



cumulative  dose,  since  that was  the only usable  measure reported.   Given



greater access to the data from these studies,  other dose measures, as well  as



models  other  than the  simple relative-risk and absolute-risk models,  could be
                                7-134
-------
studied.   It is possible  that such wide analyses would  indicate  that other



approaches are more appropriate than the ones applied here.



7.3.3  Unit Risk for Water



      The best data  available for making quantitative cancer risk estimates



for  ingestion  of  arsenic in  water  are  the data collected by Tseng  et al.



(1968).   They surveyed a  stable population of 40,421  individuals who  lived  in



a rural  area  along the  southwest coast of Taiwan and who were known  to have



consumed drinking  water  containing arsenic.   The occurrences of skin cancer



among this population, and the arsenic concentrations in their drinking water,



were measured.  Since the population was stable, the  study can be viewed as a



lifetime  feeding  study,   and  the  data  may  be used to predict  the lifetime



probability of skin cancer caused by the ingestion of arsenic.



     A model  estimating  the cancer rate as a function of drinking water arse-



nic  concentration  was generated  using  information from the above  study in its



published form, which is  a summary of data collected  by the investigators.   If



the  original data  had been available, a more exact mathematical analysis would



have been possible.



     Doll  (1971)  has suggested that the  relationship  between the  incidence  of



some site-specific cancers,  age, and exposure  level  of  a  population may be



expressed as:
                              I(x,t) =
(21)
where  x is the exposure level (which can be measured by the water concentra-



tion  in ppm),  t is the  age  of the population,  and B,  m,  k are unknown para-



meters.



     However,  the  data collected by Tseng  et  al.  (1968)  was  obtained at one



point  in time, and since skin cancer has only  a marginal  effect on the  death
                                 7-135
-------
 rate,  the obtained rates may  be  viewed  more  accurately  as the probability of
 having contracted skin cancer by time t.  The relationship between this prob-
 ability,  often  referred  to  as  the cumulative  probability density or prevalence
 F(x,t), and the incidence or age-specific  or  hazard  rate, may be expressed as:
F(x,t) = 1 - exp [-
                                                     ds].
(22)
     Utilizing equation  (21) as the  form of the  incidence rate, the prevalence
may be expressed as
                           F(x,t) = 1 - exp  (-Bxmtk),
                                             (23)
which is a Weibull distribution.
     In Table 7-36, adapted from information in Tseng, et al. (1968), estimates
are  given  of F(x,t)  for different  age  and water  concentration  groupings  for
males.  The  prevalence for females is  less  than  for males,  and therefore is
not  used to estimate risk.
     To use  this  data, specific values  for x and  t had to be obtained for the
intervals.  Where  the  intervals were closed, the midpoint was  utilized.  For
the  greater  than  0.6 ppm group, the  midpoint  between 0.6 and  the  greatest
recorded value, 1.8, was  taken, resulting in 1.2  ppm.   For  age  60  or greater,
a value of 70 was  utilized  somewhat arbitrarily,  being the  same increase  over
the  lower  level as that in  the  other  two  age intervals.  The values for (x,t)
to relate  to the  prevalence estimates are shown in parentheses  in  Table 7-36.
     From equation (23) it follows that
                                   = ln(B) + m ln(x)
                                             (24)
which  is  multiple-linear in form.   Estimating  the parameters by the  usual
least-square techniques, the following relationship is obtained:
                                7-136
-------
      TABLE 7-36.   AGE-EXPOSURE-SPECIFIC PREVALENCE RATES FOR SKIN CANCER
Exposure
in ppm3
0-0.29
(0.15)
0.30 - 0.59
(0.450)
>0.6
(1.2)

20-39
(30)
0.0013
0.0043
0.0224
AGE
40-59
(50)
0.0065
0.0477
0.0983

>60
(70)
0.0481
0.1634
0.2553
*Source:   Tseng et al.  (1968).
aRange given by authors.   Midpoint is in parentheses.
       ln( - ln[l - F(x,t)]) = 17.548 + 1.192 ln(x) + 3.881 ln(t),     (25)


which is  an excellent fit,  having a multiple correlation coefficient of 0.986
and a standard error on the exposure regression m of .138.
     Equation (25) may be expressed as
                     F(x,t) = l-exp-[2.429 x 10"8(X1'192)(t3'881)]
                              = 1-exp-CX1'192 H(t)]
(26)
If the  parameter m =  1.192  were  in  fact  equal  to  1,  then  for a  given  value  of
t, equation (26) would be "one-hit" in form.
     To test  this hypothesis (i.e., Ho: m = 1) the  student  "t" test  is used,
giving the result:

                         ^  _  1.192-1
                                 0.138
                                         = 1.391,
                                7-137
-------
which is  not  significant at the 0.1 level.  Thus, there is insufficient evi-
dence to reject the hypothesis that the dose-response relationship is "one-hit"
even at the 0.1 level.   However, a quadratic model would  be rejected at the
p<.001 level.
     Fixing no = 1, the following relationship is obtained:
                            F(x,t) = 1 - exp[-g(t)x]
(27)
Transforming this equation to  its  linear form  (as in equation 23) and obtain-
ing the least-square linear estimates of B and v, it is found that:

       g (t) = exp(-17.5393)  t3'853, where B = 2.41423 x 10"8,  k = 3.853.

     The data used to obtain these  estimates are shown  in Table 7-37, and the
goodness-of-fit is illustrated in Figure 7-11.

    TABLE 7-37.   DATA UTILIZED TO OBTAIN PREDICTOR EQUATION AND  FIGURE 7-11
ppm
Arsenic

Age at Medical
Examination

Skin Cancer
Prevalence

F(x
Observed
X
0.15


0.45


1.20


t
30
50
70
30
50
70
30
50
70

0.
0.
0.
0.
0.
0.
0.
0.
0.
Rate
0013
0063
0481
0043
0477
1634
0224
0983
2553
»
Rate
t)

-17.
Transformed Skin
Cancer
(-ln[l-
5393 +
Prevalence Rate
3.8531nt + Inx
Expected

0
0
0
0
0
0
0
0
0
Rate
.0031
.0127
.0455
.0053
.0375
.1304
.0141
.0969
.3110
Observed
6
5
3
5
3
1
3
2
1
. 64474
.03269
.00993
. 44699
. 01849
. 72368
. 78739
. 26844
.22155
Expected
6.33160
4.36341
3.06695
5.23299
3.26480
1.96834
4.25216
2.28397
0.98751
                                7-138
-------
   F(x,t) = -l
 0.0009-
 0.0025 -
 0.0067- -5.0
  0.0181- -4.0
 0.0486 - • 3.0
 0.1266- -2.0
 0.3078- -1.0
  0.6321 -1-0.0
             -2.
7.0
                   t=30
6.0
                    t=50
             t
          -1.6
           t
-1.2     -.8
 -.4
-h
.0      .4

i      I
                                                         logx
            0.135    0.202   0.301   0.449    0.670   1.000   1.492  x(ppm)
Figure 7-11. Relationship between transformed prevalence and log ppm arsenic in
water, log age.
                                   7-139
-------
      The  function
           F(x,t) = l-exp[-2.41423  x  lo"8  x  t3'853],
(28)
 is  the probability of contracting  skin  cancer by age t, given that an indi-
 vidual  had a life-time exposure  to x ppm in  his  drinking  water (and lived
 until  age  t).
     To  obtain  a unit risk  estimate,  lifetime risk  is assumed to be  approxi-
 mately  equal  to the risk to the  median  life  span  in the  absence of competing
 risk.   The unit risk is  thus obtained by substituting x  = 1 and  t = 76.2 (the
 median  U.S.  life span based upon 1976 vital  statistics  data) into equation
 (28).  This gives the result
               P(l) - l-exp[-2.414 x 10~8 x 76.23'853] = .350
(29)
The exponent is the slope estimate for cancer risks at low doses, so that:
          P(x)-.430 x     for small x, where x is in ppm.
(30)
     To express the unit in mg/kg/day exposures, it is assumed that two liters
of water  are consumed per day by  an individual  weighing 70  kg.   Under the
assumption that 100  percent of the  arsenic  is absorbed  through  the gut,  the
slope in units of mg/kg/day absorbed dose is .430 -r (.2 -r 70) = 15.8.
     A number  of potential  factors exist  that could possibly  make  the  Taiwan-
ese data unsuitable as surrogate data for the U.S.  population.  Among them are
racial, dietary, and  nutritional  differences.   Also, exposure to  ergotamine
was confounded with  arsenic exposure in  the well  water—a fact which also
could have modified the  results.   However, there is no direct evidence demon-
strating the role  of  these agents  in the carcinogenic  response  to ingested
arsenic.   Furthermore, a recent extensive review by Andelman and Barnett (1983)
                                7-140
-------
of the arsenic dose-response model developed here, demonstrates that presently
there is no  quantitative  evidence that is  inconsistent with the model.  The
Andelman and  Barnett  study  also showed that there does not appear  to  be any
population in the U.S. that could be studied that would have a reasonable power
to contradict the hypothesis that the Taiwanese dose-response model is consis-
tent with the U.S. dose response.
7.3.4  Relative Potency
     One of  the  uses  of the concept  of unit risk is  to compare  the relative
potencies of  carcinogens.   To  estimate relative potency on a per-mole basis,
the  unit  risk slope factor is  multiplied  by  the molecular weight,  and  the
resulting number  is expressed in  terms  of  (mMol/kg/day)-1.  This is called the
relative potency  index.
     Figure  7-12 is a  histogram  representing  the frequency distribution of
potency  indices  of 52 chemicals  evaluated by  the CAG as suspect carcinogens.
The  actual  data  summarized by the histogram  are presented in Table  7-38.
Where  human  data were available  for  a compound, they were used to calculate
the  index.   Where no  human  data were  available,  animal oral studies and  animal
inhalation  studies were used,  in that order.   Animal  oral  studies were selec-
ted  over animal  inhalation studies because they have been made on most of the
chemicals, thus  allowing  potency comparisons by  route.
      The potency index for  arsenic, based  on the Tseng et  al.  study,  is  2.25  x
103  (mMol/kg/day)"1.   This is derived by means of the slope estimate from the
Tseng et al.  study, which is  15(mg/kg/day)  .
      Multiplication by the molecular weight of 149.8 gives a potency  index of
 2.25 x 10+3.  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 7-12.   The
 index of 2.25 x 10+3  lies at the bottom of the first quartile of the 52 suspect
 carcinogens.
                                 7-141
-------
                                  4th        3rd      2nd        1st
                                 quartile    quartile   quartile    quartile
                                      1x10
4x10
2x10
                            246
                           Log of Potency Index
        I
        8
Figure 7-12. Histogram representing the frequency distribution of the potency
indices of 52 suspect carcinogens evaluated by the Carcinogen Assessment Group.
                                  7-142
-------
TABLE 7-38.   RELATIVE CARCINOGENIC POTENCIES AMONG 52 CHEMICALS EVALUATED
     BY THE CARCINOGEN ASSESSMENT GROUP AS SUSPECT HUMAN CARCINOGENS
Slope Molecular
Compounds (mg/kg/day)-l Weight
Acrylonitrile
Aflatoxin B,
Al dri n
Ally! Chloride
Arsenic
B[ajP
Benzene
Benzidine
Beryl lium
Cadmi urn
Carbon Tetrachloride
Chlordane
Chlorinated Ethanes
1 , 2-di chl oroethane
1, 1, 2- tri chl oroethane
1,1,2, 2- tetrachl oroethane
Hexachl oroethane
Chloroform
Chromium
DDT
Di chl orobenzi di ne
1 , 1-di chl oroethy 1 ene
Die! dri n
0.24(W)
2924
11.4
1.19x!0"2
15(H)
11.5
5.2xlO~2(W)
234(W)
4.86
6.65(W)
1. SOxlo"1
1.61
6.90X10"2
5.73x10
0.20 ?
1.42x10"
7xlO~2
41
8.42
1.69
1.04(1)
30.4
53.1
312.3
369.4
76.5
149.8
252.3
78
184.2
9
112.4
153.8
409.8
98.9
133.4
167.9
236.7
119.4
104
354.5
253.1
97
380.9
Order of
Magnitude
Potency (log-.Qx
Index Index
lxlO+1
9xlO+5
4xlO+3
gxio"1
2xlO+3
3xlO+3
4x10°
4xlO+4
4xlO+1
7xlO+2
2xlO+1
7xlO+2
7x10°
8x10"
3xlOn
3x10
8x10°
4xlO+3
3xlO+3
4xlO+2
lxlO+2
1X10+4
+1
+6
+4
0
+3
+3
+1
+5
+2
+3
+1
+3
+1
+1
+1
0
+1
+4
+3
+3
+2
+4
                             7-143
-------
TABLE 7-38.  (continued)
Slope
Compounds (mg/kg/day)-l
Dinitrotoluene
Diphenyl hydrazine
Epichlorohydrin
Bis(2-chloroethyl)ether
Bi s(chl oromethyl )ether
Ethylene Dibromide (EDB)
Ethylene Oxide
Formal dehyde
Heptachlor
Hexachl orobenzene
Hexachl orobutadiene
Hexachl orocycl ohexane
technical grade
alpha isomer
beta isomer
gamma isomer
Nickel
Ni trosami nes
Dimethyl ni trosami ne

Di ethyl ni trosami ne
Di butyl ni trosami ne
N-ni trosopyrroli dine
N-ni troso-N-ethyl urea
N-nitroso-N-methyl urea
N-ni troso-diphenyl ami ne
PCBs
Phenols
2,4,6-trichlorophenol
0.31
0.77
9.9xlO~3
1.14
9300(1)
8.51
0.63(1)
2.14xlo"2(I)
3.37
1.67
7.75xlO~2

4.75
11.12
1.84
1.33
1.15CW)

25.9(not by

43.5(not by
5.43
2.13
32.9
302.6
4.92x10
4.34

1.99X10"2
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

q*)74.1
*
q1)102.1
158.2
100.2
117.1
103.1
198
324

197.4
Order of
Magnitude
Potency Oog10)
Index Index
6X10+1
+2
1x10
gxio"1
2xlO+2
IxlO*6
2xlO*3
3X10*1
exio"1
ixio*3
5xlO+2
2xlO+1

1x10 f
3xlO~!o
5xlot,
4x10 ^
7xlO+1

2xlO+3
-uO
4x10 „
9x10 ~
2x10*
4x10 . /,
Sxioj4
1x10°
lxlO+3

4x10
+2
+2
0
+2
+6
+3
+1
0
+3
+3
+1

+3
+3
+3
+3
+2

+3

+4
+3
+2
+4
+4
0
+3

+1
     7-144
-------
                          TABLE 7-38.  (continued)
Order of
Magnitude
Slope Molecular
Compounds
Tetrachl orodi oxi n
Tetrachl oroethy 1 ene
Toxaphene
Trichloroethylene
Vinyl Chloride
Remarks:
1. Animal slopes
(mg/kg/day)-l
4.25xl05
5.31x!0"2
1.13
1.26xlo"2
1.75X10"2(I)

are 95% upper limit
Weight
322
165.8
414
131.4
62.5

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

on the linear
(log10)
Index
+8
+1
+3
0
0

multi-
•
          stage  model.   They  are  calculated  based  on  animal  oral  studies,
          except for  those  indicated  by  I  (animal  inhalation), W  (human  occu-
          pational  exposure),  and H (human drinking water  exposure).   Human
          slopes are  point  estimate,  based on  the  linear non-threshold model.

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

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


     Ranking of the  relative  potency indices  is  subject  to  the  uncertainty

involved in comparing  estimates  of  potency  for different chemicals  based on

different routes of  exposure  to  different species,  and using studies of dif-

ferent quality.   Furthermore,  all the indices are based on estimates of low-

dose  risk  using linear  extrapolation from  the observational range.  Thus,

these indices are  not valid for the  purpose  of comparing potencies  in the

experimental or observational range if linearity does not exist there.

7.3.5  Summary and Conclusions of the Carcinogenicity of Arsenic

7.3.5.1   Qualitative Summary—Human  studies  of the  effects of  arsenic from

smelters,  drinking water,  pesticide  manufacturing p"! ..:ts, and medicinals  have

                                 7-145
-------
been  conducted.   These are summarized in  Table  7-1.   Studies  of  five  indepen-



dent  smelter  worker populations  have  all  found  an  association between occupa-



tional  arsenic  exposure and lung cancer  mortality.   Several  of  the smelter



studies  have  found a dose response  both by intensity  and  by duration  of expo-



sure.   The  risk of lung  cancer mortality  in the high dose group of one study



of smelter  workers in Japan was found to  be 10 times that expected.  Some of



the recent  data from these smelter studies  suggest  that, if  the multistage



theory  of carcinogenesis  is correct,  arsenic may act  as a late-stage  carcino-



gen.   It is also possible to hypothesize  from  some of the data  that  arsenic



may be  a promoter.   Either of  these  hypotheses,  however, requires further



study.  In  addition,  some studies  of  communities  surrounding smelters have



found an association between geographic  proximity  to the smelter and lung



cancer mortality.



     Both proportionate mortality and cohort studies  of pesticide manufactur-



ing workers  have demonstrated  an excess  of  lung cancer deaths among  workers



in that occupation. One study of the population around a pesticide manufacturing



plant found that residents of the area surrounding the plant were also at an ex-



cess risk of lung cancer.   Several  case reports of arsenical  pesticide applicators



have also shown an association between arsenical  exposure and  lung cancer.



     A  study  of 40,000 persons in Taiwan  exposed to  arsenic  in  the drinking



water found  a significant excess prevalence  of  skin cancer over  that  of 7,500



other Taiwanese and residents of Matsu Island who drank water relatively free



of arsenic.   Water supplies in Chile and  Argentina were  also reported to be



the cause of arsenic-induced skin cancers.  In contrast,  studies  of populations



in the  United  States  exposed  to  relatively high  levels  of arsenic  in the



drinking water  by  U.S. standards  did not find any excess of skin  cancer.



     The difference  in  response  between  the Taiwanese and the United  States



studies may reflect the fact that the drinking water in the study area in Taiwan





                                7-146
-------
contained much higher  levels  of arsenic than did  the  drinking water in the
United States.  Furthermore,  the  Taiwanese waters also contained ergotamine-
like compounds.  Other differences between the two study areas include differ-
ences in  socioeconomic  status,  with the Taiwanese who developed skin cancer
belonging to a rather low socioeconomic strata which could have had ramifications
in terms of diet,  personal hygiene and medical care.   Racial differences may have
also contributed to the differences between the study results.  In addition to
these physical and cultural differences, the U.S.  studies were limited by small
sample sizes  and would  not  have detected  the  risk  from skin cancer that would
have been predicted, on the basis of arsenic ingestion, from a linear model of
the Taiwanese data.
     Persons  exposed to arsenical  medicinals have also been shown to be at a
risk of skin  cancer.
     Using the  International  Agency for Research  on  Cancer  (IARC) classifica-
tion scheme  for  evaluating carcinogens, the evidence  for arsenic as  a  human
carcinogen is considered sufficient.   This is evidenced by the high relative
risks, the  consistency  in findings in different studies, and the specificity
of tumor  sites (i.e., skin and  lungs).
     Consistent demonstration of arsenic  carcinogenicity in  test animals,
using different chemical  forms, routes of exposure, and various experimental
species,  has not  been  observed.    Nevertheless,  recent  data  indicate that
tumorigenicity and possibly carcinogenicity  can be demonstrated in animals if
the  retention of arsenic  in the lung  is increased.  The additional observation
that calcium arsenate  is only slowly cleared from the  lung, strongly suggests
that this agent may be  carcinogenic.   In  support of the recent animal studies,
is the observation that cultured Syrian hamster embryo cells  can be transformed
by direct exposure to sodium  arsenate.
                                 7-147
-------
7.3.5.2  Quantitative summary—Unit risks are estimated for both air and water
exposures  to  arsenic.   The air estimates were based on data obtained in five
separate  studies  involving three  independently  exposed worker populations.
Linear and quadratic dose-response models in both  the  absolute and  relative
form are fitted to the worker data.  It was found that for the models that fit
the data at the P =  .01 or better level that the corresponding unit risk esti-
                            -4              -2
mates ranged  from  1.05  x 10   to  1.36 x 10  .   However, linear models  fit
better than quadratic models  and absolute better than relative models.   Also
it was felt that  exposure to trivalent  arsenic  was more  representive of low
environmental exposure  than pentavalent  arsenic.  Restricting  unit risk  esti-
mates to  those  obtained from linear absolute models  where  exposure was  to
trivalent  arsenic  gives  a range of 1.25  x  10~   to  7.6 x 10   .  A weighted
average of the  five  estimates in this  range gave a  composite estimate of 4.29
x 10"3.
     An extensive  drinking water  study of the association between arsenic in
well water and  an  examination for skin cancer of a population who lived in a
rural area of Taiwan was  used to estimate the unit  risk for ingestion.   Using
the male population who appeared to be more susceptible, it was estimated that
the  unit  risk associated with  drinking water contaminated with 1  ng/£ of
                    ~4
arsenic was 4.3 x 10  .   To compare the air and water unit risks,  the exposure
units in both cases  were converted to mg/kg/day absorbed doses,  resulting in
unit risk estimates of 50.1 and 15.0,  respectively.
     The potency  of arsenic compared  to  other  carcinogens  is evaluated  by
noting that an  arsenic  potency  of 2.25 x 10    (mMol/kg/day)"1 lies in  the
first quartile of  the 52  suspect carcinogens that have been evaluated by  CAG.
7.3.5.3  Conclusions—Skin cancer and  lung cancer have been shown by numerous
epidemiologic studies to  have an  association  with arsenic exposure.   Arsenic
                                7-148
-------
has not definitively been found to be a carcinogen in animal studies,  however.



In applying the  IARC  criteria for evaluating a substance as to the weight of



evidence for human carcinogenicity, arsenic would be placed in group 1, which



IARC characterizes as "carcinogenic to humans."



     Using the linear  absolute  risk model, the composite estimate for cancer


                                         3
risk due to a lifetime exposure to 1 jjg/m  trivalent arsenic in the air is es-

                       -3

timated to be 4.29 x 10  .   The unit risk due to lifetime exposure to  1 ug/£ of


                                                       -4
arsenic in drinking water  is estimated to be  4.3  x 10  .   On the basis of



mg/kg/day absorbed dose, the unit-risk slope estimates for air and water are



50.1 and 15, respectively.   While  it  is unlikely that the true risks would be



higher than these  estimates, they could be substantially lower.   Compared to



other compounds on a mole unit basis, the carcinogenic potency for arsenic falls



towards the lower end of the first quartile.
                                7-149
-------
-------
                       8.  ARSENIC AS AN ESSENTIAL ELEMENT







     Mertz (1970) has set forth a set of logical criteria which trace elements



should obey in order to be considered physiologically essential for man and/or



animals.  One of  the most obvious of these, and one which  should be  readily



demonstrable, is  that  the element meet a unique requirement physiologically,



and, consequently, that  a deficiency in that element be  associated with de-



leterious effects.



     In the  case  of  arsenic, early reports attempting to show a nutritional



requirement for the  element  in animals were inconclusive  (Arsenic. NAS, 1977;



Underwood, 1977). . Part  of the problem was undoubtedly technical in  nature,



i.e., the difficulty of  carrying  out such studies  in an experimental  environ-



ment where rigorous  exclusion of  a ubiquitous element from  the diet is neces-



sary.  More  recently,  however, several carefully controlled studies have been



reported to have demonstrated nutritional essentiality for arsenic in at least



some mammalian species.



     Nielsen et al.  (1978) have  noted that deprivation of  pregnant  rats of



arsenic-supplemented diets  resulted in  offspring  showing such post-weaning



effects as  slow  growth,  enlarged spleens,  and increased red cell  osmotic



fragility.  Greater  perinatal mortality  among pups from arsenic-deprived dams



was also noted in a second experimental  group.



     In a recent review by Uthus  et al.  (1983),  the authors reported on studies



with chicks  that  suggest that arsenic influences arginine metabolism. It was



reported that arsenic  deprivation influenced the effects  of dietary arginine,



manganese and  zinc,  the  fluctuations  of which variously affected  growth,



kidney arginase,  plasma  alkaline  phosphatase,  plasma urea,  plasma uric acid
                                8-1
-------
and hematocrit.  The  authors  suggested that the  four components interacted to



affect the conversion of arginine into urea and ornithine.



     Anke et al. (1978) studied the nutritional requirements for arsenic using



goats and  mini-pigs and a semi-synthetic  diet containing  less that 50 ppb



arsenic.  Effects  attributed  to arsenic  deficiency in both species were seen



not only  in the adult animals  but  in their offspring.  Arsenic deficiency



increased the mortality of adult goats as well as altered the mineral profiles



for copper and manganese in the carcass.   Significant reproductive effects for



both arsenic-deficient goats  and mini-pigs included reduction of  the  normal



litter size.  Furthermore, the mortality of kids and piglets from the arsenic-



deficient groups was significantly higher than controls.  Manganese levels were



elevated in arsenic-deficient kids and piglets, but no perturbation of hemato-



logical indices (hemoglobin, hematocrit or mean corpuscular concentration) was



noted.  This is in contrast to the experimental observations with rats (Nielsen



et al., 1974), where decreased hematocrits, elevated iron content in spleen and



increased osmotic fragility of cells are seen.  Given the fact that the rat is



known to be an  anomalous  animal  model  for  arsenic  metabolism  (see  Chapter 4),



this difference is probably peculiar to this species.



     Schwartz  (1977)  has  noted  growth effects of  arsenite on rats fed an



arsenic-supplemented  diet,  with an optimal effect  seen  at  0.25  to 0.5 ppm.



Interestingly,  this worker noted that pentavalent  arsenic as  sodium arsenate



is less effective.



     Remaining  to  be  independently demonstrated is  a  physiological  role  for



arsenic, the  existence of any specific carrier agent  in the body, or arsenic



essentiality in man.



     A  feature  of  essential  element metabolism  is homeostatic  control of



levels  and  movement of a particular  element  i_n vivo.   From the information
                                8-2
-------
considered earlier, there is no effective absorption barrier for most soluble



inorganic arsenicals, but efficient  excretory mechanisms (kidney, hair) and



biotransformation appear to  provide some control  over any absorption-excretion



balance.
                                8-3
-------
-------
                  9.   HUMAN HEALTH RISK ASSESSMENT FOR ARSENIC

     This portion of, the report places the information in the earlier Chapters
into a  quantitative  perspective with regard  to  non-occupational  population
exposure and health effects of arsenic germane to such a population.
     Data for  levels of arsenic encountered by humans in air, water, food and
other sources, such-as cigarette smoking, were set forth in Chapter 3 and are
combined with  data on rates  of  intake  and rates of  absorption  to provide
information on the total  assimilation of arsenic on  a  daily basis.   Health
effects of  arsenic most  germane  to non-occupational  population exposures are
then summarized.   Generally,  these  are  chronic effects  associated with long-
term intake of relatively low levels of arsenic.  In the  case of hazardous
wastes, however,  some  health  effects of concern may  be  associated with acute
exposures;   therefore,  acute  and  sub-acute effects must  also  be  considered.
     The section  dealing with dose-effect/dose-response  data includes consider-
ation of various indices  of  internal exposure followed by quantitative data
for intake and population response.
     Populations  at  risk,  identified at least along  qualitative  lines,  are
included for discussion.
9.1  AGGREGATE EXPOSURE LEVELS TO ARSENIC IN THE U.S.  POPULATION
     Among  individuals of  the general population (not occupationally exposed
to arsenic), the main  routes  of exposure to arsenic  are typically via inges-
tion of  food  and  water,  with lesser exposures  occurring via inhalation.
Representative intake  figures  are  presented in Table 9-1.   Intake by inhala-
tion is augmented among smokers in proportion to the  level  of smoking.
                                9-1
-------
                TABLE 9-1.   ROUTES  OF  DAILY  HUMAN  ARSENIC  INTAKE
Route/level
Ambient air/0.006 Mg/m3(a)
Drinking waterA 10 pg/liter
Food/50 pg daily (elemental As)
Cigarettes/6 \ig in main-
stream smoke/pack^-6'
Total: < 60 pg nonsmokers

Rate
20 m3
2 liters
—
1/2 pack
1 pack
2 pack

Total intake
0.12 pg
< 20 Mg
50 Mg
O 1 1^1
s Mg
12 pg

Absorbed amount
0.036 Mg(b)
< 20 Mg(c)
40 Mg(d)
0.9 Mg(f)
1.8 Mg^fN
2.7 Mg

^National  average for 1981 (see Section 3.3.1)
^ 'Assumes 30 percent respiratory absorption (see Chapter 4).
'^Assumes total  absorption (see Chapter 4).
*• 'Assumes 80 percent absorption (see Chapter 4).
^Assumes 20 percent of cigarette content in inhaled smoke (see Chapter 4).
' 'Assumes 30 percent absorption of inhaled amount (see Chapter 4).

     Assuming a daily  ventilation rate of 20 m  ,  and  a national population
                                 q
inhalation average of  0.006 ug/m /As,  the total  daily  inhalation exposure for
arsenic can  be projected to be  approximately 0.12  ug.   Assuming 30 percent
absorption,  approximately  0.03 ug of  arsenic would be absorbed on a daily
average.
     Contribution  of  tobacco-borne arsenic to  the  respiratory burden would
depend  upon  the  rate  of cigarette smoking.  If one assumes a mass of 1 gram/
cigarette and an average tobacco value of 1.5 ppm, this yields 1.5 pg arsenic/
cigarette. With 20 percent of this amount  in mainstream smoke, the inhaled
amount for each pack of cigarettes would be approximately 6 ug arsenic, and of
this  amount, 40 percent  would be deposited in  the respiratory tract (see
Chapter 4).   Assuming  an absorption of 75 percent  of the deposited fraction,
one arrives  at  an absorption  of approximately  2 ug/pack of cigarettes or a

                                 9-2
-------
factor of  10 to 100 times  greater than intake for nonsmokers  in given  ambient


air  settings.  One may assume that the rates  of absorption  for trivalent  and


pentavalent  arsenic in the respiratory tract  are  equivalent.


     Since drinking water  arsenic is mainly  in  a soluble form (arsenate or


arsenite), virtually  all  of it is absorbed  in the GI  tract (see  Chapter  4).


Thus, assuming  an average  daily consumption  of two liters of  water  containing


at most  10 jjg As/liter as an  outside  high figure, one can  estimate that  the


total arsenic  absorbed  from drinking water would  be approximately 20 pg/day.


Most individuals  would,  in reality,  take  in much less  than  this amount, while


those in the Western U.S.  with well water supplies much higher  in  arsenic


content would assimilate proportionately more.


     Food arsenic values taken from the 1974  FDA survey indicate a daily total


dietary intake of approximately 50 ug elemental arsenic.  Based on information


presented  in  Chapter  4,  the major portion (80 percent) of food arsenic would


be absorbed  resulting in a net daily food arsenic absorption of 40 ug total.


     Thus,  a  non-smoker  would  have a total daily absorption from  all exposure


media of approximately 60  M9 arsenic/day  or  less.   Of  this, the diet would be


the major  contributor, assuming levels in water much below 10 ug/liter.   For


cigarette smokers, one would add 2 ug arsenic/pack of cigarettes smoked daily.


     If one views aggregate intake not in terms of total arsenic intake but in


terms of toxicologically significant forms  of the  element, then  much  of  the
            r\                                .-...•

dietary fraction,  for reasons  given earlier,  such  as complex  organoarsenicals


being present,  becomes relatively less important than the forms in water and


air  as well  as  in  cigarette  smoke.   Arsenic forms in  such media include:


pentavalent  arsenic  in  most water  supplies;   variable  mixtures of  tri-and


pentavalent arsenic in ambient air;  and probably an arsenic  oxide in cigarette


smoke.   From this view point,  utilizing the  examples already given above,  non-
                                9-3
-------
smokers would absorb 20 ng or less daily of toxicologically significant arse-
nic.   Heavy smokers having otherwise very low air and water exposure, conceiv-
ably could receive their major exposure via cigarettes.
9.2  SIGNIFICANT HUMAN HEALTH EFFECTS ASSOCIATED WITH AMBIENT EXPOSURES
9.2.1  Acute Exposure Effects
     Serious acute effects and  late sequelae from exposure  to  arsenic will
appear after  single  or  short-term respiratory or oral  exposures  to large
amounts of arsenic.   Available data indicate that  inorganic trivalent com-
pounds of  arsenic  are generally more acutely toxic than  inorganic  pentavalent
compounds,  which  in  turn  are more toxic  than organic  arsenic  compounds.
Serious effects will  also appear after long-term exposure to respiratory  or
oral doses of arsenic.
     The acute  symptoms  following oral exposure consist of  gastrointestinal
disturbances, which may be so severe that secondary cardiovascular effects and
shock  may  result  and cause death.  Also,  direct toxic effects  on  the  liver,
blood-forming  organs,  the central  and peripheral  nervous systems,  and the
cardiovascular  system may appear.  Some symptoms, especially those  from the
nervous system, may  appear a long time after exposure has ceased  and may  not
be reversible,  whereas  the other effects seem to be reversible.   Infants  and
young  children  especially are susceptible with regard to effects  on the cen-
tral nervous system.  The Japanese follow-up after the Morinaga  milk poisoning
showed that persisting  damage,  especially mental  retardation and epilepsy, is
a  late sequela in children  of short-term oral  exposure to  large doses of
inorganic  arsenic.   Among adults,  the  central nervous system is  not  as suscep-
tible, but peripheral neuropathy  has been  a common finding.
     Both  in adults  and children, acute oral exposure has resulted  in dermal
changes, especially  hyperpigmentation  and  keratosis, as  a late  sequela.
                                 9-4
-------
     Acute inhalation exposures have also resulted in irritation of the upper

respiratory tract,  even leading to nasal  perforations.

     Direct dermal  exposure to  arsenic may lead to,dermal changes; allergic

reactions may also  be involved.

9.2.2  Chronic Exposure Effects

     Both carcinogenic and non-carcinogenic effects are associated with long-

term exposures, which  do  not  cause any obvious  immediate  effects.   For the

purpose of this document,  such chronic effects will  be discussed in sequence

as follows:
          1.   Respiratory tract cancer
          2.   Skin cancer
          3.   Non-cancerous skin lesions
          4.   Peripheral neuropathological effects
          5.   Cardiovascular changes
     Cancer of  the  respiratory system is clearly associated with exposure to

arsenic  via  inhalation.   This  association  has been especially noted  among

workers  engaged  in  the production and usage  of  pesticides  and  among smelter

workers.  While  it  is  not  known to what  extent exposure  to  other  compounds in

industrial atmospheres has contributed  to  the  excess of  lung  cancer, the

Carcinogen Assessment  Group (CAG) has  concluded  that there  is sufficient  evi-

dence that inorganic arsenic compounds are lung  carcinogens in humans.

     Cancer of  the  skin  has been found as a dose-related effect  in a popula-

tion in  Taiwan,  with  lifetime  exposure to arsenic  in well water.   It has  also

been found among people  treated with  large doses of arsenite for skin disor-

ders.   Skin  cancer  often has a  long  latency  period on the  order  of decades,

the latency time decreasing with increasing intensity of exposure.  The CAG has

concluded that  there  is  sufficient evidence that inorganic arsenic compounds

are skin carcinogens in humans.
                                9-5
-------
     Hyperkeratosis  and  hyperpigmentation,  sometimes with precancerous changes,
have been  a common finding in persons ingesting arsenic.   These skin lesions,
as well  as  the manifest,cancer,  develop on skin surfaces  usually unexposed to
sunlight.   In  studies  in the  United States, an  association between skin lesions
or skin  cancer has  not been  demonstrated.   These studies  have  been  limited,
however, by sample sizes too  small to be able to  detect the dose response seen
in studies  outside the U.S.
     The effects on  the  peripheral  nervous system range from  sensory disturb-
ances to motor weakness  and  even paralysis.  The more severe signs  have been
noted in subacute poisonings, but more subtle changes after long-term low-level
exposure have  been found by using electromyography or measuring nerve conduc-
tion velocity.  These subclinical effects are slow in recovery and may persist
for years after cessation of  exposure.  In a study in Canada, electromyographic
(EMG) changes  were noted when water  concentrations of arsenic  exceeded 0.05
mg/1.
     Cardiovascular  effects have been noted especially in Taiwan, where Black-
foot disease  (peripheral vasculopathy)  occurred after long-term exposure  to
arsenic  in  well  water.  However, the presence  of ergotamine-like compounds
raises the  possibility of  vascular  effects from these agents.   Peripheral
vascular changes were  also found among German vintners who were exposed both
occupationally, by  spraying  arsenic-containing  pesticides,  and orally, by
drinking wine with elevated arsenic levels.  Studies on occupationally-exposed
persons have been  inconclusive  in  showing that arsenic causes an increase in
mortality from cardiac disease.
9.3  DOSE-EFFECT/DOSE-RESPONSE RELATIONSHIPS
9.3.1  General Considerations
     This section  generally attempts  to define, as presently feasible,  human
dose-effect/dose-response relationships for health effects of likely  greatest

                                9-6
-------
concern  at  ambient environment  exposure levels for  arsenic  in the United
States.  As  such, the  present section  highlights  mainly the quantitative
carcinogenic risk estimates that were derived in Section 7.3.
     The general  question  of how to define  and  employ a dose factor in at-
tempts at quantitative  assessments  of human health risk for  any  toxicant  is
highly dependent  upon:   1)  the available information  on  the body's  ability to
metabolize the agent, and 2) the assessment of the relative utility of various
internal indices of exposure.
     The time  period  over  which  a given total  intake occurs is highly impor-
tant.  For example, intake of one gram of arsenic over a period of years would
be quite different pathophysiologically  from assimilating this amount at one
time, the latter  probably  having a lethal outcome.   This time-dependent be-
havior is related  in  part  to the relative ability  of the body to  detoxify
inorganic arsenic by methylation as a function of both dose and time.
     In cases of acute and sub-acute exposure,  indicators of internal  exposure
such as  blood  or  urine  arsenic levels are probably appropriate for assessing
the intensity of exposure.
     With chronic,  low-level exposure,  however,  the available  data  would
indicate that  the total  amount assimilated is probably more important than an
indicator concentration  without  knowledge  of the total  exposure  period.   An
added problem is the background level  of arsenic found in these indicators  due
to dietary habits.  For  example, in acute exposures,  levels in blood or urine
would be  greatly  elevated  over  background values while low-level  chronic
exposures would only result in moderate increases over background.
     In regard to hair arsenic levels as an indicator of internal arsenic  ex-
posure, no reliable methods  exist  for distinguishing external contamination
levels from those accumulated via absorption  and metabolic distribution.  Hair
arsenic levels cannot, therefore, be employed as reliable indicators of either

                                9-7
-------
current or  cumulative  long-term exposures for  individual  subjects, but rather



may provide only a rough  overall  indication  of group exposure situations.



     Given  the  above  considerations and  limitations  concerning the use of



blood, urinary  or  hair arsenic concentrations as internal indices of cumula-



tive,  long-term  low-level  arsenic  exposures of concern here, the dose-effect/



dose-response relationships  summarized  below  are done so mainly in terms of



external arsenic exposure levels via either inhalation or ingestion.



9.3.2  Effects and Dose-Response Relationships



     It is difficult to define a precise acute lethal  dose of arsenic for man,



because such  exposure  situations  rarely allow accurate determination of the



effective amounts.  However,  for trivalent arsenic, the figure  is believed to



range from 70 - 180 milligrams.



     For  subacute  exposure,   it appears that  for children,  about  one  gram



assimilated over a  period  of 3-4 weeks will  induce death with severe effects



in survivors, while  for adults,  that dose will occasion significant clinical



effects.   In one poisoning episode, intake of approximately 50 milligrams over



a period  as short as two weeks resulted in clinically demonstrable effects in



adults.



9.3.2.1  Respiratory Cancer—A considerable number of studies have shown asso-



ciations between occupational  exposure  to arsenic and cancer of the respira-



tory system.  The best information available for making quantitative risk es-



timates for lung cancer are  derived from 5 sets of data  involving 4 sets of



investigators and 2 distinct exposed populations.   The 4 sets of investigators



are Brown and Chu  (1983a,b,c), Lee-Feldstein  (1983) and  Higgins (1982)--who



conducted studies on workers  at the Anaconda smelter in Montana—and Enter!ine



and Marsh (1982)—who  conducted  a study on the workers at the ASARCO smelter



in Tacoma, Washington.
                                9-8
-------
     Using an  absolute-risk  linear model and the  data from the four smelter

studies, the  lifetime  lung cancer risk, due  to  continuous  exposure  of 1  ug/

As/m3, was estimated  to range from 1.25 x  10 3  to 7.6 x 10 3.   A weighted

average of the five estimates* in this range gave a composite estimate of 4.29

x 10 3  (see  Section  7.3.2.3).   This represents  a  plausible estimate of the

upper limit  of risk—that  is, the  true  risk would  not  likely be  more than the
                                 /
estimated risk, but it could be substantially lower.

9.3.2.2  Skin  Cancer—Chronic  arsenic exposure, both  occupational and non-

occupational, is associated with a distinctive hyperkeratosis, which  is usually

followed by  a  later onset  of skin  cancer.  The best data available for making

quantitative risk estimates for skin cancer are the data collected by Tseng et

al.  (1968).   In this study, the authors surveyed a stable population  of 40,421

individuals who  lived in a rural area along the  southwest coast  of Taiwan and

who were known to have consumed drinking water containing arsenic. The occur-

rences of skin cancer among this population and  the arsenic  concentrations in

their drinking water were measured.  Since the population was stable, the  data

obtained from the study lends itself to predictions of lifetime probability of

skin cancer caused by the ingestion of arsenic.

     Using an  absolute-risk  linear model  and the data from  Tseng et  al. ,  the

lifetime skin cancer risk from drinking water containing 1 ug/liter of arsenic

was estimated  to  be  4.3 x 10 4  (see  Section  7.3.3).   It is not likely that

the true risk  for skin  cancer would be more than this estimated risk, but it

could be considerably lower.

9.3.2.3   Non-cancerous  Skin Lesions — As  noted above,  in  man, chronic oral

exposure to arsenic induces a sequence of changes in skin epithelium, proceeding

from hyperpigmentation to hyperkeratosis, characterized as keratin proliferation
*Two  risk  estimates  were derived from the  Enterline  and Marsh study based
 upon exposure periods lagged 0 and 10 years.   See Table 7-25.

                                9-9
-------
of a  verrucose  nature  and  leading,  in  some cases, to  late onset skin cancers.



These effects have  been  noted  both  in  populations which have ingested arsenic



via drinking water  and among people treated with large doses of arsenite for



skin  disorders.   In a report by  Pershagen and  Vahter (1979),  the authors,



using the  data from  a patient population exposed to  arsenic  via Fowler's



solution (Fierz, 1965), noted an increase in prevalence of hyperkeratosis with



increasing dose of arsenic.  The U.S. EPA is presently examining this informa-



tion, along with information from other studies, in order to determine whether



quantitative dose-response relationships, similar to those seen  for  skin



cancer,  can be established for these precancerous skin lesions.



9.3.2.4   Peripheral Neuropathological  Effects and Cardiovascular Changes  —



While the qualitative evidence for peripheral neurological  effects and cardio-



vascular changes  in arsenic-exposed  populations  is incontrovertible, the data



are insufficient  to establish  quantitative dose-response relationships at the



present time.



9.4  POPULATIONS AT SPECIAL RISK TO ARSENIC EXPOSURE



     In reviewing the  literature  dealing  with the acute, subacute and chronic



effects of arsenic in children and adults, the evidence suggests that children



may be  at  special  risk for the effects of inorganic arsenic under conditions



of acute or subacute exposure.



     In earlier  sections,  reference was  made to the  outbreak of pediatric



poisoning by arsenic  in  Japan  due to  the presence of  arsenic in  infant milk



formula (Hamamoto et  al. ,  1955).   From the clinical  reports published at the



time of the mass  poisoning as well as  those from follow-up studies,  a number



of signs of central nervous  system  involvement were noted both at the time of



the episode  and much  later, with  the  follow-up studies showing  behavioral



problems,  abnormal  brain wave  patterns, marked cognitive deficits, and severe



hearing loss in  some  of those children who survived the acute episode.   Some





                                9-10
-------
of these same tardive effects have also been noted in adults but appear to be
a much less constant  feature  of arsenic-induced neurotoxic effects than are
the peripheral  neuropathies.
     Because children consume more water  per body weight than  do adults,  the
daily intake of  arsenic  via  drinking water per kilogram body weight would be
greater in children.  This may  have  implications regarding chronic exposure
effects in children.   Zaldivar (1977) developed a regression equation describ-
ing this relationship.   It should be noted, however,  that serious health  ef-
fects due to chronic exposure of arsenic in drinking water have not been found
at a greater frequency in children than adults.
     Individuals residing in  the vicinity  of certain arsenic-emitting sources,
e.g., certain types of  smelters,  may be at risk for increased  arsenic intake
because of both  direct  exposure to arsenic in air and indirect exposure via
arsenic secondarily deposited from air onto soil  or other human exposure media.
The relative contribution of  such indirect exposures to increased risk of  these
individuals for  arsenic  health  effects  is difficult to define due to the lack
of information on this subject.   However,  it is most likely minimal  in relation
to the direct effects  arising from inhalation of arsenic, including lung cancer.
     As a  large  class of the general population at risk  for increased arsenic
intake, one would have to include cigarette smokers.   However,  it is not clear
to what extent some increased arsenic intake from tobacco smoke poses a speci-
fic heightened health effect  risk, although it is clear that internal indicator
levels, e.g. blood  arsenic,  are somewhat  elevated in  the  case of cigarette
smokers relative to nonsmokers.
                                9-11
-------
-------
                                10.  REFERENCES
Akland, G.  [Memo to D. Sivulka].   February 23,  1983.  Available  from:  U.S.
     Environmental  Protection Agency, Research Triangle Park,  NC;  Project  file
     no. ECAO-HA-79-5.

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

Andelman, J. B. and M. Barnett.  Feasibility  study to  resolve  questions  on the
     relationship of  arsenic in drinking water to skin cancer.  Environmental
     Protection Agency Cooperative Agreement  No.  CR-806815-02-1, 1983.

Andersen, 0.   Effects of  coal  combustion  products  and metal  compounds on
     sister  chromatid exchange  (SCE)  in a macrophage cell  line.   Environ.
     Health Perspect. 47:239-253, 1983.

Andreae, M.  0.   Arsenic speciation  in seawater  and  interstitial waters:   the
     influence of biological-chemical interactions on  the chemistry  of a trace
     element.  Limnol. Oceanogr. 24:440-452,  1979.

Andreae, M.  0.   Arsenic in  rain and the  atmospheric mass balance  of arsenic.
     J. Geophys. Res. 85:4512-4518,  1980.

Andreae, M.  0.   Biotransformation of arsenic in  the marine  environment.   In:
     Arsenic:   Industrial,  Biomedical and  Environmental  Perspectives.   W.
     Lederer and  R.  Fensterheim,  eds., Van  Nostrand  Reinhold,  New  York,  1983.

Andreae, M.  0.   Determination of arsenic species in natural waters.  Anal.
     Chem.  49:820-823, 1977.

Andreae, M.  0.   Distribution and speciation  of  arsenic in natural waters  and
     some marine algae.  Deep-Sea Res.  2_5:391-402,  1978.

Andreae, M.  0. ,  and D. Klumpp.  Biosynthesis and release of  organo arsenic
     compounds by  marine algae.   Environ.  Sci.   Techno!. 13:738-741,  1979.

Anke, M. , M. Grun,  Partschefeld,  B.  Grappel,  and A.  Hennig.  Essentiality  and
     function of  arsenic:   In:   Trace Element Metabolism in Man and Animals,
     Vol.  3, M.  Kirchgessner, ed.,  Arbeitskreses fur Tierernohrungs  forschug,
     werhenstephen, Germany, 1978.  pp. 248-252.

Archer, V.   [Personal  communication  with  D. Sivulka].  March 1, 1983.  Record
     of communication  available from:   U.S.  Environmental Protection  Agency,
     Research Triangle Park, NC; Project file no. ECAO-HA-79-5.

Arguello,  R. A.,  D. D. Cenget,  and  E.  E.  Tello.   [Regional  endemic cancer  and
     arsenical  intoxication  in  Cordoba].  Rev.  Argentina  Dermatosifilio.
     22:461-487, 1938.

Ariyoshi,  T. , and  T.  Ikeda.  On the tissue distribution and the excretion  of
     arsenic in  rats  and rabbits of administration with arsenical compounds.
     J. Hyg. Chem.  20:290-295, 1974.
                                   10-1
-------
Astrup, P.  Blackfoot disease.  Ugeskr.  Laeger 130:1807-1815, 1968.

Attrep, M.  Jr.,  and  M.  Anirudhan.   Atmospheric inorganic and organic arsenic.
     Trace Subst. Environ. Health 11:365-369,  1977.

Axelson, 0., E. Dahlgren, C.-D. Jansson, and S. 0. Rehnlund.  Arsenic exposure
     and mortality:  a case referent study from a Swedish copper smelter.  Br.
     J. Ind. Med. 35:8-15, 1978.

Baetjer, A. M. ,  A.  M.  Lilienfeld,  and  M.  L.  Levin.   Cancer and  occupational
     exposure to inorganic arsenic.  In:  Abstracts 18th International Congress
     on Occupational Health, Brighton,  England, September 14-19, 1975.  Organ-
     izing Committee, Brighton, England, 1975.  p. 393.

Bailar, J.  C.  Ill  and F. Ederer.   Significance  factors for the ratio  of a
     Poisson variable to its expectation.  Biometrics 20:639-643, 1964.

Baldridge,  C.  W.  Relationship  between antisyphylitic treatment and  toxic
     cirrhoses.  Am. J. Med. Sci. 188:685-690, 1934.

Baroni, C., G.  J.  van Esch, and  U.  Saffiotti.  Carcinogenesis tests of  two
     inorganic arsenicals.  Arch. Environ. Health 7:668-674, 1963.

Bavon, D., I.  Kunick, I. Frishmuth, and J. Petres.  Further  in vitro studies on
     the biochemistry  of the  inhibition of nucleic  acid and protein synthesis
     induced by arsenic.  Arch. Derm. Res. 253:15-22, 1975.

Baxley, M. N., R. D. Hood, G.  C. Vedel, W. P.  Harrison, and  G. M. Szczech.  Pre-
     natal toxicity of orally administered sodium arsenite in mice.  Bull. Environ.
     Contam. Toxicol. 26:749-756, 1981.

Beaudoin,   A.  R.   Teratogenicity  of sodium  arsenate  in rats.   Teratology
     10:153-158, 1974.

Beckman, G.,  L.  Beckman,  and  I.  Nordenson.   Chromosome  aberrations  in  workers
     exposed to arsenic.  Environ. Health Perspect. 19:145-146, 1977.

Bencko, V.,  and K.  Symon.   Test of environmental  exposure to arsenic and
     hearing  changes  in exposed  children.   Environ.   Health Perspect.
     19:95-101, 1977.

Bencko, V., and  K. Symon.   The  cumulation  dynamics  in some  tissue  of hairless
     mice inhaling arsenic.   Atmos. Environ. 4:157-161, 1970.

Bencko, V., B.  Benes,  and M.  Ckirt.  Biotransformation of As  (III) to As  (V)
     and arsenic tolerance.   Arch. Toxicol. 36:159-162, 1976.

Bental, E., S. Lavy, and N.  Amir.  Electroencephalographic changes  due to  arsenic,
     thallium and strychnine poisonings.  Confin. Neurol. 21(3):233-240,  1961.

Bergstrom, J.  and P. 0. Wester.  The effect of dialysis on the arsenic content
     of blood  and  muscle tissue from uraemic  patients.   Excerpta Med. Int.
     Cong. Ser. 131:38-40, 1966.
                                   10-2
-------
Bergoglio, R. M.   Mortality from cancer in regions  of arsenical  waters of the
     province of  Cordoba Argentine  Republic.   Prensa Med.  Argent.  51:994-998,
     1964.

Berteau, P. E., J. 0. Flom, R. L. Dimmick, and A. R.  Boyd.  Long-term  study of
     potential carcinogenicity of inorganic arsenic  aerosols to mice.  Toxicol.
     Appl. Pharmacol. 45:323 (Abs. No. 243), 1978.

Bertolero, F. ,  E.  Marafante,  J.  Edel-rade, R. Pietra, and E. Sabbioni.   Bio-
     transformation and  intracellular binding of arsenic in tissues of rabbits
     after i.p.  administration of 7.4As-labeled arsenite.  Toxicology 20:35-44,
     1981.

Blejer, H. P.,  and W. Wagner.  Case study 4:   inorganic  arsenic-ambient level
     approach to  the control  of occupational cancerigenic exposures.   In:
     Occupational Carcinogenesis, Proceedings of a Conference, New York Academy
     of Sciences,  New  York, New York,  March  24-27,  1975.   Ann.  N. Y. Acad.
     Sci.  271:179-186, 1976.
Blot, W. J. ,  and  J.  F.  Fraumeni.
     11:142-144, 1975.
Arsenical air pollution and cancer.  Lancet
Borgono, J.  M. ,  and R.  Greiber.   Epidemiological  study of arsenicism in the
     city of  Antofagasta.   In:   Trace Substances  in  Environmental  Health-V,
     Proceedings of the 5th Annual Conference, University  of Missouri, Columbia,
     Missouri, June 29  -  July 1,  1971.  D.  C.  Hemphill, ed. ,  University of
     Missouri, Columbia, MO, 1972.  pp. 13-24.

Borgono, P.  M.,  P.  Vincent, H. Venturino,  and A.  Infante.  Arsenic  in  the
     drinking water of  the  City of Antofogasta:   epidemiological  and clinical
     study before and after the installation of the treatment plant.  Environ.
     Health Perspect.  19:103-105, 1977.

Braman, R.  S., and C.  C. Foreback.  Methylated forms of  arsenic in the environ-
     ment.   Science 182:1247-1249, 1973.

Braman, R.  S., D. L. Johnson, C. C. Foreback, J. M. Ammons, and J. L. Bricker.
     Separation and determination of nanogram amounts of inorganic arsenic  and
     methyl arsenic compounds.  Anal.  Chem. 49:621-625, 1977.

Braun, W.   Carcinoma  of the skin  and  the  internal  organs  caused  by arsenic:
     delayed  occupational   lesions  due  to arsenic.  German  Med. Monthly
     3:321-324, 1958.

Brown, C.  C.  and K. C.  Chu.   Approaches  to epidemiologic analysis of prospec-
     tive and retrospective studies:   Example of lung cancer and exposure  to
     arsenic.  In:  Proceedings of the SIMS Conference  on Environmental  Epi-
     demiology:   Risk assessment, June  28-July 2, 1982,  Alta, Utah.   SIAM
     Publication, 1983a.

Brown, C.  C.  and K. C.  Chu.   A new method  for the analysis of  cohort studies:
     Implications of  the  multistage  theory of carcinogenesis applied to occu-
     pational arsenic exposure.   To  appear  in  Environmental Health Perspec-
     tives.   1983b.
                                   10-3
-------
Brown, C. C. and  K. C.  Chu.   Implications  of  the  multistage  theory  of  carcino-
     genesis applied  to occupational arsenic exposure.   J.  Natl.  Cancer Inst.
     70:455-463,  1983c.

Brune, 0.,  G.   Nordberg,  and P-0. Wester.   Distribution of 23 elements  in
     kidney,  liver and  lung of a control  group  in Northern Sweden and  of
     exposed workers  from a smelter and refinery.  Sci.  Total  Environ., 1980.

Buchet, J.  P.,  R. Lauwerys,  and  H.  Roels.  Comparison of several methods  for
     the determination  of arsenic compounds in water and in  urine.   Int.  Arch.
     Occup. Environ.  Health  46:11-29, 1980.

Buchet, J. P.,  R.  Lauwerys,  and H. Roels.   Comparison of the urinary excretion
     of  arsenic metabolites  after  a single  oral dose  of sodium arsenite,
     monomethyl  arsonate or  dimethyl  arsinate in  man.   Int.  Arch. Occup.
     Environ. Health  48:71-79, 1981a.

Buchet,  J.  P.,  R. Lauwerys,  and  H.  Roels.   Urinary excretion  of  inorganic
     arsenic and  its metabolites after repeated ingestion of sodium  meta
     arsenite by  volunteers.  Int.  Arch.  Occup.  Environ.  Health  48:111-118,
     1981b.                                                       ~~

Burgdorf, W., K.  Kurvink, and J.  Cervenka.  Elevated sister  chromatid  exchange
     rate in lymphocytes of  subjects treated  with arsenic.   Hum. Genet. 36:69-72,
     1977.

Burk,  D.,  and  A.  R.  Beaudoin.   Arsenate-induced  renal agenesis in rats.
     Teratology 16:247-260,  1977.

Butzengeiger, K.  H.   Chronic  arsenic poisoning.   I.  EKG alterations and  other
     symptoms observed  on the heart and vascular  system.  II.  Mucous membrane
     symptoms and pathogenesis.  Dtsch.  Arch.  Klin.  Med.  194: 1-16,  1949.

Butzengeiger, K.  H.   Peripheral  circulation disturbances with  chronic  arsenic
     poisoning.    Klin.  Wochenschr.  19: 523-527,  1940.

Byron, W.  R., Bierbower,  G.  W.  Brauwer, and W.  H. Hansen.   Pathologic  changes
     in  rats  and dogs  from  two-year feeding  of  sodium  arsenite and sodium
     arsenate.   Toxicol. Appl. Pharmacol.  10: 132-147, 1967.

Canadian Public  Health  Association Task Force on Arsenic.   Electromyography,
     Final  Report, Yellowknife and  Hay River,  Northwest Territories,  1978.
     Canadian Public  Health Association, Ottawa,  Canada, November 1978.

Challenger, F.   Biosynthesis  of organo-metallic and organo-metalloidal  compounds.
     In:   Organo-metals and  organo-metalloids (F.E. Brinckman and J.M.  Bellama,
     eds.), Symposium Series No.  82, Am.  Chem. Soc., Washington, D.C., 1978,
     pp.  1-22.

Charbonneau, S.  M., G.  K.  H.  Tarn, F.  Bryce,  and B.  Collins.  Pharmacokinetics
     and metabolism  of inorganic arsenic in  the  dog.
     Health.  12: 276-283, 1978a.
Trace Subst. Environ.
                                   10-4
-------
Charbonneau, S. M. , G.  K.  H. Tarn,  F.  Bryce,  Z.  Zadwidzka,  and E.  Sandi.  Meta-
     bolism of  orally  administered  inorganic  arsenic in the  dog.  Toxicol.
     Lett. 3:  107-113, 1979.

Charbonneau, S. M. , K.  Spencer,  F. Bryce,  and  E.  Sandi.   Arsenic excretion by
     monkeys dosed  with arsenic-containing  fish  or with inorganic arsenic.
     Bull. Environ.  Contain. Toxicol.  20: 470-477, 1978b.
Chhuttani,  P.  N. ,  L.  S.  Chawla, and T.
     Neurology 17:  269-275, 1967.
D.  Sharma.  Arsenical  neuropathy.
Ch'i, I. C. ,and  R.  Q.  Blackwell.  A controlled retrospective study of Black-
     foot disease and  epidemic  peripheral  gangrene  disease in  Taiwan.   Am.  J.
     Epidemiol.  88: 7-24, 1968.

Chung, H. C., and C. T. Liu.  Induction of lung cancer in  rats by  intratracheal
     instillation of  arsenic-containing ore  dust.   Zhonghua Zhongliu Zazhi
     4:14-16, 1982.

Cikrt, M. ,  and V. Bencko.   Fate of  arsenic after  parenteral  administration  to
     rats,  with particular reference to excretion via bile.  J. Hyg. Epidemiol.
     Microbiol. Immunol. 18: 129-136, 1974.

Cikrt, M. ,  V.  Bencko,  M.  Tich,  and  B.  Benes.   Biliary excretion  of 74As and
     its distribution in the golden hamster after administration of 74As  (III)
     and 74As  (V).  J.  Hyg.  Epidemiol. Microbiol.  Immunol. 24:384-388,  1980.

Consultants in Epidemiology  and Occupational  Health.  Arsenic  Risk Assessment
     Critique and Alternative Supplemental Submission prepared for the Arsenic
     Program Panel  Chemical  Manufacturers  Association, August  13,  1982.   OSHA
     Docket No. H-037C.  Standard for Inorganic Arsenic Post-hearing Evidentiary
     Submission  of  the Chemical  Manufacturers Association  Arsenic Panel.
Coulson, E. J.,  R.  E.  Remington, and  K.  M.  Lynch.
     the naturally  occurring arsenic  of shrimp  as
     trioxide.  J. Nutr.  10:255-270, 1935.
           Metabolism in the rat of
           compared  with arsenic
Crecelius,  E.  A.   Changes in the  chemical  speciation of  arsenic  following
     ingestion by man.  Environ. Health. Perspect. 19:147-150, 1977.

Crecelius,  E. A.  The geochemistry of  arsenic  and antimony in Puget Sound and
     Lake Washington, Washington.  Thesis:  University of Washington, Seattle,
     Washington, 1974.

Crema, A.   Distribution et elimination de 1'arsenic  76  chez la souris normale
     et cancereuse.   Arch. Int.  Pharmacodyn. 103:57-70, 1955.

Crossen, P. E.  Arsenic and SCE in human lymphocytes.  Mutat.  Res.  119:415-419,
     1983.

Cuzick, J. ,  S.  Evans, M.  Gillman, and D. Price Evans.  Medicinal  arsenic and
          internal malignancies.  Br. J. Cancer 45:904-911,  1982.
                                   10-5
-------
 Damsgaard, E., K. Heydorn,  N.  Larsen  and B.  Nielsen.   Simultaneous determina-
      tion of arsenic, manganese and selenium in human serum by neutron  activa-
      tion  analysis.   Report  No.   271.   Danish Atomic  Energy Commission,
      Roskilde,  1973.

 Day,  N. E.,  and  C.  C. Brown.  Multistage models  and primary prevention  of
      cancer.   J.  Natl. Cancer Inst. 64:977-989, 1980.

 Dieke,  S.  H.,  and C.  P.  Richter.   Comparative assays of rodenticides on wild
      Norway Rats.   I.   Toxicity.   Publ.  Health Rep. 61:672-679, 1946.

 DiPaolo,  J.  and B.  Casto.   Quantitative studies of jji vitro morphological trans-
      formation  of Syrian hamster cells by inorganic metal  salts.   Cancer  Res.
      39:1008-1013,  1979.

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

 Doll, R.  and R. Peto.   Cigarette smoking  and  bronchial carcinoma:   dose and
      time  relationships  among regular  smokers and  life-long  non-smokers.
      J. Epi.  Comm. Health  32:303-113,  1978.

 Done, A.   K. ,  and A.   J.  Peart.   Acute toxicities  of  arsenical  herbicides.
      Clin.  Toxicol. 4:343-355,  1971.

 Dubois, K.  P.,  A. L.   Moxon, and 0. E. Olson.  Further studies on  the effec-
      tiveness   of  arsenic  in  preventing  selenium poisoning.   J.  Nutr.
      19:477-482, 1940.

 Ducoff, H.  S. ,  W.  B.   Neal,  R.  L.  Straube,  L.  D.  Jacobson, and A.  M. Brues.
      Biological  studies  with arsenic76.   II.   Excretion  and tissue localiza-
      tion.  Proc. Soc. Exp.  Biol. Med.  69:548-554,  1948.

 Durrant, P. J., and B. Durrant.   Introduction to Advanced Inorganic Chemistry.
     3rd Ed., Longmans, Green  and Co.,  Ltd.,  London,  1966.

 Dutkiewicz,  T.   Experimental studies  on  arsenic  absorption routes  in  rats.
     Environ. Health.   Perspect. 19:173-177,  1977.

 Edmonds, J.  S., and K. A.  Francesconi.   Methylated arsenic from marine fauna.
     Nature 265:436, 1977.

 Edmonds, J.  S., K. A.  Francesconi,  J.  R.  Cannon,  C. L. Raston,  B.  W. Skelton,
     and A. H. White.   Isolation, crystal  structure and synthesis  of arsenobe-
     taine, the arsenical  constituent of  the western  rock lobster Panulirus
     longipes cygnus George.  Tetrahedron  Lett. 18:1543-1546, 1977.

Eldred,  R. ,  L.  Ashbaugh,  T. Cahill, R.  Flocchini  and M.  Pitchford.  Sulfate
     levels in  the southwest during the  1980 copper smelter strike. JAPCA 33:
     110-113, 1983.                                                        ~~

Enter!ine, P. E. and G. M.  Marsh.   Mortality studies  of smelter workers.  Am.
     J.  Ind. Med. 1:  251-259, 1980.
                                   10-6
-------
Enterline, P. E. and G. M. Marsh.  Mortality  among  workers  exposed to arsenic
     and other substances in a copper smelter.  Am.  J. Epidemic!.  116: 895-910,
     1982.

Falk, H. , G. G. Caldwell, K. G.  Ishak,  L.  B.  Thomas,  and H.  Popper.   Arsenic-
     related hepatic angiosarcoma.  Am. J. Ind. Med. 2:43-50, 1981a.

Falk, H., J.  T.  Herbert, L.  Edmonds, C. W. Heath Jr., L. B. Thomas, and H. Popper.
     Review of four cases of childhood hepatic angiosarcoma—elevated environ-
     mental  arsenic exposure in one case.  Cancer 47:383-391, 1981b.

Feldman,  C.   Improvements in  the arsine  accumulation-helium glow detector
     procedure  for  determining traces of  arsenic.   Anal.  Chem. 51:664-669,
     1979.

Ferm, V.  H.   Arsenic  as  a  teratogenic  agent.  Environ. Health Perspect.
     19:215-217, 1977.

Ferm, V.  H., and S. Carpenter.   Malformations induced by sodium arsenate.   J.
     Reprod.  Fertil. 17:199-201, 1968.

Ferslew,  K.  E. ,  and G. T. Edds.  Effects  of  arsanilie acids on  growth, serum
     enzymes, hematologic values, and  residual arsenic in young  swine.  Am.  J.
     Vet. Res.   40:1365-1369, 1979.

Feussner, J. R.,  J.  D. Shelburne, S.  Bredehoeft,  and H. J.  Cohen.  Arsenic-
     induced bone marrow toxicity:  ultrastructural and  electron-probe analysis.
     Blood 53:820-827,  1979.

Fierz,  U.  Catamnestic investigations of  the side  effects  of therapy of  skin
     diseases with inorganic arsenic.  Dermatologica  131:41-58,  1965.

Flessel,  C.  P.   Metals as mutagens.   In:  Inorganic  and Nutritional  Aspects
     of Cancer, G.W. Schrauzer,  ed.   Plenum Press,  New York, 1978, pp. 117-128.

Fong, K., F. Lee, and  R. Bockrath.  Effects of sodium arsenate  on  single-strand
     DNA  break  formation and post-replication repair in E.  coli following UV
     irradiation.  Mutat. Res.  70:151-156, 1980.

Fornace,  A.  J. , and  J. B.  Little.   DNA-protein cross-linking  by chemical
     carcinogens in mammalian cells.   Cancer  Res. 39:704-710, 1979.

Fowler,  B.   Toxicology of environmental  arsenic.   In:   Toxicology of Trace
     Elements.  R. A.  Gayer  and M. A.  Mehlman, eds.   Halstead Press,  New  York.
     pp.  79-122, 1977.

Francis,  J. ,  B. Brower, W.  Graham,  0. Larson, J.  McCaull,  and H. Vigorita.
     National  Statistical  Assessment  of  Rural Water Conditions (Vol. II),
     Dept. of Rural Sociology,  Cornell  University,  1982.

Frank,  G.   Neurologische und psychiatrische  Folgesymptome  bei   akuter Arsen-
     Wasserstoff-Vergiftung.  J.  Neurol.  213:59-70, 1976.

Franke,  K.  W. ,  and A.  L.  Moxon.  A comparison of the minimal fatal  doses of
     selenium,  tellurium, arsenic  and vanadium.   J.  Pharmacol. Exp. Ther.
     58:454-459, 1935.

                                    10-7
-------
 Freeman, J. W.,  and J. R. Couch.  Prolonged encephalopathy with arsenic poison-
      ing.   Neurology 28:853-855, 1978.

 Ganther, H. E.,  and C. A. Baumann.  Selenium metabolism.  I.  Effects of diet,
      arsenic and cadmium.  J. Nutr.  77:210-216, 1962.

 Ganther, H. E.,  and H. S. Hsieh.  Mechanisms for the conversion of selenite to
      selenides in mammalian tissues.  In:  Trace Element Metabolism in Animals.
      2.   (W.  G.   Hoekstra,  J.  W. Suttie,  H.  E.  Ganther and W. Mertz, Eds.).
      University  Park Press, Baltimore, 1974.  p. 339.

 Garb, L. G.,  and C. H. Mine.  Arsenical  neuropathy:  Residual effects follow-
      ing acute industrial exposure.   J.  Occup.  Med.  19:567-568, 1977.

 Gerhardt,  R.,  J. Hudson,  R.  Rao and  R.  Sobel.   Chronic renal insufficiency
      from  cortical  necrosis induced by arsenic  poisoning.  Arch.  Intern. Med
      138:1267-1269, 1978.

 Gilbert, F., R.  Duncan,  W.  Lederer and J. Wilkinson.   Effects of chemical  pre-
      servatives  on  the  health  of  wood  treating  workers in  Hawaii,  1983
      (unpublished).

 Ginsberg,  J. M.   Renal mechanism  for  excretion and transformation of arsenic
      in  the dog.  Am.  J.  Physiol.  208:832-840,  1965.

 Ginsberg,  J. M.  and W. D.  Lotspeich.   Interrelations  of arsenate and phosphate
      transport in the  dog kidney.   Am. J. Physiol.  205:707-714,  1963.

 Gurley,  L.  R. , R. A. Walters,  J. H.  Jett, and R. A. Tobey.   Response of  CHO cell
      proliferation  and histone phosphorylation  to  sodium arsenite.   Toxicol.
      Environ.  Health 6:87-105, 1980.

 Hamamoto,  E.  [Infant arsenic poisoning by powdered milk].  Jap.  Med. J.
      1649:2-12,  1955.

 Hanlon,  D.  P., and  V.  H.  Ferm.  Placental permeability of arsenate ion during
      early  embryogenesis in  the hamster.  Experientia   33:1221-1222, 1977.

 Hara,  I.,  K. Hashimoto,  K. Miyazaki,  and K. Sunada.   Case  studies  of poly-
      neuritis  believed to have been  caused by arsenical  poisoning.   TR-79-0524,
      Labor  Health Department of  the  Osaka Public Health  Research  Center, Osaka
      University.   Sangyo  Igaku 11(1):84-90,  1968.

Harrington,  M. ,  J.  P.  Mcdaugh,  D.  L.  Morse, and J. Housworth.  A study of a
      population  exposed  to  high concentrations  of  arsenic in well water  in
      Fairbanks, Alaska.  Am. J.  Epidemiol.  108:377-385,  1978.

Harrison, J. W.  E. ,  E. W.  Packman,  and D. C.  Abbott.   Acute oral  toxicity  and
      chemical  and physical properties  of  arsenic trioxides.   A.M.A.  Arch.  Ind.
      Health  17:118-123, 1958.

Harrison, W. P.,  J.  C.  Frazier,  E.  M.  Mazzanti,  and R. D.  Hood.  Teratogenicity
     of disodium methanarsonate and sodium dimethyl arsenate  (sodium cacodylate)
      in mice.  Teratology 21:43A, 1980.
                                   10-8
-------
Henry, F. T., and T. M. Thorpe.   Determination  of  arsenic  (III),  arsenic  (V),
     monomethyl  arsonate  and  dimethyl  arsinate by differential pulse polaro-
     graphy  after  separation by  ion-exchange chromatography.  Anal.  Chem.
     52:80-83, 1980.

Hernberg, S.  Incidence  of cancer in population with exceptional exposure to
     metals.  In:   Cold  Spring  Harbor  Conferences  on Cell  Proliferation,  Vol.
     4, Origins  of  Human  Cancer,  Book  A,  Incidence of Cancer in Humans,  H.  H.
     Hiatt,  J.  D. Watson, and J.  A.  Winsten,  eds., Cold Spring Harbor Labora-
     tory, Cold Spring Harbor, New York, 1977.  pp.  147-157.

Heyman, A.,  J.  B. Pfeiffer,  Jr.,  R.  W.  Willett,  and H.  M.  Taylor.   Peripheral
     neuropathy  caused by arsenical  intoxication.    A  study of 41 cases  with
     observations on the effects of BAL (2,3-dimercapto-propanol).   N. Engl.
     J. Med. 254:401-409,  1956.

Heydorn,  K.   Environmental  variation  of arsenic  levels  in  human  blood
     determined by  neutron activation  analysis.  Clin.  Chim.  Acta. 28:344-356,
     1969.

Heywood,  R. , and R. J. Sortwell.  Arsenic intoxication in the Rhesus monkey.
     Toxicol. Lett. 3:137-144, 1979.

Higgins,  I., K. Welch and C.  Burchfiel.  Mortality of  Anaconda smelter workers
      in  relation to arsenic and  other exposures.   Ann Arbor,  MI.,  Dept.  of
      Epidemiology,  Univ.  of Michigan.   1982.

Higgins,  I.  Arsenic and  respiratory cancer among  a sample of Anaconda Smelter
     Workers.   Report  submitted  to  OSHA  in  the comments of  the  Kennecott
      Minerals  Company  on the  inorganic arsenic rulemaking  (Exhibit 203-5),
      1982.
 Hill,  A.  B.,  and  E.  L.  Faning.   Studies  on the
      handling inorganic compounds  of arsenic.
      factory.   Br.  J.  Ind.  Med.  5:1-6, 1948.
incidence of cancer in a factory
I.   Mortality experience in the
 Hindmarsh,  J.  T.,  0.  R.  McLetchie,  L.  P.  Heffernan,  0.  A.  Haynie,  H.  A.  Ellen-
      berger,  R.  F. McCurdy, and H.  J. Thiebaux.  Electromyographic abnormali-
      ties  in  chronic  environmental  arsenicalism.   J.  Anal.  Toxicol.   1:270-276,
      1977.

 Hine, C.  H.,  S.  S. Pinto, and K. W. Nelson.  Medical problems associated with
      arsenic  exposure.   J.  Occup.  Med.  19:391-396,  1977.

 Holden,  H. E.   Comparison  of somatic and  germ cell  models for cytogenetic
      screening.   Appl.  Toxicol.  2:196-200,  1982.

 Holland,  J. W.  Arsenic.   In:   A Textbook of Legal  Medicine and  Toxicology,
      Vol  2, F. Peterson and W.  S.  Haines,  eds., Philadelphia:  W. B.  Saunders
      & Company,  1904.

 Holland,  R.  H.  and  A.  R.  Acevedo.  Current status  of  arsenic  in American
      cigarettes.   Cancer 19:1248-1250, 1966.
                                    10-9
-------
 Holland,  R.  H., M.  S.  McCall,  and H.  C.  Lanz.   A study of inhaled arsenic-74
      in man.   Cancer Res.  19:1154-1156, 1959.

 Holmqvist,  I.   Occupational arsenical  dermatitis.  A study among employees at
      a copper ore smelting work including investigations of skin reactions to
      contact with  arsenic  compounds.   Acta.  Derm.  Venerol.  31:(Suppl.  26)1-214
      1951.                                                  —

 Hood,  R.  Toxicology of prenatal  exposure to arsenic.   In:   Arsenic:   Industrial,
      Biomedical  and  Environmental  Perspectives.   W.  Lederer and R.  Fensterheim,
      eds., Van Nostrand Reinhold,  New York,  1983.

 Hood,  R.  D.,  and  C.  T. Pike.   BAL alleviation  of arsenate-induced terato-
      genesis  in  mice.   Teratology  6:235-238, 1972.

 Hood,  R.  D., G.  T. Thacker,  and B.  L.  Patterson.   Effects in the mouse and rat
      of prenatal exposure to arsenic.   Environ.  Health Perspect.  19:219-222,
      1977.                                                         —

 Hood,  R.  D.,  and S.  L.  Bishop.  Teratogenic effects of sodium arsenate  in
     mice.  Arch.  Environ.  Health  24:62-65,  1972.

 Hood,  R.  D; ,  G. C.  Vedel,  M. J. Zaworotko,  and  F.  M. Tatum.   Distribution of
     arsenite  and methylated metabolites  in  pregnant  mice.   Teratology 25:50A,
     1982a.                                                             —

 Hood,  R.  D., G.  T. Thacker,  B.  L.  Patterson, and G. M.  Szczech.   Prenatal effects
     of oral versus  intraperitoneal sodium arsenate in  mice.   J.  Environ. Pathol.
     Toxicol.  1:857-864, 1978.

 Hood,  R.  D. , W.  P. Harrison, and G. C.  Vedel.  Evaluation of arsenic metabolites
     for  prenatal  effects in the  hamster.   Bull.  Environ.  Contam.  Toxicol.
     29:679-687, 1982b.

 Hueper, W.   A  quest into  the  environmental  causes of  cancer of the  lung,
     Public Health  Monograph No.  36.    U.S.  Department  of Health, Education,
     and Welfare, Public Health Service,  1955.   pp. 27-29.

 Hueper, W. C.  Experimental  studies in  metal carcinogenesis.   VI. Tissue reactions
     in rats  and  rabbits  after parenteral  introduction of suspensions  of
     arsenic,  beryllium,  or asbestos  in  lanolin.   J.  Nat.  Cancer Inst.  15:
     113,  1954.                                                           ~~

 Hueper, W. C.  and  W.  W. Payne.   Experimental studies  in metal  carcinogenesis.
     Chromium,  nickel,  iron, arsenic.   Arch. Environ.  Health.  5:  445, 1962.

 Hunter, F. T., A. F. Kip, and J. W. Irvine,  Jr.  Radioactive tracer  studies  on
     arsenic  injected  as  potassium  arsenite.   J. Pharmacol.  Exp.  Ther.
     76:207-220, 1942.

Hutchinson, J.   On  some examples  of  arsenic keratoses  of the skin and  of
     arsenic cancer.  Trans. Pathol. Soc. (London) 39:352-363,  1888.
                                   10-10
-------
Inamasu,  T.,  A.  Hisanaga and N.  Ishinishi.
     calcium arsenate retention in the  rat
     tion.   Toxicol.  Lett. 12: 1-5, 1982.
Comparison  of  arsenic  trioxide  and
lung after intratracheal instilla-
International Agency for  Research on Cancer.  IARC monographs on the evalua-
     tion of the carcinogenic risk of chemicals to man. Vol. 2.  Some inorganic
     and organometallic compounds.   World Health Organization, Lyon, France,
     1973.

International Agency for Research on Cancer.  IARC monographs on the evaluation
     of the  carcinogenic  risk of  chemicals  to man.   Vol.  23.   Some  metals and
     metallic compounds. World Health Organization,  Lyon, France, 1980.

Irgolic, K. J. , E. A. Woolson, R.  A. Stockton, R. D. Newman, N. R.  Bottino, R.
     A. Zingaro,  P.  C.  Kearney, R. A. Pyles, S. A. Maeda, W. J. McShane,  and
     E. R.  Cox.  Characterization of arsenic compounds formed by Daphnia  magna
     and Tetraselmis cheiri from inorganic arsenate.   Environ. Health Perspect.
     19:61-66, 1977.

Irgolic, K. Speciation of arsenic compounds  in water supplies.  U.  S. Environ-
     mental Protection Agency, Health Effects Research Laboratory,  Cincinnati,
     Ohio.   EPA-600/S1-82-010, 1982.

Irgolic, K. [Personal communication with  D.  Sivulka].  March 10, 1983.  Record
     of communication  available from:   U.S. Environmental  Protection Agency,
     Research Triangle Park, NC; Project  file no. ECAO-HA-70-5.

Ishinishi,  N.,  Y.  Kodama, E. Kunitake, K.  Nobutomo, M. Urabe, T. Inamasu,  Y.
     Suenaga,  and T. Hitano.  Arsenic and  arsenic compounds.   Ninon Rinsho
     31:1991-1999,  1973.
Ishinishi,  N. ,  Y.  Kodama, K. Nobutomo, and A.  Hisanaga.
     mental study on carcinogenicity of arsenic trioxide
     Health Perspect.  19_:191~196>  1977.
               Preliminary experi-
              in rat lung.  Environ.
Ishinishi,  N. ,  M.  Mizunoe, T.  Inamasu and A.  Hisanaga.   Experimental study on
     carcinogenicity  of beryllium oxide and  arsenic  trioxide to the lung of
     rats  by an  intratracheal  instillation.   Fukuoka Acta  Med.  71: 19-26,
     1980.

Ishinishi,  N. ,  and A. Yamamoto.   Discrepancy between epidemiological evidence
     and  animal experimental  results.   J. Uoeh. 5  (Suppl.):109-116, 1983.

Ivankovic,  S. ,  G.  Eisenbrand,  and R. Preussmann.  Lung carcinoma induction in
     BD rats  after a  single intratracheal  instillation  of an arsenic-containing
     pesticide  mixture formerly  used in vineyards.   Int.  J.  Cancer 24:786-788,
     1979.

Japanese  Pediatric Society.  Summary of report of  activities of the Morinaga
     arsenic-tainted  powdered  milk  poisoning  investigation.   TR-124-74.   May
     26,  1973.

Jelinek,  C.  F. ,  and P. E.  Corneliussen.   Levels of arsenic  in  the United
     States food supply.   Environ.  Health Perspect.   : 1:83-87 (1977).
                                    10-11
-------
 Jenkins,  R.  B.  Inorganic  arsenic  and the nervous system.   Brain 89:479-498,
      1966.

 Johnson,  D.  L.,  and  R.  S.  Braman.   Alkyl-and inorganic arsenic in air samples
      Chemosphere 6:333-338,  1975.

 Johnson,  R.  D.,  D.  Manske, D.  New and D. Podrebarac.  Pesticide, heavy metal,
      and  other  chemical  residues  in infant and toddler  total  diet samples
      (II).  August,  1975  -  July, 1976.   Pesticides Monit. J.  15:  39-50,  1981b.

 Johnson,  R.  D. ,  D.  Manske  and D.  Podrebarac.  Pesticide,  metal,  and other
      chemical  residues  in  adult total  diet  samples (XII).   August,  1975  -July,
      1976.  Pesticides  Monit.  J. 15: 54-69,  1981a.

 Jung, E.,  B.  Trachsel and  H.  Immich.  Arsenic as  an inhibitor  of the  enzymes
      concerned in cellular  recovery  (dark repair).   Ger. Med. Mon.  14:614-616,
      1969.

 Jung, E.  and  B.  Trachsel.  Molekularbiologische Untersuchungen zur Arsencar-
      cinogenese.   Arch. Klin.  Exp.  Derinatol. 237:  819-826,  1970

 Kadowaki,  K.   Studies on  the  arsenic contents in  organ tissues of  the normal
      Japanese.   Osaka City Med. J.  9:2083-2099,  1960.   (In Japanese with
      English summary.)

 Kagey,  B.  T., J.  E.  Bumgarner,  and J. P. Creason.  Arsenic  levels in maternal-
      fetal tissue sets.  Trace Subst. Environ. Health 11:252-256, 1977.

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

 Kanisawa,  M:,  and H.  A. Schroeder.   Life term studies on the effects of arsenic,
      germanium, tin  and vanadium on spontaneous tumors 'in mice.  Cancer Res
      27: 1192-1195,  1967.

 Kanisawa,  M.,  and H.  A. Schroeder.   Life term studies on the effect of trace
      elements  on  spontaneous tumors  in  mice and rats.  Cancer Res. 29:892-895,


 Kanstra, L.  D. ,  and  C.  W.  Bonhorst.  Effect of arsenic on  the expiration of
      volatile  selenium compounds by  rats.   Proc.  S. Dak. Acad.  Sci  32-72
      1953.                                                             —*   '

 Kelynack,  T.  N. ,  W.  Kirkly,  S. Delepine, and  C.  H.  Tattersall.  Arsenical
      poisoning from beer-drinking.   Lancet  2:1600-1602,  1900.

 Klaassen,  C.  D.  Biliary  excretion  of  arsenic in  rats,  rabbits,  and doqs
     Toxicol.  Appl. Pharmacol.  29:447-457,  1974.

Knoth, W.  Arsenbehandlung.  Arch. Klin.  Exp.  Derm.   227:228-234,  1966.

Kraybill,  H.  F.  Carcinogenesis  induced by trace contaminants in  potable
     water.  Bull. N.Y. Acad. Sci. 54:413-427,  1978.
                                   10-12
-------
Kroes, R. , M.  J.  van Logten, J. M. Berkvens, T. de Vries and G. J. Van Esch.
     Study on  the carcinogenicity  of  lead  arsenate  and  sodium arsenate and on
     the possible  synergistic  effect  of diethylnitrosamine.   Food  Cosmet.
     Toxicol. 12:  671-679, 1974.

Kuratsune, M.,  S. Tokudome,  T.  Shirakusa,  M.  Yoshida, Y.  Tokumitsu,  T. Hayano
     and M.  Seita.   Occupational  lung cancer among copper smelters.   Int. J.
     Cancer.   13:552-558, 1974.

Kyle, P. A.,  and  G.  L.  Pease.  Hematologic aspects of  arsenic intoxication.
     'N.  Eng.  J. Med.  273:18-23, 1965.

Lakso, J. U., and S.  A.  Peoples.  Methylation of inorganic arsenic by  mammals.
     J.  Agric.  Food Chem. 23:674-676,  1975.

Lakso, J. , L.  Rose, S. Peoples  and D.  Shirachi.  A  colorimetric method for the
     determination of arsenite, arsenate,  monomethylarsonic  acid,  and  dimethyl-
     arsinic  acid in biological and  environmental  samples.   J. Agric. Food
     Chem. 27:1229-1233,  1979.

Landau,  E. ,  D. J. Thompson,  R.  G.  Feldman, G. J.  Goble, and W. vL Dixon.
     Selected  non-carcinogenic effects  of industrial  exposure  to  inorganic
     arsenic,  EPA 569/6-77-018, U.S.  Environmental  Protection Agency,  Washing-
     ton, DC,  July 1977.

Lander,  J.  J. , R. J.  Stanely,  H.  W.   Sumner, D. C.  Boswell, and R.  D. Aach.
     Angiosarcoma  of the liver associated with Fowler's solution (potassium
      arsenate).  Gastroenterology  68:1582-1586, 1975.

Lanz H. ,  Jr., P.  C.  Wallace, and J.  G.   Hamilton.   The metabolism of arsenic
      in  laboratory animals  using As74 as a tracer.   Univ.  Calif. Publs.  Pharmacol.
      2:263-282, 1950.

Lao, R.  C.,  R.  S.  Thomas, T.  Teichman, and L.  Dubois.   Efficiency of collection
      of  arsenic trioxide in high volume  sampling.   Sci.  Total Environ. 2:373-379,
      1974.

Larramendy,  M.  L., N.  C. Popescu,  and J. DiPaolo.   Induction by inorganic metal
      salts  of sister chromatid exchanges and chromosome aberrations in human and
      Syrian  hamster  strains.  Environ. Mutagen. 3:597-606, 1981.

 Larsen,  N.  A., H. Pakkenberg,  E.  Damsgaard, and K.  Heydorn.  Topographical
      distribution of  arsenic,  manganese,  and  selenium in the  normal  human
      brain.   J. Neurolog. Sci.  42:407-416, 1979.

 Le Quesne,  P.  M., and J. G.  McLeod.   Peripheral neuropathy  following  a single
      exposure to arsenic.  J. Neurol. Sci. 32:437-451,  1977.

 LeBlanc, P. J., and A.  L.  Jackson.   Arsenic  in marine  fish  and invertebrates.
      Mar.  Pollut. Bull.  4:88-90, 1973.

 Lee-Feldstein, A.  Arsenic   and respiratory cancer  in  man:  follow-up of  an
      occupational study. In: Arsenic: Industrial, Biomedical  and  Environmental
      Perspectives.  W.  Lederer and R. Fensterheim, eds., Van  Nostrand Reinhold,
      New York, 1983.

                                    10-13
-------
 Lee, A. M.,  and  J.  F. Fraumem*, Jr.  Arsenic  and  respiratory cancer in man:
      an occupational study.   J. Natl. Cancer Inst.  42:1045-1052, 1969.

 Leitch, A., and E.  L.  Kennaway.  Experimental production of cancer by arsenic
      Brit.  Med. J.  2:1107-1108, 1922.  (Cited in IARC, 1980)

 Leonard, A.,  and R.  R. Lauwerys.  Carcinogenicity,  teratogenicity, and mutageni-
      city of arsenic.   Mutat.  Res.  75:49-62, 1980.

 Lerman, B.  B., N. Ali, and D. Green.  Megaloblastic, dyserythropoietic  anemia
      following arsenic  ingestion.   Ann.   Clin.  Lab.  Sci.  10:515-517, 1980.

 Lerman, S.  A., and  T.  W.  Clarkson.   The  metabolism  of arsenite and arsenate by
      the rat.   Fundam. Appl.  Toxicol. (in press, 1983).

 Leslie, H., and H.   Smith.   Self-poisoning by the abuse  of arsenic-containing
      tonics.   Med.  Sci.  Law  18:159-162,  1978.

 Levander,  0.  A.  Metabolic  interrelationships  between arsenic and selenium
      Environ.  Health.  Perspect.  19:159-164,  1977.

 Levander,  0.  A., and  C.  A.  Baumann.  Selenium  metabolism.   VI.   Effect of
      arsenic  on the  excretion  of selenium in the bile.   Toxicol.  Appl.  Pharmacol
      9:106-115, 1966.

 Liebscher,  K., and  H.  Smith.   Essential  and nonessential  trace elements.  A
      method of determining whether an element is essential or nonessential in
      human  tissue.   Arch.  Environ.  Health 17:881-890,  1968.

 Lindh,  U.,  D.  Brune, G.  Nordberg, P. 0.  Wester.  Levels  of antimony, arsenic,
      cadmium,  copper,  lead,  mercury, selenium,  silver,  tin and zinc in bone
      tissue of industrially  exposed workers.  Sci.  Total Environ. 16:109-116
      1980.                                                         — ,

 Lofroth, G., and B.   N. Ames.   Mutagenicity of inorganic  compounds in  Salmonella
      typhimurium: arsenic, chromium,  and  selenium.  Mutat.  Res. 54:65-66,  1978.

 Lu,  F-J., ^M-H. Tsai,  and L-H.  Ling.  Studies  of fluorescent compounds  in
      drinking  water  of Blackfoot disease  indemic areas: 1. The toxic effects
      of fluorescent  compounds  on  the  chick embryo.  J. Formosan Med.  Assn  76-
      58-63, 1977a.                                                         —

 Lu,  F-J.,  M-H. Tsai,  and K-H  Ling.   Studies on fluorescent  compounds  in
      drinking water of Blackfoot  endemic  areas.  2.   Isolation and identifica-
     tion of  fluorescent  compounds.  Tai-wan I  Hsuch Hui Tsa Chin 76:209-17
     1977.  Chem.  Abs.  87:141054d,  1977b.                          ~~

 Lu,  F-J,  M-H.  Tsai,  and K-H.  Ling.   Studies on fluorescent  compounds  in
     drinking water of areas endemic  for  Blackfoot disease.  3.   Isolation and
     identification   of fluorescent  compounds.   Tai-wan  I  Hsuch  Hui  Tsa  Chih
     77:68-76, 1978.   Chem. Abs. 89:152,200r, 1978.

Lubin, J. H. ,  L.  M.  Pottern,  W.  J.  Blot, S.  Tokudome, B.  J.  Stone and J.  F.
     Fraumeni, Jr.   Respiratory  cancer  among copper smelter  workers: recent
     mortality statistics.  JOM 23:  779-784, 1981.

                                   10-14
-------
Luchtrath, H.  Die Leberzirrhose bei chronischer Arsen  vergiftung  der winzer.
     Dtsch.  Med.  Wochenschr.  97:21-22, 1972.

Lundgren; K. D.  Damage to  respiratory organs  in workers  in  a  smelting  plant.
     Nord. Hyg. Tidskr. 3:66-82, 1954.

Lyon, J.  L., et  al.   Arsenical air pollution and lung cancer.  Lancet  2:869,
     October 22,  1977.

Mabuchi, K., A. Lilienfeld, and L.  Snail.  Lung cancer among pesticide  workers
     exposed to  inorganic  arsenicals.   Arch. Environ.  Health  312-319,  1979.

Mahaffey, K. R. ,  and B.  A. Fowler.  Effects of concurrent administration of
     lead,  cadmium,   and  arsenic in the rat.   Environ.  Health Perspect.
     19:165-171,  1977.

Mappes,  R.   [Experiments on  excretion of arsenic in urine.]   Versuche zur
     Ausscheidung von Arsen  in  Urin.    Int.  Arch.  Occup.  Environ.  Health
     40:267-272,  1977.

Marafante,  E., J.  Rade,  R. Pietra,  E.  Sabbioni, and F.  Bertolero.  In:   M.
     Anke,  H.-J.  Schneider, and C.  Bruckner.   Eds.   Proc. 3rd Symp. on Trace
     Elements.    Arsenic.   Abteilung  Wissenschaftliche  Publikationen  der
     Friedrich-Schiller Universitat, Jena, 1980.  pp. 49-55.

Masahiki, 0.,  and A.  Hideyasu.   Epidemiological  studies on the Morinaga powdered
     milk poisoning  incident:  Final  report of the  joint project team from
     Hiroshima and  Okayama Universities.  Nihon eiseigaku zasshi  [Jap. J.  of
     Hyg.]  27:500-531, 1973.

Massmann, W.,  and H.  Opitz.   Experimental^  Untersuchungen uber Ekg-Veranderunger
     bei  chronischer Arsenvergiftung.   Z. Kreislaufforsch.  43:704-713, 1954.

Matanoski,  G., E. Landau,  and E. Elliott.   Epidemiology Studies.   Task I-Phase
     I.   Pilot Study of Cancer Mortality Near an Arsenical Pesticide Plant in
     Baltimore.   EPA-560/6-76-003,  U.S.  Environmental   Protection  Agency,
     Washington,  DC,  May 1976.

Matanoski,  G. , E. Landau, J.  Tonascia, C. Lazar, E.   Elliot, W. McEnroe and K.
     King.   Cancer mortality  in an industrial  area  of Baltimore.  Environ. Res.
     25:  8-28, 1981.

McBride, B. , H.  Merilees,  W.  Cullen and W.  Pickett.   Anaerobic  and aerobic
     alkylation  of  arsenic.   In:   Organometals and  Organometalloids.  F. E.
     Brinckman and J. M.  Bellama,  eds.  Symposium Series No.  82, Am.  Chem.
     Soc.,  Washington DC,  1978.   pp. 94-115.  (Cited in Vahter, 1981).

McCabe, L.  J., J.  M.  Symons,  R.  D.  Lee, and G. G.  Robeck.  Survey of community
     water  supply systems.   J.  Am.  Water Works Assoc. 62:670-687, 1970.

McCann, J.  , E.  Choi, E.  Yamasaki, and  B. Ames.  Detection  of carcinogens as
      mutagens  in the Salmonel1 a/microsome test:  Assay of 300 chemicals.  Proc.
      Natl.  Acad. Sci. (U.S.) 72:5135,  1975.
                                    10-15
-------
 McCarroll, N.  E. ,  C.  E.  Piper, and  B.  H.  Keech.   An E.  coli  microsuspension
      assay for the detection of DNA damage induced by direct-acting agents and
      promutagens.  Environ. Mutagen. 3:429-444, 1981.

 Mealey, J., Jr.,  G.  L.  Browne!!, and W.  H.  Sweet.  Radioarsenic in plasma,
      urine, normal tissues, and intracranial neoplasms.  Arch. Neurol. Psvchiatr
      81:310-320, 1959.
 Mertz, W.   Aspects of nutritional trace element research.
      Soc.  Exp.  Biol.  29:1482-1488, 1970.
Fed.  Proc.  Fed.  Am.
 Milham,  S. , Jr.  Studies of morbidity  near a copper  smelter
      Perspect.  19:131-132,  1977.
   Environ. Health
 Milham,  S.,  Jr.,  and T.  Strong.   Human arsenic exposure  in  relation to a
      copper smelter.   Environ.  Res. 7:176-182, 1974.

 Mizuta,  N. , M. Mizuta,  F.  Ita, T. Ito, H. Uchida, Y. Watanabe, H. Akama, T.
      Murakami, F.  Hayashi, K.  Nakamura, T.  Yamaguchi, W.  Mizuia, S.  Oishi, and
      H.  Matsumura.  An outbreak of acute arsenic poisoning caused by  arsenic-
      contaminated  soy sauce (shoye).   A clinical  report of 220 cases.  Bull
      Yamaguchi Med.  Sch.  4:131-50, 1956.

 Morris,  J. S., M.  Schmid, S. Newman, P. J. Scheuer,  and S. Sherlock.  Arsenic
      and noncirrhotic portal hypertension.   Gastroenterology  64:86-94,  1974.

 Morton,  W.,  G. Starr, D.  Pohl, J. Stoner, S.  Wagner,  and P.  Weswig.  Skin
      cancer  and water arsenic  in  Lane  County, Oregon.   Cancer (Philadelphia")
      37:2523-2532,  1976.

 Moxon, A.  L. , and  K.  P.  DuBois.   The influence of arsenic and certain other
      elements  on  the toxicity  of  seleniferous grains.   J. Nutr. 18:447-457
      1939.                                                        —

 Munro, I.  C.   Naturally occurring toxicants  in foods and  their significance.
      Clin. Toxicol. 9:647-663,  1976.

 Munro, I.  C. , S.  M.  Charbonneau,  E.  Sandi, K.  Spencer, F.  Bryce, and H.  C.
      Grice.   Biological  availability of arsenic from fish.   Toxicol. Appl
      Pharmacol. 29:111,  1974.   (Abstr.  #92.)

 Mushak,  P. _K. _Dussauer and E. L. Walls.   Flameless  atomic  absorption  (FAA) and
      gas-liquid  chromatographic studies  in arsenic  bioanalysis.   Environ
      Health. Persp. 19:  5-10, 1977.

 Nadeyenko,  V.  G. ,   V.  G.  Lenchenko,  S.  B.  Genkina,  and T.  A. Arkhipenko.
      Influence  of  tungsten, molybdenum,  copper and arsenic in  intrauterine
      development of the  fetus.   TR-79-0353.   Farmakologiya i  Toksikoloqiya
      41:620-623, 1978.

Nagamatsu, K.,  and A.  Igata.   Dominant  and non-dominant arsenical  neuropathy.
     TR-79-0523, Kagoshim University.   Rinsho  Shinkei 15:1-4,. 1975.
                                   10-16
-------
Nakamura, I., T.  Arao,  K.  Inoue, T. Ono, Y. Ishii, T. Sato, K. Kawakami, and r
     H. Kuwahara.   Study  on  the effect of  arsenic on human bodies, Part I.
     Japanese Public  Health  Association  (Translated for  EPA by SCITRAN,  Santa
     Barbara, Calif., EPA translation No. TR-120), 1973.
                                                     )
Nakamuro, M., and Y.  Sayato.  Comparative studies of chromosomal aberrations
     induced by trivalent and pentavalent arsenic.  Mutat.  Res. 88:73-80, 1981.
National Academy of Sciences.
     DC, 1977.
Arsenic.  National Academy of Sciences, Washington,
National Institute  for  Occupational  Safety and Health.   Criteria for a Recom-
     mended  Standard....   Occupational  Exposure  to Inorganic Arsenic.   New
     Criteria - 1975.   HEW  Publication No.  (NIOSH) 75-149,  U.S.  Department of
     Health, Education, and Welfare, Washington, DC, 1975.

Natusch, D.  F.  S.,  J.  R.  Wallace,  and C.  A.  Evans, Jr.   Toxic trace elements:
     Preferential concentration in respirable  particles.  Science  183:202-204,
     1974.

Nelson, H.  A.,  M.  R. Crane,  and  K.  Tomson.    Inorganic  arsenic  poisoning in
     pastured  feeder lambs.   J.  Am.  Vet. Med. Assoc.  158:1943-1945,  1971.

Nelson, W.  C., M. H. Lykins,  J. Mackey, V. A.  Newill, J.  F.  Finklea, and  D.  I.
     Hammer.  Mortality among orchard workers exposed to lead arsenate spray:
     a cohort study.  J.  Chron. Dis.  26:105-118, 1973.

Neubauer,  0.   Arsenical cancer:   a  review.   Br.  J. Cancer  1:192-251,  1947.

Newman, J.  A.,  V.  E. Archer,  G.  Saccomanno, M. Kuschner, 0. Auerbach, R. D.
     Grondahl, and J. C.  Wilson.  Histological  types of bronchogenic carcinoma
     among  members  of  copper-mining and  smelting  communities.   In:  Occupa-
     tional  Carcinogenesis,  Proceedings of  a  Conference, New York Academy of
     Sciences,  New  York,  New York,  March 24-27,  1975.   Ann. N.Y.  Acad. Sci.
     271:260-268, 1976.

Nielsen, F.  H. , D.  R.  Myron,  and E.  0. Uthus.  Newer trace elements—vanadium
     (V) and arsenic (As) deficiency signs and possible metabolic roles.   In:
     Trace  Element  Metabolism in Man  and Animals, Vol. 3,  M.  Kirchgessner,
     ed. ,  Arbeitskreses fur Tierernahrungsforschung, Weikenstephen, Germany,
     1978.   pp. 244-247.
Nielsen,  I.  H.,  S.  H.  Givand,  and  D.  R.  Myron.
     ment for  arsenic  by  the  rat.   Fed.  Proc.
     1974.
                  Evidence  of  a  possible  require-
                Fed. Am.  Soc.  Exp.  Biol.  34:923,
 Nishioka,  H.  Mutagenic activities of metal compounds in bacteria. Mutat. Res.
      31:  185-189,  1975

 Nordenson,  I.,  G.  Beckman,  L.  Beckman,  and S. Nordstrom.   Occupational  and
      environmental  risks  in and around a smelter  in  northern Sweden.  II.
      Chromosomal  aberrations  in workers  exposed  to  arsenic.   Hereditas
      88:47-50,  1978.
                                    10-17
-------
 Nordenson,  I.,  and L.  Beckman.   Occupational  and environmental risks in and
      around a smelter in northern Sweden.   Hereditas 96:175-181,  1982.

 Nordenson,  I.,  S.  Salmonsson,  E.  Brun,  and G.  Beckman.   Chromosome aberrations
      in  psoriatic patients treated with  arsenic.   Hum.  Genet. 48:1-8, 1979.

 Nordenson,  I.,  A. Sweins, and L.  Beckman.  Chromosome aberrations in cultured
      human  lymphocytes exposed to trivalent and pentavalent arsenic.   Scand.  J.
      Work Environ.  Health 7:277-281,  1981.

 Nordstrom,  S.,  L. Beckman, and I.  Nordenson.   Occupational  and environmental
      risks  in  and  around a smelter  in northern Sweden.  I.  Variations  in
      birthweight.   Hereditas 88:43-46,  1978a.

 Nordstrom,  S. ,  L. Beckman, and I.  Nordenson.   Occupational  and environmental
      risks  in and around a smelter in northern Sweden.   V.  Spontaneous abor-
      tion among female employees  and  decreased birth weight  in their  offspring.
      Hereditas  90:291-296, 1978b.

 Nordstrom,  S. ,  L. Beckman, and I.  Nordenson.   Occupational  and environmental
      risks  in and around a smelter in  northern Sweden.  VI.   Congenital  Mal-
      formations.   Hereditas 90:297-302, 1978c.

 Nordstrom,  S. ,  L. Beckman, and I.  Nordenson.   Occupational  and environmental
      risks  in and around a smelter in  northern Sweden.  III.   Frequencies  of
      spontaneous  abortion.  Hereditas 88:51-54,  1978d.

 Nurse, D. S.   Hazards of  inorganic arsenic.   Med.  J. Australia 1:102,  1978.

 Oberly, T. J.,  C.  E.  Piper, and D.  S. McDonald.  Mutagenicity  of metal  salts in
      the  L5178Y mouse lymphoma assay.  Toxicol.  Environ.  Health.  9:367-376,
      1982.

 Occupational  Safety  and  Health Administration.  Occupational  exposure  to
      inorganic  arsenic:   Final standard.    (43  FR 19589, May 5, 1978; 29 CFR
      1910.1018),  1978.

 Odanaka,  Y. ,  0.  Matano, and S. Goto.   Identification of dimethylated arsenic
      by gas chromotography-mass spectrometry in  blood, urine,  and  feces of  rats
     treated  with ferric methanearsonate.   J.   Agric. Food Chem.  26:505-507,
     1978.                                                        ~~

Ohno, H., F.  Hanaoka,  and M.  Yamada.  Inducibility of sister chromatid exchanges
     by heavy-metal ions.  Mutat.  Res. 104:141-145,  1982.

Ohta, M.  Ultra-structure of  sural nerve  in a case  of  arsenical  neuropathy.
     Acta Neuropathol. (Berl.) 16:233-242,  1970.

Okamura,   K.  ,  T. Ota,  K.  Horiuchi,  H.  Hiroshima, T. Takai,  Y.  Sakurane,  and  T.
     Baba.  Symposium on arsenic  poisoning by  powdered  milk.   (2).  Diagnos.
     Ther. (Shinryo)  9:240-249, 1956.

Osato, K.   Effects of oral  administration of arsenic  trioxide during the
     suckling stage of rats.    Fukuoka Acta Med.  68:464-491, 1977.
                                   10-18
-------
O'Shaughnessy, E.,  and  G.  H.  Kraft.  Arsenic poisoning:   Long-term follow-up
     of a nonfatal case.  Arch. Phys. Med. Rehabil. 57:403-406, 1976.

Osswald, H. ,  and K.  I.  Goerttler.   Leukosen  bei  der Maus nach diaplacent arei
     und post natalei Arsenic-application.  Verh. Dtsch. Ges.  Pathol.  55:289-293,
     1971.

Ott, M. G. ,  B.  B. Holder, and  H.  I.  Gordon.   Respiratory cancer and occupa-
     tional exposure  to arsenicals.  Arch. Environ. Health  29:250-255,  1974.

Overby, L.  R.,   and  R.  L.  Fredrickson.   Metabolic  stability of radioactive
     arsanilic acid in  chickens.   J.  Agric.  Food  Chem.   11:378-381,  1963.

Pagano, G., A.  Eposito, P. Bove,  M.  de Angelis, A. Rota,  E. Vamvakinos, and
     G. G.  Giordana.   Arsenic-induced  developmental  defects  and  mitotic
     abnormalities  in  sea urchin  development.   Mutat.  Res. 104:351-354, 1983.

Paton,  G.  R.  and A. C.  Allison.   Chromosome damage in  human cell  cultures
     induced by  metal salts.  Mutat.  Res.  16:332-336,  1972.

Pelfrene, A.  Arsenic and  cancer:   the  still unanswered  question.  J.  Toxicol.
     Environ. Health 1:1003-1016,  1976.

Penrose, W.  R.,  H.  B.  S.  Conacher, R.  Black, J.  C. Meranger, W.  Miles,  H.  M.
     Cunningham,  and W.  R. Squires.   Implications of inorganic/ organic inter-
     conversion  on  fluxes of arsenic in marine  food webs.   Environ.  Health
     Perspect. 19:53-59, 1977.

Peoples, S. A.   Arsenic toxicity in cattle.  Ann. N.Y. Acad.  Sci. 111:644-649,
     1964.

Peoples, S.  A.   The metabolism of arsenic in man and  animals.   In:  Arsenic:
     Industrial,  Biomedical,  and  Environmental  Perspectives.   W.  Lederer and
     R. Fensterheim, eds., Van  Nostrand Reinhold, New  York,  1983.

Perry,  K.,  R.  G. Bowler,  H.  M. Buckell, H. A. Druett, and R. S.  F.  Shilling.
     Studies  in  the incidence of  cancer in a factory  handling inorganic com-
     pounds of  arsenic.  II.   Clinical  and environmental investigations.  Br.
     J. Ind. Med. 5:6-15,  1948.

Pershagen,  G.  Lung  cancer mortality, occupational  exposure and smoking habits
     in a  region surrounding a smeltery.   In:   Proceedings  from an Interna-
     tional  Symposium on the Control  of Air Pollution in the Working  Environ-
     ment,  Stockholm, 6-8  September 1977,  Geneva, International Labour Office,
     1978.

Pershagen,  G. ,  and M.  Vahter.  Arsenic.  A  toxicological  and epidemiological
     appraisal.   Libertryck,  Stockholm, Sweden,  SNV PM-1128, 1979.

Pershagen,  G.,  C-G. Elinder, and  A-M.  Bolander.  Mortality  in a region sur-
     rounding an arsenic emitting  plant.  Environ.  Health Perspect.  19:133-137,
     1977.

Pershagen,  G. ,  B. Lind and  N-E.  Bjorklund.   Lung  retention and toxicity of
     some  inorganic arsenic  compounds.   Environ. Res.  29:425-434, 1982.

                                    10-19
-------
 Pershagen,  G.,  G.  Nordberg, and N-E.  Bjorklund.   Carcinomas of the respiratory
      tract  in  hamsters given arsenic trioxide and/or benzo(a)pyrene by the pul-
      monary route.   Environ.  Res.  (in press, 1983).

 Petres,  J. , and A.  Berger.   Zum einfluss anorganischen  arsens  auf die DNS-
      synthese   menschlicher lymphocyten  in  vitro.   Arch.  Derm.   Forsch.
      242:343-352,  1972.                   ~~

 Petres,  J., and M.  Hundeiker.   [Chromosome pulverization induced iji vitro in
      cell  cultures  by  sodium  diarsonate.]   Arch.   Klin.  Exp.  Dermatol.
      231:366-370,  1968.

 Petres,  J.,  D.  Baron,  and M.  Hagedorn.   Effects  of arsenic cell  metabolism and
      cell  proliferation:   cytogenetic  and  biochemical  studies.   Environ.
      Health Perspect.  19:223-227,  1977.

 Petres,  J., K.  Schmidt-Ullrich, and  U.  Wolf.   Chromosomenaberrationen  an
      Menschlichen  lymphozyten  bei chronischen  Arsenchader.   Dtsch.  Med.
      Wochenschr.   95:79,  1970.

 Pimentel, J. C., and F.  Marques.   Vineyard  sprayer's lung.   A new occupational
      disease.   Thorax  24:678-688,  1969.
Pinto,  S.  S. ,  and B.  M. Bennett.   Effect of arsenic trioxide  exposure
     mortality.  Arch.  Environ.  Health  7:583-591,  1963.
on
Pinto, S.  S. ,  and C.  M.  McGill.   Arsenic trioxide exposure in industry.  Ind.
     Med. Surg. 22:281-287,  1953.

Pinto, S.  S., M.  0.  Varner, K.  W.  Nelson,  A.  L. Labbe, and  L.  D.  White.
     Arsenic trioxide absorption and excretion  in  industry.   J.  Occup. Med.
     18:677-680, 1976.

Pinto, S.  S. , P.  E.  Enterline,  V.  Henderson  and M.  0. Varner.   Mortality
     experience in relation  to a measured arsenic trioxide  exposure.   Environ.
     Health Perspect. 19:127-130,  1977.

Pinto, S.  S. ,  V.  Henderson,  and  P.  E.  Enterline.  Mortality  experience of
     arsenic - exposed  workers.    Arch.  Environ. Health 33:325-331,  1978.

Poma, K., N. Degraeve, M. Kirsch-Volders, and C.  Susanne.   Cytogenetic analysis
     of bone marrow  cells  and spermatogonia of male mice after iji vivo treat-
     ment with arsenic.  Experientia 37:129-130,  1981b.

Poma, K. , N. Degraeve, and M. Kirsch-Volders.  A  combined action  of  arsenic and
     ethyl methanesulfonate  (EMS)  in somatic and  germ cells of mice.   Mutat.  Res.
     85:295, 1981a.

Poma, K., M. Kirsch-Volders, and C. Susanne.  Cytogenetic investigation  in  mice
     of As, Hg, and EMS administration on their own and  in  combination.  Mutat.
     Res.  97:213, 1983.

Poma, K.,  and C.  Susanne.  Influence of thiol-inhibiting substances  on the  cyto-
     genetic effects  of  ethyl methanesulfonate  (EMS)  on mice  treated jm vivo.
     Mutat.  Res.  97:244, 1982.

                                   10-20
-------
Popper, H. ,  L.  B. Thomas, N. C. Telles, H. Falk, and I. J. Selikoff.  Develop-
     ment of hepatic angiosarcoma in man induced by vinyl chloride, thorotrast
     and arsenic.  Am. J. Pathol. 92:349-376, 1978.

Prystowsky,  S.  D., G. J. Elfenbein, and S. I. Lamberg.  Nasopharyngeal carcinoma
     associated with long-term arsenic ingestion.  Arch. Dermatol. 114:602-'603,
     1978.

Ray-Bettley, F., and J. A. O'Shea.  The absorption of  arsenic  and  its relation
     to carcinoma.  Br. J. Dermatol. 92:563-568, 1975.                       /

Regelson, W. ,  U. Kim,  J.  Ospina,  and J.  F. Holland.  Hemangioendothelial
     sarcoma of  liver from  chronic  arsenic intoxication by Fowler's solution. \
     Cancer 12:514-522, 1968.

Rencher, A.   C. ,  M.  W. Carter, and  D.  W.  McKee.  A  retrospective  epidemio-
     logical study  of mortality  at  a  large western copper smelter.  J.  Occup.
     Med. 19:754-58, 1977.

Reymann, F.  , R.  Miller, and A. Nielsen.   Relationship between arsenic intake
     and internal  malignant neoplasms.   Arch.  Dermatol.  114:378-381, 1978.

Reynolds, E. S.   An account of  the epidemic outbreak of arsenical poisoning
     occurring in beer-drinkers  in the north of England and Midland Counties
     in 1900.  Lancet 1:166-170,  1901.

Ridgeway, L. P., and D. A.  Karnofsky.   The effects of metals on  the chick
     embryo:   Toxicity  and  production of  abnormalities in development.   Ann.
     N.Y. Acad.  Sci.  55:203-215,  1952.

Roat,  J. W. , A.  Wald,  H.  Mendelow, and K.I. Pataki.   Hepatic angiosarcoma
     associated  with  short-term  arsenic ingestion.  Am.  J.  Med. 73:933-936,
     1982.

Rogers,  E.  H. , N.  Chernoff, and  R. J. Kavlock.  The  teratogenic potential  of
     cacodylic acid in the  rat and mouse.   Drug Chem.  Toxicol. 4:49-61,  1981.

Rom, W. N., G.  Varley,  J. L. Lyon and S.  Shapkow.   Lung cancer mortality among
     residents living near  the El Paso smelter.  Br.  J.  Ind. Med.  39:  269-272,
     1982.

Rossman, T.  G. ,  M.  S.  Meyn, and  W. Troll.   Effects  of arsenite on DNA repair
     in Escherichia col i.   Environ. Health Perspect.  19:229-233, 1977.

Rossman, T.  G. ,  D.  Stone,  M.  Molina and  W.  Troll.   Absence  of arsenite  muta-
     genicity  in E. coli and Chinese  hamster cells.   Environ.  Mut. 2:371-379,
     1980.

Rossman, T.  G.   Enhancement of UV-mutagenesis by low concentrations of arsenite
     in  E.  coli.  Mutat.  Res.  91:207-211,  1981.

Rossman, T. G.   Progress report:  EPA  project no. 808482-02,  pg 8,  1983.
                                    10-21
-------
Rossman,  T.  G.,  M. S.  Meyn,  and  W.  Troll.   Effects  of sodium arsenite on the
     survival of UV-irradiated  Escherichia  coli:   Inhibition of a recA-dependent
     function.  Mutat.  Res. 30:157-162,  1975.

Rossner,  P.,  V.  Bencko, and  H. Havrankova.   Effect  of the  combined action of
     selenium and  arsenic  on  suspension  culture  of mice fibroblasts.   Environ.
     Health. Perspect.  19:235-237,  1977.

Rossner,  P.   Mutagenic effect of sodium arsenite in Chinese hamster cell line
     Dede.  Mutat. Res. 96:65-66, 1977.

Roth, F.  The sequelae  of  chronic arsenic poisoning  in Moselle  Vintners.   Ger.
     Med. Mon.  2:172-175,  1957.  Roth F. Uber den Bronchialkrebs Arsengescho-
     digter Winzer.  Virchows Arch.  331:119-137,  1958.

Rowland,  I. R., and M.  J.  Davies.   In vitro  metabolism of inorganic arsenic by
     the  gastro-intestinal microflora of the rat.  J.  Appl.  Toxicol.  1:278-283,
     1982.

Rozenshtein, I. S.  Sanitary  toxicological assessment  of low concentrations of
     arsenic trioxide  in  the atmosphere.  Hyg.  Sanit.  (USSR) 34:16-22,  1970.

Rudnai, P., and M. Borzsony.  Tumor-inducing effect  of arsenic  trioxide  treat-
     ment in CFLP  mice.  Magy.  Onkol. 25:73-77,  1981.

Sabbioni, E. ,  E.   Marafonte,  F. Bertolero,  and  V. Foa. Inorganic  arsenic:
     metabolic patterns and  identification  of arsenic-binding  components in
     the  rabbit.   Proc. Int. Conf.  Management Control  Heavy Metals  in  the
     Environment.   London, 1979.  Pp. 18-21.

Schrauzer, G. N.,  and D.   Ishmael.   Effects  of selenium and of  arsenic on the
     genesis of spontaneous  mammary tumors  in inbred  C-H  mice.  Ann. Clin.
     Lab. Sci. 4:441-447,  1974.                         6

Schrauzer, G. N.,  D.  A. White, and  C. J. Schneider.   Cancer mortality corre-
     lation studies.   IV.   Associations with dietary  intake and  blood levels
     of certain trace elements, notably  selenium  antagonists.   Bioinorg.  Chem.
     7:35-36, 1977.
Schrauzer, G.  N. ,  D.  A.  White,  J.  E.  McGinness,  and C.  J.  Schneider.   Arsenic
     and cancer:   effects  of joint administration of arsenite and selenite on
     the genesis  of  mammary adenocarcinoma in
     Bioinorg. Chem.  9:245-253, 1978.
inbred female C-H/St  mice.
Schroeder, H. A., M. Mitchener.  Toxic effects of trace elements on the repro-
     duction of mice and rats.  Arch. Environ. Health 23:102-106, 1971.

Schwartz, K.  Essentiality versus toxicity of metals.  In:  Clinical Chemistry
     and Chemical Toxicology  of Metals, S. S. Brown,  ed.,  Vol.  1,  Elsevier,
     Amsterdam, 1977.  pp. 3-22.

Selby, L. A.,  A.  A. Case, G.  D. Osweiler, and H. M. Hayes, Jr.  Epidemiology
     and toxicology of arsenic poisoning in domestic animals.  Environ. Health
     Perspect. 19:183-189, 1977.
                                   10-22
-------
Silver, A.  S. ,  and P.  L.  Wainman.   Chronic arsenic poisoning following use of
     an asthma  remedy.  J.A.M.A. 150:584-585,  1952.

Singh,  I.   Induction of reverse mutation  and  mitotic  gene  conversion  by  some
     metal compounds in Saccharomyces cerevisiae.  Mutat. Res.  117:149-152,  1983.

Sirover, M. A.  Effects of metals  in ut vitro  bioassays.  Environ.  Health.  Per-
     spect. 40:163-172, 1981.

Small,  H.  and C.  McCants.   Residual  arsenic in soils and concentration in to-
     bacco.  Tobacco Sci. 6:34-36, 1962.

Smith, T.  J., D. J. Eatough, L. D. Hansen,  and N.  F. Mangelsen.  The chemistry
     of sulfur  and arsenic in  airborne  copper smelter particulates.   Bull.
     Environ. Contam. Toxicol.  15:651-659,  1976.

Smith,  T.  J. ,  E.  A. Crecelius, and J. C.  Reading.   Airborne arsenic exposure
     and excretion  of methylated arsenic compounds.  Environ. Health.  Perspect.
     19:89-93,  1977.

Sommers, S. C. , and R. G. McManus.  Multiple arsenical cancers  of  the  skin  and
     internal organs.  Cancer 6:347-359, 1953.

Southwick, J. ,  A.  Western,  M.  Beck,  T.  Whitley,  R. Isaacs,  J.  Petajan, and C.
     Hansen.  Community  health associated with arsenic  in  drinking water in
     Millard  County,  Utah.   Health Effects Research  Laboratory, U.S.  Environ-
     mental Protection Agency,  Cincinnati,  Ohio.   EPA  Report No. EPA-600/1-81-064.
     Available  from:  NTIS, Springfield, VA; PB82-108374.   1981.

Sram, R. and V. Bencko.  A contribution for the evaluation  of the  genetic risk
     of exposure to arsenic.  Cs.  Hyg. 19:  308-315,  1974.

Sram, R. J.   Relationship  between  acute  and chronic  exposures  in mutagenicity
     studies in mice.  Mutat. Res. 41:25-42, 1976.

Sunderman, F. W. ,  Jr.   A review of the carcinogenicities of nickel, chromium
     and  arsenic   compounds  in  man  and animals.   Prev.  Med.  55:279-294,
     1976.

Suta, B. E.   Human exposure to atmospheric arsenic.   Final Report.  Stanford
     Research Institute  CRESS  Rpt. No.  50, EPA  Contract No.  68-01-4314 and
     68-02-2835, 1980.

Szuler, I. M. ,  C.  N.  Williams, J. T.  Hindmarsh,  and  P.-D. Hosoon.  Massive
     variceal hemorrhage secondary to presinusoidal  portal  hypertension due to
     arsenic poisoning.  CMA J. 120:168-171, 1979.

Tallin, Y. , and  D.  T.  Bostick.  The  determination of arsenic  and arsenicals.
     J. Chromatogr. Sci. 13:231-237, 1975.

Tarn, K. H., S. M.  Charbonneau,  F.  Bryce, and G. Lacroix.  Separation of arsenic
     metabolites in  dog  plasma and urine  following  intravenous injection of
     74As.   Anal.  Biochem. 86:505-511, 1978.
                                   10-23
-------
Tarn, K.  H. ,  S.  M. Charbonneau,  G.  Lacroix,  and F.  Bryce.   Confirmation of
     inorganic  arsenic  and dimethylarsinic acid in urine and plasma of dog by
     ion-exchange and TLC.  Bull. Environ. Contam. Toxicol.  21:371-374,  1979a.

Tarn, K.  H. ,  S.  M.  Charbonneau,  E.  Bryce, and G. La  Croix.   Metabolism of
     inorganic  arsenic  (74As) in humans  following oral  ingestion.   Toxicol.
     Appl. Pharmacol. 50:319-322, 1979b.

Tamura,  S.   Study of arsenic metabolism  (Report  20) -  influence of  arsenic
     agents  on  the brain  of  rats in the  developmental  stage.   TR-79-0356.
     Folia Pharmacol. Japn. 74:1-14, 1978.

Tay, C-H., and  C-S.  Seah.  Arsenic poisoning  from anti-asthmatic herbal  pre-
     parations.   Med. J. Aust. 2:424-428,  1975.

Terada,  H. ,  K.  Katsuta,  T.  Sasakawa,  T.  Saito,  H.  Shrota, K. Fukuchi,  E.
     Sekiya, Y. Yokoyama,  S.  Hirokav/a,  G. Watanabe,  K.  Hasegawa,  T. 0. Shina,
     and E. Sekiguchi.  Clinical  observations  of chronic toxicosis  by arsenic.
     Ninon Rinsho 18:2394-2403 (EPA Tr. No. TR106-74), 1960.

Terada,  H.   Clinical observation of chronic  toxicosis  by arsenic.   Nihon
     Rinsho 18:118-127, 1960.

Thomas,  M. D. ,  and T. R.  Collier.   The concentration of arsenic in  tobacco
     smoke determined  by a rapid titrimetric  method.   J. Ind.  Hyg. Toxicol.
     27:201-206, 1945.

Thompson,  R.  J.  The collection and measurement of airborne  arsenic.  WHO
     Publ. Air  Pollut. Meas.  Tech. Part II, 126-131, 1976.

Tkeshelashvili, T.  K.,  C.  W. Shearman,  R.  A. Zakour,  R.  M.  Koplitz, and
     L. A. Loeb.  Effects  of  arsenic, selenium and chromium on  the  fidelity of
     DMA synthesis.  Cancer Res.  40:2455-2460,  1980.
Tokanehara, S. ,  S.  Akao, and S. Tagaya.
     toxicoses  caused by powdered  milk.
     (EPA translation No. TR 110-74).
Blood findings in infantile arsenic
Shonika  Ninsho  9:1078-1084,  1956.
Tokudome, S. ,  and M.  Kuratsune.  A cohort  study  on mortality  from  cancer  and
     other causes among workers at a metal  refinery.   Int. J.  Cancer  17:310-317,
     1976.

Tseng, W.  P.   Effects and dose-response relationships of  skin  cancer and
     Blackfoot disease with arsenic.   Environ. Health Perspect.  19:109-119,
     1977.

Tseng, W. P., H., M. Chu, S. W. How, J. M. Fong, C. S.  Lin, and S. Yeh.   Preva-
   •  lence  of skin cancer  in  an  endemic area of  chronic  arsenalicalism in
     Taiwan.  J. Natl. Cancer  Inst. 40:453-463, 1968.

Tsutsumi, S. ,  and K.  Kato.  Effects  of  dimercaprol  or thioctic  acid on the
     distribution  and excretion  of 74As in rats.  Bull.  Tokyo Dent. Coll.
     16:69-74, 1975.
                                   10-24
-------
U.S. Environmental Protection Agency.  Ambient Water Quality  Criteria  Document
     for Arsenic.  Cincinnati, OH.  1980.

U.S. Environmental Protection Agency.  National  Revised  Primary  Drinking  Water
     Regulations, Advanced  Notice of Proposed Rulemaking,  Federal  Register,
     Vol. 48, No. 194:45502-45521  (October 5), 1983.

Uldall,  P.,  H.  Khan,  J.  Ennis,  R.  McCallum and T.  Crimson.   Renal damage from
     industrial arsine poisoning.  Br. J. Ind. Med. 27:372-377,  1970.

Underwood,  E.  J.   Trace  Elements in  Human  and Animal  Nutrition,  4th ed.
     Academic Press, N.Y., 1977.

Union Carbide  Corporation.   Review of the Environmental Effects of Arsenic.
     Oak  Ridge National  Laboratory,  Oak Ridge,  Tenn.   ORNL/EIS-79,  1977.

Urakabo, G. , A.  Hasegawa, and S.  Nakaura.   Studies on the fate of poisonous
     metals  in experimental animals  (V).   Body retention  and  excretion  of
     arsenic.  J. Food  Hyg.  Soc.  Jpn. 16:334-336,  1975.  (In Japanese with
     English summary.)

Uthus,  E. ,  W.  Cornatzer and F.  Nielsen.  Consequences of arsenic deprivation
     in  laboratory  animals.   In:  Arsenic:   Industrial,  Biomedical  and
     Environmental Perspectives.   W.  Lederer  and  R.  Fensterheim, eds.,  Van
     Nostrand Reinhold, New York,  1983.

Vahter, M.   Biotransformation  of trivalent and  pentavalent inorganic  arsenic
     in mice and rats.  Environ. Res.  25:286-293, 1981.

Vahter, M. ,  and  H.  Norin.  Metabolism of   As-labeled trivalent and pentava-
     lent inorganic arsenic in mice.  Environ. Res. 2_l:446-457,  1980.

Vahter, M., and J. Envall.  Iji vivo reduction of arsenate in  mice and  rabbits.
     Environ. Res. 32:14-24, 1983.

Vahter, M., E.  Marafante, A. Lindgren, and L. Dencker.   Tissue distribution  and
     sub-cellular binding of  arsenic  in marmoset monkeys  after ingestion of
     74As-arsenite.   Arch. Toxicol. 51:65-77, 1982.

Vainio, H., and M. Sorsa.  Chromosome aberrations and their relevance  to  metal
     carcinogenesis.   Environ. Health Perspect.  40:173-180, 1981.

Valentine, J. , H. Kang,  and G. Spivey.  Arsenic levels  in human blood, urine
     and  hair  in response  to exposure  via  drinking  water.   Environ.  Res.
     20:24-32,  1979.

Vallee,  B.  L. , D. D. Ulmer,  and W.  E.  C. Wacker.   Arsenic toxicology and
     biochemistry.  Arch. Ind. Health 21:132-151, 1960.

Villar, T.  G.   Vineyard sprayer's lung.   Clinical aspects.   Am. Rev.  Respir.
     Dis.  110:545-555, 1974.

Walsh,  L.  M. ,  and D.  R.  Keeney.  Behavior  and  phytotoxicity of inorganic
     arsenicals in soils.   In:   Arsenic Pesticides, ACS Symp.  Series  7,  Am.
     Chem. Soc., Washington, DC, 1975.  p. 35.

                                   10-25
-------
Walsh,  P.  R.,  R.  A. Duce,  and J.  L.  Fasching.   A sampling technique for the
     determination  of  participate and vapor-phase  arsenic  in  the  atmosphere.
     Air  Pollut.  Meas.  Tech., Part II,  Publ.  World Meteorol.  Org.  No.  460,
     140-149, 1977a.

Walsh,  L.  M.,  M.  E.  Sumner, and D.  R.  Keeney.   Occurrence and  distribution of
     arsenic in  soils  and plants.   Environ.  Health Perspect. 19:67-72, 1977b.

Wan, B.,  R. T.  Christian,  and  S. W. Soukup.  Studies of  cytogenetic  effects  of
     sodium arsenicals on  mammalian cells  in vitro.  Environ.  Mutag.  4:493-498,
     1982.

Welch,  K., I. Higgins, M.  Oh,  and C. Burchfiel.   Arsenic exposure, smoking and
     respiratory cancer  in copper smelter  workers.  Arch. Env. Hlth.  37:325-335,
     1982.

Wen, W.  N.,  T-L.  Lieu,  H-J. Chang,  S.  W.  Wuu,  M-L.  Yau,  and K. Y.  Jan.  Base-
     line  and  sodium arsenite-inducted sister  chromatid exchanges in cultured
     lymphocytes from patients with Blackfoot disease  and healthy  persons.   Hum.
     Genet. 59:201-203,  1981.

Wentworth, W.  U.S.  EPA, Office of Drinking Water, State Programs Divis.ion,
     Compliance Monitoring Data  from  Federal  Reporting Data  System, 1983.

Westhoff,  D. D.,  J.  R.   Samaha, and A. Barnes, Jr.  Arsenic  intoxication as  a
     cause of megaloblastic anemia.  Blood 45:241-246, 1975.

Westb'b', G., and M. Rydalv.  Arsenic levels in foods.   Var Foda 24:21-40,  1972.
     (In Swedish with English  summary.)

Whanger,  P.  D.,  P.  H.  Weswig,  and  J.  C.   Stoner.   Arsenic  levels  in Oregon
     waters.  Environ.  Health  Perspect. 19:139-143, 1977.

Whittemore, A.  S.  The age  distribution of human  cancer  for  carcinogenic expo-
     sures of varying intensity.  Am. J. Epidemic!. 106:418-423, 1977.

Wildenberg,  J.   An  assessment of experimental  carcinogen-detecting  systems
     with special reference to inorganic arsenicals.   Environ. Res.  16:139-152,
     1978.

Willhite,  C. C.   Arsenic-induced axial skeletal  (dysraphic) disorders.   Exp.
     Mol. Pathol. 34:145-158,  1981.

Winkler, W. 0.   Identification and estimation of  the arsenic residue  in livers
     of rats ingesting arsenicals.  J. Assoc. Off.  Anal.  Chem. 45:80-91,  1962.

Wood, J. M.  Biological  cycles for toxic elements  in the environment.   Science
     183:1049-1052,  1974.

Woods,  J.  S.,  and  B. A.  Fowler.   Effects  of  chronic arsenic exposure on hema-
     topoietic function  in adult mammalian liver.  Environ. Health  Perspect.
     19:209-213, 1977.

Woolson,  E.  A.  Fate  of arsenicals in different  environmental  substrates.
     Environ. Health Perspect. 19:73-81, 1977.

                                   10-26
-------
Wool son,  E.  A.   Generation of  dimethylarsine from soil.   Paper  presented at
 '  .  16th.. Meeting,  Weed Science Society  of America, 1976.   Paper No. 218.
World- Health Organization.   Environmental  Health  Criteria 18:
     International Programme  on  Chemical  Safety, Geneva, 1981.
                                                         Arsenic.
Yamashita, N. ,  D.--Makoto, M..  Rtshio,  H.  Hojo,  and M.  Tanaka Masato.  Current
     state of,"Kyoto, children  poisoned by arsenic-tainted Morinaga  dry  milk.
     TR-74-108.-  Nihon eiseigaku  zasshi  [Japanese J.  of Hyg.]  27:364-399,
     .1972, .-••''.                                 '    .   ,  ~~ :

Yamauchi,  Hi,  and Y.  Yamamura.   Urinary  inorganic arsenic and methyl arsenic
     excretion  following  arsenate-rich seaweed  ingestion.  Jpn. 'J. Ind;  Health
     21:47-54,  1'979. .  (Engl.  summary,  Tables and Figures).
YoshlkaWa, 6-116/ IK) 6 24 REGION NO. 4
                                    10-27
-------
-------