ASSESSMENT OF

TECHNOLOGIES
FOR DETERMINING
CANCER RISKS FROM
THE ENVIRONMENT

 

JUNE 1981

 
life expectancy of a person born in the U.S. in 1950 was 68 years. By. 1977 it
had risen to 73 years, and 80% of those dying today are over 60 years of age.

Cancer, although it kills at all ages, primarily affects the elderly.

Cancer has a major impact on the nation’s economy, from the personal costs
of treatment and lost income, to public expenditures for screening programs,
public education, and cancer research. The costs of cancer are not merely
economic, though these are enormous. Social costs have taken on increasing
prominence in recent years, and include more than the obvious pain and suf fering
of the victim. Relatives and friends of victims, and care givers all may suffer
direct consequences of the victim’s morbidity and mortality. Social isolation,
economic dependence, lost personal and business opportunities, and many
undesirable and unwanted alterations in lifestyle are inevitable. Serious
emotional and psychological problems requiring professional attention are not
uncommon among victims and their family members, often producing irreversible
changes in family structure and relationships. The costs of these social factors

are not directly quantifiable, but some progress has been made in methodologies

{
to measure them. Severity of pain and suffering can be measured, t least in ”
relative terms, by the medication required for relief. Costs of psychiatric care ne

may be used as surrogates for emotional and psychological stress. Other
"shadow-pricing” mechanisms have been used, and a number of profiles have been
developed to consider many social factors together (Granger and Greer, 1976;
Elinson, 1974). There is no question that social costs are enormous, and /
improved methodologies will paint a more accurate picture of the impact of cancer i
on its victims and on society as a whole. (For a review of some methodologies

for valuation, see OTA, 1980.)

A common measure of disease is the number of years of life lost due to
premature mortality. This takes into account both the number of deaths and the
age at which people die. The death of a younger person will contribute more

: . fp ee X x np > Caviawee 5 aap
a Leth debe Anaamreny 5 TM, S —Oulbet ton be vanes

teen Raoul a5 rt} 4, la cen i
Lp uA We PAV LAER AN QUAN TA sR Sp os A hoe.
4 yi AB G . a A pre.

cae tm een nese a Lo.
forces.

All individuals exposed to the same dose of a carcinogen do not develop
cancer, indicating the involvement of individual susceptibility or host factors.
The genetic contribution may be minimal or may predominate. Certain familial and
genetic disorders are known to increase the risk of developing cancer. Daughters
of breast cancer patients have a higher breast cancer risk than women without
this family history, though many other factors affect the probability of
developing the cancer. Individuals with deeply pigmented skin have a lower risk
of skin cancer induced by sunlight. Retinoblastoma, a usually fatal malignant
disorder of the retinal cells occurring usually before the age of three, has a
well-defined hereditary pattern. Individuals with multiple polyposis of the
colon, an inherited trait, are at an increased risk of colon cancer. There is
also a group of familial disorders manifesting cellular abnormalities that

increase the risk of cancer: Bloom’s syndrome, Fanconi’s anemia, and the
e

immunologic deficiences (Fraumeni, 1973).

Even in these cases, however, the malignancies are not necessarily
completely spontaneous, and actions taken may prevent some of them. A case in
point is xeroderma pigmentosum, a genetic disorder predisposing to multiple skin
malignancies. Individuals with this defect develop numerous cancers and die at a
young age, usually of leukemia or lymphoma, if exposed to even moderate amounts
of sunlight. Affected individuals who have been completely sheltered from

exposure to sunlight, the precipitating factor, however, have developed no

malignancies (ref.). Oo \ Se

Table 4-1 lists several cancers that occur as inherited traits or as
TERRESTRES TRS

complications of an inherited precursor state. All of these conditions together

are believed to account for not more than an “7 y small percentage of all

fale an wen
uf - a r arty by tu

rh tec Mi, TAA.

cancer deaths (Knutsen, 19).
s

regular cigarette smokers and those among lifetime non-smokers is so extreme
. lung cancer rates can be accounted for almost totally by cigarette smoking. z

that it is not likely to be an artifact of the epidemiologic °
method. Doll and Peto (1981), calculate that the increase in male and female
wT

in which large numbers of people have been asked what they normally smoke and

These findings on the effects “of tobacco on cancer are derived from studies ' ‘
a
.
they are then followed for several years to determine the causes of any deaths

that may occur. TableJ-] presents data from the first 13 years of the largest of

 

these studies, in which the smoking habits of one million Americans were
ascertained in 1959 by Dr. E.C. Hammond on behalf of the American Cancer Society
(ACS) (unpublished). The data show that deaths from lung cancer occurred almost
12 times as frequently in male one-pack-a-day snokers as compared to male

non-smokers. Deaths from oral cavity, bladder and pancreatic tumors occurred in

 

the smoking population 6, 3 and 2 times as frequently, respectively, as in the
non-smokers. It should be borne in mind that these elevated risks would probably g
have been even higher if the people who had quit smoking during the course of the
study were eliminated from the analysis. Many who reported a history of smoking
regularly had quit by 1967, and others quit years later but this was not
accounted for in the data (Hammond, 1980, Prev. Med.). Deaths from cancers at

other sites were “t, found to be significantly affected by smoking.

pete to cancef rates seen an the ACS study are almost exactly mirrored in

a comparison of veterans who were cigarette smokers in 1954 or 1957 and veterans

    
  
  
  

who said they had never smoked regularly. Rogot and Murray (1980) found lung

 

cancer deaths occurred 11.3 times as frequently among smokers; oral cavity canter
deaths, 7 times; bladder cancer deaths, 2 times; pancreatic cancer deaths, 2
times; and deaths from cancers at other sites, 1.3 times. Other studies / om

Great Britain (Doll and Peto, 1976) and other countries (Surgeon Genera, 9

1980) show similar elevated cancer death rates ectong smokers. i yogi

yell nae
 
  
 

consumed as spirits (ref.). The apple-baged drinks that are consumed in

Northwest France are believed parti

Pure alcohol is not by itself mutagenic or carcinogenic by any of the
laboratory tests thus far devised, although many alcoholic drinks are found to be

positive in short-term tests for mutagenicity. Given the good correlation

Mutation Research, 65 (1979) 229—259 ments of alcoholic
© Elsevier/North-Holland Biomedical Press
meer riske Alcohol
)
i ‘between extrinsic
“
MUTAGENIC, CANCEROGENIC AND TERATOGENIC EFFECTS OF it are responsible
ALCOHOL

and larynx.

GUNTER OBE and HANSJURGEN RISTOW
Institut fiir Genetik, Arnimallee 5-7, D-1000 Berlin 33 (Germany)

isive alcohol

(Received 12 January 1979) jouth (excluding
(Revision received 29 March 1979) . . _-
(Accepted 5 April 1979) ’ and perhaps Ben

f the upper
Summary

Jensen, 1977).
Alcohol is mutagenic? cancerogenic and teratogenic in man. Ethanol is mu-
tagenic via its first metabolite, acetaldehyde. This is substantiated by the find- & esophagus in
ings that acetaldehyde induces chromosomal aberrations, sister-chromatid
exchanges and cross-links between DNA strands. Methanol, a contaminant of ective study of male
many alcoholic beverages, is also mutagenic via its metabolite, formaldehyde.
In addition, different indirect pathways may lead to mutations by alcohol. The effects for smoking
cancerogenic activity of alcohol remains unverified by modern standard carci- . °
nogenicity tests. Ethanol and other alcohols, as well as aldehydes, inhibit RNA the mouth and

1

synthesis in cells and in cell-free transcriptional systems. A reduction of cellular .
RNA synthesis may play an important role in the mutagenic, carcinogenic and 1ld be eliminated if

teratogenic activity of alcohol.
CApUSULCD LU Garvie ee ee cen oon ‘mate was made by

Schottenfeld (1980) for tobacco/alcohol sites (762). These sites combined
represent “approximately 36% of cancer deaths for all sites. Feldman, et al.
(1975) found that the risk of head and neck cancer was 6 to 15 times greater in
heavy drinkers who smoked than for nondrinkers and nonsmokers. Nonsmoking
drinkers had a "slightly" higher risk (around 1.5) than total abstainers while

nondrinking or light-drinking smokers had 2 to 4 times the risk.

Breslow and Enstrom (1974) correlated average annual age-adjusted cancer

foaa {-17
the United States (Cole, 14—) In animals, caffeine is shown to potentiate
the effect of carcinogenic substances (Donovan, ne) ") and whether it has

similar properties for humans is not yet known.

Whether other naturally occuring carcinogens exist in food is left to
speculation, but on present evidence, naturally occurring carcinogens are not

regarded as an important cause of cancer in the United States.
Cc. Carcinogens or Precursors Produced by Cooking

Another possible source of carcinogens is their production in cooking.
Humans are the only animals which cook their food, and it has been known for many
years that carcinogenic chemicals such as benzo(a)pyrene and other polycyclic

hydrocarbons are produced by pyrolysis when meat or fish is broiled or smoked or

. an . - ,
when food is fryed in fat which has been used repeatedly. / Sugimura (1977) Cnn
‘ | Le =
demonstrated that broiling also produces powerful mutagens that cannot be ye
acpunted for by the production of benzo(a)pyrene akone. Commoner (]{—) and SRI

(14—) have shown that mutagens are produc ed,“by cooking to relatively low

2 y {
. “4 . 2 :
temperatures between 100-200 C. 3 hire par AIY~ af 3 hae
Aged cow Ana Ares <f 7 (me
Many epidemiologists have sought to rélate the consumption of various cooked "4

Are VAN

foods to the development of gastric cancer, but none has succeeded in doing so

convincingly. Few people.eat more broiled foo s than Americans (ref.), and with | Jf ‘

gastric cancer rapidly diminishing in incidence i the U.S., it is unlikely that

, L.. rs 4 a Ly WO a ?
Reagneeg nl BO fee |

GRE A '
this source is important, Whige the method of cooking e impgrtant, its
f

effect is not subjét + quantification. pany Oe .
ar rrr
a4 ,7" ™
/

D. Adventitious Carcinogens

A less obvious source and one that was overlooked altogether until the early
1960’s, is the production of carcinogens by microorganisms in stored food. There

is now good evidence for believing that aflatoxin, a product of the fungus

4-30
een

|

|

i
epidemiologic data. Many of these studies were reviewed by the National Academy

mee

of Sciences (NAS) Safe Drinking Water Committee (NAS, 1977; 1978; 1979, 1980) as
mandated by the Congress. These reports serve as references for many issues

concerning the health effects of drinking water contaminants.

Crump and Guess (1980), in a draft report for CEQ, review five of the recent
case-control epidemiologic studies on cancer risk associated with drinking water
in this country. Inadequactes are identified with each of the studies but most
suggest an elevated cancer risk when the rates for persons living in areas with
chlorinated water are compared to those for persons in areas with unchlorinated
water. The most consistent association found is with rectal cancer. None of the

studies permitted linking individual risks with individual exposures. cope

ex Geeks 11 hn, Ler but logens and 3

‘ng water. Based on

y Ud brary hone ofuss [atoms ancer risks to

+ m/f/liter for each of

4 Kev fotean Ans \ eyes ye? risk estimates for

bige: Urfe Meet Fre fo ba eAAbeted Pt

or 100-fold
Qin Bhs oer) poe s are due to the
interval estimates
e and 1,2

ve since been shown

bromodichloromethane and chlorodibromomethane, found in most drinking water
systems surveyed by EPA, have not been appropriately tested for carcinogenic
properties. They have, however, been shown to be mutagenic in the Ames test
(Simmon and Tardiff, 1978). Presumably, additional substances found in drinking

water will also be shown to be carcinogenic as more chemicals are tested. NCI

4s
\.
[7
[248

Consumer products such as detergents and other surfactants, hair dyes and

Consumer Products

other cosmetics, solid or foam plastics, paints, dyes, polishes, solvents,
fabrics, and even the processed paper and the printer’s ink in the present volume
are a class of agents which are so numerous that it is only possible to echo the
uncertainty with which pollutants were discussed in the previous section. It is

possible that some of these products are already causing, unnoticed, a number of

today’s cancers, and it is quite possible that, after prolonged exposure to then,

 

some substantial risks will be detected in the future. For example, in mouse _-.
. ae Le a a
skin carcinogenesis experiments, surfactants (e.g-, Tween 60) Are potent pret
: : IO”
promoting agents; permanent hair dyes contain cubs hances such as 2,4 ‘
diaminoanisole which can damage DNA, and some components of hair dyes are ae

carcinogenic to laboratory rodents. Many of the ponomers from which plastics are
made are carcinogenic in animals, and the monomer inevitably slightly contaminate
the finished products. Many of the halogenated solvents in common domestic and
office use can cause mouse liver tumours. (For many consumer products, the type
of laboratory and human evidence is insufficient for determining whether they

pose a cancer risk.

At this time, it is difficult if not impossible, to assess the contribution
of consumer products to the overall cancer rate. Doll and Peto attribute "less

than 1%" of all cancer deaths to such products, but they stress that there is too

much ignorance for complacency to be justified. Many industrial products have

been introduced so recently that even if they do prove hazardous their effects

would not yet be apparent.

‘ ?

an,
~

for American women, thought to be attributable to improved nutritional status

(Miller and Bulbrook, 1980). The effect of this on future breast cancer rates is
uncertain. Early studies in rats correlated body size, more than age, with onset
of menarche (Kennedy & Mitra, 1963). Observations in humans, including a recent
look at menarche and amenorrhea*in ballet dancers (Frisch, et al., 1980), provide

additional evidence that lean body mass is related to later menarche.

Later age at menopause brings increased risk. Women with natural menopause
after age 55 have about twice the risk of developing breast cancer as do women
with natural menopause before age 45 (MacMahon, Cole and Brown, 1973). Although
increasing age is an important risk factor for the development of breast cancer,

‘erent populations.

rial Ca. yu c5 \creases throughout

mecreases until middle

vo 73). This
ast cancer may have

prart BAe: M c Cosy nd one acting
SA ee do < : YOG question,
{/rs [8° the breast cancer

dence examined here

Fiebre hans wud Vr Prrnwk ct to modifying

 

 

pt erated comets 5 no consensus about
oY. [0S 7 (4974) aSeS.

 

Breast cancer risk has a strong familial component that has not been
entirely explained by lifestyle similarities among relatives. Anderson (1971)
categorized different types of risks for women with different familial histories
of breast cancer. Relatives of women with unilateral disease have a risk of 2 to
3 times that of the general population; and relative of women with bilateral

breast cancers have a much higher risk.

to¢ f

fet
Go
Sarees |
rattction oe gta
tation Tp PO

Infection particularly viral, has long been thought a cause of cancer, but

 

Statistical evidence does not support the idea that cancer is a contagious
disease. People who come into close contact with cancer patients, such as
nurses, doctors, and spouses of patients, are at no higher risk of developing the
disease than others. Reports have occasionally been published of the occurrence
of an unusually large number of cases of some rare type of cancer in a small
community, but such clusters can be expected to occur pertodically by chance
alone in a population as large as that of the United States. It is more
plauaible that viruses that are transmitted from one person to another are
important in the development of some types of cancer, but they probably they are
widespread in the community. A variety of other factors determine whether

exposure to the virus leads to the development of disease, which probably happens

in only a small proportion of those exposed.

The strongest evidence to implicate a virus in cancer causation concerns two
types of cancer that are rare in the United States -- Burkitt’s lymphoma and
nasopharyngeal carcinoma. In both cases, the causative agent is believed to be
the Epstein-Barr virus, a DNA herpesvirus which occurs ubiquitously and is known
to be the specific cause of infectious mononucleosis. It is postulated that the
viral DNA integrates into the genetic material of a human stem cell and that cell
becomes the parent of a malignant clone. Seroepidemiologic data and the
detection of Epstein-Barr viral DNA in lymphoma cells supports the association
between the virus and these two cancers. Burkitt’s lymphoma occurs mainly in
children in central Africa and New Guinea. The unusual geographical distribution
suggests that the virus may act as a co-carcinogen and that additional factors,

such as immunosuppression from malaria may be involved.

Nasopharyngeal carcinoma, found in the Far East, also is associated with the

Epstein-Barr virus, but the association is not as strong as with Burlett’s

4-15
limited. Studies of patients in mental hospitals (Clemmesen and Hjalgrim-Jensen,
197 ; Buldwin 1979) are not support ive of an increased risk. Psychological
stress does have a recognized importance in causing people to smoke, drink,
overeat, and partake in other harmful activities which may indirectly increase

their risk of cancer.
New Cancer Assocations

Hazards exist today which may not have caused any cancers, but which may do
so in the future. A timely exampie are hazardous wastes that have been
improperly disposed of in areas commonly termed "dumps." EPA has estimated that
there are more than 50,000 dump sites containing hazardous waste that are not
being properly operated. Of these, they estimate that 30,000 pose a significant
health risk. The carcinogenic potential of the myriad of chemicals in these
dumps is unknown at present. (OTA is conducting an assessment of non~nuclear
industrial wastes which will look at health risks, among other things, to be

completed in late 1982.)
x

Development of new chemicals has been booming. They are introduced into
commerce at the rate of about 400 per year at present. The ability of the EPA to
adequately evaluate these additions is limited. Some potential hazards will
undoubtedly be identified through the Premanufacturing Notices required by
Section 5 of TSCA, but new hazards may well be released. Exposures will most
likely be through pollution, occupation, consumer products, foods, or other

As fou
routes already described. (a A

\ vO
Sources of carcinogens yet to be imagined are difficult to discuss, except od

to say that we best be on the lookout.
Table lL.

Method

1. Molecular
structure
analysis

2. Short-term

tests
3. Bioassay
4. Epidemi-
ologic

+ dun pop aegoer eaten ( Comey)
peotion frais ~ oh yeh of own

Organism
sused

s {.

Nott

Bacteria,
yeast,
cultured
cells,
intact
animals

Intact
animals
(rats,
mice)

Humans

Time re-
quired

Short
(days)

-- Wakes

Generally
few weeks
(range 1
day to 8
months)

Ca. 5
years

Months to
lifetimes

Basis for test

Chemicals with
like structures
interact simi-
larly with DNA

Chemical inter-
action with
DNA can be
measured in
biological
system

Chemicals that
cause tumors in
animals may
cause tumors in
humans

Chemicals that
cause cancer can
be detected in
studies of human
populations

Result

Structure resembles
(positive) or does
not resemble (negative)
structure of known

carcinogen

Chemical causes

(positive) or does
not cause (negative)
a response known to

be caused by
carcinogens

Chemical causes

(positive) or does
not cause (negative)
increased incidence

of tumors

Chemical is associ-
ated (positive) or is
not associated (nega-
tive) with an increased
incidence of cancer

General Classification of Tests Available to Determine Properties Related to Carcinogenicity

Conclusion, if result
is positive

Chemical may be
Hazardous, That deter-
mination requires
further testing

Chemical is recognized
as a potential
carcinogen

Chemical is recognized

as a carcinogen in that
species and as a potential
human carcinogen

Chemical is recognized
as a human carcinogen

wf onmenlal eae
years), which dictate the length of a lifetime exposure experiment, and a large
amount of information about the genetics, breeding, housing, and health of these

animals. Rats and mice are cheap to buy, feed, and house.

Primates are sometimes used for certain toxicological testing. They are
certainly more like humans than rodents but their supply is limited. They are
expensive, live up to 25 years, and require large areas for housing. Despite

these difficulties, NCI now maintains about 600 monkeys for carcinogenicity

testing at a cost of about $500,000 a year (R. Adamson, personal communication).

 

ogs lie between rodents and monkeys

—

are more like primates in costs. At try Bry Son 5 .
. yf
oT Qariderres tS ont :

Differences in metabolism, bioaccumulation, and excretion between rodents
and humans are valid reasons for questioning rodent results; however, these
differences should be documented before they are used to negate test results.
There is no question that further research in the comparative biochemistry and
physiology of man and rodents is necessary, but the comparisons will ultimately
be limited by restrictions on what can be determined by experimentation in
humans.

General Objection 4. Some test animals or organs of test animals are
exquisitely sensitive to carcinogens, and such sensitivity invalidates use of
results from such animals.

Criesemer and Cueto (1979) have analyzed the results of testing 190
chemicals in the NCI Bioassay Program (see discussion in "Expert Reviews of
Bioassays," below). They identified 35 chemicals which were “strongly
carcinogenic" in either the rat or the mouse and non-carcinogenic in the other
species. Of the 35, 18 were positive in the mouse and negative in the rat, and
17 were positive in the rat and negative in the mouse, which indicates that
neither animal is much more often the sensitive species. However, 12 chemicals
caused mouse liver tumors, no other lesion in the mouse and no lesions in rats.

Taken by themselves these results suggest that the mouse liver is a sensitive

\
—
 

basis for such estimates, and all of them show that a significant porportion of

tested chemicals have been classified as carcinogenic.

A definitive answer to questions about what chemicals are carcinogens
depends on testing every chemical, and that 1s beyond the capacity of the
bioassay system. Tomatis (1977) reported that 828 chemicals were under test
worldwide in 1975, and that 317 were repeat tests of chemicals for which, in his '
opinion, adequate data already existed. The 828 chemicals did not include all
chemicals under test in private or commercial laboratories; he did not estimate

that number.

Finding more and more chemicals to be carcinogenic in bioassays raises
important policy questions and may force a decision to place carcinogens in order
for possible regulation or voluntary reductions. It is not apparent how to deal

with a large number of carcinogens without ordering them as to their riskiness. ic

ana rh prmoent pant ten f
en abe VASkS

7 - tA fb, Lele oho fitomt
eb an Coed spt af my BOTH

‘

BIAS
prxpordien Bows cots.
ope gn Lvererel

_—_— Oy