&EPA
United States
Environmental Protection
Agency
Office of Health and
Environmental Assessment
Washington, DC 20460
EPA/600/9-90/031b
August 1989
Research and Development
Presentation of Risk
Assessments of
Carcinogens
Appendix B:
List of Case Studies
-------
-------
CONTENTS
APPENDIX B
Introduction
List of Case Studies and Ordering Information
9
Excerpts from the Case Studies Illustrating the Attributes:
B. 1 General Attributes
B.2 Hazard Identification
B.3 Dose-Response Evaluation
B.4 Exposure Assessment
B.5 Risk Characterization
-------
INTRODUCTION
T
his appendix is intended to convey important background for risk
analysts. The report of the Study Group includes a discussion of the
desirable attributes of a risk assessment and refers to examples of
these attributes in the case studies. This appendix consists of excerpts
taken from the case studies to illustrate the attributes identified. The
reader may wish to refer to the complete case study for a more
comprehensive view of the context from which a particular example
was taken. Ordering information for the case studies is contained in
this appendix.
The appendix is organized according to the sequence of the report
discussion: i.e., general attributes and specific attributes for hazard
identification, dose-response evaluation, exposure assessment, and
risk characterization.
In selecting the case studies and the excerpts contained here, the
Study Group considered the nature of presentation but did not form
judgments on the conclusions reached by the authors of the assess-
ments. The reader should also note that the assessments were pre-
pared prior to mid-1987; thus, the information excerpted is not
necessarily current. .
-------
I
RISK ASSESSMENT
CASE STUDIES
ORDERING INFORMATION
To obtain complete copies of the risk assessments analyzed by the'
Study Group, the following organizations should be contacted:
A. DEHP (Di- [2-ethylhexyl] phthalate)
Turnbull, D. and Rodricks, J.V. (1985). Assessment of Possible Car-
cinogenic Risk to Humans Resulting from Exposure to Di (2-ethylhex.yl)
phthalate (DEHP). J. Amer. Coll. Toxicol. 4, 111-145.
Contact: Joseph Rodricks
Environ Corporation
The Flour Mill
1000 Potomac Street, N.W.
Washington, D.C. 20007
(202) 337-7444
B. ("Dioxin") TCDD (2,3,7,8 - Tetrachlorodibeniodioxin I
Environ Corporation (May 1987). Dioxin and Its Human Health
.Significance. Prepared for the National Council of the Paper Industry
for Air and Stream Improvement, Inc.
Contact: William J. Gillespie
National Council of the Paper Industry for Air and
Stream Improvement, Inc.
260 Madison Avenue
New York, NY 10016
(212)532-9349
C. Ethylene Oxide
Sielken, R. L. (1987). A Time-to-Response Perspective on Ethylene
Oxide's Carcinogenicity. In Risk Assessment of Environmental and
Human Health Hazards: A Textbook of Case Studies. Dennis
Paustenbach, Ed., Wiley & Sons, New York (1988).
Contact: Robert L. Sielken
Sielken Incorporated
Suite 210
3833 Texas Avenue
Bryan, Texas 77802
(409) 846-5175
-------
RISK ASSESSMENT
CASE STUDIES
ORDERING INFORMATION
D. Formaldehyde
• EPA (April 1987). Assessment of Health Risk to Garment Workers and
Certain Home Residents from Exposure to Formaldehyde. Office of Pesti-
cides and Toxic Substances, Washington, D.C.
Contact: TSCA Assistance Office (TS-799)
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
(202) 382-3790
E. Formaldehyde
OSHA (December 4, 1987). Occupational Exposure to Formaldehyde;
Final Rule. Federal 'Register Volume 52, No. 233, pp. 46173-46237.
Contact: Lisa Odoms
American Industrial Health Council
Suite 300
1330 Connecticut Avenue, N.W.
Washington, D.C. 20036
(202) 659-0060
I. Lead
Wallsten, T.S., and Whitfield, R.G. (December 1986). Assessing the
Risks to Young Children of Three Effects Associated With Ekvated Blood-
LeadLevels. Publication of ANL/AA-32, Argonne National Labora-
tory. Prepared for EPA Office of Air Quality Planning and Stan- "
dards, Washington, D.C.
Contact: Les Grant
Environmental Criteria and Assessment Office
(MD52)
U.S. Environmental Protection Agency
3200 Highway 54
Research Triangle Park, NC 27711
(919) 541-4173
-------
RISK ASSESSMENT
CASE STUDIES
ORDERING INFORMATION
G/H. Methylene Chloride
EPA (July 1987). Technical Analysis of New Methods and Data Regarding
Dichloromethane Hazard Assessment. EPA/600/8-87/029A.
— and—
Update to the Health-Assessment Document and Addendum for Dichlorom-
ethane (Methylene Chloride): Pharmacokinetics, Mechanism of Action, and
Epidemiokgy. EPA 600/8-87/030A. Office of Health and Environ-
mental Assessment, Washington, D.C.
Contact: Marie Pfaff
U.S. Environmental Protection Agency
401 M Street, S.W., (RD 689)
Washington, D.C. 20460
(202)382-7345
I. Para-dichlorobenzene
EPA (February 13, 1987). Exposure Assessment. In Assessment of
Human Cancer Risks from Para-dichlorobenzene B-l-16. Office of Toxic
Substances, Washington, D.C.
Contact: Risk Analysis Branch
Office of Toxic Substances
U.S. EPA .
401 M Street, S.W.
Washington, D.C. 20460
(202)382-3832
J. Red Dye No. 3 (FD & C Red No. 3; erythrosine)
FDA (July 1987). An Inquiry Into the Mechanisms of Carcinogenic Action
ofPD&C Red No. 3 and Its Significance for Risk Assessment. FD&C Red.
No. 3 Peer Review Panel, Washington, D.C.
Contact: Office of the Director
. National Center for Toxicological Research
1 NCTR Drive
Jefferson, Arkansas 72079-9502
(501)541-4517
-------
RISK ASSESSMENT
CASE STUDIES
ORDERING INFORMATION
K. Tetrachloroethylene
Bogen, K.T., Hall, L.C., McKone, T.E., Layton, D.W., and Patton,
S.E. (April 1987). Health Risk Assessment of Tetrachloroethylene (PCE)
in California Drinking Water. Publication UORL-1583, Lawrence
Livermore National Laboratory. Prepared for the California Public
Health Foundation, Berkeley, California.
Contact: Joseph Brown . • •
California Department of Health Services
2151 Berkeley Way
Berkeley, CA 94704
(415) 540-3191 .
-------
* B.I
GENERAL ATTRIBUTES
Attribute \ The.'scope and objectives of the report are explicitly stated.
SOURCE Case Study J. Red Dye No. 3 (Pages 1-3).
Note After presenting contextual background, the authors clearly outlined
the objectives and scope on Pages 2 and 3 of the excerpt.
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW FANEL
AH INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH
Dr. Scan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA-
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH,
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NCTR/FDA
July, 1987
-------
July, 1987
CHAPTER 1 - INTRODUCTION
A. The Recent Use and Certification of FD&C Red No. 3 (R-3)
FD&C Red No. 3 (R-3) is a color additive that is permanently listed
-------
July, 1987
of the Certified,Color Manufacturers' Association (CCMA), proposed that the
FDA postpone the closing date for the provisional listings of R-3, and take
no action with respect to the permanent listing of R-3 (9), based on the
assertion that R-3 should.be regulated using the general safety require-
ments, rather than the anticancer clauses, of the Act.
The main points of the .petitioner's submission can be summarized as
follows:
1) The weight of scientific information indicated that R-3 was not a
primary carcinogen or tumorigen which acted directly on the genetic
material to initiate a tumorigenic process;
2) R-3 appeared to be a non-direct or secondary carcinogen which acted
indirectly to increase the incidence of tumors;
3) Additional studies were then currently underway which would further
support these conclusions; and
4) If the weight of scientific information supported the first two
points, then R-3 should be considered to act through a secondary
mechanism and not be regulated similarly to an agent acting through
a primary mechanism of carcinogenesis.
The recent White House Office of Science and Technology Policy (OSTP)
consensus document on chemical carcinogens (10) suggested that relevant
biological and biochemical data, including information from studies of
mechanism, be incorporated into evaluations of cafcinogenicity. In accord
with this, FDA has considered that it may need to use a spectrum of
approaches based on mechanism to determine the potential health impact of
the use of regulated substances.
C. FDA Commissioner's Charge to the Peer Review Panel,
Therefore, the FDA Commissioner directed that a Peer Review Panel of
scientists with pertinent training and expertise be constituted to provide
scientifically valid answers to the following three questions:
1) Do the data indicate a secondary mechanism of action with respect
to R-3 and allow determination of any potential risk posed by human
exposure to this color?
2) If not, then 'what further studies or analyses are, necessary to
resolve these issues and provide an adequate basis for risk assess-
ment?
-------
July, 1987
3) In the interim, what concerns relative to human health, if any,
would be posed by continued use of this additive while such studies
and analyses are conducted and evaluated?
To accomplish the Panel's charge, pertinent scientific information con-
tained in the following was reviewed and evaluated including: 1) the sub-
missions by the various product sponsors (FDA Color.Additive Petition [CAP]
96); 2) the recent peer review of the R-3 risk assessments (4); 3) the
relevant published literature; 4) interviews of some 'of the principal
researchers involved in specific scientific investigations; and 5) the re-
sults of additional studies presented to the Panel by Drs. S. Schwartz and
S. Irigbar. It should be understood that most of these studies were not de-
signed to examine direct effects by R-3 on the target organ, but, instead,
were specifically aimed at demonstrating an indirect mechanism of interest.
D. Some Major Considerations in the Panel's Evaluation
*
It is worthwhile to discuss, briefly, five major issues which impact on
providing the answers to the Commissioner's questions.
First, a point discussed in the FDA's 1973 Proposed Regulation on
Selenium in Animal Feed (11), and a precedent set in the subsequent 1974
Finalized Regulation (12) concerning carcinogenesis, are of importance.
The preamble to the Proposed Regulation (11) states that -
i?
"The various anticancer clauses contained in the Act [the specific
clauses of the act are noted in Footnote #1 below] were predicated
on the theory that, since we do not know the mechanisms of carcin-
ogenesis, even one molecule of a carcinogen should not be allowed
into the food supply."
The discussion continues -
"The anticancer clauses do not apply in the case of an agent which
[1] occurs naturally in practically all foods, [2] is used in a
manner such that the natural level in the food is not increased,
'-
[3] has a definite hepatotoxic effect/no effect level, and [4] has
a possible carcinogenic effect which is associated only with the
hepatotoxic effect."
Footnote # 1: Sees. 409[c][3][A], 512[d][1][H], 706[b][5][B], 72 Stat.
1786, 82 Stat 345, 74 Stat. 400; 21 U.S.C. 348[c] [3] [A], 360[d] [1] [H],
376[b][5][B] .
-------
-------
B.I
GENERAL ATTRIBUTES
Attribute 2 The report's content is laid out impartially, with a balanced
treatment of the evidence bearing on the conclusions.
SOURCE Case Study H. Methylene Chloride (Pages 71-87).
Note This case study focused, to a considerable extent, on two alternative
approaches to dose-response evaluation. The attached excerpts are
illustrative of balanced treatment of the evidence. This balance was
generally apparent throughout the report.
-------
United Slates
Environmental Protection
Agency
Office of Health and
Environmental Assessment
Washington DC 20460
EPA/600/8-87/030A
July 1987
External Review Draft
Research and Development
EPA
Ujpdate to the Health Assessment Document
and Addendum for Dic±0.oromethane (Methylene
Chloride): Pharmacokinetics, Mechanism of
Action, and Epidemiology
Review
Draft
(Do Not
Cite or Quote)
(Permission to
reproduce granted
by EPA)
Notice
This document is a preliminary draft. It has"not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
8.3. COMPARISON OF METHODS 1 AND 2
.As with any risk estimation that involves extrapolation,
many assumptions affect the actual risk numbers, and Methods 1
and 2 are certainly no exception. Both use pharmacokinetic
information in an.attempt to reduce uncertainties inherent in
*• .
risk assessments. It is commonly, although erroneously,
conceived "that by incorporating pharmacokinetic information into
a risk assessment, magical reductions of uncertainity are
achieved. Actually, the examination and application of
pharmacokinetic data and models for this compound have revealed
something quite different. As one becomes more familiar with
developing and using this type of information, new and often
more complex questions arise. The utility of pharmacokinetics
is, in fact, this very point. It allows for a systematic
analysis of a chemical's disposition in the body, an important
component of the risk assessment. In applied-dose approaches,
assumptions are frequently made which, although sometimes based
on empirical evidence, are often inflexible and thus in error at
some conditions of the human exposure. For example,.absorption
fraction is. frequently set at some arbitrary value determined
from some empirical evidence or from assuming the "worst case" of
100% absorption. Pharmacokinetic modeling, when properly
performed, is able to account more logically and realistically
for amounts absorped on a time basis. Pharmacokinetic models
seek to account for instantaneous concentrations and changes in
those concentrations that are due not only to changes in exposure
71
-------
conditions, but changes in the physiologic responses as well.
As observed in earlier discussions and in the HRAC (1987)
report, while many uncertainties are reduced, several of the "old
problems" remain, and in fact, new challenges arise. Only with
continued work and trial applications will the science continue
to mature. Only two possible methods have been applied here, but
given the body of evidence and the development of the science, at
this time these two possibilities are considered the most
reasonable. « '
Methods 1 and 2 employ many of the same assumptions, and yet
vary in some very significant ways. Although the actual
calculated numbers are almost identical (within a factor of 4 for
liver and much less for lung), the methodologies are quite
different. When in error even some of the common assumptions
have different implications depending on the method chosen.
However, the differences are mainly in the "last step" of the
risk assessment process, that is, how to actually use delivered
dose to calculate a risk number.
A major and fundamental assumption that EPA has made for
both methods is that the physiologically based pharmacokinetic
model used by Andersen et al. (1986, 1987) is a reasonable method
for describing and predicting the disposition of DCM and its
metabolites in human tissues. This would include acceptance of
9
the model's structure and input parameters. The HRAC (1987)
report raises several important questions that are deemed
important, and future elucidation for purposes of methodology
72
-------
development will be necessary. However/ for the present, EPA has
applied the model with some minor changes. It was felt that the
uncertainity raised by questions regarding model structure were
no greater than those raised by a conventional applied-dose risk
assessment. In fact, because the model is able to quantitatively
describe numerous physiologic and biochemical processes, it is
highly probable that model structure questions pose less
uncertainty than the traditional approach. The HRAC is less
certain about some of the input parameters, such as the metabolic
rate constants. The consequences of errors in these could be
great, and the impact may be somewhat different depending on
whether Method 1 or Method 2 is employed.
There are three major sources of uncertainity with the
metabolic scheme and parameters in the model. First, the model
assumes that any carbon dioxide observed at low doses is being
produced from the MFO pathway. The implication is that- the
carbon dioxide observed by several investigators at low doses is
still compatible with the assumption built into the model that
the GST pathway is virtually inactive at low doses. If incorrect
it would mean that the GST pathway is active at low doses, where
the model is predicting that it is not. Both methods would be in
significant error in predicting risk at low doses
(underprediction).
A second uncertainity common to both Methods 1 and 2 are the
values of the input parameters which apportion metabolism by both
pathways between the liver and lung. As discussed in the HRAC
73
-------
(1987) report, there is great concern-over the values estimated
•for these parameters. The pharmacokinetic model is quite
sensitive to these parameters, and thus any error would be
reflected in model predictions. Such error would be significant
in both methods (over- or underprediction).
The third uncertainty regarding metabolism is with the
values of the metabolic rate constants. As discussed in the HRAC
(1987) report, most questions remain with the value determined
for the first-order rate constant for the GST-mediated pathway.
Determined by allometeric scaling and "curve fitting" of the
model to exposure chamber data, the value of this parameter is
uncertain. In fact, data from CEFIC (1987b) indicate that the
value selected by Andersen et al. (1986, 1987) is in error.
Significant questions also remain regarding the methodology and
results of the CEFIC experiments. It is reported that
experiments are being conducted which may reduce some of the
uncertainty with regard to this rate constant. Method 1 is more
sensitive to this parameter than is Method 2; thus, any error
would result in greater error in the risk number calculated using
Method 1. However, because of this sensitivity, if the value of
this parameter is established more accurately, Method 1 would
better reflect the impact of such findings.
As described previously, numerous uncertainties are
associated with the assumptions that have been made to develop a
risk assessment using pharmacokinetic information for DCM. Apart
from the major generic difference between Methods 1 and 2, there
74
-------
is also the uncertainty associated with the metabolic rate
constant of the GST-mediated metabolic pathway. The risk
calculation resulting from Method 1 would be greatly lowered by
significant reductions in the estimate of this parameter.
Also, one might want to allow for some possible minor role
of metabolism by the MFO-mediated pathway in the carcinogenic
process. No current evidence suggests a contribution by this
«
pathway, and a large role is ruled out by the very low tumor
response in the National Coffee Association drinking water *
bioassay (NCA, 1982a, b; 1983), in which the MFO metabolism was
saturated at a level similar to those associated with the highly
tumorogenic exposures in the NTP inhalation bioassay (NTP, 1985,
1986). However, even a small contibution to DCM's
carcinogenicity by MFO metabolism at high doses might have a
meaningful impact on low-dose risks, since the proportion of the
dose metabolized by this pathway increases at low exposure
levels. One might, for example, hypothesize that 5% of the
carcinogenic risk to mice at the bioassay doses resulted from MFO
metabolism. (Any higher contribution begins to conflict with the
observed lack of correlation of MFO metabolism—and clear
correlation of GST metabolism—with tumor incidences.) Such a
small contribution to tumor production by MFO metabolites would
not have a major impact on the human risk estimates as they have
currently been calculated using Method 1. If, however, the human
GST metabolic rate constant were greatly reduced from the present
estimate, resulting.in much lower predicted human risk from this
75
-------
pathway, that same hypothetical 5% contribution from MFO
metabolism would have a far greater impact on the total human
risk estimate at low doses. Neglecting a contribution of as
t
little as 5% by the MFO-mediated pathway towards carcinogenesis
would, under those circumstances, greatly .underestimate the risk.
If new data indicate that human GST activity towards DCM is much
less than the estimate used here, then a reevaluation of the
«
assumptions would be necessary. More confidence in the
assumption that the MFO path does not contribute to
carcinogenicity and greater certainty in the values of the
appropriate metabolic parameters will be required before the
concomitant reduction in the risk estimates would be accepted as
appropriate.
Another question that arises, regardless of method, is upon
which organs are risk estimates to be based? The pharmacokinetic
approach gives information regarding specific organs. Site
concordance of tumor production between animals and humans is not
normally assumed in performing risk estimates. It is not clear
how to extrapolate for the entire human (all organs) when risks
have been calculated for specific organs by using pharmacokinetic
knowledge. One possible solution is to select the organ with the
highest risk number and apply this to the whole body. This could
result in an overestimation of the risk for many organs but would
ensure that no underestimation would occur due to a lack of
knowledge about an oversensitive or a highly metabolic tissue.
Alternately, if individual organ risks occur independently, they
76
-------
could be mathematically summed. Both Methods 1 and 2 could do
either of these. However, in the case of DCM, because of the
comparative insensitivity of Method 2 to the GST metabolic
*
parameter, even organs with several fold greater metabolic
activity than the lung would not be expected to have a risk far
different from that calculated here. The results of Method 1,
however, are more difficult to apply to other organs. A several
fold change in the GST level (as might be observed in other
organs) would result in a different value for the risk number.
without knowing, specific GST activities towards DCM in other
tissues, it is difficult to ascertain the impact of such an
uncertainty. Although there is no clear evidence of
carcinogenicity in organs other than the lung or liver, there
are some findings that raise concern about this issue. Benign
mammary gland adenomas and salivary gland tumors developed in
rats (NTP, 1985, 1986). The HRAC (1987, Chapter 6) discussed
pancreatic tumors in workers exposed to DCM. Although these
tumors may not be significant, some note should be taken.
It is clear that, once estimates or measurements of internal
dose at the sites of toxic action are obtained, many difficult
issues must be resolved as to how to use such data in the
extrapolation of risk from experimental animals to humans. The
problem is not confined to DCM, nor does it result from any
shortcoming in the information on the pharmacokinetics of this
compound. It is a general problem, reflecting the lack of • >-
understanding of the pharmacodynamics of carcinogenesis.
77
-------
As discussed earlier, there are many.difficulties in using
metabolic differences in species to modify a carcinogenic risk
assessment. Extrapolation between species involves many factors,
including metabolism and pharmacokinetics. The ability to
elucidate a species difference in one contributing component does
not necessarily indicate what, if any, adjustments should be made
to the overall extrapolation. It does not necessarily provide
more certainty than the empirical process currently used; in
fact,: making the necessary assumption may introduce new
questions.
Method 1 advocates the adjustment of the applied-dose risk
extrapolation by the degree to which humans (at lower doses) and
the bioassay rodents metabolize different proportions of their
applied doses at the internal site of carcinogenic action. In
the present case, this method leads to a risk reduction of 8.8-
fold from the level estimated in EPA's previous applied-dose risk
assessment, in Method 1, the observed pharmacokinetic ,
differences between species are to be compared with those
expected to emerge as a result of differences in physical.size
and the rates of physiological processes in rodents and humans.
Method 2, which leads to a risk reduction of 2.1-fold,
advocates the adjustment of the applied-dose risk extrapolation
only by the degree to which the proportion of the applied"dose
that is metabolized differs from high human doses to low human
doses; any species difference in the proportion of the dose that
is metabolized is ignored as a basis for determining human
78
-------
carcinogenic potency. Instead, the interspecies component of
extrapolation"" is carried out as would be done if using applied
dose. The reasoning is that, in addition'to the effect of
species differences in metabolism, there are expected (but
unknown) differences in the carcinogenic responsiveness of the
tissues to a given delivered dose. Pharmacokinetic data
illuminate only the metabolic differences. It may be, for
example, that greater sensitivity to carcinogens in humans
"compensates" for lower metabolic activation of the applied dose.
The justification for using the surface area correction on the
applied dose during the species-to-species extrapolation rests on
tradition. Empirical comparison of carcinogenic potencies in
humans (determined directly from epidemiologic data) with those
from experimental animals shows the surface .area scaling . •
relationship to be a reasonable estimator for many compounds,
although other chemicals show potencies that differ from the
expectation based on this relationship by orders of magnitude.
Because of the problem of specifying interspecies
differences in tissue responsiveness to carcinogens, both Methods
1 and 2 can only give a relative adjustment to the applied-dose
calculation of human risk. That is, incorporation of
pharmacokinetic information can only raise or lower the "dose
delivery" .component of interspecies extrapolation relative to its
appearance in the applied-dose procedure, while the component.
representing "responsiveness" or pharmacodynamic differences
between experimental™ animals and humans remains problematic and
79 ' »
-------
continues to be based on assumptions retained from the former
«* • ' '
applied-dose procedure. Methods 1 and 2 differ*, chiefly in the
way that assumptions from the applied-dose extrapolation
procedure are retained when data on the pharmacokinetic component
are available.
Method 1 is based on the conclusion that, given the most
reasonable scaling of key physiological variables across species,
delivered dose is expected, a priori, to be the same proportion
of applied dose in rodents and in humans. That is, differences
in body size and physiological rates between rodents and humans
do not, in themselves, lead to an expectation of differences in
the delivered doses of metabolically activated carcinogens. If
the proportion of a dose that is metabolized is in fact the same
across species, then the applied dose serves as a good surrogate
measure for the delivered or internal dose of a carcinogen, and
both dose measures will result in the same risk extrapolation.
Thus, the surface area correction, as traditionally used in the
applied-dose procedure at EPA, corresponds to an assumption about
(and correction for) interspecies sensitivity differences rather
than about metabolic differences. It is the factor by which
human risk is assumed to exceed mouse risk for a given dose
(applied or internal). This same assumption about relative
sensitivity is retained when a pharmacokinetic analysis of the
proportion of a dose that is metabolized replaces the prior
assumption of equality of dose delivery across species, implicit
in the use of applied dose in extrapolation.
80
-------
In the present case, for example, the surface area
correction between mouse and human doses is a factor of 12.7.
According to the model used by Andersen and Reitz, the species
difference in metabolism is such that humans (at low exposure
levels) metabolize about one-ninth as much of their applied dose
'via the GST pathway as do mice at 2000 or 4000 ppm. (Most of
this difference is due to high- to low-dose differences that
result from the saturation of the competing MFO pathway at the
high bioassay exposures experienced by mice—the interspecies
difference at the same applied dose is quite small.) According
to Method 1, the lower metabolic activation of DCM in humans
e
implies that the carcinogenic potency difference between humans
and mice is only one-ninth as large as it was previously thought
to be, before the metabolism data were available. The
carcinogenic potency in humans (expressed in units of applied
dose) is only one-ninth as large as the value based on applied
dose.
Method 2, in contrast, suggests that no reasonable
assumption can be made about the effect of allometric scale on
metabolic differences among species. Under this view, any
magnitude of species difference in metabolism seems equally
probable a priori, and so there is no prior assumption against
which to compare empirical data on the actual difference.
Instead, it is presumed that, for a given dose level, the
combined effect of metabolic and sensitivity differences is given
by the surface area correction on applied dose. No explicit
81
-------
assumption about the species difference in sensitivity is made;
in fixing the magnitude of the combined effect, however, a value -
of the sensitivity component is assumed implicitly. For example,
in the present case the pharmacokinetic model estimates that
humans at high doses metabolize 4.5-fold less of their delivered
dose in the lung and 1.5-fold less in the liver than do mice at
equally high doses. By assuming.that the overall interspecies
factor is 12.7, Method 2 implicitly assumes that these metabolic
deficits are compensated for by greater human sensitivity of
57.2-fold in lung (1/4.5 x 57.2 = 12.7) and 19.1-fold in liver
(1/1.5 x 19.1 * 12;7). Low-dose human risks are adjusted by the
i *
degree to which the proportion of the applied dose that is
metabolized via the GST pathway is different than at these high
human doses. That is, delivered dose is used only for the
extrapolation within species, where the question of interspecies
difference in sensitivity does not arise.
Thus, the crux of the difference between the two methods is
whether or not a reasonable prior assumption about the -expected
species differences in metabolism can be made before
pharmacokinetic data are available. If a prior expectation can
be specified, when pharmacokinetic data become available, one may
replace that 'assumption with data (which may show the assumption
to have been inappropriate for that compound). The same
assumption about species differences in sensitivity is applied in
all cases. If, on the other hand, no prior expectation about
pharmacokinetic differences between species can be specified,
82
-------
there is no way to know whether the observed differences are
bigger or smaller than usual. The applied dose is therefore
used to extrapolate across species, and the sensitivity
assumption is adjusted to make its combined effect with the
observed metabolic differences come out to be equal to the
surface area correction, since it is assumed that the combined
effect scales in this way.
The choice between Method 1 and Method 2 has not been an
easy one, and has been made only after considerable debate and
discussion both within EPA and with representatives of other
federal regulatory agencies. The attributes of each method that
have been considered include their relative conservatism in the
face of uncertainty, their sensitivity to errors in the
underlying assumptions and estimates of the pharmacokinetic model
used by Andersen and Reitz, their correspondence to previous
practice, their ability to incorporate current understanding of
metabolism, however imperfect, into the risk extrapolation
process, and/ of course, the plausibility of the assumptions upon
which they are founded.
EPA concludes that Method 1, which extrapolates risk across
species and from high to low doses based on the amount of
metabolism of DCM by the GST pathway, is the most advisable basis
for use of current pharmacokinetic information. The evident
importance of differences in metabolism among rats, mice, and
hamsters to DCM's carcinogenic potency in these species makes the
use of metabolic differences desirable in the estimation of human
83
-------
risk. While acknowledging that many factors in addition to
pharmacokinetics influence species differences in carcinogenic
potency, EPA concludes that it is reasonable to modify risk
extrapolation from experimental animals to humans by the degree
to which the species manifest different degrees of metabolic
activation of their applied doses at the site of carcinogenic
action. The absolute levels of human risk that.are estimated
remain uncertain, as always, owing to the lack of knowledge about
the contributions of the other, non-pharmacokinetic factors to
the relative carcinogenicity of DCM in rodents and human's. The
need to retain assumptions about the role of such factors should
not, in EPA's opinion, dissuade us from examining the potential
contribution of'such factors as can be experimentally examined.
The choice of Method 1, the choice of 'the GST pathway as the sole
route to carcinogenic activation, and the choice of. the model
used by Andersen .and Reitz as a means of its estimation have been
made because, in EPA's judgment, they represent the most likely
and plausible interpretation of the data at hand.. Each choice is
made in the face of some uncertainty, and the interpretation of
the resulting estimate of the carcinogenic potency of DCM in
humans must be tempered with the knowledge that further data may
lead to other choices and different risk estimates becoming more
defensible.
The unit risk for continuous inhalation of 1 ug/m3 of DCM is
thus estimated as 4.7 x 1CT7. For comparison, the applied-dose
extrapolation leads to a value of 4.1 x ICT* (which is 8.8 times
84 ,
-------
higher), and the use of the same metabolic data, but
extrapolating to human risk using Method 2, results in a value of
1.8 x 10-6 (which is 2.1-fold lower than the applied-dose method
and 3.8-fold higher than Method 1).
It should also be noted that both Methods 1 and 2 are
presented as modifications of the method of risk extrapolation to
humans commonly employed by the EPA and CPSC, that is, with the
surface area correction used as an interspecies correction
factor. If one instead uses^the body weight basis for defining
equally risky applied doses in animals and humans (as is done by
the FDA), then the estimated human risk by all methods would be
12.7 times lower. That is, the applied dose procedure would lead
to a unit risk estimate of 3.2 x lO'7, while modifications of .
this unit risk by accounting for metabolism would yield unit
risks of 3.7 x 10~8 for Method 1 and 1.4 x 10~7 for Method 2.
These numbers are 12.7-fold, 8.8 x 12.7 = Ill-fold, and 2.1 x
12.7 = 26.7-fold, respectively, below the published EPA unit risk
of 4.1 x 1CT6 based on applied doses scaled by surface area.
Andersen et al. (1986, 1987) and Reitz et al. (1986) argue that,
because interspecies differences in metabolic and physiologic
parameters have been accounted for by the pharmacokinetic model,
there is no longer a need for any interspecies correlation
factor. Implicit in this view is the assumption that the
interspecies correction factor on applied dose is used solely to
account for species differences in metabolism, and that metabolic
differences completely account for differences in carcinogenic
85
-------
potency of a compound in animals and humans. The EPA takes the
view that this is not the case, since it ignores the contribution
of non-pharmacokinetic factors that influence a species'
responsiveness to a given internal dose.
Lifetime extra risks over background from continuous and
constant low-level exposure to DCM may be estimated by
multiplying the vapor concentration by the internal unit risk
value. However, the EPA's analyses of the model used by Andersen
and Reitz indicate that," if vapor concentrations exceed 100 ppm
or so for any part of an exposure/ substantial nonlinearities
begin to appear that tend to invalidate the assumptions allowing
the unit risk to be used. Under such conditions the MFO pathway
begins to show saturation, resulting in disproportionally more
DCM being available to GST metabolism, which results in
disproportional increases, in internal dose. Exposures involving
high vapor concentrations can have estimated risks that are
several fold above the levels implied by the "equivalent" time-
weighted average exposure. The reader is also reminded that the
unit risk assumes a breathing rate of 20 m3/day. Occupational
exposures, or other exposures occurring during more-strenuous-
than-average activity, will consequently have risks somewhat
underestimated.
Although EPA feels that it is warranted to use species-to-
species pharmacokinetic and metabolic information to adjust
estimates of human risk based on animal data, the absolute levels
of estimated human risk remain uncertain, owing to the unknown
86
-------
contribution of species differences in sensitivity to a given
internal dose of carcinogen. EPA recommends that intensive
efforts be made to develop information' on the pharmacodynamics of
carcinogenesis that could be used in the risk assessment process
in the future. One approach, which may elucidate the magnitude
and variability of. the pharmacodynamic factor for various species
comparisons, is to obtain pharmacokinetic information in both
animals and humans for known human carcinogens. This would allow
an implicit determination of the pharmacodynamics for humans
relative to various rodent species, since the contribution of
pharmacokinetics and the relative potencies of applied doses
could be estimated from available data.
87
-------
-------
B.I
GENERAL ATTRIBUTES
Attribute 3 The risk assessment presentation includes a description of any
review process that was employed, acknowledging specific
review commentary.
SOURCE Case Study E. Formaldehyde (Pages 46237-46247). .
Note These excerpts summarize public review of regulatory alternatives
followed by response by the regulatory agency (OSHA). The review
comments reflect a mixture of scientific review and commentary on
regulatory impacts.
-------
Friday
December 4, 1987
Part I!
Department of Labor
Occupational Safety and Health
Administration
29 CFR Parts 1910 and 1926
Occupational Exposure to Formaldehyde;
Final Rule
-------
Federal Register / Vol. 52. &o. 233 / Friday. December 4. 1987 / Rule* aad Regulations
48237
the apparel industry an average of 30
workdays were lost in 1985 for each
case of skin disease reported to the
Bureau of Labor Statistics.
The incidence rates of reportable
occupational illness related to skin
disease or disorders for 1985 was 10.9/
10,000 full-time workers in textile
finishing (except wool), SIC 2260, and
4.4/10.000.in the apparel industry, SIC
23. These rates are substantially above
low risk industries, such as finance,
insurance, and real estate, where the
rates is 1.0/10,000 full-time workers.
OSHA has no estimate on the
incidence rate of formaldehyde-induced
skin disease in the workplace. However,
about 4 percent of alfpatients tested in
allergy clinics were found to be .
sensitized {Exs. 42-75; 42-93]. If the
percent of employees.who would
become sensitized to formaldehyde
without the use of personal protection is
similar to the incidence seen in allergy
clinics, then provision of protective
clothing to 269,700 employees exposed
to formaldehyde would prevent 10,790
cases of allergic dermatitis annually.
Moreover, numerous cases of non-
allergic skin irritation, which would not
have consequences as serious as allergic
dermatitis, would also be prevented (see
OSHA's Regulatory Impact Analysis).
OSHA believes that the risk of skin
diseases and disorders among workers
who come into contact with
formaldehyde solutions and
formaldehyde-bearing solids is
significant and that this risk will be
substantially reduced by the new
standard for formaldehyde.
In a few cases, information was
sufficient to quantify the risks of skin
diseases and disorders in specific
industry sectors. These cases could be
used to determine the likely benefits of
full compliance with the revised
formaldehyde standard since improved
work practices and housekeeping along
with the use of adequate protective
clothing and equipment should virtually
eliminate skin diseases and disorders.
Available literature on dermatitis
(both irritation and sensitization) was
generally derived from "problem areas"
where NIOSH HHEs have been
requested or where investigators were
examining known hazards. The
information available is generally on
skin disease of a serious nature and
does not distinguish between irritation
and sensitization. There are five studies
of textile dermatitis [Exs. 77-11; 78-84:
85-20; 85-23; 85-24). Incidences ranged
from 3 to 58 percent of those
administered queetionnaries with some
correlatioa between incidence rates, the
amount of releaseafale formaldehyde ia
the fabric, and airborne ct>ncerrtratkrn»
of formaldehyde. (Heat and humidity
also influence release of formaldehyde,
as does the age of the doth and whether
or not it has been washed.) Twenty-six
percent of 874 workers were affected.
Airborne levels ia all bat one case were
below QJj ppm indicating a need to
provide protection from dermal contaST
regardless of the airborne level of
formaldehyde.
There are also reports of adverse
health effects in persons who handled
tissue-preserving solutions {Exs. 42-ltJl;
42-123; 73-86D; 78-53]. Dermatitis, either
irritant or sensitizing, was found in 28 to
37 percent of these workers. (The effects
described were more severe than those
generally seen in the apparel workers,
even though most wore gloves some of
the time.) A total of 47 cases in 140
individuals was reported, for an overall
incidence rate of 33.8 percent, well
above figures reported in the overall
occupational setting and well above the
incidences observed In dermatology
clinics [Ex. 42-75]. Available literature
indicates that the incidence of asthma
and bronchitis, as well as the incidence
of dermal Irritation and sensitrzation,
may be quite high in this group (see
Health Effects section).
Summary of the Significance of the
Risk: OSHA has determined that the
existing standard for formaldehyde
poses significant risk to employees of
cancer, sensory irritation, dermatitis,
and asthma. Full compliance with afl of
the provisions of the revised standard (1
ppm TWA, 2 ppm STEL, and aaciHary
provisions) will substantially reduce
these risks to a level that can be viewed
as safe for the worker. These findings
have been made using the approach
which OSHA has used in setting other
standards for'toxic substances siace the
Benzene decision and are consistent
with the formaldehyde ruleraaking
record.
VIII. Summary of Regulatory Impact and
Regulatory flexibility Analysis
Executive Order 12291 (48 FR 13197,
February 19,1981) requires that a
regulatory analysis be conducted for
any ijile having major economic
consequences on the national economy,
individual industries, geographical
regions, or levels of government. In
addition, the Regulatory Flexibility Act
of 1980, 5 U.S.C. 601 et seq., requires
OSHA to determine whether a
regulation will have a significant impact
-------
46238 . Federal Register / Vol. 52. No. 233 / Friday. December 4,1987 / Rules and Regulations
on a substantial number of small
entities.
Consistent with these requirements.
OSHA has prepared a Regulatory
Impact and Regulatory Flexibility
Analysis (RIA) for the formaldehyde
standard. This analysis includes a
profile of the industries that are covered
by the standard, an estimate of the
number of exposed workers, a review of
the nonregulatory alternatives, and
assessments of the technological
feasibility, costs, benefits, and overall
economic impacts of the final standard.
This RIA is available at the OSHA
Docket Office.
Based upon an analysis of the record,
OSHA has determined that the
industries affected by the revised
standard can be grouped into three
classes, according to the potential
exposure levels. Tier One, which covers
approximately 36,000 affected
establishments and approximately
412,000 exposed workers, consists of the
industries where some firms have
workers who are currently exposed
above either the 1 ppm PEL or 2 ppm
STEL. This group is comprised of
foundries, laboratories, funeral homes,
and industry sectors engaged in the
manufacture of: (1) Hardwood plywood.
(2) particleboard, (3) fiberboard, (4)
furniture, and (5) formaldehyde resins.
Tier Two, which includes approximately
29,000 affected establishments and
approximately 1.1 million exposed
workers, consists of the industries
where some firms have workers who are
currently exposed between the 0.5 ppm
action level and the 1 ppm PEL and
where no firms have employees exposed
above either the 1 ppm PEL or 2 ppm
STEL This group is comprised of textile
finishing and industry sectors engaged
in the manufacture of: (1) Apparel, (2)
formaldehyde, and (3) plastic molding.
Tier Three consists of 24 industries
where some workers are currently
exposed above 0.1 ppm and where no
employees are exposed above the 0.5
ppm action level. This group covers
approximately 47,000 establishments
and approximately 676.000 workers.
Table 5 presents OSHA's estimate of the
number of affected establishments and
employees for each of the affected
industry sectors. Establishments are
grouped by the highest exposure found
within the establishment.
Based on the rulemaking record,
OSHA has determined that compliance
with the revised standard is
technologically feasible. Exposures in
the Tier One industries can be reduced
to below the PEL and STEL through the
increased use of ventilation (either local
or general), and the subsitution of low
emitting urea formaldehyde resins
(LEUF). No industry representatives
contended that such controls could not
achieve these limits. Personal protective
equipment (e.g., gloves, goggles,
respirators, etc.), monitoring badges, and
hygiene equipment (e.g., eye wash and
emergency shower) are also readily
available. Medical resources will not be
a problem because the medical
surveillance questionnaire, of which an
example is provided in nonmandatory
Appendix D to the standard, may be
• administered under the supervision of,
and not necessarily by, a physician.
Also, the medical exams, when required,
consist of standard tests. Finally, the
requirements of several provisions (e.g.,
recordkeeping, the development of an
emergency plan, and employee training)
can be met using in-house personnel.
OSHA estimates that the annual cost
of the revised standard will be'
approximately $64.2 million. Medical
surveillance ($12.9 million), and the
installation, maintenance and operation
of the required engineering controls in
Tier One ($27.6 million) are the two most
expensive provisions, accounting for 20
percent and 43 percent of total annual
costs, respectively. Table 6 provides a
breakdown of industry compliance costs
by exposure level of individual
establishments.
OSHA has determined that
compliance with the revised standard is
economically feasible for firms in each
of,the industry sectors based upon an
industry-specific analysis of revenue
and profit ratios. Table 7 displays total
annualized compliance costs for each
industry as a percentage of annual
industry sales and profits. The annual
compliance costs are less than one-half
of one percent of revenues in all sectors
expect for the fiberboard sector, where
this ratio is 1.65 percent. Impacts of this
magnitude will not adversely impact the
viability of most firms in these
industries. Where data^allowed. OSHA
also examined profit levels of the
industries affected by the standard and
determined that in all but a few of the
wood product sectors, costs would not
exceed 3 percent of profits. For those
industries, improved growth spurred by
the reduced value of the U.S. dollar will
permit the industry to pass through
some of the costs of compliance to the
purchasers of these products.
-------
Federal Register / VaJ. 52. No. 233 / Friday. December 4, 1987 / Rules *aA Regulations
TABLE 5.—NUMBER OF AFFECTED ESTABLISHMENTS .AND EMPLOYEES
SIC and industry
4Wr3M WWi CxpoMrn From 0.1 1o Above 1:0 ppm)
ifV\ — -Hardwood otvwood ...,......-„.,.--•-,,—„.,„..,.-,
7nr(A , ,
2BI9—
-------
46240
" J*
Federal Register / Vol. 52. No. 233 / Friday, December 4. 1987 / Rules and Regulations
TABLE 6.—COMPLIANCE COSTS BY INDUSTRY AND PROVISION
InduUry
With Exposure* Abovt 1
PPM
Hardwood plywood . ..........
Parfcleboard ...„._ ........ ..
Ffcerboard. . .
Furniture ™..«™..™™.™........™..
Rwir* ,„„„.„..„„.... .
Foundries:.«w».,.....»..,,,..,..,,...,.
Labor atories .™....™.^._.._™....
Funeral services...
SUbtOUJ. .„...:......„...
Wttti Exposure* Between
03 and 1 PPM
Hardwood plywood
PH\&QtXWd....,~ ....................
Fiberboard .
Furnfojre Mn.«.HMMH...H.......u,m.
RO$k^H«MM»,
-------
Federal Register / Vol. 52, No. 233 ./ Friday. December 4. 1987 J Rules and RqgiAaitiena 4S241
TABLE 7.—COMPLIANCE COSTS AS A PERCENTAGE OF REVENUES AND PROFITS
.SIC and industry
H»i<**ulS*»aat
006. flQ7 LnborntoafiS ... «M...............,..,_........_.,.,T..-t,,,-T[, T - -
226— 3«KU« filling ~ - .-. -..- -.-..-.
*rj— .^uppral -r - . ...... ..„_,
3O7JB"~PI0fttic moWBnfl
2*38— Softwood ptywxxJ
CZI Pipit milli .. i. ,., ,,.....„.,.,.
2531 — Papftrboard nulls -..
2642~Envotopos. j.-r.i... .. ... .........U..... ,,,...^..,, ... ^ ! ^ _. _a . _. _m .
2653— Corrugated and wlid "fiber boxoo . 4... ......„,, ...... ... . ..
2866— Cyclic crudes, dyos and pigments
2051 Paints piornonts •• ...
2891 Adhosivos and Malantfl .„.„.« .„......._.«... *.....« -. ,..- .'.
3291 Abraaiv® products - -. « '.
TTM MinttiJ wool JnauWian .„,».„ , . . ..„,),,„.,,.. ,
1t634 fl^rtric homowartu ind tent •
38M f
-------
46242 Federal Register / Vol. 52. No. 233 / Friday. December 4. 1987 / Rules and Regulations
approximately S5.7 million. Further, the
use of personal protective equipment by
about 270.000 workers, in conjunction
with other regulatory requirements, such
as medical surveillance and training, is
estimated to prevent about 11,000 cases
of dermatitis, which will provide
approximately $35.5 million in annual
savings. Table 8 presents OSHA's
estimates of the number of employees
exposed to various levels of
formaldehyde, the reduced risk of
cancer provided by the standard, and an
enumeration of respiratory and
dermatitis benefits derived from the
standard.
IX. Environmental Impact
The National Environmental Policy
Act of 1969, 42 U.S.C. 4321 et. seq.,
requires OSHA to determine whether
this standard will have a significant
impact on the environment.
Formaldehyde, by nature, is volatile
and does not remain soluble in water for
any length of time. Consequently,
formaldehyde is not considered to be a
potentially hazardous wastewater
effluent. Similarly, formaldehyde in
waste products such as sludge would
vaporize rapidly, thereby having the
potential to affect air quality more than
waste disposal. In any event, OSHA
does not expect that the proposed action
would significantly increase the amount
of waste containing formaldehyde.
Although more emissions will
TABLE 8
theoretically be removed from the
workplace to the outside environment,
because of the nature of formaldehyde
and its ability to dissipate rapidly, no
incre'ase in the amount of formaldehyde
in the ambient air is anticipated. It is not
expected that compliance with the'
revised standard .will add significantly
to the levels of airborne formaldehyde
that already exist in the environment as
a result of automobile exhaust, cigarette
smoking, and emissions from fabrics
treated with formaldehyde-rbased resins.
pressed wood products and other
sources. Similarly, it is believed that the
levels of ozone (a formaldehyde
decomposition product) in the ambient
atmosphere, will not increase
significantly as a result of this action.
A. Cancers Avoided by Reducing the PEL to 1.0 ppm
[Maximum Likelihood Estimate (MLE) and Upper Confidence Level (UCL)]
Industries
Hardwood Ptywood „ .„
Pofticteboard
Rbwtooi/d ..,.....,„..........._.,...
KOSiO&.*.«*iM*<.H* <«.««•.«•*»••.•*•»...•..«•...•.•...».....•........«.««.
Fooodrioa............... ._.„.._,...„.......
Labor* too«» .....!..._....._...-..„....„......._.,................
Total
Cancws Avoided:
3-sUg» (MLE)......... .„ .....
-.5-slifl9 (UCL)..........™..,...., "i™"!.!"™.
Employees exposed at different levels
Parts Per Million
Total
455
301
230
1.031
385
4.509
_6,289
13,180
1.25
455
255
164
•873
192
2,116
2,070
6.125
2.16
0.08
8.11
1.75
0
46.
66
158
193
1.058
1,437
2,958
3.36
0.18
7.89
2.25
0
0
0
0
0
463
690
1,153
2.91
0.22
4.68
2.75
0
0
0
0
0
232
288
520
2.44
0.25
2.88
3.50
0'
0
0
0
0
,320
230
550
5.37
0.79
4.49
4.00
0
. 0
0
0
0
320
1,554
1.874
27.40
4.98
19.45.
Estimated cancers avoided
3-stage
(ML6)
. 0.16
0.14
0.13
0,49
0.29
12.00
30.41
43.62
43.62
'8.48
47.51
5-stage
(MLE)
0.01
0.01
0.01
. 0.02
0.01
1.60
4.83
6.48
5-stage
(UC*>
0.60
0.46
0.39
1.58"
0.77
14.72
28.98
47.51
B. OTHER ANNUAL BENEFITS
(Relited to PPE, Medical Surveillance.
TraWng, etc.]
Rttpinitocy — „..
DtrmattJs
Total — ;„,.
cases
avoided
5,911
10,780
16,70t
vahMOf
benefits
5.650.844
35,469.359
41,120,203
OSHA concludes that, as a result of
this action, there will be no significant
impact on the general quality of the
human environment outside the
workplace, particularly in terms of
ambient air quality, water quality, or
'solid waste disposal. No comments
made at the public hearing or submitted
to the record contradict this conclusion.
X. Summary and Explanation of the
Standard
The following sections of the
preamble discuss-the individual
provisions of the final standard for
formaldehyde. Each section includes an
analysis of the record evidence and the
reasons underlying OSHA's adoption of
these individual regulatory '
requirements.
Overview.
The final standard applies to all
Occupational exposures to
.formaldehyde, including those resulting
-------
Federal Register / Vol. 52. No. 233 / Friday. December 4. 1987 / Rules and
46243
from formaldehyde gas, solutions, and
materials releasing formaldehyde. The
standard contains two permissible
exposure limits-(PELs) for formaldehyde:
an 8-hour time-weigh ted average (TWA)
exposure limit of 1 ppm and a 15-minute
short-term exposure limit (STEL) of 2
ppm.
Engineering and work practice
controls are the preferred methods of
compliance, respirators are required
where engineering and work practice
controls are not feasible and respirators
are necessary to achieve compliance
with the PELs. In addition, employers
must perform employee exposure
monitoring, unless they can demonstrate
by means of objective data that such
monitoring is not necessary in their
workplaces. Other requirements in the
standard include provisions to establish
regulated areas and medical
surveillance programs. The final rule
specifies how affected employers can
comply with OSHA's generic hazard
communication standard (29 CFR
1910.1200). The final rule also specifies
how affected employers can comply
with other existing OSHA standards (29
QER 1910.132, Protective equipment; 29
CFR 1910.133, Eye and face protection;
and 29 CFR 1910.134, Respiratory
protection) that may be relevant to
workplaces where employees are
exposed to formaldehyde.
Several provisions have been
modified from- the proposed standard in
response to testimony and comments
submitted to the record, the availability
of new data and information, or the
need to clarify language. The reasons for
these changes are described below. The
requirements of the final rule are
considered both necessary and •
appropriate to provide adequate
protection to employees exposed to
formaldehyde.
Proposed Regulatory Alternatives A and"
B
In the. preamble to the proposed rule
[50PR 50468-69], OSHA noted that the
regulatory options before the Agency
were broad, ranging from a change in
the PELs. for formaldehyde in 29 CFR
1910.1000, Table Z-2, to adoption of a
full health standard with the provisions
found in standards for carcinogens. The
preamble indicated that OSHA's choice
of regulatory strategy would depend on
the Agency's decision to regulate
formaldehyde as an irritant, a
carcinogen, or as both. OSHA invited
the public to comment on "the entire
spectrum of regulatory possibilities"
before the Agency.
If OSHA adopted the Alternative A
strategy—regulation of formaldehyde as
an irritant—the preamble stated that "it
might be sufficient to lower the
permissible exposure limits for
formaldehyde contained in Table Z-2
* and to rely on existing sections of
the general industry standards, such as
§§1910.132,1910.133,1910.134 and
1910.1200, to provide the supplementary
coverage needed for employee
protection" [50 FR 50468].
OSHA also stated that if available
evidence was sufficient to conclude that
"formaldehyde is an animal carcinogen
and should be treated for regulatory
purposes as a potential occupational
carcinogen," Alternative B would be
appropriate [50 FR 50469]. Proposed
Alternative fr recognized
formaldehyde's role as a potential
human carcinogen, skin sensitizer and
irritant, and eye and respiratory system
irritant. Proposed Alternative B was a
comprehensive OSHA health standard,
containing PELs and provisions for
employee monitoring, medical
surveillance, protective equipment and
clothing, hygiene facilities, and hazard
communication.
The final rule regulates occupational
exposures to formaldehyde on the, basis
of formaldehyde's potential
careinogenicity, as well as its ability to
aet as a strong irritant and sensitizer of
the skin, eyes, and respiratory system
(see the Health Effects section, above).
Many industry participants favored
lowering the PELs for formaldehyde
through adoption of Alternative A [Exs
77-2; 77-15; 77-33; 77-34; 80-1; 80-4; 80-
5: 80-6; 80-15; 80-18; 80-23; 80-24; 80-26;
80-30; 80-38; 80-40; 80-43; 80-47; 80-55;
80-58; 80-59; 80-61; 80-63; 80-64; 80-67;
80-68; 80-69; 80-71; 80-72; 80-76; 80-77;
80-78; 80-79; 80-80; 80-83; 80-85; 80-86;
80-89; 80-100; 80-132; 80-261; 80-303; 86-
7; 140; 150; 171; 201-9; Tr. 5/13/86, p. 62:
Tr. 5/13/86, p. 137]. The reason most
often presented to explain why these
rulemaking participants preferred
Alternative A to Alternative B was that,
in their .opinion, occupational exposure
to formaldehyde presented no risk of
cancer [Exs. 77-2; 80-1; 80-5; 80-6; 80-
30; 80-58; 80-61; 80-68; 80^-71; 80-72; 80-
76; 80-77; 80-83; 80-85; .80-86; 80-89].
Typical of these comment* was the
statement submitted by John T. Barr,
Manager of Air Products and Chemicals:
We support regulatory alternative A: We
do not betjeve that current occupational
exposures to formaldehyde present a risk of
cancer to humans, but we believe that proper
steps need to be taken to assure that this safe
condition continues (Ex. 80-1, p. 3].
Some participants felt occupational
exposures to formaldehyde at current
leVels present only a risk of "minor"
irritant effects [Exs. 77-33; 80-43; 80-55:
80-59; 80-67; 80-80; 80-303; Tr. 5/13/86.
p. 66].
Some participants believed
Alternative A was attractive because it
would be easy to implement [Exs. 77-34,
80-5; 80-23; 80-26). For example, Leon J
Mamch, Technical Director of the
Louisiana-Pacific Corporation, a
manufacturer of medium-density
fiberboard, endorsed the adoption of
Alternative A because, in his view, it-
would "protect the public and serve as a
workable approach for industry" fEx
80-5, p. 1]. l
The costs associated with Alternative
B's ancillary provisions, i.e.,
requirements such as emergency plans,
employee monitoring, and hazard
communication, were emphasized in the
submittal of Norman Newhouse. an
Illinois lumber dealer. Mr. Newhouse
favored adoption of Alternative A
because thecosts of Alternative B were
likely to be "considerable" [Ex. 80-132].
Other commenters favored a simple
revision of the Z-2 table limits for
formaldehyde because they felt that
other OSHA standards, such as the
generic hazard communication standard
and §§1910.132.1910.133, and 1910.134.
already provided formaldehyde-exposed
workers with adequate protection {Exs.
80-47; 80-71; 80-79; 140].
Michael Farrar, Vice President of the
American Paper Institute and the
National Forest Products Association.
trade associations representing the pulp,.
paper, paperboard. and solid wood
industries, urged the adoption of
Alternative A [Ex. 80-63, p. 6] so that
formaldehyde would not have to be
treated as a carcinogen under OSHA's
generic Hazard Communication
standard (29 CFR 1910.1200). However.
as is discussed elsewhere in more detail.
formaldehyde clearly should be
regarded as a potential human
carcinogen (see Health Effects and
Significance of Risk discussion above).
Employers will have to treat
formaldehyde in accordance with the
Hazard Communication standard's '
requirements for carcinogens since it
meets the criteria set out in that
standard. Regulating formaldehyde only
as an irritant would be grossly
inconsistent with the record and with
the intent of the Hazard Communication
standard.
Four principal beliefs appear to
underlie the preference of several
rulemaking participants for the adoption
of Alternative A: (1) Employers should
be allowed the maximum amount of
discretion in determining conditions in
their workplaces: (2) formaldehyde is
not a carcinogen; (3) formaldehyde's
transient irritant effects do not pose a
-------
46244 Federal Register I Vol. 52. No. 233 / Friday. December 4. 1987 / Rules and
significant risk of material health
impairment to«Kpt»eand Signtficanoe-of Risk portions
of this preamble, fcowever, (he evidence
in OSR/Vs rococd clearly .establishes
that formaldehyde fx»ses a significant
riskof •uUeristl topainaent of health at
the old PELs of 3 ppw 48-honr TWA), 5
Mxn (oeiJing), -and M ppni (pea?k). In
addition, evidence in the record
indicates that merely reducing -airborne
crpttmr. as proposed in Alternative A,
would not address the rick of skin
diseases aari c*har irri trtion .and
scnsitizatioo caused by dermai contact
with -fotatridfihyde.
The evidence in Ae record as a whote
does cot «Mpport Ae adoption of
Alternative A. fa .short, the Agency has
food AafcJSj Some esnfdoyer* have not
exercised Appropriate discretion to
protect iaeir-eHyiloyee*; {2}
formaldekjwb is a potential human
carciw^erc 43) &rn»WeJ^de-HKteced
irritation aad aeasiiiralioH can
comtitute material ianwirmest of feeahh;
andfd) acoBBjireberrtwe OSHA
standard i» .needed to jpnotect workeis.
OSHA kwodacted Alteaatrws A
because the Agenqr beKeses inat a
chfiaga JH 4be PELs alone will not
adequately protect the iseal&»f
forroakieb yde-«xpased waiters,
.a certified
Several rulemaking participants
pointed out lhal formaldehyde meets
established criteria for classification as
a potential carcinogen. For example. Dr.
Michael Silverstein, Assistant Director
of the Health and Safety Department,
UAW. asserted that "formaldehyde
musttre regulated as a presumptive
human carcinogen" because
"indisputable animal carcrrmgemcity
data * * * [are] sufficient to «nder
such a ludfmenT [Tr. 5/M/86, pp. Wl-
200). Ms. Jamie Osihen, Project
Coordinator -af the Ihrited Fwniture
Workers of America, also urged OSHA
to promulgate a comprehensive
standard because formaldehyde is a
carcinogen [Tr. 5/1S/98, p. 2j. Margaret
Seminario, Associate Oirectw of the
AFL-ClO's Department of.Occupa«0Kal
Safety, Health, and Social Security,
reported fca* ihe APL-CJO ooaeiders
formaldeiiyde a potential occupational
carcinogen aoader terms of OSHA"*
cancer policy [29 . « tNIOSH); Tr. 5/14f«6. p. 196
view that formaldehyde «kauUl he
handled as a.carcinc|gefl in occupational
envirosmenU. ftHOSH reoonaoejided
that formaldehyde expawwe "be
controlled to the lowest feasible luait"
on the basis of its rJ
'•urnittire Workers -of Araerrca)].
*" as a
,carcin«gen £Ex. 77-U..p, 3j. JSOOSH
submitted these comments in fulfillment
of its atalutoxy jespojosibilities «nder the
OSH Act, Which xeijuire the Institute .to:
dea/eJsp criteria deoihifwith to«jc
matariak aod han»iul phyaical afento ^nd
substances which will describe exposure
levels that are safe for various periods of
emjftoyment, indlodhigljut not limited to the
exponire levels at Which no -employee -wffl
suSer .imptntsi be riifc «r innctional
capaottiM -
50.jip.a-y.
In -addition, OSHA's review erf tbe
record .Eodicatei that seduction of (be
exposure limits in Table £-<2 Jilaae
weiM mot provide protection «gaimt
dermal sensiiizatwm ao4 •tteer nan-
canQBT.tfects. OSMA bases tfai«
conckiBfcm, in part, «n 'the itafltiknony -of
expemb at flie xulemalnngaieariiig. fw
example, Ei>eard A. &Mnc4t, MJD.,
DiMctor^f the Center *w Occupational
and Enraonmentad Heal* «ft ? ofais
Hopkins University, stated tfeafc
Regulatory Ahernalive A * * * would
ignore or discount Oil joiner -effects «ff
formaldehyde terin«ng-burns to fl>e eye,
senaHiartJan-df *e fldn, tH'ttatSwofthe skm,
and indeed tite carcS»ogen5cfty fif
formaldehyde in experimental •ni
has led to a presampttonift* ff« a
h«amn«cra«p«tiomd -carcinogen.
does JMt oratnm «iy,ynorai»i«H»to decieaM
skin ej^potwae, ani J»ri»cwe«rcartHjil»f«ir
levels of -fesnaMehyde wiH jiave 4»Hlei«r m
effect on protecting ibe AiB.an-direot-ooBtocI
with sources of fQJTOaldeijrde [TX.5J&/M. ff.
68-69].
t&flf\ fJWTD *»OTC»t«»J »,»»**»*»•. — •*» —
comments wgBrtling the prapos«d
regulatory alternatives, andfcas, 4n
* addition, weighed ihe «vi4asee aa Jbe
health effects associated with
workplace exposures to formaldehyde.
The Agency concludes, .that only-a
comprehensive standard can decrease
the risks associated with aH of fhese
adverse iieatth effects; a-regulatory
alternative that simply lowers fhe PEL
cannot protect workers against the
broad range of formaldehyde-induced
health effects.
The ancillary provisions contained in
a Mllieatth standard ace eapeciaHy
appropriate in the regulation of An
irritant and sensUizer such as
formaldehyde. For example, training and
personal protective equipment pro vide
important worker protection when
irritants, which by definUiox ane
corrosive in action, inflamiqg the meiet
mucous BurJaces .of the body JEx. 7J-a7a,
-------
Federal Register / Vol. 52. No. 23* / Friday. December 4. 1987 / Rules and Regulations
46245
P- 20], are being handled in the
workplace. The inclusion of such
requirements is.consiftent with section
6f b)(7} of the OSH Act, which specifies
that: .
Any standard promulgated under this
subsection shall prescribe tht uae of labels or
other appropriate forms of warning as are
necessary to insure that employees are
apprised of all hazards to which they are
exposed, relevant symptoms and appropriate
emergency treatment, and proper conditions
and precautions of safe use or exposure.
Where appropriate, such standard shall also
prescribe suitable protective equipment and
control or technological procedures to be
used in connection with such hazards and
shall provide for monitoring or measuring
employee exposure al such locations and
intervals, and in such manner as may be
necessaiy for the protection of employees. In
addition, where appropriate, any such
standard shall prescribe the type and
frequency of medical examinations or other
tests which shall be made available, by the
employer or at his cost, to employees
exposed to such hazards in order to most
effectively determine whether the health of
such employees is adversely affected by such
exposure.
Based on the record as a whole, the
Agency concludes that regulation of
formaldehyde as a potential human
carcinogen, an irritant, and a sensitiaer
is both necessary and appropriate to
protect workers' health and functional
capacity. The best available evidence,
discussed above in the health effects
section, clearly dictates promulgation of
a final rule that will addresa
formaldehyde's potential adverse
effects, including skin, eye, and
respiratory irritation: dermal and
pulmonary sensitization; and cancer.
Paragraph (a)—Scope and Application
Like the proposal, the final standard
applies to all occupational exposures to
formaldehyde, including those in general
industry, maritime, and construction.
The language in paragraph (a)(l) has
been altered from that of the proposal to
clarify that the standard applies to a
single, specific chemical entity,
formaldehyde, with a Chemical
Abstracts Service Registry number of
5O-00-O.
Employee exposure to formaldehyde
can occur in a number of ways. For
example, formaldehyde production
workers-are potentially exposed'to
formaldehyde gas; workers who handle
formaldehyde solutions.
paraformaldehyde. or materials
containing resins or glues that have
releasable formaldehyde may also be
exposed to gaseous formaldehyde.
These solid and liquid products may
also present a risk from dermal contact
with formaldehyde. The final standard
is designed to protect against all
hazards from formaldehyde exposure,
regardless of. the source. (Of couse, the
standard does not necessarily protect
against all of the hazardous ingredients
in a mixture). -
The Agency's determination that all
occupational exposures should be
covered by the final rule is" consistent
with evidence in the record that the risk
of formaldehyde exposure is related to
the degree of exposure rather than to the
operation, workplace, or segment of
industry in which such exposure occurs.
OSHA's position on the appropriate
scope of the standard is unchanged from
the proposal, and the Agency received
only a few comments on this subject.
The proposed rule explicitly stated
that OSHA intended the standard to
apply to the construction industry. The
proposal also noted that OSHA's
Advisory Committee on Construction
Safety and Health had requested that
OSHA develop a separate standard for
formaldehyde for the construction
industry or that, as an alternative
measure, the Agency adopt "the most
protective standard available" [50 FR
50414],
Despite OSHA's request for
information on how the standard should
be modified to consider the unique
characteristics of the construction
industry [50 FR 50413], OSHA received
very little information on this industry
and has reached and has reached the
conclusion, based on the Agency's
evaluation of the types of construction
jobs known or suspected of having some
potential for formaldehyde exposure,
that the impact on this industry is small.
For.example. Scott Schneider, an
industrial hygienist with the United
Brotherhood of Carpenters and Joiners
of America testified that there are
several construction jobs where there is
potential formaldehyde exposure,
including laying floors in a confined
space when glues containing
formaldehyde are used, blowing of urea-
formaldehyde foam insulation into
walls, and cutting and sanding wood
products containing formaldehyde [Tr.
May 14.1988, pp. 137-138]. It is apparent
that in the past construction workers
could have received substantial
exposures to formaldehyde. However,
negative publicity surrounding the use of
UFFI has virtually eliminated this use of
formaldehyde. Recent evidence on the
generation of "particleboard aerosol" by
a sanding process conducted under
laboratory conditions indicated all
airborne formaldehyde concentrations
below 1 ppm [Ex. 201-5A], showing that
under the more typical intermittent
exposure situation during actual
carpentry, exposures should be well
below the action level of 0.5 ppm. With
the exception of industrial construction.
whch can also be hazardous because of
the presence of formaldehyde from the
operation, there appears to be little
:' impact that the formaldehyde standard
will have on the construction industry.
OSHA's analysis of the data available
suggests that most construction
activities result in worker exposure well
below 0.5 ppm. To the extent that there
•are any unique operations, such as
construction-related maintenance, or an
increase in the use of formaldehyde-
releasing resins in a confined area, the
general industry standard is being
applied to construction {see 1910.19).
A representative of the maritime
industry, Hal Draper, observed that:
it would be impractical if not impossible for
me marine cargo handling industry to meet
the requirements of the proposed
formaldehyde standard due to the uae of
.casual labor, and mobile work sites with
remote possibility of employee exposure to
formaldehyde [Ex. 80-53, p. 2J.
OSHA has chosen not to write a
separate standard for the construction
or maritime industries, both because
commenters did not suggest any specific
modifications to the standard to adapt it
to the mobile worksite environment
typical of these sectors and because the
Agency believes that the final standard
is flexible enough to present few
compliance burdens for employers in
these sectors.
The standard has been tailored so
that certain provisions become
inapplicable or have only limited
applicability when employee exposure
is low. For example, routine medical
surveillance is triggered by employee
exposures above the action leVel. Thus,
the standard is more stringent where
employee exposures to formaldehyde
present higher risk and becomes more
flexible in situations where exposure
and risk decrease. The final rule has
been structured so that compliance
burden imposed by the standard is
directly to the potential hazard posed by
occupational exposure to. formaldehyde
in each particular employment setting.
The Agency believes that, because of
this approach, no significant compliance
burden will be imposed on construction
or martime employers whose employees,
in general, are exposed to formaldehyde
only at concentrations believed by
OSHA to be well below the PELs.
In the proposal. OSHA expressed the
intent to fully cover laboratory uses of
formaldehyde under the standard [50 FR
50470]. Further, it was proposed that the
limited coverage of laboratories under
Hazard Communication be enlarged in
the formaldehyde standard so that
-------
482*6 Federal Rogigtec / Vci. 52. No. 233 / Friday. December 4. 1987
and
laboratories using focmaldebyde would
be subjected to all -of the Hazard
Communication proviitons rather -than
Ihc limited provisions foand ia 28 CFR
more fully in ihe Hazard •Corarausicatioa
section of Ihe txamnsy and explanation,
OSHA has .decided not to enlarge the
hazard commuckatian coverage of
laboratories ID this final rule.
The Standard Oil Company, which
has several production and 'laboratory
facilities where formaldehyde is used,
requested «pecia4 •consideration of
laboratories. According -to Standard Q&
The StawlardOtt Company does not
believe ikai laboratory uecekplBces should be
subjected to Ihe same req»pemen)a Mother
workplaces within the scope of the proposed
standard.
In t»ntr«9t t« typical manufacturing
locations, wo* practices, quantities of
formaldehyde handled and exposure controls
are vasUy-fiffferent for a laboratory. Small *
quanUtkrt'of formaldehyde ere-u«d irnnost
quality oonlrd an3 research laboratories; 1he
formaldehyde is handled ty highly trained
technicians *od chemists: .the .exposure to
formaldehyde, is usually below sensory
irritation levels: the duration ol (he work tasV
Is short. wrasUy fastmg only a matter of
minutest and tome Ufcs require handling
fornwldcfcs, de in « 3«&»tsttty -exhaust hood.
Based cat MtsamsJl jonsuntf«f fenoaMehyde
usad.She concentration of-fQaxeMebydeiin
Iheiabiiocd odtwiat is mUirMl^Ex-ao-SB.
pp. 2-3 of attached caaunentsj.
In its ieaaripticrL Staatdszd Oil
presents the case of a typical iatotatory
where 4he rets of any oae cheaacal, im
this case formaldehyde, is-very
incidental and a minor part iof ihe
overall exposure potential Xheoe are
undoukJ*dly many tach chcutnatances
where fscmaldehyde mofariiocre. in SMOOT
quaoiifieB.jrre used as one of masjy
reagents .or wfar^re very sas^l amaxrds of
formsiWefcyde JBIC peesent at
prcsernttives. l&ese are psecisah/ the
circuctatances that tke Texic Substances
in Laboratories proposal (SI f£ 26860,
July 24, t0a£} «±leinpiad te address;
namely the aoacmtiEte use of small
amounts cif amccj»as toxic .substances.
OSHA is Trmdtni «f tirepotBntial for
overlap 'between rrmcediEss leqnired in
laboratories by the focmaldehyde
standard scd the yrocedtngs imrtar the
laboratory «tandard. la feaHring ihe
laboratory -staadaud, OSHA will make
every effort to assure lhat Acre ace cot
conflicts or dtmlicartive resnurementB.
However, OSHA has Identified one
laboratory use of foraaaidebyde where
the seventy of the exposures to
employees have tended to be even
greater ifcan -the typicaJ ex^nMures tbtct
occur withia §eaer*l industry. This is
the tue of aoltkisas otataming
formaldehyde tie preserve tisane and ihe
subsequent handbag of «och (tissues.
Exposed employees are laboratory
workers and teachers in histology,
pathnksgy, and anatomy Laboratories.
Evidence submitted to OSItA's record
[Exs. 78-20; 78-54; 85-29; 91; 128J dearly
confirm ihat work in Bach laboratories
may result iniamthoe exposure to
formaideByde, boA by inhalation and
by dermal contact. Furthesmane,
numecous examples of exposure to
extremely high airborne concentrations
of formaldehyde were found {Exs. 42-85;
42-9ft 1E5-17: ITS-IB], as was •evidence
of formaldefayde-inducari irritation and
skin diaocdexs «nd senBhizaition
«; 85-28].
rea
.
These jtrafalems were Strand '.despite the
very high levdl jrf training and education
of some of the indmdu«l< who were
being expoeefl.
While many tidk^ laboratories have
installed adequate engmeeruig ccmtrok
and leqDtDe -their employees to observe
good work practices, trthers have not
done so. Dr. Melvin First, of the Harvard
School of ftibiic Healtii *nd an expert
witness in iatdustniail iiygMne and
engineerkig contasls, testified "tiist 17
percent of a group of 637 personal
samples from hospital iabaralories were
above&e .8-iour TWA oft ppm. in Or.
First's opinion, these results showed
that a "a^niScant nainber -of hospital
laboratoides are using poe shelter accommodated
under Ibis final Tale wn Occapationai
Expcmre io Formalrieiryde .or the more
general Toxic .Substances in
LabGrstories mie wown it is
promtdgaded. listsetavett, any category «f
* laboratory thai as e»enu«ally exempted
from the Toxic Sdsstaaces io
LaberstarifiB «tBsndand wail
automatically be covered by tfei«
standard to the caokeat -there axe
occupational expcaares tto
formaldehyde. T4w ««aie wifl te
considered f urflier as part -of fee
promulgation sf the Toxic Substances fa
Laboratories regulation. OSHA believes
that most laboratories'{except histology,
pathology, and anatomy) are already in
compliance with the provisions al 'the
formaldehyde standard. However, ito
avoid imposing start-up cos** {or other
laboratories wider this -standard which _
may be unnecessatry if-fhe formaWBhyde
standard is sirsperseded by fhe .general
standard for laboratories. OSHA is
extending the compliance dote lor .other
laboratories to September 1,1088, «t
w hich .time -the L*bar«tory standard is
expected to -be dn effect •Qhrea the
seventy of the health effects projected
to occur at the existing 3 ppm TWA, 3
ppm ceiling, and 10 ppm peak, "however,
OSHA is requiring aH laboratories to lie
in compliance with .the newPELi at 1
ppm as a TWA and .2 ppmyfts a.STEL
Clearly, formaldehyde is A very toxic
substance which must be handled
extremely carefully.. White many other
labs ace presently m OoiBptianoe «rith
this standard, -some are «ot, fmdOSHA
is hesitant to create an open-ended
period where •some jrflbe "rtlier"
laboratory employees -wail have .110
protection from some of ihe advecse
effects of formaldehyde, •MpaeiaHir
those .proteotioBa in this ade toigganed
by concern *bout tba d*amal •effects
associated with fonnaktefejaie exposure.
Therefore, should the Toxic Substtscaa
in 'Laboratories final mde »«rt be is .effeot
by September 1, IBSa this ntle (2BOFS
1910.104&J will become (eifactisre far such
other laboratories-so (that tisear
employees will .be appnsfHimiely
protected.
The Society of the Plastics Industry.
Inc. (SPI) and the E.I. duPont de
Nemours 'Company requested an
exemption for the thermoplastic acetal
molding industry from the *cope «(f beteves It
would be mapppapriate forOSHA 4e
grant fhe Tequested e»eeinpfien. OSHA
notes, hovrevw, Ifeert
-------
Federal Register / Vol. 52. No. 233 / Friday.. December 4, 1987 / Rules and Regulations
46247
below the action level, the compliance
burden will be minimal.
. The Scott Paper Company urged
OSHA.to. exempt paper product* from
the scope of the standard's hazard
communication requirements because
/'the use of paper products under normal
conditions will not result in exposure of
workers to formaldehyde at levels even
remotely approaching" the action level
or PELS proposed by OSHA [Ex. 80-62,
pp. 1-2]. Where paper products emit
only trivial amounts of formaldehyde,
the final rule's hazard communication
provisions will not be triggered.
Furthermore, the final standard exempts
employers whose workplaces contain
only mixtures or solutions composed of
less than 0.1 percent formaldehyde or
materials incapable of releasing
formaldehyde at concentrations at or
above 0.1 ppm from compliance with the
hazard communication provisions. Use
of this regulatory approach means, in
effect, that the great majority of
downstream uses of paper products will
effectively be exempted from the
standard. Evidence in the record,
however, indicated that certain workers
involved in the production of paper
products may be exposed to significant
levels of formaldehyde [Ex. 149,
Appendix A]. In addition, the United
Paperworkers International Union
(UPIU) testified that many industrial
hygiene practices and procedures used
in this sector-are inadequate [Ex. 149,'
Appendix B]. Accordingly, it would be
inappropriate for OSHA to exempt this
industry from coverage.
The American Furniture
Manufacturers Association (AFMA)
believed that the occupational
exposures resulting from downstream
uses of formaldehyde-bearing products
are sufficiently different from those of
formaldehyde producers to warrant
separate ancillary requirements [i.e.,
other than the PELs). The AFMA felt
that the proposed standard's protective
clothing, emergency, and waste disposal
provisions were inappropriate for
furniture manufacturers [Ex. 80-68, pp.
5-6]. In general, OSHA agrees with
AFMA that these provisions are much
less important in a plant that assembles
furniture that they are where large
quantities of formalin solution are being
handled. This does not mean that it is
necessary to design industry-specific
standards. The general standard's
provisions require only that the
employer provide the protection needed
given the individual circumstances.
A representative of the maritime
industry. Hal Draper of the West Gulf
Maritime Association, asked .that the
final r-'e sontain an exemption for "the
storage, transportation, distribution, or
sales of formaldehyde in intact
containters" from all of the provisions of
the final rule except the hazard
communication and emergency
requirements [Ex. 80-53, p. 2]. Mr.
Draper noted that such a "partial
exemption" would be in keeping with
similar language in OSHA's proposed
benzene standard [50 FR 50512] and in
other OSHA standards [Ex. 80-53, p. 2].
With the exception of the need to make
an objective determination that the
containers are, in fact, intact, the
formaldehyde standard already
provides the relief requested by Mr.
Draper. Employers whose only use of
formaldehyde involves intact containers
would have minimal obligations under
the standard.
On the basis of OSHA's analysis of
these comments, the Agency believes
that, at the present time, a single
formaldehyde standard applicable to all
occupational exposures to formaldehyde
is most appropriate and protective of
employee safety and health! OSHA
believes that all exposed employees,
including those exposed only
infrequently, should be provided with
some basic protection because some of
the adverse effects of exposure to
formaldehyde are acute and arise after
only a few minutes of exposure.
The major change to the scope and
application section of the standard since
the proposal is the deletion of the
proposed exemption for (i) Liquid
formaldehyde solutions containing less
than 0.1 percent formaldehyde, and (ii)
solid materials made from or containing
formaldehyde that are incapable of
releasing formaldehyde into the
workplace air.
OSHA received many comments on
this proposed exemption [Exs. 80-21; 80-
34; 80-35; 80-37; 80-54; 80-56; 80-59; 80-
63; 80-64; 80-69; 80-72; 80^-303; Tr. 5/8/
86, p. 18] from participants requesting
that OSHA extend the proposed 0.1
percent exclusion for liquid-
formaldehyde solutions to solids
containing a similar percentage of
formaldehyde [Exs. 80-54; 80-56; 80-59;
80-64; 80-69; 80-72; 80-303]. The major
concern expressed was that, without an
exclusion, employers would be required
to label articles such as textiles, apparel,
envelopes, and other common items
routinely used by consumers [Exs. 80-56;
80-59; 80-63; 80-72; 8O-303]. Evidence in
the record indicates, however, that a 0.1
percent exemption for solids which
continue to release formaldehyde over
long periods of time is inappropriate and
potentially dangerous. For example, all
textiles would be exempt under such a
definition according to representatives
of the National Cotton Council [Tr. 5/
14/86, p. 76], but the record contains
numerous reports of formaldehyde-
induced illnesses among garment
workers (see, for example, the NIOSH
HHEs (Exs. 78 and 85) cited in Health
Effects). OSHA's generic Hazard
Communication standard clearly cover-
textiles used in apparel manufacture,
and the record in this rulemaking clepr'
supports such coverage.1
OSHA's response to these
commenters has involved two changes
to the final rule. First, the Agency has
refined the definition of formaldehvde
so that it is clear that the standard
applies only to formaldehyde: it does
not coyer all of the other substances that
may be present in a mixture. Even
though OSHA's generic Hazard
Communication standard exempts
components of mixtures present in
concentrations of less than 0.1 percent
by weight, an employer is still obligated
to recognize the minor component as
hazardous if employees are exposed at
airborne concentrations over either PEL.
The best method available to assure that
such hazardous exposures are not
occurring is the use of employee
exposure monitoring, or at the least,
objective data, which the formaldehyde
standard assures will be collected for
formaldehyde. Accordingly, the
exemption has been moved to paragraph
(m](l)(i), which defines, for the purposes
of hazard communication,'
"formaldehyde gas, all mixtures or ,
solutions composed of greater than 0.1
percent formaldehyde, and materials •
capable of releasing formaldehyde into
the air under any normal condition of
use at concentrations reaching or
exceeding 0.1 ppm" as a health hazard.
Although use of this definition of a
formaldehyde health hazard will not
ensure that every employer is alerted to
the presence of each and every product
that contains even a trace of
formaldehyde, OSHA believes that a
manufacturer, importer, or distributor of
a formaldehyde-bearing product would
be aware if a product is capable of
emitting sufficient formaldehyde to pose
a health hazard.
Paragraph (b)—Definitions .
In the final standard, the definitions of
"Assistant Secretary". "Authorized
persons", and "Director" remain
unchanged from the proposal.
The definition of an "action level" as
half the PEL, calculated as an eight-hour
time-weighted average (TWA), is also
essentially unchanged from the
proposal. An action level is an exposure
limit above which the monitoring and
annual training provisions of the
-------
-------
GENERAL ATTRIBUTES
Attribute 4
The key findings of the report are highlighted in a concise
executive summary.
SOURCE Case Study J. Red Dye No. 3 (Pages vi, vii).
Note The excerpt illustrates a very concise executive summary for an
assessment of complex information relating to a somewhat limited
objective and scope. <°
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared, by: .
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH .
Dr.. Stan-G. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries - ,
Dr. -Paul Lepore, ORA/FDA
' Dri Angelo Turturro, NCTR/FDA
July, 1987
-------
July, 1987
EXECUTIVE SUMMARY '
A Peer Review Panel of scientists drawn from a number of different
agencies was directed to make an inquiry into whether the data indicated
that FD&C Red No. 3 (R-3) had a secondary mechanism of carcinogenesis,
whether the potential human risk of the color could be determined, and
whether' additional studies to address important questions in determining
risk should be performed.
In spite of the difficulties, of the task, the Panel made a "best"
effort to address the charge and make a reasonable estimate of human risk
based on this effort. Because of the complexity of the task, consideration
of the conclusions of the report without an appreciation of their
scientific context invites misunderstanding.
The Panel developed a working definition for a secondary mechanism, of
carcinogenesis which was specifically directed to R-3. Examination of the
pharmacokinetic studies using the color in rats and humans led to a series
of conclusions, including: absorption is similar in ra.t and man, and is low
(less than 2%). Analysis of short-term tests indicate no evidence for any
DNA interaction relevant to mammalian systems, and give no suggestion of
potential direct mechanisms of genetic toxicity. However, there is a
light-activated toxic activity. Evaluation of long-term toxicity. studies
indicated that R-3 induces thyroid follicular tumors in rats in a two-
generation study, with adequate negative studies in mice.
Because the toxic effect was on the thyroid, the effect of R-3 on
thyroid economy was evaluated. R-3 had a number of effects in vivo,
including an elevation of Thyroid Stimulating Hormone (TSH) and thyroxine
levels. A number of possible mechanisms of this action were considered, as
well as other mechanisms for inducing tumors. As part of understanding .the
biological relevance of the results in the animal tests to man, the rela-
tionship of thyroid tumors in rat to those in man was explored, as was the
special role that iodide, both as contaminant and as a product of R-3
metabolism, plays in the toxic effect of R-3. .
Using a weight-of-evidence approach, the Panel found sufficent evidence
for animal oncogenicity and limited evidence of animal carcinogenicity.
vi
-------
.July, 1987
Characterizing the practical situation , in humans, an average exposure
of 1.41 mg/d was estimated in adults, with analysis of the limits on human
5:
exposure. The use of these exposure estimates in defining .risk was
explored in a very comprehensive fashion, exploring basic questions in
assessing risk.
In summary, the Panel could not come to any conclusion concerning the
exact mechanism by which R-3 induced thyroid tumors in rats. However,
there was little evidence that R-3 operates through any mechanism which was
? - •
inconsistent with the working definition of a secondary mechanism of carci-
nogenesis, qualified by the acknowledged limitations that the definition
has in describing carcinogenic mechanisms. A variety of ways of deriving
risk estimates using this information are discussed and analysed.
vii
-------
B.I
GENERAL ATTRIBUTES
Attribute 5 77^ report explains clearly how and why its findings differ
from other risk assessment reports on the same topic.
SOURCE Attribute J. Red Dye No. 3 (Pages 56-63).
Note The excerpts illustrate how the authors' conclusions differ from
those of other studies. Section C lists important differences in expo-
sure estimations based on more recent data. Section D discusses
differences in tumor incidences reported by various pathologists.
Section F discusses conclusions on a no-observable effect level that can
be drawn from various studies.
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW FANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH
Dr. Scan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA -
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NCTR/FDA
July, 1987
-------
July, 1987
cancer. Further, the above definition may not be applicable to all possi-
ble mechanisms that the Panel may consider to be secondary. However, the
above is a working definition that facilitates further discussion in this
chapter. ,
B. Pharmacokinetics ,
The animal and human studies on the bioavailability of ingested R-3,
described in Chapter 2, confirm older observations of poor absorption (131-
133). As shown in these absorption studies, the intestinal absorption of
orally given R-3 is less than 2% for" rats and humans. It is worthy to
mention that the definitive rat data were only recently accessible. Thus,
these studies were not available to the Color Additive Review Panel in its
assessment of the risk from 6 dyes for external use (4). If the risk from
external uses of R-3 is re-evaluated based on these studies, it is likely
to be increased and, thus, may need to be recalculated in light of the new
information.
C. Hormonal Effects of R-3 ' . • . •
As discussed in the previous- chapter, T~ and T, are thyroid hormones
subject to a feedback regulatory system that includes the pituitary, the
hypothalamus, and the hormones TSH and'TRH. A decrease in Ta or T^ levels
evokes an increase in TSH. A TSH-increasing effect has also been seen
after administration of iodide and a number of antithyroid drugs, which are
known to suppress the formation and release of thyroid hormones. The
studies discussed in Chapter 5 have 'shown that R-3 induces an elevation of
both T^ and TSH in rats. Evidence for an action of R-3 through TSH in
rats would gain in strength if a decrease in T, is observed. Such an
effect was seen in the 2 semichronic studies (109,113). The observation of
an increased, total T- and a decreased percentage of free T~ in one study
(123) is so complicated by the factors discussed in Chapter 5, Section G.3.
that it can be considered to demonstrate only .that TRH provocation
increases following R-3 exposure.
For humans, more detail is provided below.
1. Human Studies
a) Ingbar _et_ _al_. (27) administered a single dose of 75-80 mg R-3 to
5 persons, who were pretreated with a large dose of iodide to block the.
56
-------
July, 1987
uptake of iodide by the thyroid. Within a few days, TSH levels showed a
statistically significant increase. It is not possible, however, to attri-
bute with certainty this effect to R-3, since the concomitant dosing with
iodide may have caused the TSH increase as well. No changes were seen in
the T, and T, levels. This study is discussed more extensively in Chapter
2 in terms of R-3 absorption and disposition.
b) Witorsch ^t _al. (134) (Study Wl) exposed males orally to 20, 60,
and 200 mg R-3/d for 2 weeks (10 males per dose group). The differences
between the means of Day 15 and Day 1 TSH levels were -0.03, 0.29, and 0.51
UU/ml at doses of 20, 60, and 200 mg/d of R-3, respectively. The average
increase in the TSH level for the 200 mg/day group was statistically signi-
ficant. No statistically significant changes were seen in the levels of T~
and T^ in this dose group, or of TS, T^ and TSH in the lower dose groups.
In the 200 mg/d dosesgroup, the daily urinary excretion of iodide was about
one mg. Provocation with intravenously administered TRH resulted in a
large and statistically significant increased TSH response in the 200 mg/d
dose group, but not at lower dose levels.
c) In a second study, Witorsch et_ al. (135) (Study W2) exposed 9
males and 9 females to an oral dose of 0.75 mg iodide-, twice a day, for 2
weeks. In addition, on day 1 (dosing of iodide started day 2), and on day
15, a dose of 0.5 mg TRH was administered intravenously to study the TSH
response. A significant decrease was observed in the levels of T~ and T,,
but no change was seen in free T_. TSH did increase with the iodide dose,
level, but this increase was not statistically significant.. The TSH
response on the TRH provocation was significantly larger on day 15 (after 2
weeks of iodide exposure) than on day 1. The total urinary excretion of
iodide was 1 to 1.1 mg per day, and thus the same as seen in Study Wl.
There was no difference in the results between males and females.
2. Analysis
From the above described studies, it could be inferred that 200 mg
R-3 significantly increases the human TSH levels, but, there were no statis-
tically significant increases in average TSH levels for the 20 and 60 mg/d
groups in Study Wl. Interpretation of the outcomes of this study is diffi-
cult, however. The mean baseline TSH value is different for each group,
and the baseline for the 200 mg dose group is significantly lower than for
20 and 60 mg doses. In addition to the difference in baseline values,
57
-------
July, 1987
heterogeneity is also evident from the large variance observed in the 20
and 60 mg groups. Moreover, the increased TSH in the 60 mg group is
entirely attributable to the response of 2 individuals, showing an increase
2 to 3 times higher than the other subjects in this or other groups. This
may indicate either measurement variation or an abriormal thyroid function.
One should -be aware that any differences between Day 1 and Day 15, even in
the 200 mg/d group, are still small in view of the TSH assay variance.
Also, since the baseline TSH level at the 200 mg is close to the assay
detection limit, it is probable that errors in results based on the
difference between the baseline and another value is likely to be higher
than in the other dose levels where the baseline level is* well above the
assay detection level.
Given the small number .and apparent heterogeneity of the test subjects,
and in view of the small, if any, increase in the R-3 level, (which, is well
within the range of normal values), the Panel feels that this study does
not allow an assessment of whether or not a dose-response effect is pre-
sent. Since even the 200 mg/d dose-group had final TSH values (at day 15
after dosing) not different from those of the 20 and 60 mg/d groups, it may
be even questioned whether the study has shown an effect on TSH at all. It
. . « '
is quite possible that the TSH increase seen in this group is an artifact
, caused by the abnormally low starting values. No information was available
to exclude this possibility. There is a lack of quality assurance and con-
trol, which could have included information on the health staus of .the
volunteers, the time of day the blood samples were taken, the basal meta-
bolic rate, the use of medication with potential effects on this rate or on
TSH, and on laboratory performance. The study design could have been
improved by randomizing the treatment over all study participants, and by
measures that ensure equal study and laboratory conditions. Therefore, it
appears that the most that can be concluded from this study is that a NOEL
seems to be present at 20 and 60 mg R-3/d, but that the study design did
not. exclude higher or lower NOEL '"s. '..'-•
-------
July, 1987
A dose of 1.5 mg iodide, effectively equal to that available from 200
mg R-3 (judged on the equal amounts of iodide excreted in the urine),
appears sufficient to reduce TS and T^ in humans if the iodide is given
alone, without concomitant exposure to R-3. Howevert> the increase in TSH
with exposure to that amount of iodide is small and not statistically
significant. In the PRI study (109), male rats showed increased TSH in all
treatment groups, although statistical significance was not reached for the
purified R-3-only group. The Panel does not share PRI's conclusion that
the significant effect in the Nal-orily group is due to chance. The obser-
vation that this group showed the largest increase is consistent with the
demonstration that the purified R-3-only group showed the least increase.
Since adding more than 80 ppm of Nal in the diet did not result in a larger
effect, that level may reflect some-limit, such as saturation. However,
inferences from non-significant differences between groups, albeit consis-
tent, should be judged with caution. Female rats exhibited a significant
TSH increase in the R-3 + Nal dose group only. Strangely enough, the
"serial assays" of both male and female rats did not show significant TSH
changes in any of the treatment groups. This sampling group was designed
to evaluate longitudinal effects by avoiding inter-assay variation (all
samples were frozen until .the end of the study, and testing for TSH was
done in one run). When this is done, there was no effect with time at all.
It is difficult for the Panel .to understand the results of the PRI
study. It seems that R-3 affects the T3 and T^ balance, but not the TSH
level, while the reverse is true for Nal. The TSH outcomes of the study
can be explained by the effect of the iodine (as impurity or as part of the
molecule) component of R-3. Statistically significant decreases in T3 and
increases in T, levels were observed in male and female rats treated with
R-3 fortified with Nal, not in- the groups treated with Nal or R-3 only.
An aspect not discussed by the researchers conducting the semichronic
rat studies is the difficulty in interpreting the' longitudinal results in
light of the role of TSH in thyroid hormone economy. Although the studies
done by PRI and Ingbar _et_ al. (109,121,122) did show changes in TSH, T3 and
T, levels, it appears that the changes occurred within the first months of
observation. It is not explained why the hormone levels disclosed a tend-
ency to go back to normal at the end of the six to seven month studies,
rather than maintaining their peak level in a state of equilibrium. With
59
-------
July, 1987
this in mind, it. is unknown what to expect of the hormone levels of a
second generation after, two years in the bioassays. Moreover, because of
the rapid response, the first month rat values are likely to be an
'inadequate reference for hormonal changes in studies involving humans
exposed for only 15 days (Study Wl and Study W2).
At present, it appears that the peripheral conversion of T, to T3 has a
central role in rat thyroid economy. This conversion has been confirmed in
brain, pituitary, liver, and kidney (91). R-3 inhibited this process _in
vitro in liver cells of the same cohort that showed the increased T and
4
TSH (124). The inconsistency in the effects of R-3 in vivo on the hormone
levels may be attributable to a dual effect of R-3. On one hand, R-3
lowers the TS level by inhibiting (or reducing) its peripheral production
from T
4.
As
deiodination is reduced, the T4 level is. likely to
increase. A decreased. T3 will induce 'an elevation of TSH through a feed-
back mechanism. This elevation will stimulate the thyroid to produce and
release more TA , further increasing T4 blood levels. On the other hand,
R-3 has iodide as a metabolite and as impurity. This will tend to lower
the release of T4 and T3 by the thyroid, and . to increase TSH. The actual
hormonal effect of R-3 should then be thought of as a function of the
equilibrium state of the different effects of R-3 proper and iodide.
Whether TSH and/or the thyroid hormones will increase, decrease, or remain
unchanged is, according to this hypothesis, a matter of the balance between
the amount of free iodide, the deiodination rate of R-3, the metabolic
conversion rate of T4 to T3 in the liver, the conversion rate in the pitui-
tary, and the susceptibility of the thyroid to stimulation by iodide. This
balance is complicated by the clinical .observation that the thyroid
suppression by iodide may be only temporary and an escape from suppression
may occur (136).
In summary, in the intact animal, R-3 appears to act on thyroid economy
through an inhibition of. peripheral metabolic conversion of T, to T~,
resulting in a reactive TSH increase. The latter response is increased by
the effect of the iodide impurity and the iodine component in the R-3 mole-
cule. Because there is an increased TSH-response on the 'TRH-provocation
with high doses of R-3, it is possible that R-3 increases the sensitivity
of the pituitary TSH response to thyroid hormone levels as well. Since
there is evidence of an effect on thyroid economy, and since the effect (an
60
-------
July, 1987
increase in TSH elaboration) appears to be important in the induction of
thyroid follicular neoplasia, and since the. target organ is the thyroid,
therefore, available evidence supports the conclusion that the tumorigenic
effect of R-3 is the result of an indirect effect on the thyroid, probably
involving TSH. This is further supported by finding little evidence of R-3
genetic toxicity (Chapter 3).
D. Pathology . .
The microscopic distinction between adenomas and carcinomas of the
thyroid is often difficult, which may lead to differences in diagnosis
among pathologists. The slides of the IRDC studies have been reviewed by
several pathologists with differing results (see Table 3). Although the
original reports of the IRDC studies indicated only an oncogenic effect of
R-3 in male rat thyroid follicular cells, and only at the 4% dose, a more
comprehensive evaluation of the available data has presented another
picture. As discussed in Chapter 7, there is very weak evidence that there
is a carcinogenic effect of R-3 in male rats.
The Panel has evaluated the information on ultramicroscopy of the thy-
roid and agrees that there is some evidence that the increased , abnormal-
lysosomal 'bodies may represent a compound-related effect. However, the
Panel's evaluation, as well as that of pathologists from FDA and the study
contractors, is based on a non-quantitative judgment. Confirmation of this
feature requires a quantitative method of evaluation, such as image
analysis or cytophotometry.
E. Maximum Tolerated Dose in a Chronic Toxicity Bioassay
The Panel has considered the issue of the effect of.exceeding the Maxi-
mum Tolerated Dose (MTD) on the validity of the tumor response data with
regard to their relevance to risk estimation. The OSTP document (10) has
defined the MTD as "the highest dose .... which, given for the duration ,of
the chronic study, is just high enough to elicit signs of minimal toxicity
without significantly altering the animal's normal lifespan due to effects
other than carcinogenicity." The problem is how to define "minimal toxi-
city." If this definition is broadened to include any histomorphologic or
pathophysiologic effect, many would find the definition unacceptable for
chronic toxicity testing. Using the broadened definition, the 4% dose
61
-------
July, 1987
level of the IRDC study (74) did exceed the MTD, as shown by the electrons-
microscopic lesions of the thyroid and in the weight loss in R-3 dosed
animals (Chapter 5). Exceeding the MTD could have resulted in metabolic
mechanisms ultimately responsible, for tumor formation not occurring at the
lower doses. Using the OSTP definition, bioassay results at a dose which
may be above the MTD might be acceptable as long as these toxic effects do
not interfere with the tumorigenic mechanism, e.g., by inducing premature
mortality .or triggering a metabolic mechanism that would not have occurred
at a lower dose level. From this viewpoint, the 4% dose level in the IRDC
study did not exceed the MTD. The Panel's opinion is that the possible
exceeding of the MTD in this experiment does not alter the positive finding
of oncogenicity since excess tumors have also been observed;below the MTD.
Risk estimates will be presented for data both inclusive and exclusive of
the 4% test results.
Another complication in the interpretation of the MTD for R-3 is that
the data concerning -oncogenicity result from multigeneration studies. In
addition to the possible intrauterine exposure of the fetuses to R-3 from
the moment of conception until the fetus regulates its own hormone levels,
there are also the consequences of exposure to altered maternal levels of
T4, T3 and TSH. The fetus is also likely to be exposed to elevated levels
of-iodide. Even slightly increased TSH levels may disturb the development
of the fetal growth regulation system. It is, thus, conceivable that doses
of R-3 not exceeding the *MTD in the mother, may grossly exceed a threshold
beyond which the growth regulation system of the fetus is affected. This
effect may be the cause of tumor formation. That the R-3 doses in the IRDC
studies were fetotoxic is beyond doubt. Doses lower than the lowest IRDC
dose have caused fetal absorption and decreased litter size In rabbits
(80). Rats fed 4% R-3 had offspring with fewer and smaller pups (73).
However, it should also be recognized that exposure of .the fetus from '
conception through maternal exposure to R-3 mimics human exposure. The
Panel is of the opinion that the issue of an MTD in multigeneration studies
-needs further investigation. jSome arguments have been presented above per-
taining to fetal intoxication as a triggering mechanism. Although fetal
toxicity is likely to have occurred even at doses lower than 4%, the lack
of factual data on hormone levels in fetuses or newborn rats renders these
arguments speculative, albeit' based on scientific reasoning.
62
-------
July, 1987
F. No-Effect Levels of R-3
No observed effect levels (NOELs) are often used to set limits to
exposure levels for non-neoplastic health effects. Because R-3 may work
through secondary mechanisms which are not by themselves directly neo-
plastic, and which may have thresholds, it is useful to try to define a
NOEL for its effects. It should be clear that the value of a NOEL depends
on a number of factors, such as the toxic potency of the chemical, the size
of the study population, the number and spacing of doses, etc.
None of the rat studies discussed in this report demonstrated a NOEL
for the effects on T3> T4 or TSH, not even at the lowest subchronic dietary
R-3 dose level of 0.5% (approximately 250 mg/d), for which hormone esti-
mates were available. In humans, as noted above, Study Wl found a
statistically significant increase in TSH from, a dose of 200 mg/day, but
not at 20 or 60 mg. However, as stated earlier, analysis of this study is
complicated by'apparent heterogeneity among the 3 dose groups. Because of
the lack of significant differences, the 20 and 60 mg/d groups could be
taken as the NOEL. Ingbar et al.. (27) found no effects in humans at any of
the dose levels from 5 to 25 mg/d administered for up to 4 weeks, but the
same authors reported a TSH-increasing effect of a single, oral dose of 75-
80 mg R-3 (28).
From the above, the Panel concludes that a human NOEL for R-3 hormonal
effects, i.e., the highest level ,of R-3 that does not cause changes in
hormone levels that may possibly lead to tumorigenesis, can be set at 20 to
60 mg/d. Alternatively, one could set the NOEL at a higher level, perhaps
200 mg/d, and use a safety factor to account for the poor quality of the
study which "defines" -the NOEL.- Because of this uncertainty, the Panel
will give results based on several possible NOELs.
G. Relevance of Animal Toxicity to Humans
As stated in the OSTP document (10), when evaluating human carcino-
genicity, in the absence of adequate human data it is reasonable, for
practical purposes, to regard chemicals for' which there is sufficenf
evidence of carcinogenicity in animals as if they present a carcinogenic
risk to humans. However, there are some human data that suggest that
thyroid tumors in humans may have a pathogenefic basis different from that
in rats. • First, the annual age-adjusted incidence rate of all types of
63
-------
B.I
GENERAL ATTRIBUTES
Attribute 6 The report explicitly and fairly conveys scientific uncertainty,
including a discussion of research that might clarify the degree
of uncertainty.
SOURCE Case Study D. Formaldehyde (Pages 7-14 to 7-23).
Note This report illustrates a treatment of uncertainty. Potential research
to clarify uncertainty is not discussed.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
7-3. Uncertainty in Riafc Estimates
Model-derived risk estimates should be viewed m the prcoec
context. The upper bound estimate should not be viewed as -a
point estimate of risk. As the Guidelines state (s?A, 1936):
"the linearized multistage procedure leads to a plausible upper
limit to che risk that is consistent with some proposed
mechanisms of carcinogenesis. Such an estimate, however, does
not necessarily give a realistic prediction of the risk. The
true value of the risk is unknown, and may be as low as zero."
Other factors are also important.
As Table 7-2 illustrates, there is a wide range between the
MLE and upper bound estimates, approximately 4 or 5 orders of
magnitude. This illustrates the statistical uncertainty of.the
estimates generated due to the input data from the study used,
which in this case is highly non-linear. For instance, the
individual risks for apparel .workers range from 1 X 10"3 [31] zo
6 X 10" CBl]. In addition, it has been shown that the MLE _s -
sensitive to small changes in response data when the response is
very nonlinear in the experimental range. For instance, the dose
giving a risk of 1 X 10'6 (MLE) varies significantly due to small
changes in the response data of the Kerns et ai. (L983) study
(Cohn, 1985b). The following illustrates this:
Response at 2 ppm
.(malignant)
1. 0 (actual)
2. 1/1,000
.3. 1
Dose for Risk of
1 X 10"6 (MLE)-
0.67 ppm
0.0022 ppm
0.0006 ppm
7-14
-------
Ten perturbations of the squamous ceil carcinoma data for
the Fischer 344 rats were selected by slight alteration in one of
the dose-response proportions or the,elimination of a dose level
from the study in an attempt to show sensitivity to these
perturbations was examined by modeling. These estimates appear
in Appendix 5. It was found that, in general, slight
perturbations of the data do not significantly disturb the
predictive power of the model for upper bound estimates. This is
not the case for MLEs. Only extreme perturbations significantly
affect upper bound risk estimates. Consequently, when modeling
data that are very non-linear, one should not place great
certainty on MLE estimates. In addition, model choice can lead
to uncertainty. As Appendix 3 illustrates, there is a wide
divergence in risk estimates obtained using the CUT rat data.
Independent background, tolerance distribution models such as,
the probit, logit, and Weibull, produce estimates indicating
virtually zero risk (probit predicts zero risk). The independent
and additive background gamma-multihit models produce similar
results. However, when additive background models are used risk
estimates are much higher, with the multistage model giving the
highest riafca. As discussed in section 7.1, the linearized
multistage procedure was used for primary risk estimation.
As discussed above, the major contributor to the uncertainty
seen in the risk estimates using the multistage model is the .
steep dose-response seen in the Kerns et al (1983) study. There
were no carcinomas at 2 ppm, 2 at 5.6 ppm, and 103 at ,14.5 PPm,
7-15
-------
which is a 50-fold increase for only a 2.5 times increase in
dose. If changes in respiratory rate are taken into account (the
rats at 14.3 ppm are receiving the equivalent of a 12 ppm
exposure—use of this data leads to no significant change in
estimated risks at exposures of concern) (Grinstaff, 1985), there
is a 50-fold increase for 'only a doubling of the dose.
HCHO's ability -to cause rapid cell proliferation, cell
killing and subsequent restorative cell proliferation, its
ability to interact with single-strand DNA (during replication),
interfere with DNA repair, its demonstrated mutagenicity, and the
fact that the dose was delivered to a finite area may help
explain the abrupt increase in the response. However, none of
these factors demonstrate the presence of a threshold or minimal
risk at exposures below those that cause significant
nonneoplastic responses such as cell proliferation, restorative
cell growth, etc. For instance, although HCHO causes varying
degrees of cell proliferation in the nasal mucosa of rats due to
HCHO exposure, it must be remembered that there is a natural rate
of cell turnover in this tissue. While it is low in comparison
to HCHO induced increases, it does provide the opportunity for
HCHO to react with single-strand DNA during cell replication,
possibly resulting in a' mutant cell which, if proper conditions
are met, could result in a neoplasm. While an event such as this
may be rare, it is not unreasonable when one considers that the
•>
opportunities for this event to occur are great due to the
immense number of cell-turnovers which may lead, to defects ;in
7-US
-------
some cells of the population of the individuals exposed. Even
so, the marked nonlinearity of the response introduces
considerable uncertainty into any discussion of the possible
mechanism of HCHO induced carcinogenicity at exposures below the
experimental range. ;~
The .different responses seen in the animals tested also
leads to a degree of uncertainty. Although rats, mice, and
hamsters have been tested in long-term bioassays, only in rats
have statistically significant numbers of neoplasms been
observed. Only two carcinomas were seen in mice at the highest
dose in the CUT study, but the nature 'of this response is
complicated by the fact that mice are able to reduce their
breathing rate to a greater extent than rats. If this effect is
accounted for, the "dose" mice received at 14.3 ppm is
approximately that which the rats received at 5.6 ppm, where two
carcinomas were observed. Consequently, on a "dose" received
basis, rats and mice may be equally sensitive to HCHO. Although
no neoplasms were seen in the hamster study, a number of factors
may be responsible. Firs't, there was poor survival. About 40%
of the 88 hamster died before eighty weeks, and only 20 hamsters
survived ninety weeks or more. If a response comparable to that
of the CUT-study were expected, 25% or five of the hamsters
surviving ninety weeks or more would have had tumors. However,
the duration of the study may not have permitted them to be
grossly visible. Second, the limited pathology protocol may not
have been able to detect small tumors. And third, the dosing
7-17
-------
regimen and physiologic factors* (changes in breathing rate) may
have been factors (see section 4.1).
Although the foregoing helps explain some of the species
differences observed, there remains the possibility that other,
unknown, factors may be important. However, in any event, no
data have been developed to show that humans would respond
differently to ECHO than rats and data exist showing that rats
and monkeys respond similarly to HCHO when nasal irritation and'
squamous metaplasia are used an endpoints.
It. is often useful to compare lifetime excess risks
estimated from the epidemiologic studies to "those risks estimated
from animal data. Tables 7-4 and 7-5 and Figure 7-1 present such
a comparison. Estimated lifetime excess risk can be determined
for either occupational or domestic exposure to HCHO. This
comparison assumes that exposure- to HCHO is associated with an
increase in neoplasms at one site only and that the site-specifie
excess risk observed in the epidemiological study is the excess
above a risk of one for the study population relative to the U.S.
population (Margosches and Springer, 1983). Hence, lifetime
excess risks based on the epidemiological studies, are calculated
by multiplying the excess risk observed in the epidemiologi.c
study by the site-speed fie mortality ratio. 1980 mortality data
t,
are used in this calculation.
The estimated lifetime excess risks were based on-
significant associations observed in the Blair et al. (1986),
Vaughan et al. (1986a,b), Hayes et al. (1986), Stroup, Harrington
and Oakes (1982), and Harrington
7-L8
-------
Table 7-4.
Upper Bound Risk Estimates Based on
the CUT Data for Given Exposures to HCHO
Exoosure
Level (ppm)
Animal Based
Upper Bounda
Resin Worker
0.24
1.4
Furniture Worker
Pathologists
Mobile Home
Residents
(10 years)
0.1
1.3
3.2
0.19
5 X 10
X 10
1 X 10
2 X 10
6 X 10
3 X 10
-4
-3
-4
-3
-3
a Based on the linearized multistage model and the rat data
from Kerns et al. (1983).
7-19
-------
Table 7-5
Estimated Lifetime Excess Risks
Calculated from the Epidemic-logic Studies
Exposure Author
Resins Blair et al.
Site
Lung
Risk
Ratio
1.32b
Nasopharynx 2.0C
Resin, Glue Vaughan et.al.
HCHO & Wood . Hayes et al .
Pathologists Harrington &
Shannon
Harrington &
Oakes°
Anatomists Stroup
a Estimated lifetime excess risk
Nasal
Cavity &
Sinuses
Nasal
Cavity &
Sinuses
Leukemia
Brain
Brain
= (RR-1) *
3'. 8
1.9C
2.0
3.31.
2.7
Estimated Lifetime
Excess Riska
2 X 10"2
8 x icr4
7 X ID'4
2 X 10~3
2 X 10~2
1 X 10"2
8 X 10~3
*•=•
# of site-specific deaths
proprotion
— c
of site specific
^aths __
Mortality proportion based on 1980 deaths.
Analysis -of white male wage workers with greater -than 20 years latencv
and HCHO exposure above Oppm-year.
- . '
Analysis o.f white male wage worker with HCHO exposure greater than OD
year . ,.,'••
7-20
-------
Pathologists
10
3.2 ppm
1 I
-.5
.-4
«-3 J ,
,-2
10
10
10
Stroup, Harringtons Harrir.o-
Brain" Oakes, sShannc:
Brain Leukeni
Kesin Workers
10
-5
0.24 ppm .1. 4 ppm
,
10
-4
Vaughan
et al. ,
SNC*
ao
Blair et al. ,
Nasopharynx
10
Blair et al,
•Lung- 7.0 yr
• latency
Pttrniture Workers
10
-5
Mobile Hone
Residents
10
-5
0.1 pom
1.3 ppm
10
-4 '
10
-3
10'
Kayes et al.
SMC, Controlled
for high wood dust
exposure
0.19 ppm
10'
10
-3
10'
Traurrhan et al. ,
Masoioharvnx
Fiqure 7-1. Comparison of the upper bound risks based,on the animal
data to estimated lifetime excess risks based on the epidemiological
studies. Animal-based upper bound risks for the identified exposure
lavel to HCHO are above -the line. The estimated excess lifetime
risks based on the observed excesses in site-specific neoplasms are
-------
and Shannon (1975) studies. For example, when one examines lifetime
risks from exposure to resins, the estimated lifetime excess risk
associated with the 35% increase in lung cancer among white males
with a greater than 20 years latency reported by Blair et al. (1986)
would be 2 X 10~2 and the estimated lifetime excess risk associated
with their reported 200% increase in nasopharyngeal cancers would be
8 X 10~. ,* The 280% increase observed by Vaughan et al., (as
reported in SAIC, 1986) for nasal sinus and cavity neoplasms in
conjunction with exposure greater than 10,000 hours to resins,
glues, and adhesives gives an estimated lifetime excess risk of 7 X
10~ . The upper bound risk for an exposure of 0.24 ppm HCHO based
on the animal data is 5 X 10~4, and for an exposure of 1.4 ppm
HCHO, would be 3 X 10~3.
Comparing the results reported by Hayes et al. (1986) is
more complicated since Hayes et al. do not delineate the exposed
population. However, if one chooses an exposure group, such as
furniture workers who may be exposed to both wood dust and HCHO,.
one can make some observations. The reported exposure for
furniture workers ranges from 0.1 ppra to 1.3' ppm HCHO as an /
8-hour, time-weighted-average. Upper bound risks based on the
animal data associated with these exposures are 1 X 10"4 and
2 X 10~3, respectively. Using the 90% increase in nasal cavity
and sinus risk observed in analyses which controlled for high
wood dust exposure, the estimated lifetime excess risk based on
the Hayes et al. study would be 2 X 10~3.
Thus, when individual tumor types are examined, one can see
'. /
that the upper bounds are not indicating larger excesses than
7-22
-------
seen in certain studies given uncertainties about exposure.
Although HCHO's potential carcinogenic effects are not expected
to be limited to one site in humans because humans do not
necessarily breathe through their noses as rats do, the analysis
described above provides a check of the risks derived from animal
data and those seen in human studies.
Finally, a factor that can have a major bearing on
population risk estimates is the quality of the available
exposure data. Assumptions made in reporting exposure levels can
have a major impact.' For instance, it is not uncommon during a
monitoring exercise to find a number of samples that are below
the detection limit of the analytical technique used. Thus, when
a mean exposure level is calculated it should be realized that if,
the nondetectable (ND) samples are counted as 0 the calculated
mean will understate the actual situation. Conversely, if the MD
samples are counted as the limit of detection, the mean win
overstate the true situation. Another factor that can skew
exposure estimates are changes in non-governmental exposure limit
recommendations and the number of years over which the data are
collected. Since a number of years of exposure data are often
used to calculate means, it is possible that the mean will be ,
weighted by samples taken prior to changes in voluntary exposure
limits. Thus, the reported mean could be substantially
overestimating the true situation. For instance, in the garment
industry, HCHO levels have apparently been falling since the late
70's and early 80's as a result of increased concern and a
7-23
-------
B.2
HAZARD IDENTIFICATION
Attribute 1
All relevant information is presented and reviewed.
SOURCE Case Study B. TCDD (Pages v-xvi)
Note These excerpts are' from the executive summary of a very compre-
hensive assessment. This illustrates the many types of information
given detailed consideration in the full report.
SOURCE Case Study H. Methylene Chloride (Pages 5-9).
SOURCE Case Study J. Red Dye No. 3 (Pages iii-v).
• Note This report considers a large body of information. The Table of
Contents is included in this appendix to illustrate the types of data
considered.
-------
ncasl
special report
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT. INC, 260 MADISON AVENUE. NEW YORK. N.Y. 1001.
EXECUTIVE SUMMARY
DIOXIN: A CRITICAL REVIEW OF ITS DISTRIBUTION,
MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,
AND THE SETTING OF ACCEPTABLE EXPOSURE LIMITS
SPECIAL REPORT NO, 87-07
MAY 1987
-------
— y, -
C. The Synthesis and Discovery of "Dioxin"
Unlike many environmental chemicals which have value in
manufacturing or commerce, no use has been discovered for
dioxin.* Although it was first synthesized in the heyday of
German organic chemistry near the turn of the century, mention
of it practically disappeared from the scientific literature
until 1957. In that year, an outbreak of chloracne, an
occupational skin disease caused by exposure to certain
chlorinatediorganic chemicals, prompted a dermatologist and a
chemist to investigate the chemicals in a plant in Germany
After a number of false starts, they identified dioxin as a
contaminant of trichlorophenol, a chemical used in the
manufacture of various pesticides, the most important of which
was 2,4,5-T (2,4,5-trichlorophenoxy acetic acid), an herbicide.
The chemists at the German plant devised new production
methods^ that reduced dioxin contamination of trichlorophenol
and eliminated chloracne from,the plant's workforce. Within a
decade or so, similar changes were made in other plants
worldwide, and chloracne as a consequence of exposure during
trichlorophenol manufacture was largely controlled.
D- Human.Effects of Dioxin •
1.
Effects .from Occupational
Exposures to High Levels of Dioxin
The chloracne discovered in the German plant was
associated with routine exposures from leaks and spills
during the production process: Greater exposures resulted
from industrial accidents. At one time or another in the
United States,. Denmark, England, France, and Germany,
large pressurized kettles (autoclaves) blew out their
seals spraying trichlorophenol and, we now know, dioxin
across workrooms and workers. Chloracne was common in men
directly exposed and in men who cleaned up the messes. In
addition, some liver pathology, dizziness, disabling aches
and pains (nervous system damage), reduced sex drive, and
The word "dioxin" is used to refer to
als° called
-------
- vi -
other adverse effects were seen in some of those workers.
With the exception of chloracne, which still persists in
some men exposed more than 35 years ago, the other
symptoms abated over time. The presence of chloracne
among these men is convincing evidence of high-level
exposure.
Numerous studies of more than 1,000 workers who were
exposed to high levels of dioxin during trichlorophenol
manufacture and accidents have failed to show elevated
cancer, heart disease, reproductive difficulties, or
premature mortality. An excess of stomach cancers in one
occupationally exposed group is statistically significant,
but no excess of that cancer has been seen in other
exposed groups. These overall negative findings are
especially important in judging dioxin's effects on humans
because there is no question that those workers were
exposed. Many other studies and proposed studies of the
possible effects of dioxin on humans are hobbled or
forestalled by the impossibility of deciding who was and
who was not exposed, and if they were, to how much.^
2. Dioxin Exposure at Seveso, Italy
On June 10, 1976, an industrial accident spewed
dioxin and other chemicals into the air, and the resulting
chemical cloud drifted over the town of Seveso, Italy.
Some effects were immediately apparent; vegetation browned
and died, and small wild animals sickened and died.
The suddenness of the accident compounded with
incomplete plans for handling- such a disaster resulted in
many residents remaining for up to two weeks in the area
contaminated by "fallout" from the cloud. Before
evacuation was completed, residents could have been
exposed directly to the cloud, to chemical fallout from
it, and through ingestion of contaminated garden
vegetables. The exposures were sufficiently high to cause
chloracne in more than 100 children. This is the only
exposure situation, except for the high-level occupational
exposures, in which dioxin caused chloracne.
During the next year, reports appeared stating that
the exposures had caused abortions and an increase in
birth defects. In addition, some alterations in human
biochemistry were associated with the exposures. In the
-------
- vii -
ten years since the accident, the importance attached to
the reports of abortions and birth defects has decreased.
In 1976, therapeutic abortions were illegal in Italy,'and
some 30 presumably exposed pregnant women went to
Switzerland for abortions rather than risk delivering a
child with birth defects. Suggestions,have been made that
secret induced abortions may have been reported as
"spontaneous" abortions, contributing to an apparent
increase. The .study of birth defects was hampered by poor
historical data in that region of Italy making comparisons
with "normal" rates dependent on guesswork. After
Studying all the reports from the exposed population, the
Government of Italy and an advisory group of international
experts have decided that chloracne was the only human
health effect.
Seveso was characteristic of environmental exposures
in that a cross-section of the human population—children,
the elderly, women and men—were all exposed. It was
unusual in that it was a single high-level exposure. The
question of a possible increase in cancer rates as a
result of the accident cannot be eliminated; 10 years may
not be sufficiently long for the development of cancer,
but the Seveso population continues to be carefully
monitored. If there is some health effect that is not yet
manifest, it should be detected in the future.
3. Herbicide Sprayers
Chemical industry workers and the children of Seveso,
• both of whom developed chloracne were clearly exposed.
Although their exposure levels are unknown, it can be
assumed that their one-time or repeated exposures were
greater, in general, than those of people who did not
develop chloracne. Epidemiologic studies in those highly
exposed people have failed to produce convincing evidence
of diseases other than chloracne being related to dioxin
exposure.
The evidence for an association between dioxin
exposure and human cancer that has attracted the most
attention is an excess of cancers in herbicide sprayers
The associations were made in two "case control studiesi"
in which the histories of men who had specific cancers
(the cases) were compared to the histories of men who did-
not have the specific cancers (the controls). The cases
more often reported exposures to herbicides, and those
-------
- Vlll -
studies showed associations between cancers classified as
soft tissue sarcomas and lymphomas and herbicide use in
Swedish lumberjacks. Similar case control studies in New
Zealand, Finland, and another one in Sweden faiTed to
confirm the associations. Furthermore, a "cohort study,"
in which the incidences of those cancers in Swedish
agricultural and forestry workers were compared to the
incidences in workers in other industries failed to detect
an excess. The strength of the association in the Swedish
case control studies makes it impossible, to disregard
completely those results. At the same time, absence of
confirmatory studies makes them less than convincing.
Studies of another group of herbicide sprayers revealed an
excess of stomach cancers. Although that finding
parallels a report from one chemical plant population, it
is not complemented by other studies of herbicide sprayers
or chemical workers where this effect has not been found.
The most famous group of herbicide sprayers is
probably the Ranch Hands, the Air Force unit that sprayed
Agent Orange in Vietnam. All of those 1,200 or so men
have had thorough medical and psychological examinations,
which are to be repeated at 3-year intervals for 20
years. As of late 1985, the death rate among Ranch Hands
was lower, but not statistically so, than among a
comparison group of Air Force personnel who were not
exposed to herbicides. There is no difference in cancer
rates between the two groups. There are no soft tissue
sarcomas among Ranch Hands, and one in the comparison
population. Sixteen years have passed since the last
Agent Orange spraying and about 20 since the peak spray
years. The absence of excess disease or early mortality
among Ranch Hands so far argues that no excesses will be
found.
Herbicide sprayers, whether lumberjacks or Ranch
Hands, are probably an intermediate exposure group. They
are, judging from the absence of chloracne among sprayers,
less exposed than the chemical plant workers, and they are
almost certainly more exposed than people who may have
been "environmentally" exposed through being around one or
a few application areas.
-------
- ix -
4. Environmental Exposures to Dioxin
a. Environmental Exposure and Claims.of Miscarriages
In 1978, EPA announced that it was considering
suspending all uses of 2,4,5-T as well as other
pesticides made from trichloroghenol. At the time of
the announcement, available data about the toxicity
of 2,4,5-T and dioxin were limited to results from
animal studies and fragmentary reports from studies
. of industrially exposed workers. That changed
significantly when an Oregon high school teacher
wrote letters to Federal agencies reporting her
investigation of a possible association between
2,4,5-T spraying and miscarriages in women living
near Alsea,;Oregon. The EPA commissioned a quickly
done epidemiologic study that appeared to confirm the
association. Acting on those results, EPA in
February 1979 declared most 2,4,5-T uses an "imminent
hazard" and issued an emergency suspension of those
uses.
Scientists critiqued and criticized the EPA's
epidemiologic study, and it is now accorded scant
credibility. The surest test of a scientific study's
value is whether or not scientists cite it in
subsequent discussions. The Oregon miscarriage study-
has almost disappeared except when referenced' as the
cause of the emergency suspension of uses of 2,4,5-T.
b. Agent Orange -
In the 1960's, a new market opened up for
2,4,5-T. The United States military settled on a
50:50 mixture of 2,4,5-T and 2,4-D
(2,4-dichlorophenoxy acetic acid) as the best
herbicide to denude the jungles and destroy certain
food crops of the enemy in Vietnam. That mixture,
called Agent Orange because of the color-coded band
on the drums in which it was shipped, became the best
known herbicide in history.
. From the initiation of the chemical spray
program in Vietnam, some Vietnamese, both North and
South, claimed that the chemicals had caused birth
defects and miscarriages. The laboratory results
showing that dioxin caused birth defects in animals
-------
- x -
created enough concern that a series of Congressional
hearings culminated in suspension of Agent Orange
spraying in 1970.
A team of American scientists that visited
Vietnam during the war years found it impossible to
verify that Agent Orange had caused birth defects or
miscarriages. Those health effects are notoriously
difficult to study—they are not always recorded and
diagnostic criteria differ according to time and
place. Furthermore, record keeping in the midst of a
war is necessarily poor. The team concluded that the
evidence was not persuasive, but that the many
problems with records and information precluded
coming to a definite decision.
%
Ground troops in Vietnam could have been exposed
to Agent Orange, and some veterans claim their
illnesses as well as birth defects among their
children stem from those exposures. A study by the
Centers for Disease Control (CDC) failed to find any
excess of birth defects among children of Vietnam
veterans. The same study found associations between
3 birth defects and "opportunities for exposure to
•Agent Orange," but the authors of the study as well •
as many reviewers discount the reported
associations. The link between "opportunities for
exposure" and the likelihood of exposure is so
tenuous that the associations are suspect. An
equally good explanation for the apparent
associations is-that they are due to chance.
Congress, in 1979, mandated a study of the
health of veterans who were exposed to Agent Orange.
The study has not yet begun because of the to-date
impossibility of deciding which veterans may have
been exposed and which were not. However, the recent
development of techniques to measure dioxin in about
100 milliliters of blood possibly provides a method
to verify exposure. If the method proves out, the
study of possible effects -of Agent Orange exposure on
veterans may get underway in 1987.
-------
- xi -
c. Exposures in Missouri
In Missouri, oily wastes from trichlorophenol
manufacture were sprayed on unpaved roadways and in
trailer parks and in horse arenas. The confirmation
of soil concentrations greater than 1 part per
billion (ppb) in Times Beach was the stated reason
the Federal government purchased the town.
Epidemiologic studies have found no increased
disease incidence among former residents of Times
Beach nor among other Missouri residents that lived
,at or near sites with high dioxin levels. However, a
recent CDC study reported that there may be '
dioxin-related impairment of the immune system in
people who lived in a trailer park where
trichlorophenol wastes were used for dust
suppression. Soil concentrations of dioxin at that
site ranged up to 1,100 ppb, which is consistent with
;possible high exposures. The CDC study depended on
measurements of skin reddening as an indication of
immune system competence. Problems in
interpretations of those skin tests resulted in the
investigators discarding data from 61 percent of the
"exposed" population and 32 percent of the
"unexposed" control population. The missing data, of
course, might contribute to or account for tlie
reported differences in the immunologic
characteristics of the two populations. Furthermore,
the reported immune deficits are of such magnitude
that clinically detectable diseases would be expected
to be more common in the exposed people. No disease
excesses have been seen which casts doubt on the
validity of the reported observations.
Studies of the exposed populations in Missouri
continue, and additional information will be
forthcoming. Whatever the results of those studies,
they will likely be the primary source of data about
environmental exposures in the United States. It is
unlikely that people anywhere else in the United
States were environmentally (as opposed to
occupationally) exposed to more dioxin.
-------
- xii -
E. Animal Studies of Dioxin Toxicity
Almost all our knowledge of the toxicity of dioxin comes
from animal studies. Although much has been learned, much
research remains to be done to understand dioxin's effects in
animals. For instance, the mechanism by which it kills, remains
unclear. Neither is it understood why the lethal dose varies
5,000-fold between guinea pigs '(0.6 ng/kg) and hamsters
(>3000 ng/kg). What is known is that lethally exposed
animals "waste away," gradually losing weight, and die after
two weeks or so. One model for dioxin's lethal action is that
it alters a biological set point' causing the animal to eat so
little that it cannot sustain itself. In addition, alterations
in Vitamin A levels and thyroid functioning have been suggested
as important in the wasting syndrome and death. The .decreased
food consumption is sufficient to cause death, but the
mechanism that causes the decrease remains obscure.
The wasting syndrome, in contrast to effects on specific
organs, is the only observed,effect of dioxin intoxication in
some species. In other species, various organ systems are
affected, but the toxicity seen in those organ systems does not
appear to be 'sufficient to cause death. • .
The human exposures most likely to have been immediately
life threatening were those in industrial accidents. However,
there have been no reports of dioxin-related deaths". In at
least one dioxin-contaminated workplace, rabbits were used as
biomonitors. Even though workers had been in the same area
with no ill effects except chloracne, the animals died. These
observations may mean that humans are less sensitive to the
lethal effects of dioxin, but that must be a guarded
conclusion. Workers were clothed, reducing their contact with
contaminated surfaces and atmosphere, and the animals licked
and cleaned their fur. Those differences in opportunities for
exposure complicate drawing conclusions about relative
sensitivity.
Possible chronic effects from dioxin exposure, especially
cancer and reproductive health effects, are of greater concern
than acute effects. Investigations of exposed human
populations have failed to produce convincing, consistent
evidence that dioxin has caused those effects, but they have
been seen in laboratory animals. -
In the 1960's, Congress and the Executive Branch directed
attention at possible health risks from pesticides. In
response to those concerns, the National Cancer 'Institute (NCI)
-------
contracted for the testing of some 30 pesticides in laboratory
animals. The pesticides were tested for carcinogenicity and
teratogehicity (the properties of causing cancer and birth
defects, respectively). In 1969, the herbicide" 2,4,5-T emerged
,as a potent animal teratogen, causing cleft palates in
laboratory mice exposed in the uterus. Within a short time,
further studies showed that the teratogenic activity- resided in
the dioxin contaminant in 2,4,5-T, not in the herbicide
itself. Even especially cleaned-up 2,4,5-T, with dioxin
contamination reduced to 1/30 normal, remained a potent animal
teratogen.
The most important finding for assessing the risk of
dioxin was the discovery .that it causes cancer in laboratory
animals. Federal agencies assess cancer risks using methods
different from those used for other diseases. As a result, at
low exposure levels, the predicted risk of cancer is greater
than the risk of other diseases.
F. The Carcinogenicity of Dioxin
There is no doubt that dioxin causes cancer in laboratory
animals. The most sensitive organ is the liver of female rats;
the liver of male rats is less sensitive toward the
carcinogenicity of dioxin. This pattern of sexual difference
in carcinogenicity is commonly seen in rats, but the pattern
also shows that despite its potency, dioxin is not equally
potent in all animals. That sex hormones are probably involved
in carcinogenicity in the rat Liver has been shown by
experiments with females from which the ovaries have been
removed. They do not develop cancer following exposure to
dioxin.
A growing body of scientific data supports the idea that a
mutation (or at least a "mutation-like event") is the first
step in carcinogenesis. That step is also called
"initiation." Whether or not an initiated cell progresses to a
cancer cell, each of its daughter cells inherits the initiated
DNA. These initiated cells, although harboring a change in
their DNA, are not recognizably different from other cells.
However, given appropriate conditions, such as exposure to
agents called "promoters," these cells ccn become cancerous.
Promoters, unlike initiators, are not believed to interact
directly with DNA, and they have little effect on non-initiated
cells. They can, however, hasten progress from the initiated
state to the cancerous state. Initiators, because they
interact with DNA, are "genotoxic;" promoters are not.
-------
- xiv -
Many experiments have examined the potential of dioxin to
cause mutations. Almost all of those experiments have been
negative, and the few positive reports have not been replicated
despite repeated tries. Committees of experts in various
countries, including advisors to the EPA, have concluded that
there is no convincing evidence for dioxin being mutagenic.
Therefore, there is no evidence for it being an initiator of
carcinogenesis.
If dioxin is not an initiator, what role does it play in
carcinogenesis? Despite-intensive efforts directed at that
guestion, only partial answers are available, The most
productive research for generating a testable hypothesis about
dioxin's mechanism has focused on interactions between the
chemical and certain "receptor" molecules in the cell.
Animals and humans are exposed to many foreign
("exogeneous") chemicals. Many plants synthesize toxic
chemicals that inhibit growth of parasites or cause sickness in
animals that eat them; spoiled meat and vegetables can be
contaminated with putrification products (less of a problem in
this age of refrigeration and food preservatives, but certainly
a problem for our ancestors), cheeses and other fermentation
products contain large numbers of complex organic chemicals,
and smoke from wood fires and other sources are loaded with
chemicals. Animal and human cells have the genetic information
to produce enzymes that metabolize those foreign chemicals and
eliminate them from the body. A group of enzymes called the -
"P-450 enzymes" or "mixed function oxidases" (MFOK or., "aryl
hydrocarbon hydroxylases" (AHH) are synthesized in response to
the entry of foreign chemicals into the cell. They are
generally studied in the liver or in liver cells because that
organ is the site of much of the metabolism of foreign
chemicals.
There are several classes of P-450 enzymes. One class is
responsive to pqlyaromatic hydrocarbons (PAHs). Exposure of an
animal to such a chemical causes increased synthesis of the
P-450 enzymes. The mechanism of the induction of synthesis is
binding of a PAH or related chemical to a protein called the Ah
receptor. The chemical-Ah receptor complex then interacts with
a specific location on the DNA to "turn on" the P-450 enzyme
genes. The products of the genes are P-450 enzymes which
degrade or otherwise metabolize the foreign compounds.
-------
- xv -
Dioxin is a potent dnducer of the PAH-indueible P-450
enzymes. Metabolism of dioxin by P-450 enzymes produces
products that are, less toxic than the parent compound, and the
current consensus holds that dioxin itself, and not a
metabolite, is the active toxic agent.
j» • v' ,
The induction of P-450 enzymes is a reversible event.
Once the level of the inducer, whether a PAH or dioxin, falls
below a critical concentration, the Ah-receptor protein is not
bound and does not interact with DNA.
s
A current model for the carcinogenicity of dioxin is that
the dioxin-Ah receptor complex acts to "turn on" a second
battery of genes, different from the P-450 enzyme genes, in
some cells or under certain conditions. The.toxic effects of
dioxin appear to be'limited to certain cells. For instance,
dioxin causes cancer in initiated liver cells and skin cells of
hairless mice, but not in uninitiated liver cells or in skin
cells of normally haired mice.
The exact nature of the second battery of genes is not
known.. Neither is the action of the gene products. Currently,
some scientists are investigating the possibility that dioxin
disrupts the normal regulation of cell growth. Dioxin also
interferes with the immune system of laboratory animals, and
certain immune deficiencies, could facilitate cancer
development. The possibility that the promotion activity of
dioxin involves inhibition of the immune system is also a focus
of active investigation.
Whatever- the nature of the genes controlled by the
dioxin-receptor. complexes, what is known about the seguence of
events is consistent with the idea that there is a threshold
for dioxin's role in carcinogenesis. A sufficient
concentration of the complex is necessary to turn on and keep
turned on the genes to produce enough gene products to cause -
some critical biochemical reactions. Those reactions probably
have a threshold; until a certain number of reactions
accumulate, damage should be reversible. Although not all
aspects of this mechanism are known, there is substantial
experimental support for it.
Classification of carcinogens as initiators or promoters
is based on observations from many different experiments.
Dioxin has been found to cause cancer in laboratory animals at
0.01 micrograms/kilogram body weight of the test animal/day
(hereafter vig/kg/day) administered over a lifetime. , However,
-------
-XVI -
it has not been found to interact with DNA or to cause
mutations. It is not an initiator and therefore, can be
classified as a promoter. The mechanism of dioxin promotion
activity is not completely understood.
The promotion model for the action of' dioxin can be used
to explain cancer caused in laboratory animals. Liver tumors
arise "spontaneously" in rats that are not"treated with any
chemical. Therefore, the presence of initiated cells does not
depend on a specific treatment. The excess of tumors seen in
dioxin-treated animals could result from the chemical hastening
the process from initiated cell to cancer cell so that more
tumors developed in the exposed animals' lifetimes.
G. The Importance3 of the Decision as To
Whether or Not a Chemical is a Carcinogen
The decision about whether or not a chemical is a
carcinogen is the most important step in estimating what level
of exposure is associated with human risk and what level of,
exposure is permissible.
It is assumed that thresholds exist for most toxic effects
and that doses below the threshold will not cause harm. For
instance, a chemical that causes liver damage in laboratory
animals is a potential cause of liver damage in humans.
However, the laboratory studies can also establish that there
are doses below which the .chemical does not cause liver
damage. At those doses, the chemical might be metabolized and
excreted from the body before toxic effects are manifest or the
toxic effects may be so few that fio damage is detectable.
There is, of course, some possibility that some humans would be
more sensitive to the chemical than are the test animals.
Allowances are made for those possible differences:
First, a No Observed Adverse Effects Level (NOAEL) is
determined in laboratory animals. In practice, that is
generally the highest dose that does not cause a detectable
adverse effect. Then the NOAEL is divided by a "safety factor"
of 100 or 1,000 to set an Acceptable Daily Intake (ADI) level
for humans. The saf.ety factor of 100 incorporates a factor of
10 to allow for possible differences in sensitivity between
animals and humans and a factor of 10 to allow for differences .
in sensitivity in the human population. Depending on the
quality of the experiment, how long the animals were exposed
and observed for instance, an additional factor of 10, based on°
judgement, may be incorporated. The NOAEL plus safety factor
-------
EPA
United States
Environmental Protection
Ag«ncv
•Off
-------
2. OVERVIEW OF DICHLOROMETHANE CANCER HAZARD/RISK ISSUES
In May of 1985, EPA's Office of Toxic Substances found that
DCM met the criteria for priority review under section 4(f) of
* v
the Toxic Substances Control Act (TSCA). Underpinning the 4(f)
decision was the conclusion that DCM should be considered a
probable human carcinogen, as defined by EPA's Guidelines for
Carcinogen Risk Assessment.
\.
In assessing the cancer hazard posed by exposure to DCM,
primary consideration was given to the evidence of
carcinogenicity from the NTP's animal studies (1985). The NTP
carcinogenesis bioassays clearly demonstrate that DCM is
oncogenic in two species of laboratory animals, rats and mice,
exposed at different dose levels via the primary route of human
exposure to DCM, inhalation.
In the mouse bioassay, DCM induced a dose-dependent,
statistically significant increase in liver and lung adenomas and
t*to • *
carcinomas in male and female mice exposed through inhalation for
a lifetime at concentrations of 2000 or 4000 ppra. Tumor
incidences were as follows: at 2000 ppm,' 30/48 female mice and
27/50 male mice developed lung tumors; 16/48 female mice and
24/49 male mice developed liver tumors. At 4000 ppm, 41/48
female mice and 40/50 male mice developed lung tumors; 40/48
female mice and 33/49 male mice developed liver tumors.
In the rat bioassay, DCM induced a statistically significant
increase in benign mammary gland tumors, of a type not expected
-------
V
,0^
,.c°
^
tf #
<*\
^'^
~ <*
U° O ^
U0° j
•cP'
"A
• »''
?
^. CO
^v^;
0-°
?e^ ^A
^V^ Xk.»*
^^ e
* A.O
«?t
6-*
t.»
O*
-------
to progress to malignant tumors (McConnell et al., 1986), at the
two highest doses in female rats exposed, at 1000, 2000, or
4000 ppm. Male rats developed mammary gland fibroadenomas at
4000 ppm, but only at a marginally significant rate. The NTP
interpreted their study as showing clear evidence of animal
carcinogenicity, and data from the NTP bioassay on mice are the
basis of the regulatory agencies' estimates of human risks at
expected human exposures.
A study of Syrian golden hamsters exposed to DCM at inhaled
doses of 500.to 3500 ppm was negative, but several chronic
studies of mice and rats, including inhalation studies by Dow
Chemical Company (1980, 1982) and a drinking water study by the
National Coffee Association (NCA) (1982 a, b; 1983), reported an
increase in tumors in rats and mice at sites corresponding to the
sites observed in the NTP bioassay. One of the Dow studies
*
(1980) (inhalation at 1500 to 3500 ppm) reported an increase in
salivary gland sarcomas in male rats. These tumors have hot been
repeated in other studies. Results of the DOw and NCA studies,
conducted at doses below those used in the NTP bioassays, were
not statistically significant, with the exception of"the salivary
gland tumors in male rats.
Based on an estimated risk comparison with the NTP bioassay
data, EPA concluded that despite the lack of statistical
significance, the results of the Dow and NCA studies were not
clearly inconsistent with those of the NTP bioassays. For
example, comparing the NCA and NTP unit risk numbers (estimated
-------
using the multistage model) for liver tumors in male mice, the
95% upper confidence limit (UCL) for the NCA study was estimated
to be 0.78 x 10~3; the UCL derived from data on male mice in the
NTP Study was 0.195 X 10~3 (U.S. EPA, 1985b).
At the time of the 4(f) decision, data on humans exposed to
DCM in the workplace were considered to be inadequate for judging
carcinogenic potential. Data from two epidemiologic studies did
not show evidence of a significant increase in deaths from lung
or liver cancer in exposed workers, but these studies had
insufficient statistical power to detect increased risks as
predicted using the upper-bound estimate derived from the NTP
bioassay on mice.
Based on the evidence, EPA concluded that DCM should be
classified as a probable human carcinogen, group B2. This
classification signifies that evidence of animal carcinogenicity
as provided by the NTP bioassays is sufficient, but data from
human studies are inadequate. CPSC, FDA, and OSHA, after
reviewing the DCM database, came to similar conclusions.
In response to EPA's 4(f) announcement in 1985 and the
initiation of investigations by CPSC, OSHA, and FDA, a number of
comments and studies were submitted to the federal agencies
advancing reasons why the results of the NTP bioassay on DCM in
rats and mice should not lead to the conclusion that DCM presents
a high risk to humans. The major criticisms of the preliminary
assessments suggest that (1) current DCM risk estimates
overestimate risks to humans because they ignore species
-------
differences in metabolism and pharmacokinetics; or (2) the
carcinogenic response shown by mice is unique to that species,
i.e., the mechanism by which DCM causes cancer in mice is not
expected in humans. .
Addressing these criticisms calls for a brief review of DCM
metabolism. DCM is metabolized in mice, .the species which showed
a clear carcinogenic response, by two routes; one mediated by the
cytochrome P-450 oxidative system [often referred to as the mixed
function oxidase (MFO) pathway], and the.other by the
glutathione-S-transferase system (also known as the GST pathway).
Both pathways may be active in mice at low doses, but at higher
doses the MFO pathway becomes saturated and the metabolic load is
increasingly shifted to the alternative GST pathway. Recent
studies (CEFIC, I986e) indicate that the GST pathway is less
active in rats, hamsters, and humans than in mice.
Arguments against the. conclusion that DCM presents a risk to
humans take the position, in general, that the carcinogenicity of
DCM is due to reactive metabolites produced by the GST metabolic
pathway, and that this pathway is significantly active only
following saturation of the MFO pathway, i.e., only at high
doses. Further, the GST pathway is assumed to be the sole
carcinogenic pathway,and to be far less active in humans than in
mice, the test species in which malignant tumors have been
observed. Finally, some hypothesize that the metabolites of the
GST pathway are not reactive with DNA, but initiate cancer in
mice through some alternative mechanism such as specialized cell
8
-------
toxicity or increased, cell turnover, events unlikely to occur at
low doses and .possibly irrelevant to humans. One might conclude
from these assumptions that the human risk for developing tumors
from exposure to DCM is very low, that it may not exist below
some threshold level, or that there may be no risk to humans
whatsoever.
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM.OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart,„ NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH
Dr. Scan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr.. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr.. Angelo Turturro , • NCTR/FDA
July, 1987
-------
July, 1987
TABLE OF CONTENTS
SECTION
PAGE
ACKNOWLEDGMENTS , ii
TABLE OF CONTENTS iii
EXECUTIVE SUMMARY vi
CHAPTER 1 - INTRODUCTION • 1
A. The Recent Use and Certification of FD&C Red No. 3 (R-3) 1
B. Issues Confronting the FDA on R-3 1
C. FDA Commissioner's Charge to the Peer Review Panel 2
D. Some Major Considerations in the Panel's Evaluation 3
E. Conclusion 6
CHAPTER 2 - CHEMICAL DISPOSITION AND METABOLISM 7
A. Chemistry " 7
Figure 1 7
B. General Considerations for Chemical Disposition Studies 8
C. Studies of the Disposition of R-3 in Rats 9
1. Study 1: Oral Administration of R-3 to Male and Female Rats 9
la. Study 1 Phase I - Excretion Balance of Orally
Administered R-3 in Rats. 10
Ib. 'Study 1 Phase II - Absorption and Distribution of
Orally Administered R-3. , ' 10
2. Study 2: Intravenous Administration of R-3 to Male Rats 12
2a. Study 2 Phase I - Excretion Balance of Intravenously
Administered R-3 in Male Rats. \ 12
2b. Study 2 Phase II - Distribution and Blood Clearance of
Intravenously Administered Rr3 in male Rats. . 13
3. Discussion of Studies 1 and 2 in Rat . 13
D. Studies of Disposition in Humans 15
1. Study 3: Oral Administration of R-3 to'Humans . 15
2. Study 4: Oral Administration of 131I-R-3 to Humans ._ 15
E. Summary . 16
CHAPTER 3 - SHORT-TERM TESTS FOR GENETIC TOXICITY 18
C '
A. Mutagenicity . . 18
B. DNA/Damage Repair 19
C. Chromosomal Damage . 19
D. Transformation 20
E. Discussion of Test Results ,20
F. Activity of Metabolites and Contaminants 21
G. Significance of Results 22
iii
-------
July, 1987
CHAPTER 4 - ANIMAL TOXICITY STUDIES
A. Introduction
B. Long-term Carcinogenicity Studies
1.
2. *
3.
4.
Rats
Mice
Gerbil
Dog
C. Reproductive Effects
1. Rat
2. Rabbits
. D. Behavioral Effects
E. Analysis
Tables 1-3 '
CHAPTER 5 - THYROID ANATOMY, FUNCTION, AND PATHOLOGY
A. Introduction
B. Thyroid Anatomy
C. Thyroid Function . ,
D. Thyroid Hormone Production
Figure 2
E. Pathology of Alterations of Growth of Follicular Epithelium
F. TSH and Oncogenesis
G. Effects of R-3
1. Pathology Findings in Studies with R-3 in Rats
2. Discussion of the Pathology and Toxiciology Findings
3. Changes in Thyroid Economy
H. Other Possible Mechanisms
1. Binding to Brain
2. Binding to the Pituitary
3. Direct Effect on Follicular Cells '.
4. Increase in Sensitivity to Iodide
5. Binding to TBG and TBPA in Humans
I. Summary
Tables 4-11
CHAPTER 6-- EXPOSURE AND TOXICITY CONSIDERATIONS RELEVANT TO
RISK ESTIMATION
A. Introduction
B. Pharmacokinetics
C. .Hormonal Effects of R-3
1. Human Studies =
2. Analysis
Pathology
Maximum Tolerated Dose in a Chronic Toxicity Bioassay
No-effect Levels of R-3
Relevance of Animal Toxicity to Humans
Human Exposure to- R-3
Table 12 . .
I. Weight-of-Evidence for Carcinogenicity
J. Summary
24
24
24
24
25
26
26
26
26
27
27
27
28
33
33
33
33
34
35
36
37
39
39
40
41
44
44
44
45
45
45
46
47
55
55
56
56
56
57
61
61
63
63
66
67
70
72
.iv
-------
July, 1987
CHAPTER 7 - ISSUES IN QUANTITATIVE RISK ESTIMATION
A. Introduction
B. Animal Tumor Data
1. Follicular Cell Tumors
Tables 13-15
2. C-Cell Tumors
3. Summary
C. Dose-Response Models /AriTv
D. No-Observed-Effect-Level (NOEL) and Acceptable Daily Intake (ADI)
E. R-3 Risk Estimates ,
F. Discussion
Tables 16,17
Figures 3 and 4
CHAPTER 8 - RISK CHARACTERIZATION
A. Introduction
B. 'Characterization of Assumptions
C. Analysis
D. Conclusions
ADDENDUM - SUGGESTED STUDIES
REFERENCES
APPENDICES ' . ' '
APPENDIX A - Charter, Agenda, and. Minutes of the Meetings
held by the Panel
73
73
74
74
75
78
78
78
79
80
82"
84
86
87
87
88
94
96
98
102
Al
APPENDIX B - Administrative Record of Panel materials
Bl
-------
B.2
HAZARD IDENTIFICATION
Attribute 2 The report highlights critical aspects of data quality.
SOURCE Case Study D. Formaldehyde (Pages 1-7 to 1-14).
Note Contextually, this excerpt was preceded by a treatment of non-
cancerous effects and followed by a discussion of additional cancer-
related information.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S, Environmental Protection Agency
-------
1.2.1. Studies of Humans
The EPA has examined 28 epidemiologic studies relevant to
formaldehyde. Three of these studies, two cohort* (Blair et al.,
1986; 1987 in press; .Stayner et al., 1986) and one case-control2
(Vaughan et al., in press), were well conducted and specifically
designed to detect small to moderate increases in formaldehyde-
associated human risks. Each of these three studies observed
statistically significant associations between respiratory site-
specific cancers and exposure to formaldehyde or formaldehyde-
containing products. These associations are noteworthy since
during, inhalation, tissues in the nose, nasal sinuses, buccal
cavity (mouth), pharynx,3 and lungs come into direct contact with
formaldehyde. In each of the above three studies, the
populations studied were also undoubtedly exposed to other
chemicals and these exposures may have contributed to the
observed increases in cancer risk. Only the study by Vaughan
et al. (1986a,b) controlled for smoking and alcohol consumption.
A cohort study follows a group of exposed individuals for a
specified time period and measures the incidence of site-specific
deaths. The observed number of site-specific deaths which
occurred in the time period are compared to the number of site-
specific deaths which would be expected based on jnortality rates
of a standard population. - ,
2 A case-control study identifies cases with the .disease of
interest and controls who do not have the disease. The cases and
controls are compared with respect to past exposure.
^ The pharynx is the passage between the nasal cavity and the
larynx. The nasopharynx, hypopharyrtx, oropharynx, and
laryngopharynx comprise the pharyngeal region.
1-7
-------
The Blair et al. (1986; 1987 in press) cohort study observed
significant excesses in lung and nasopharyngeal cancers among U.S.
workers occupationally exposed to formaldehyde at 10 industrial
sites. Blair et -al. (1986), however, argued that the lung cancer
excesses provided little evidence of an association with
formaldehyde exposure since the lung cancer risk did not increase
consistently with either increasing intensity or cumulative
formaldehyde exposure. EPA, after reviewing the data, has
concluded that the, significant excesses in total lung cancer
mortality, in analyses either with or without a latency period
equal to or greater than 20 years, and together with nasopharyngeal
cancer mortality among formaldehyde-exposed workers are meaningful
despite the lack of significant trends with exposure.
Misclassification of exposure (or lack of specificity between
exposure categories) and categorization of deaths into four
exposure levels which lowers the power to detect small increases in
risk, may have accounted for the observed lack of a significant
dose-response relationship. -The significance of these findings is
*
reinforced by the fact that the site of the tumors seen in humans
(the nasopharyngeal region) is similar to that seen in animals.
Blair.et al. (1987) performed further analyses of the
nasopharyngeal cancers regarding exposure to formaldehyde and
particulates. For those workers with particuiate exposure,•the
trend between increasing nasopharyngeal risk and increasing
cumulative formaldehyde exposure was not statistically significant,
however, the authors concluded that formaldehyde and particulates
appeared to be a risk factor for nasopharyngeal cancer.
-------
The Stayner et al. (1986) cohort study reported statistically
significant excesses in mortality from buccal cavity tumors among
formaldehyde-exposed garment workers. The standardized mortality
ratio (SMR), a ratio of the observed number of deaths to an age-
adjusted number of deaths expected in the group, was highest among
workers with a long duration of employment (exposure) and follow-
up period (latency). A significant excess in deaths from cancer
of the tonsils was also reported, but there were too few-deaths to
examine any trends with exposure.
Results from the case-control study by Vaughan et al.
(19S6a,b) showed a significant association between nasopharyngeal
cancer and having lived 10 or more years in a "mobile home".
Persons for whom this association was drawn had lived in mobile
homes that were built in the 1950s to 1970s. This study also
reported significant associations between sinonasal cancer and
orohypopharyngeal cancer and exposure to resins, glues, and
adhesives (SAIC, 1986).4 No significant trends were found in
cancer incidence at any of these sites with respect to
occupational formaldehyde exposure; however, the risk estimates
> . '
for the highest exposure level and cancers of the orohypo- and
naso-pharynx appeared elevated. As stated earlier, however, this
population, like the two previously discussed, .was also
undoubtedly exposed to other chemicals which may have contributed
to the observed increases in cancer risk.
^Several residential and occupational characteristics were _a
priori selected as likely surrogates for formaldehyde exposure,
Among these were mobile home residency and occupational resins,
glue, and adhesive exposure.
1-9
-------
'EPA previously had reviewed 25 other epidemiologic studies.
These studies had limited ability (lower power) to detect small to
moderate increases in formaldehyde-related risks due to (1) small
sample sizes; (2) small numbers of observed site-specific deaths;
and (3) insufficient follow-up. Even with these potential
limitations, six of the 25 studies (Acheson et al., 1984a; Hardell
et al., 1982,; Hayes et al. , 1985; Liebling et.al.', 1984; Olsen et
al., 1984; Stayner et al., 1985) reported significant associations
between excess site-specific respiratory (lung, buccal cavity, and
pharyngeal) cancers and exposure to formaldehyde.'
The Olsen et al. (1984), Hayes et al. (1986), 'and Hardell et
i
al. (1982) studies reported significant excesses of sinonasal
cancer in individuals who were exposed to both formaldehyde and
wood-dust, or who were employed in particleboard manufacturing
where formaldehyde is a component of the resins used to make
particleboard. Only the. Hayes et al. (1986) and Olsen et'al.
(1984) studies controlled for wood-dust exposure; the detection
limits in both studies,.however, exceeded corresponding expected
excesses in the incidence of sinonasal tumors and, therefore, no
significant excesses were likely to have been observed.
The Acheson et al. (1984a) study conducted in the United
Kingdom supports the results of Blair et al. in that, when
compared to mortality rates of the general population, significant
excesses in mortality from lung cancer were observed in one of six
formaldehyde resin producing plants in England. A trend of
borderline significance with dose'was observed for this one
plant. Acheson et al. concluded that the increases in mortality
1-10
-------
from lung cancer were not related to formaldehyde exposure since
the elevation and trend were not statistically significant when
*»
compared with local lung cancer rates. EPA believes that the
risks and trends from analyses using local lung cancer rates as
the comparison risks appeared sufficiently increased for
corroborative use. *
The remaining two studies reported-significant excesses- of
buccal cavity cancer among garment workers in 3 plants (Stayner e.t
al., 1985) and excesses of buccal cavity and pharyngeal cancer
among formaldehyde resin workers in 1 plant (Liebling et al.-/
1984). Portions of the Liebling et al. (1984) and Blair et al.
(1986, 1987) studies overlapped as did portions of the two Stayner
et al.* (1985; 1986) studies. However, the non-overlapping
portions and improved design of "the more recent studies (i.e.,
Blair et al. 1986, 1987; Stayner et al. 1986) reinforce the
conclusions of the earlier studies.
Analyses of the remaining 19 epidemiologic studies have
indicated the possibility that observed leukemia and- neoplasms of
the brain and colon may be associated.with formaldehyde
exposure. The biological support for such postulates, however,
has not yet been demonstrated.
l-ll
-------
Based on a review of these studies, EPA has concluded that
there is "limited" evidence to indicate that formaldehyde may be a
carcinogen in humans.5 Nine studies reported statistically
significant associations between site-specific respiratory
neoplasms and exposure to formaldehyde or formaldehyde-containing
products. This is of interest 'since inhalation is the primary
route of exposure in humans. Although the common exposure in all
of these studies was formaldehyde, the epidemioiogic evidence is
categorized a's "limited" primarily due to possible exposures to
other agents which may have confounded the findings of excess
cancers.
1.2.2. Studies in Animals
. **
The principal evidence indicating that formaldehyde causes
cancer in animals comes from studies conducted by the Chemical
Industry Institute of Toxicology (CUT) (Kerns et al., 1983) and'
those by Albert et al. (1982) and Tobe et al. (1985). The CUT
study was a well, conducted, multidose inhalation study in rats and
mice. In this study, a statistically significant increase in
malignant tumors (i.e.,*squamous cell carcinomas) was seen in the
nasal cavities of male and female rats dosed at 15 ppm. In
addition, a small increased incidence of squamous cell carcinoma,
while not statistically significant, was seen in male mice.
* EPA's Guidelines for Carcinogen Risk Assessment define limited
evidence of carcinogenicity in humans as indicating that "...a
causal interpretation is credible, but that alternative
explanations, such as chance, bias, or confounding, could not
adequately be excluded."
1-12
-------
Because this type of nasal lesion is rare in mice, these data can
be considered to have biological importance. Benign tumors (i.e.-,
polypoid adenomas) were seen in male rats .in the CUT study at all
dose levels and in female rats, exposed to 2 ppm of formaldehyde.
Notably, the dose-response curve for the benign tumors in this
study was not linear; the tumor incidence was highest at 2.0 ppm
and decreased at higher doses.
Tobe et al. also observed a statistically significant
increase in the numbers of squamous cell carcinomas in the same
strain of male rats as was used in the CUT study. Albert et al.
reported a statistically significant elevation of the same
• ' • „ .
malignant tumor type in male rats of a different strain. In both
the Tobe et al. and Albert et al. studies benign squamous cell
papillomas were seen. This observation was in contrast to the
CUT study in which polypoid adenomas were the only benign tumors
observed. Hamsters have been tested in long-term inhalation
studies (Dalbey, 1982) but no increased incidence of tumors was
seen in formaldehyde-treated animals. However, deficiencies in
the study design and poor survival.limit the interpretation of the
results from these, studies.
Additional studies in animals that indicate an association
between exposures to formaldehyde and cancer are those by Dalbey
(1982) in which formaldehyde enhanced the production of tumors
induced by a known animal carcinogen (i.e., diethylnitrosamine);
Mueller et al. (1978) in which formalin (a water solution of
formaldehyde) produced lesions in the oral mucosa of rabbits which
showed histological features of carcinoma in situ; and studies by
1-13
-------
Watanabe et al'.' (1954; 1955) in which injections of formalin and
hexamethylenetetramine (from which formaldehyde is.-liberated in
vivo) produced sarcomas (malignant tumors) and one adenoma (benign
tumor) at the site of injection.
Based upon a review of these studies, EPA has concluded that
there is "sufficient" evide'nce of carcinogenicity of formaldehyde
in animals by the inhalation route.6 This finding is based on the
induction by formaldehyde of an increased incidence of a rare, type
of malignant tumor (i.e., nasal squamous-cell carcinoma) in both
sexes of rats, in multiple inhalation- experiments, and in multiple
species (i.e., rats and mice). In these long-term laboratory
studies, tumors were'not observed .beyond the initial site of nasal
contact nor have other.mammalian in vivo tests shown effects at
distant sites. • •
1.2.3. Additional Supportive Evidence . .
Other relevant information which is considered in carcinogen
assessments include results from short-term tests designed to
measure effects of a chemical on DNA. Tests for point mutations,
numerical and structural chromosome aberrations, DNA
damage/repair, and in vitro cell transformation provide evidence
for the potential mechanisms of carcinogenicity. A battery of
6 EPA'3 Guidelines for Carcinogen Risk Assessment define
sufficient evidence of carcinogenicity from studies in
experimental animals as indicating that ".. .there is ^increased
incidence of malignant and benign tumors: (a) In multiple
species or strains; or (b) in multiple experiments (preferably
with different routes of administration or using different dose
levels); or (c) to an unusual degree with regard to incidence,
site or type of tumor, dose-response effects, as well as
information from short-term tests or on chemical structure..
1-14
-------
B.2
HAZARD IDENTIFICATION
Attribute 3 A weight-of-the-evidence approach is presented for judgment
as to the likelihood of human carcinogenic hazard and
includes a clear articulation of the rationale for the position
taken.
SOURCE Case Study J. Red Dye'No. 3 (Pages 70-72).
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED'NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH'
Dr. Stan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NCTR/FDA
July, 1987
-------
July, 1987
I. Weight of Evidence for Carcinogenicity
Relative to many chemicals, there is substantial information on the
exposure and toxicity of R-3. Interpretation of these data depends on the
assumptions used and the weight given to various factors in their analysis.
The Panel has come to the following conclusions:
a. There is some evidence that certified R-3 has an oncogenic effect
on rats, i.e., it increases the incidence of benign thyroid folli-
cular adenomas.
b.
There is equivocal evidence that certified R-3 is carcinogenic in
male rats. The effect is only seen by one reviewer, and it is only
marginally significant. In addition, interpretation of the
important response 4% dose level is complicated by evidence that
the MTD may have been exceeded. This may have triggered a
mechanism not occurring at lower doses and, thus, not at human
exposure levels.
Further, the above effects of certified R-3 have been observed in
two-generation rat studies and have not been reproduced in other
studies in rat and other species. While some of these studies were
inadequate in design,, at least two negative mouse studies .were of
sufficient size and duration to be adequate one-generation studies.
There is some suggestion that R-3 is not a tumorgen in a one-gener-
ation test. Lack of reproducibility in the tumorigenic effect
weakens confidence that R-3 is a human tumorgen.
c. Short-term tests of R-3 have been negative for mutagenicity in bac-
teria or mammals. Thus, there is evidence that tumor formation by
R-3 is not through a mechanism directly affecting the genome.
d. There is some evidence that the tumorigenic and hormonal effects of
R-3 are exerted by its iodine component, either molecularly bound
or present as an impurity. The reported R-3 induced alterations in
thyroid economy, with the exception of the effect on the T, - T,
balance, and s.ome of the results in short-term tests, are well
known properties of iodide. There is insufficient evidence that
the tumorigenic effect of R-3 is attributable to R-3 proper".
70
-------
July, 1987
e. There is some evidence that, in rats, the tumorigenic effect of
certified R-3 is mediated through the hypothalamus-pituitary-
thyroid axis in which TSH has a central role.
f. There is insufficent evidence that the mechanisms controlling the
hypothalamus-pituitary-thyroid axis in man are qualitatively
different from those" in rats. Although the human baseline TSH
level is about 1/10 of that in rats and although TSH levels in
males are not higher than in females, (unlike rats), available data
on quantitative differences in the response to R-3 exposure are
insufficent to allow conclusions be drawn pertaining to a differ-
ence between humans and rats in the sensitivity to R-3 and iodine.
g. There is some evidence that thyroid cancer in humans may not be
related to TSH levels. The higher incidence of tumors in women
compared to man, different than in rats, also suggests a different
mechanism than that in rat.
h. There have been no epidemiologic studies on R-3'. However, the low
incidence of thyroid cancer in combination with the low tumorigenic
potency of R-3, as predicted by the rat studies, would require a
descriptive study size far beyond what is reasonably feasible. It
is unknown' whether there is a sufficient number of people 'exposed
to high dose levels to conduct an occupational follow-up study.
Because of the low potency, more cancer cases would be .required for
case-control studies than are usually available.
i. There is some evidence that the human exposure level from R-3 in
food and drugs, estimated to be 1.41 mg/d, is below the NOEL of 20
to 60 mg/d for an increase in TSH levels in humans. However, if
TSH does not play a role in the genesis of human thyroid tumors, it
can be questioned what the value for tumorigenesis is of a human
NOEL for a TSH effect.
j. In summary, there is limited evidence for certified R-3 to be
*
oncogenic in humans•
71
-------
• • • July, 1987
J. Summary
The Panel concludes from the preponderance of available evidence that
R-3 probably acts on the hormone economy by processes occurring outside the
thyroid, with no evidence for a direct mechanism. Evidence for an indirect
or secondary mechanism includes the demonstrated absence of relevant
genetic toxicity, the association between elevated TSH and tumorigenesis,
and the association between R-3 exposure and TSH elevation. Following the
Panel's ad-hoc definition, R-3 is, thus, considered to act predominantly
through a secondary mechanism. For the estimation of the risk, it has been
shown that the intestinal absorption of R-3 is poor, and that the absorp-
tion rate in humans approximately equals that in rats. For humans, a NOEL
is taken to be 20 to 60 mg/day, although the data for these values are of
poor quality. The NOEL is not necessarily equivalent to a threshold dose.
For the U.S. population as a whole', the average per capita availability of
R-3 in food products and drugs is estimated to be 1.4 mg/day.
72
-------
-------
B.2
HAZARD IDENTIFICATION
AttMuUte *l The report identifies research that would permit a more
confident statement"about human hazard.
SOURCE Case Study J. Red Dye No. 3 (Pages 98-101).
' Note The excerpt from the report outlines suggested studies on mechanism
of action, which would be useful in clarifying both hazard and dose- •
response extrapolation.
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF .CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH
Dr. Scan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NCTR/FDA
July, L987
-------
July, 1987
ADDENDUM - SUGGESTED STUDIES
Since the exact mechanism(s) of action of R-3 on the rat thyroid which
could result in activation of secondary mechanisms such as increased TSH
and T^ levels and decreased Tg levels are not known, additional studies
which would be useful to understand .the mechanisms of R-3 oncogenicity and
help in extrapolation are outlined here.
1.
2.
131
A radioactive scan of I labeled R-3 to study the possible
localization in thyroid should be carried out at a higher dose
level than in .the studies performed by the petitioner. The
material should be. pure, however, with no contaminating radio-
active iodine. Since there is dehalogenation, appropriate
controls should be used to distinguish R-3 localization and iodine
localization. More advanced techniques, such as nuclear magnetic
resonance and deuterium-labeled compounds, should be considered.
The effect of the sex hormones on the thyroid should be assessed
since there are' differences in the two sexes. For instance, in
males the incidence' of thyroid neoplasms was significantly in-
creased. Animal experiments provide support for the concept of
sex-influenced development of thyroid tumors. In studies of
irradiation-induced thyroid tumors in rats, it was reported that
more than twice as many male rats developed thyroid adenomas 'and
carcinomas as females (161). In other studies where only male
rats were used, X-ray or fadioiodine induced, tumor incidences were
enhanced when serum TSH .levels were elevated by treatment of '
animals with.methylthiouracil or propylthiouracil (162). Hoffman
et al. (163) found that the incidence of irradiation induced thy-
roid tumors in rats could be significantly reduced by castration
prior to irradiation and the effect of castration can be reversed
by administration of testosterone. Although the results varied
with the age at administration, testosterone increased the inci-
dence of tumors up to 100%. The authors also showed that TSH
levels in both irradiated intact^ and castrated rats treated with
testosterone were higher than in untreated intact males. The
98
-------
July, 1987,
effect of testosterone on the induction and/or progression of rad-
iation-associated thyroid tumors may be through a direct or
indirect mechanism. It has been shown that ,TSH may enhance thy-"
roid tumor development by growth-promoting actions, therefore,
androgens may indirectly modulate growth of radiation-induced
thyroid tumors through TSH. Since, in the R-3 studies, TSH levels
were suspected of being elevated, it would be important to Jcnow
whether or not there was a testosterone-mediated effect for the
sex-specific carcinogenic action. There are similar considera-
tions for estrogen effects.
3. Further studies using hypophysectomized animals could be very
useful in elucidating the mechanism. If hypophysectomized animals
treated with a tumorigenic dose of R-3 did not develop tumors,
this would be further support for an indirect action using TSH.
Recent studies are exploring aspects of this.
4. A chronic bioassay where R-3 is given at a tumorigenic dose and in
which some agency restores T- levels would be definitive proof of
mediation through R-3. It is our understanding a study similar to
this is underway.
5. The obvious occurrence of abnormal lysosomes in R-3 treated ani-
mals needs to be evaluated in a quantitative fashion, since it may
form the basis of an inferred dose-response relationship. Such
information 'would not only be useful in understanding the mechan-
ism of action of R-3, but it may also provide an alternative to
tumor incidence in the calculation of the risk.
6. One area totally ignored is the effect of R-3 on the fetus. In a
multi-generation study, the critical effect may be on the fetus,
especially since R-3 influences reproductive endpoints. A series
of studies should characterize this important area, e.g., studies
of fetal pathology and physiology.
7. A useful area would be to investigate the longitudinal effect of
99
-------
. July, 1987
R-3 on human hormone levels, with particular emphasis on
randomizing dose levels, sampling blood at the same time of the
day, and prescreening for pre-existing thyroid pathology. The
dose levels of interest are 50, 100, 250, and 1000 mg/d for up to-
one month. Hormone levels should be determined at weekly
intervals. The actual assays should be deferred until they can be
done in one batch. Compounds to be tested are commercial R-3 and
iodide at dose levels comparable to what would be available from
the doses of commercial R-3.
8. It is of crucial importance to assess interspecies differences in
sensitivity to R-3 in a quantitative fashion. Tests of two weeks
should be done in humans and rats in a strictly standardized way
with regard to time of the day, day of blood sampling, randomizing
of dosing, exclusion or control of iodide in the diet from other
sources. For humans, compounds to be tested and dose levels are
as in proposal 7. For rats, dose levels should be the same on a
mg/kg dose basis, but should also include .an extra dose level of
70 and 350 mg/kg-d, which is comparable to 5000 and 2500 mg/d in
humans, to account for possible greater sensitivity of the rat.
9. It would provide important information to conduct a study
evaluating the predictive ability of various procedures for
extrapolating risk, viz., biologically-based mathematical models
and the NOEL approach. There is currently a substantially im-
proved data base on human carcinogens and human risk of cancer at
high doses in animals and low "and moderate doses in humans.
*
10. Conduct a large scale epidemiologic study among people occupa-
* tionally exposed to high levels of iodine or R-3. Exposure to R-3
may include skin contact, since the absorption of R-3 through the
skin is of the same order of magnitude as through the gut mucosa.
Preferably, the study design should be that of a case-control
study nested in a retrospective cohort study.
11. Conduct a 'new survey of food" consumption pattern, taking care that
100
-------
... July, 1987
representative samples are taken of the food products (including
iodine-containing commodities), named by the interviewees, for
testing on the content of R-3 and iodine. The analysis .should
cover amount of food consumed, variety of products within a given
category, and amount of R-3 and iodine ingested.
101
-------
B.3
DOSE-RESPONSE
EVALUATION
AtfriDUIe I Valid data sets and plausible models for high-to-low dose and
interspecies extrapolation are presented in dose-response
modeling. •
SOURCE Case Study H. Methylene Chloride (Pages 71-87).
Note See General Attribute 2 in this Appendix./
SOURCE Case Study D. Formaldehyde (Pages 1-23 to 1-28).
Note The excerpts illustrate the consideration of several data sets. In this
case, only one model was selected for extrapolation.
SOURCE Case Study A. DEHP (Pages 111-145).
Note This report illustrates all three of the attributes for dose-response
evaluation. The report is included in its entirety to. provide contextual
background. The report is especially illustrative of consideration of
data sets of different types (bioassays, genotoxicity, species vari-
ations, and possible mechanisms). Several modeling approaches are
presented, along with strengths and weaknesses. The authors present
a range of risk estimates. The authors were unable to indicate a
quantitative central tendency of the potency estimates but indicated
qualitative considerations in weighing the range of risks.
Some other pertinent portions are found on Pages 121-127, 129-130,
and 138.
-------
DOSE-RESPONSE
EVALUATION
Attribute 1 (continued)
SOURCE Case Study B. TCDD (Pages v to xvi).
Note See Hazard Identification Attribute 1 in this Appendix.
SOURCE Case Study C. Ethylene Oxide (Pages'9-23).
Note This report illustrates the evaluation of several data sets and models
with particular emphasis on time-to-response modeling.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
- from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
less intense exposure. In addition, the cellular effects are
expected to be reversible once formaldehyde exposure is
eliminated.
1.4.2. Cancer Dose-Response Assessment
In principal, data from studies of humans are preferred for
making numerical risk estimates. However, as is" often the case,
the available epidemiologic data on formaldehyde were not suitable
for low dose quantitative cancer risk estimation, mainly because
of a lack of adequate exposure information in the studies.
Accordingly, results from studies in animals were used to estimate
low-dose human cancer risk. In addition, even though the
epidemiologic studies were not suitable for quantifying a dose-
response curve, those studies with observed statistically elevated
cancer risks provided some support for the animal-based predicted
upper bound risk. This comparison, while yielding valuable
information to the assessment, should be viewed with caution since
exposure levels in these epidemiologic studies were subject to
some variation. .
•
1.4.2.1. Selection of Data
Of the carcinogenicity studies with formaldehyde in animals,
EPA has selected the CUT study in rats as the best study for
cancer risk extrapolation. This study was well,designed, well
conducted, included multiple doses, and .used a large number of
animals par dose. ° • _
Each of the remaining inhalation studies suffered from
various limitations which precluded their use in quantitative risk
assessment. The CUT study in mice showed a limited tumor
1-23
-------
response oaly at. the highest dose of formaldehyde, while the
Albert et al. (1982) study had only a single formaldehyde-exposed
group. -Although the Tobe et al. (1985) study contained multiple
dose groups, a tumor response was seen only at the highest dose,
and the number of animals per group was relatively small. Lower
cancer risks than those estimated from the CUT study in rats
would have been predicted had the Agency been able to use the CUT
study in mice for risk extrapolation, while higher cancer risks
would have been estimated had the results from the Tobe et al.
(higher by a factor of ten) or Albert et al. studies been used.
Two types of nasal tumors were obs-erved in the CUT study in
rats, squamous cell carcinomas (malignant tumor) and polypoid
adenomas (benign tumor). EPA's risk assessment relied only on the
malignant tumor data of the.CUT study to predict human cancer
risks because: -(1) the malignant tumor response in formaldehyde-
exposed rats was definite and unequivocal in both .males and
females, whereas the frequency of benign tumors reached
.>>
statistical significance only when the incidences, in males and .
females were pooled; (2) the malignant tumor response in the CUT
study in rats showed an increasing dose-related trend, while-the
benign tumor response showed a decreasing trend; (3) unlike the
benign tumor response which was not confirmed by the other rat
inhalation studies, similar malignant tumor types were found both
in all rat and mouse inhalation studies with formaldehyde and in a
study of acetaldehyde, a close structural analogue of formaldehyde.
The appearance of benign nasal tumors in rats following
inhalational exposure .to formaldehyde in the CUT study
1-24
-------
contributes to the qualitative weight-of-the-evidence that
formaldehyde may pose a carcinogenic hazard, but because of the
attendant uncertainties they were not included in the
quantitative estimate of human cancer risk. Had the Agency
chosen to use the benign tumor response in the quantitative
estimation of human cancer risk, the predicted values would have
been about ten-fold greater than those reported in Section 1.4.3
using the malignant tumor response alone.
1.4.2.2. Choice of Mathematical Extrapolation Model
Since risks at low exposure levels cannot be measured
directly either by experiments in animals or by epidemiologic
studies, a number of mathematical models have been developed to
it
extrapolate from results at high doses to expected responses at
low doses. The Office of Science and Technology Policy (OSTP)
published principles on model selection which states that:
"No single mathematical procedure is recognized as the most
appropriate for low dose extrapolation in carcinogenesis.
When relevant biological evidence on mechanism of action _.
exists, the models or procedures employed should be
consistent with the evidence. When data and information are
limited, however, and when much uncertainty exists regarding
the'rwchanisra of carcinogenic action, models or procedures
which incorporate low dose linearity are preferred when
compatible with the limited information."
Data relevant to selecting a model for extrapolation of
cancer risk associated with exposure to formaldehyde were
reviewed? some of the biological information support a direct
1-25
-------
relationship between exposure and carcinogenicity while other
data are consistent with a non-linearrresponse. The Agency,
however, did not conclude that enough information was available
to propose an extrapolation model for formaldehyde that was
different from the one recommended by the OSTP and EPA's
Guidelines for Carcinogen Risk Assessment (i.e., linearized
multistage procedure). The Agency has presented various other
models for comparative purposes.
Biologic evidence on mechanism of action, which can,aid in
model selection, largely is inferred from a variety of types of
J"
studies. These are limited'and suggestive of several mechanisms
i
for formaldehyde. Mutagenicity studies suggest a direct
relationship (i.e., a linear one) between exposure to
formaldehyde and carcinogenicity. Thus, the ability of
formaldehyde to cause point mutations, chromosome aberrations and
DNA damage is consistent with the chemical's ability to initiate
the carcinogenic reaction.
The steep curvilinearity of the rat nasal carcinoma dose-
response data in the CUT study in rats suggests, however, that
cancer development is greatly accentuated above ce'rtain con-
centrations. In keeping with this observation are the results of
experiments.on DNA synthesis and cell proliferation following *
short-term formaldehyde exposures and the conversion of normal
mucosal cells to squamous cell epithelium (squamous metaplasia)
following longer exposures which indicate that certain toxic
effects are only noted above certain formaldehyde
concentrations. Any relationship between cell proliferation
1-26
-------
following formaldehyde exposures and the carcinogenic process is
currently unknown. Likewise, although squamous metaplasia may
represent a step in the formation of squamous cell carcinoma, its
specific role is uncertain. No lesions that,may represent stages
in a continuum between the-squamous metaplasia and carcinoma were
identified in the CUT study. ,
The CUT also conducted molecular dosimetry experiments
attempting to relate ambient exposures to formaldehyde with
tissue-specific levels of formaldehyde-DNA adducts. Use of the
data generated by these experiments in risk extrapolation models
yields lower estimates of risk, sometimes significantly'lower
than use of the experimental doses. The CUT data have been
reviewed by EPA scientists and a review panel of non-government
scientists to determine whether or not they should be used in the
quantitative risk assessment. Both groups concluded that the
study had several shortcomings which preclude its use in
modifying the doses used in quantitative risk assessment, and
they provided three reasons for their conclusion. First, the
experimental methodologies must be validated to assure that the
experimental assumptions were scientifically sound and that the
formaldehyde-DNA-protein complexes were identified properly;
second/ the single intracellular target used in the study may be
inadequate; and third, and perhaps most important, the use of an
acute exposure model in the CUT study may not be appropriate
because chronic, not acute exposure is most relevant to risk
i ' ?
assessment. •
1-27
-------
Different"extrapolation models fit the observed data
reasonably well but there are large differences, among them in the
risks calculated at low doses. EPA's Guidelines for Carcinogen
Risk Assessment state, however, that goodness of fit to the
observed tumor data,by a given model is not an effective means of
discriminating among models. In the absence of compelling
biological evidence on the mechanism of action, as in the case
3. —.
for formaldehyde, EPA's guidelines specify that the linearized
multistage procedure will be used, with the possible presentation
of various other models for comparative purposes. The analysis
showed that of the models examined, only -the one-hit model
produced higher risk estimates (about ten fold higher). .
Studies show that non-human primates and rats respond
similarly to formaldehyde exposure,. Accordingly, an interspecies
scaling factor was not used in the risk extrapolation. This
position was supported by the Consensus Workshop on
Formaldehyde. Consequently, the response of rats and humans was
judged to be the same at equivalent exposure levels and
durations. However, if a conversion factor, such as nasal
surface area, had been used the estimated human cancer risks
would have been about an order of magnitude higher.
1-28
-------
JOURNAL OF THE AMERICAN COLLEGE OF TOXICOLOGY
Volume 4, Number 2, 1985
Maiy Ann Llebert, Inc., Publishers
Assessment of Possible Carcinogenic Risk
to Humans Resulting from Exposure to
Di(2-ethylhexyl)phthalate (DEHP)
D. TURNBULL and J.V. RODRICKS
ABSTRACT .
The purpose of this work was to estimate the degree of risk that might be associated with
human exposure to low levels of the plasticizer di(2-ethylhexyl)phthalate (DEHP). DEHP
is a common component, sometimes at high concentrations, of poiyvinyl chloride (PVC)
plastics and was recently reported by the National Toxicology Program (NTP) to be carci-
nogenic in rats and mice, inducing hepatocellular tumors in both species. This work was
also designed to illustrate an approach to risk assessment that attempts to incorporate all
available biological data. Based on the dose-response data generated by the NTP bioas-
says, we have performed extrapolations of risk to low dose levels using several proce-
dures, including some that incorporate inferences from the available data that shed light
on the likely relationship between dose level and risk at low dose levels. In drawing these
inferences, consideration was given to such factors as genotoxicity, metabolism and phar-
macokinetics, and physiological and biochemical effects of DEHP that might reveal its
mechanism of action. The relative merits of each of the various risk estimates are de-
scribed, based on current understanding of DEHP's mode of biological action. It is con-
cluded that DEHP's mechanism of carcinogenicity in rodents most likely involves its abil-
ity to induce peroxisome proliferation and related enzymatic changes, although other
mechanisms cannot be excluded. If humans and rodents are assumed to be at the same
risk at the same daily dose level of DEHP, application of the various low dose extrapola-
tion models leads to the prediction that the daily dose resulting in a lifetime risk of no
more than 1 in 1 million would be between 1.5 and 791 mg/kg per day, with the most
likely figure being 116 mg/kg per day. If the carcinogenicity of DEHP is dependent upon
its pattern of metabolism, however, it would be inappropriate to extrapolate from rodents
to man without qualification because of the major quantitative differences in metabolism
in rats, mice, and primates, including man. One of the major differences in metabolism of
DEHP between rats and mice and primates is in production of a metabolite whose level
may be an indicator of the level of peroxisomal activity and, hence, if the peroxisome pro-
liferation theory of DEHP carcinogenicity is correct, of carcinogenic risk. However, the
substantial doubt that exists regarding the applicability of rodent carcinogenicity data to
humans must be expressed in qualitative terms.
Environ Corporation, Washington, D.C.
Ill
-------
TURNBULL AND RODRICKS
INTRODUCTION
Di(2-ETHYLHEXYL)pHTHALATE (DEHP), the structure of which is shown in Figure 1 along with
those of its two primary hydrolysis products, is a widely used plasticizer for polyvinyl chloride
(PVC). It has been estimated that 1188 million pounds of plasticizers were used in PVC in 1980, and
of this total, 30% (about 356 million pounds) was DEHP.'" PVC plastics may contain up to 40%
DEHP by weight and are widely used in consumer products, such as imitation leather bags and fur-
CH2CH3
Di(2-ethylhexyl)phthalate (DEHP)
0
C-0-CH0CHCH0CH-_CH.CH,
/ 2. 2. Z
Mono(2-ethylhexyl)phtha.late (MEHP)
CH2CH3
2-Ethylhexanol (2-EH)
FIG. 1. Chemical structures of DEHP, MEHP, and 2-EH.
112
-------
RISK ASSESSMENT OF DEHP
nishings, wallpaper, lawn furniture, rainwear, swimming pool liners, flooring, footwear, children's
toys, containers and tubing for transfusions of blood and blood products.(1> Since DEHP is not
chemically bound within the PVC, there is at least the potential for widespread human exposure to
DEHP as a result of migration out of the plastic. Human exposure to DEHP is of concern particu-
larly in light of the recent report by the National Toxicology Program that DEHP at high dietary lev-
els is carcinogenic in rats and mice.U)
Estimation of the magnitude of human exposure to DEHP is, however, extremely complex be-
cause of the wide range of items that contain DEHP and the uncertainty regarding how much of the
DEHP content of a PVC item to which someone was exposed would reach that individual and how
much would be absorbed. No attempt has been made to estimate human exposure in this paper.
Rather, our purpose is to describe the available data pertinent to the assessment of carcinogenic risk
resulting from exposure to DEHP and to present estimates of risk per unit of dose (or no-observed-
effect levels), which can subsequently be combined with estimates of human exposure to estimate
human risk. :
The conduct of this risk assessment follows the recommendations of the National Academy of Scir
ences.'31 It starts with a critical review and evaluation of the literature pertaining to the carcinogenic
properties of DEHP. This is followed by a review of data that might shed light on the underlying
mechanism(s) of tumor induction in animals. This may assist in dealing with the next two compo-
nents of risk assessment: dose-response evaluation and interspecies extrapolation. The approach to
risk assessment used here is one in which several estimates Tare presented, along with a discussion of
their relative degrees of support based on current understanding of DEHP's biological behavior. We
avoid presenting only worst-case estimates but also attempt to avoid overstating the degree of cer-,
tainty associated with the other estimates presented.
The broad outline of this risk assessment process is presented in Figure 2. Each component of this
process is discussed in relation to available data on DEHP.
REVIIW DATA ON
CARCINOGfNICITY
STRfNCTHS & WEAKNESSES
, . OF ANIMAL DATA
Ml CHAN ISM .OF CANCtR
INDUCTION IN EXPERIMENTAL
ANIMALS
(includes all underlying
processes)
METABOLIC SHIFTS
HIGH-TO-LOW DOSE
Lifetime risk per
unit or avoraqe.
daily lifetime)
exDOSuro
Li Tetimo risk per
unit or average,
da ily Iifettme
exposure
Lifetime risk per
unit of average,
da ily Ii fetime
exposure
THRESHOLD DOSE FOR
CANCER INDUCTION
THRESHOLD MODEL
- EXPERIMENTAL OR
ESTIMATED NOEL
FIG. 2. Broad outline of carcinogenesis risk assessment.
113
-------
TURNBULL AND RODRICKS
* CARCINOGENICITY EVALUATION
Few studies are available that evaluate possible adverse effects of chronic human exposure to
DEHP, and none of these specifically address the possibility of an association between DEHP expo-
sure and cancer. (4"S) The only evidence of earcinogenicity of DEHP comes from the recent NTP bio-
assay.<2) Earlier negative studies involving exposure of smaller numbers of animals to lower doses re-
vealed no significant carcinogenic effect. <7~10) These studies do not conform to current standards for
carcinogenicity bioassays,'11' however, and although they demonstrate that DEHP is not a potent
carcinogen, they were probably insufficiently sensitive to have detected an effect of DEHP of the
small magnitude seen in the NTP bioassay.
In the NTP bioassay,(2> DEHP was fed in the diet for 103 weeks to groups of 50 B6C3F1 mice and*
50 F-344 rats of each sex at each of 2 dose levels, 3000 and 6000 ppm in mice and 6000 and 12,000
ppm in rats. Using the food consumption data presented by NTP, these feeding levels approximately
correspond to the following average doses: 320 and 670 mg'/kg per day in male rats, 390 and 770
mg/kg per day in female rats, 670 and 1300 mg/kg per day in male mice, and 800 and 1800 mg/kg
per day in female mice. In rats, mortality was not affected by treatment, but there was a dose-related
decrease in body weight throughout the study for males at both doses and for females at the high
dose. Food consumption was reduced slightly in all treatment groups.
The incidence of liver tumors increased in treated rats, as shown in Table 1. Using the Cochran-
Armitage test, there was a significant dose-related trend in the incidence of combined carcinoma or
neoplastic nodules in male rats (P = 0.002) and female rats (P < 0.001) and in carcinomas alone (P
= 0.002) and neoplastic nodules alone (P = 0.03) in female rats. Also, by direct comparison using
the Fisher's exacktest, there was a significant increase in the incidence of hepatocellular carcinoma
(P = 0.003) and neoplastic nodules (P = 0.028) in high dose female rats and in combined carcinoma
and neoplastic nodules in low dose (P = 0.012) and high dose (P < 0.001) female rats and in high
dose male rats (P = 0.01). No other tumor type was significantly increased in incidence in rats.
In mice, there was a dose-related decrease in body weight gain in female mice from Week 25 to the
end of the study, but food consumption was within 4% of the control level in both treated groups.
No positive trends in mortaility were noted, but the low dose female mice had significantly shortened
survival compared to controls. Overall, survival to the end of the study varied between 50% and
78% in the various groups.
The incidence of liver tumors was increased in treated mice, as shown in Table 2. By the Cochran-
Armitage test, there was a significant dose-related trend in the incidence of hepatocellular carcinoma
(P = 0.018) and combined carcinoma and adenoma (P = 0.002) in male mice and of the same tumor
types in females (P < 0.001 for both). By direct comparison using Fisher's exact test there were sig-
nificant increases in hepatocellular carcinoma in high dose male mice (P = 0.022) and low dose (P =
0.006) and high dose (P < 0.001) females. Combined carcinoma and adenoma were significantly in-
creased in all groups of mice fed DEHP (P = 0.013 in low dose males, P = 0.002 in high dose males,
P = 0.001 in low dose females, and P < 0.001 in high dose females). Of the 57 treated mice of both
TABLE 1. INCIDENCE OF LIVER TUMORS IN RATS IN NTP BIOASSAY OF DEHP
Control
Low Dose
High Dose
Male
Female
Male
Female
Male
"Significantly greater than corresponding control by Fisher's exact test (P < 0.03).
114
Female
Hepatocellular
carcinoma
Neoplastic
nodules
Combined
1/50
2/50
3/50
0/50
0/50
0/50
1/49
5/49
6/49
2/49
4/49
6/49a
5/49
7/49
12/49*
8/50*
5/50*
13/50a
-------
RISK ASSESSMENT OF DEHP
TABLE 2. INCIDENCE OF LIVER TUMORS IN MICE IN NTP BIOASSAY OF DEHP
Control
Low Dose
High Dose
Male
Female
Male
Female
Male
Female
Hepatocellular
carcinoma
Neoplastic
nodules
Combined
9/50
6/50
14/50
0/50
1/50
1/50
14/48
11/48
25/48*
7/50*
•
5/50
12/50*
19/50*
10/50
29/50*
17/50*
1/50*
18/50*
Significantly greater than corresponding control by Fisher's exact test (P < 0.05).
sexes having hepatocellular carcinoma, 20 (12 males and 8 females) had pulmonary metastases. None
of the controls had pulmonary metastases. No other type of neoplasm was increased in incidence in
treated mice of either sex.
Northup et al.(l2) have suggested that the maximum tolerated dose (MTD) was exceeded in all
treatment groups except low and high dose male mice and low dose female rats because body weight
gain was depressed by more than 10% in all other treated groups. NTP,<2) however, have pointed
out that the 10% weight differential is only a guideline and that the primary reason for not exceeding
the theoretical MTD is to avoid excessive early deaths, which might prevent tumor development, and
to avoid pathological changes other than neoplasia that might be involved^in secondary mechanisms
of carcinogenesis. NTP concluded that both of these goals were fulfilled in the case of the DEHP
bioassay. However, as will be discussed in more detail later, there is support for the hypothesis that
peroxisome proliferation, which occurs in rats and mice at the doses of DEHP used in the NTP bio-
assay, "3il4) is involved in a secondary mechanism of cancer induction. At low doses where peroxi-
some proliferation does not occur, one would not expect cancer to develop.
In both mice and rats, only liver tumors were increased in incidence. The relevance of liver tumors
in rodents to humans has been questioned, particularly because of the high and variable spontaneous
incidence of liver tumors in various strains of mice'15-16' and the high spontaneous incidence in_
the livers of rats of preneoplastic cells that can be stimulated by promoting agents to produce
tumors. <17M9> Such high incidences of preneoplastic cells are not known to exist in the human
liver, which, therefore, would not be susceptible to enhancement of tumor incidence by such a
mechanism.
Taking these factors into account is is concluded that under the conditions of the NTP bioassay,
DEHP at very high dietary levels was carcinogenic to mice and rats of both sexes. There does, how-
ever, remain some question of the relevance of these findings to lower dose levels and to humans.
This topic is discussed in detail in a later section.
Strength of evidence of carcinogenicity
As described above, the only evidence that DEHP is carcinogenic comes from the NTP bioassay in
which tumors at a single histogenic site (hepatocytes) were increased in incidence in rats and mice.
Earlier, less sensitive studies found no carcinogenic effects, and there is no evidence from epidemio-
logical studies that DEHP is carcinogenic. Several schemes have been developed for assessing the
strength of evidence that a particular chemical is a human carcinogen. <20~22) In all 3 of these classifi-
cation schemes, DEHP is assigned to a low'category because the evidence of its carcinogenicity is rel-
atively weak, since it comes only from experimental animals, based on neoplasms occurring at a
single histogenic site, induced at a relatively high dose level, with no supporting evidence of genotox-
icity (see following section). This fact should be taken into account when the significance of the risk
is considered.
115
-------
IURNBULL AND ROUR1CKS
GENOTOXICITY
The overwhelming preponderance of evidence indicates that DEHP and its metabolites, MEHP
and 2-ethylhexanol, are not genotoxic (Table 3), though a few studies have reported positive results,
mostly with MEHP. These are discussed briefly below.
DEHP »
DEHP has been found nonmutagenic .in at least 8 separate Ames assays. Only Tomita et al.<30)
have reported positive results, and their data are not convincing: results of only a single dose were re-
ported, and the increase was less than 2-fold. Tomita et al.<30) also reported increased chromosomal
aberrations and transformation in embryonic cells from Syrian hamsters treated trarisplacentally
with DEHP at 3.75-15 g/kg administered by gavage. It might be noted that the proportion of nor-
mal diploid cells in all cultures including the controls was'low (40-70%), suggesting chromosomal in-
stability or technical deficiencies in experimental procedures.
The significance of the apparent positive results in the transformation assay is unclear, since the
increase is small, and it would be desirable to repeat the study to show that the results were not due
simply to variations in background transformation-frequency in different animals. This test system is
not directly comparable to the 3T3-system used by Barber et al.,U9) since the latter is entirely in vitro
and the former involves treatment in vivo, permitting full metabolic activity. The possibility of a dif-
ference between in vivo and in vitro systems is also raised by the weakly positive results in dominant
lethal assays in mice reported by Singh et al.(44) and Autian.(4S> Interpretation of effects in the study
by Singh et al.(44) is difficult because mice treated at the high.dos&(25.56 ml/kg) had reduced fertil-
ity, which might be expected, since high doses of DEHP cause testicular degeneration in rats'50-5"
and mice.'2' Hence, increases in early fetal deaths may be an indication of testicular toxicity rather
than mutagenicity. Effects at lower doses (1-10 ml/kg) reported by Autian(45) are not statistically
significant.
Albro et al.<47) reported "association" of radioactivity with DNA when 14C-DEHP with label in
the ethylhexyl moiety was fed to rats. No association occurred when the label was in the phthalate
group or when DEHP was saponified to phthalic acid and free 14C-ethylhexanol before being fed.
The lack of an effect with 14C-ethylhexanol suggests that labeling is not due only to incorporation of
label into a general metabolic pool. However, Von Daniken et al.,(48) while finding similar associa-
tion of radiolabel from DEHP with DNA under similar circumstances, also found label in DNA
when 14C-ethylhexanol was given orally. They presented evidence that the results with DEHP were
caused by metabolism and incorporation of label into nucleotides and not due to covalent DNA
binding of the type seen with genotoxic carcinogens. More recent work by Albro et al.<52) also indi-
cates that association of label from 14C-DEHP is due to catabolism of the ethylhexyl moiety and in-
corporation de novo into normal DNA nucleotides.
MEHP
Once again most of the reported positive results come from Tomita et al.<30) They report a dose-
related increase in toxicity to rec" compared to rec* Bacillus subtilis. This seems to be a genuine ef-
fect, since consistent results were obtained over a range of doses (100-500 rng/disc). These authors
also report an apparent dose-related increase in revertants in Salmonella (TA.100) and Escherichia
coli (WP2 B/r) treated with MEHP in suspension. The results are reported in terms of revertants/
survivor. This method of reporting results can be misleading unless the protocol used is appropriate.
For example, the standard Ames assay with Salmonella involves plating the bacteria in the presence
of a small amount of histidine. If histidine is incorporated following suspension treatment, a treat-
ment that is simply toxic can appear to be mutagenic, since the number of spontaneous revertants
appearing is governed by the number of cells that the histidine can sustain not by the number of
116
-------
RISK ASSESSMENT OF DEHP
TABLE 3. SUMMARY OF GENOTOXICITY TESTS ON DEHP AND ITS METABOLITES
Type of Test
Mammalian cell
mutagenicity
DNA damage
In vitro cytoge-
netics
Results
Reference
DEHP
Bacterial muta- DEHP negative in standard Ames assay ± S9 23
genicity DEHP reported "nonmutagenic in bacterial test" 24
DEHP at up to 1000 ^g/plate negative in Ames assay with * 25
strains TA98 and TA100
DEHP at up to 10 /xl/plate negative in Ames assay ± S9 26
DEHP at 0.1-10.0 mg/plate negative in 4 strains in Ames 4
assay ± S9
DEHP at lO'MO'2 M negative in Ames strains TA98 and 28
TA100 ± S9
DEHP at 0.15-150/tg/plate negative in 5 strains in Ames 29
assay ± S9
DEHP reported positive at 5 mg/plate in Salmonella strain 30
TA100 with S9; no data on other doses reported
DEHP at 50-2000 /tg/plate negative in Ames strains TA98 31
and TA100
Urine from rats given DEHP by gavage at 2 g/kg per day 32
for 15 days was negative in Ames assay ± S9 and ± j8-
glucuronidase and arylsulfatase
DEHP at 25-10,000 pg/ml negative in V79/HGPRT assay 33
± S9
DEHP at 7.8-250. nl/ml negative in mouse lymphoma TK 29
assay ± S9
DEHP at 0.016-1.0 /tl/ml without S9 and 0.067-5.0 /d/ml 26
with S9 negative in mouse lymphoma TK assay
No induction of unscheduled DNA synthesis (UDS) in pri- 34
mary rat hepatocytes in vitro with DEHP up to 10
mg/ml
DEHP negative in rat hepatocyte UDS assay 29
No increase in alkaline elution of DNA from hepatocytes • 35
of rats given DEHP at 500 mg/kg per day for 14 days
and no induction of UDS in rat or human hepatocytes
treated in vitro at 0.1-10 mM or in hepatocytes from,
rats fed DEHP at 1200 ppm for 30 days
No significant effect of DEHP at up to 75 /tg/ml on chro- 36
mosome aberration frequency in human lymphocytes
No significant effect of DEHP at up to 60 /tg/ml on chro- 37
mosome aberration frequency in G0 human lymphocytes
or human fetal lung cells
No chromosome damage in human lymphocytes treated 38,
with DEHP at 0.16 mg/ml
No significant increase in chromosome breaks but slight in- 39
crease in SCE (not dose related) in Chinese hamster cells
at up to 1 mM
No increase in chromosome aberration assay in CHO cells 40
at up to 2 mM (limit of solubility)
No increase in chromosome aberrations in Chinese hamster 41
cells
117
-------
TURNBULL AND RODRICKS
TABLE 3. (CONTINUED)
Type of Test
Results
Reference
In vivo cytoge-
netics
Dominant lethal
assay
Cell transforma-
tion
In vivo DNA
binding
Bacterial muta-
genicity
No increase in chromosomal aberrations in peripheral lym- 42
phocytes of workers exposed to DEHP for average of 22
years
DEHP at 5, 1.7, and 0.5 g/kg per day for 5 days by gavage 43
caused no increase in chromosome aberrations in bone
marrow cells in male rats
DEHP negative in micronucleus test in mice at 5 g/kg- 29
single dose and 5 g/kg per day for 5 days
Apparent increase in chromosomal aberrations in cells cul- 30
tured from hamster embryos treated with DEHP trans-
placentally at 7.5 and 15 g/kg
DEHP at 25.56 mg/ml (2/3 IP LD50) in male mice caused 44
reduced fertility, reduced litter size, and increased early
fetal deaths in first 3 weeks after treatment; slight effects
at 1/2 and 1/3 LD50
Slight but not significant increases in early fetal deaths at 45
1-10 ml/kg
DEHP at MTD (9.86 g/kg) and 1/2 and 1/4 MTD by ga- 46
vage daily for 5 days "was negative in dominant lethal
assay in mice
No increase in transformation in mouse 3T3 cells in vitro 29
at 0.875-21.0 nl/ml without metabolic activation or at
6.25-100 /d/ml with rat hepatocytes to provide activation
Apparent increase in frequency of transformation in cells 30
of hamster embryos treated transplacentally at 7.5 and
15 g/kg
Some incorporation of radiolabel into liver DNA in rats 47
fed "C-DEHP with label in ethylhexyl moiety but not if
labeled in phthalate moiety; nature of attachment to
DNA unclear
Similar results to those of Albro et al.,(47) but found to be 48
caused by incorporation into nucleotides via intermedi-
ary metabolism
MEHP
MEHP "showed DNA-damage-provoking activity in B. 24
subtilis and mutagenicity in E. coir" no details
MEHP negative in Ames test in TA98 and TA100 at up to 25
1 mg/plate ± S9
MEHP negative at 0.002-0.2 /J/plate in Ames test 26
MEHP negative at 1.03-1030 /tg/plate ± S9 in Ames test. 29
with 5 strains of Salmonella
Apparent dose-related increase in revertants in E. coli 30
(WP2) and slight increase in Salmonella (TA100) without
S9
MEHP negative at 50-2000 j^g/plate in Ames strains TA98 31
and TA100 ± S9
118
-------
RISK ASSESSMENT OF DEHP
TABLE 3. (CONTINUED)
Type of Test
Results
Reference
Mammalian cell
mutagenicity
DNA damage
In vitro cytoge-
netics
In vivo cytoge-
netics
Dominant lethal
assay
Cell transforma-
tion
Bacterial muta-
genicity
Mammalian cell
mutagenicity
DNA damage
In vitro cytoge-
netics
MEHP negative at 0.013-0.32 /J/ml in mouse lymphoma 26
TK assay ± S9 ,
MEHP negative at 0.081-1.25 mM in CHO/HGPRT 40
mutagenicity assay
Apparent dose-related increase in differential inhibition in 30
repair-deficient bacteria (Rec assay)
MEHP at "toxic and nontoxic levels" negative in primary 34
rat hepatocyte UDS assay
MEHP induced significant increases in chromosome aber- 40
rations in CHO cells at 0.8-1.75 mM but no increase in
SCEsatO.7-1.3 mM
Apparent increase in SCEs in V79 cells at 25-50 /tg/ml 30
MEHP at 0.14, 0.05, and 0.01 g/kg per day for 5 days 43
caused no significant increase in chromosome aberra-
tions in bone marrow cells
Apparent reproducible increase in the frequency of micro- 29
nucleated cells in bone marrow of female mice given IP
doses of MEHP at 125 mg/kg per day on 2 successive
days; no effect with a single dose or in males
Apparent increase in chromosome abnormalities in cells 30
from hamster embryos treated transplacentally at 375-
1500 mg/kg
MEHP at 50, 100, and 200 mg/kg per day for 5 days by • 46
gavage was negative in assay in mice
MEHP negative in mouse 3T3 transformation test at 25- 29
120 nl/ml without metabolic activation and at 5-125
nl/ml with rat hepatocytes to provide activation
2-Ethylhexanol .
Negative in Ames assay ± S9 at 0.01-1.0 /ul/plate 26
Negative in 5 strains in Ames assay ± S9 at 0.002-1.8 29
jd/plate
Small (less than 3-fold) increase in mutation frequency 49 •
matched by reduction in survival with 2-EH at 0.5-
1.5 mM
Negative in Ames strains TA98 and TA100 ± S9 at 0.1- 28
10 mM
Urine from rats given 2-EH by gavage at 2 g/kg per day 32
for 15 days was negative in Ames assay ± S9 and ±
/3-glucuronidase/arylsulfatase
Negative in mouse lymphoma assay at 0.018-0.24 ul/ml 26
± S9
2-EH negative "at nontoxic and toxic levels" in primary rat „ 34
hepatocyte UDS assay
Negative in CHO chromosome aberration assay at 1.5- 40
2.8 mM
119
-------
TURNBULL AND RODRICKS
TABLE 3. (CONTINUED)
Type of Test
Results
Reference
In vivo cytoge-
netics
Cell transforma-
tion
2-EH negative in micronucleus test in mice treated once or 29
twice with 456 mg/kg per day IP
2-EH negative in bone marrow chromosome aberration 43
assay in rats at 0.21, 0.07, and 0.02 g/kg per day for
5 days
Negative in mouse 3T3 cells at 48-225 nl/ml without meta- 29
. bolic activation and at 0.011-0.162/il/ml with rat hepa-
tocytes for activation
viable cells plated. Unfortunately, details of the procedures used were not reported; therefore, this
finding cannot be evaluated.
A dose-related increase in chromosome aberrations and transformation in hamster embryo cells
treated transplacentally at 375-1500 mg/kg of MEHP was also reported by Tdmita et al.<30) Our
comments on these findings are the same as those we made in connection with the work of these in-
vestigators on DEHP. They also report a small increase in SCEs in Chinese hamster V79 cells treated
in vitro with MEHP at 25-50 mg/ml for 24 hours and mention —but give no details of—an increase
in mutations in the azaguanine/thioguanine resistance system with V79 cells in vitro and Syrian ham-
ster embryo cells treated transplacentally.
- Overall, the report of Tomita et al.(30) suffers from a lack of details about the procedures used.
Without this and some of the supporting data that are'missing from the report, a detailed evaluation
is impossible. However, some of the reported positive results are in systems sufficiently different in
endpoint (B. subtitis rec assay) or method of treatment (transplacental assays) from other published
negative results that they can not simply be considered inconsistent (e.g., if intact mammalian me-
tabolism is needed for expression of activity, Tomita's positive results with transformation do not
necessarily conflict with the negative results found by Barber et al.(29) in 3T3 cells in vitro).
Yagi et al.<24) report in .an abstract that MEHP "showed DNA-damage provoking activity in B.
subtilis and mutagenicity in E. coli," but no details are given. Some of these authors are from the
same laboratory as Tomita et al.,<30) and they may be reporting data from the same studies.
Phillips et al. <40> reported an increased frequency of chromosomal aberrations in Chinese hamster
(CHO) cells in vitro treated for 2 hours with MEHP at 0.8-1.75 mM. In particular, there was a dose-
related increase in chromatid interchanges in addition to increases in chromatid and chromosome
breaks. These does levels were in the toxic range and reduced cell survival to between about 65% to
less than 10% of control values. Unlike Tomita et al.,<30) Phillips et al.(40) found no increase in SCE
in CHO cells treated with MEHP, but Phillips et al. used CHO cells and treated for 2 hours, whereas
the positive results reported by Tomita et al.(30) involved V79 cells treated for 24 hours.
Barber et al.<29) observed a slight, reproducible increase in the frequency of cells with micrdnuclei
among polychromatic erythrocytes in the bone marrow of female mice given IP doses of MEHP at
125 mg/kg per dose on 2 successive days. No effects were seen in males given this treatment or in
either sex given a single dose of 125 mg/kg. Although the increase was-significant when compared to
a concurrent solvent control group, the frequency, observed was within the range of historical con-
trol values. ' , *
Albro et al.(47) found "association" of .label with DNA when 14C-MEHP labeled in the ethylhex-
anol moiety was fed to rats. Labeling was less than resulted with an equivalent amount of 14C-
DEHP. The nature of this "association" is not known for certain, but the study by Von Daniken et
al.(48) mentioned above in relation to DEHP suggests that incorporation of label via intermediary
metabolism may be responsible.
120
-------
RISK ASSESSMENT OF DEHP
2-Ethylhexanol (2-EH)
The only suggestion of positive results with 2-ethylfiexanol comes from an azaguanine-resistance
assay in Salmonella.<49) There was a slight, apparently dose-related increase in azaguanine-resistant
mutants in bacteria treated with 0.5, 1.0, and 1.5 mM EH. The increase in frequency of mutants/
survivor was small (less than 3-fold) and was matched by a reduction in survival, however, so that
there was no increase in the absolute number of mutants at any dose. Such a response is not gener-
ally considered conclusive evidence of mutagenicity.
Discussion
As we have noted and as is evident from' Table 3, the overwhelming preponderance of evidence
suggests that DEHP, MEHP, and 2-EH have little or no propensity for direct interaction with and
alteration of DNA. A few studies suggest genotoxic activity, but, as described above, in most of
these cases serious questions exist regarding methodology or the significance of the results. How-
ever, there are a few findings that cannot be dismissed, and it appears that there remains some uncer-
tainty regarding the capacity of at least MEHP to display some degree of genotoxic potential in some
systems.
If the proposed mechanism of DEHP carcinogenesis is correct (see next section), one would expect
genetic damage to be induced in vivo (or in in vitro systems with peroxisomes present) at doses capa-
ble of« inducing peroxisome proliferation. In this sense DEHP/MEHP may be threshold genotoxic
agents, which would not be expected to display genotoxic activity if the test systems used did not
provide for the presence of peroxisomes or if the DEHP/MEHP doses were insufficiently high to
cause peroxisome proliferation.
MECHANISM OF TUMOR INDUCTION IN RODENTS
Recent investigations into possible mechanisms of chemical carcinogenesis have emphasized that
more than one biological process may be involved in cancer development.<53) The influential work of
the Millers'54-"' emphasized the fact that a large proportion of known carcinogens either are them-
selves, or are metabolized to, electrophiles. These react with and damage nucleophilic sites in the .
cell, particularly DNA, and are, therefore, genotoxic. Damage to DNA provides a basis for explain-
ing the permanent nature of the neoplastic state on the basis of a heritable (at the cellular level) alter-
ation in the genome.
Although the ability to damage DNA seems to be an important property of many carcinogens —
the so-called genotoxic carcinogens —some substances appear to increase the incidence of tumors in
experimental animals without interacting directly with DNA. Examples include tumor promoters
and cocarcinogens, such as phorbol esters, that increase the incidence of tumors when applied in
conjunction with a genotoxic carcinogen, and some hormones, particularly estrogens.'53' Also in-
cluded in this group of epigenetic or nongenotoxic carcinogens are such chemicals as saccharin,
DDT, and perhaps carbon tetrachloride.'53'
As discussed in the previous section, DEHP has no substantive genotoxic activity and, therefore,
belongs in the general class of nongenotoxic carcinogens. Several hypotheses have been proposed to
explain the mechanism of action of various nongenotoxic carcinogens.(S3) Thus, immunosuppressive
drugs may permit tumors to develop from transformed cells that would normally be detected and de-
stroyed by the body's immunological system. Some chemicals may alter the activation or detoxificaT
lion of other carcinogens so that higher levels of an active metabolite are present.
There is considerable support for the hypothesis that DEHP belongs to a diverse class of
nongenotoxic carcinogens whose mechanisms of action involve induction of peroxisome prolifera-
tion. '"•"' In this section, the support for this hypothesis and its implications concerning the shape
of the dose-response curve for tumors in rodents and, hence, its implications for assessing the risk of
tumor development at low doses are discussed.
121 ,
-------
TURNBULL AND RODRICKS
There is one other aspect of the behavior of DEHP that influences the dose-response*relation, and
this is the relationship between administered dose and target-site dose over a range of dose levels
(i.e., dose-related pharmacokinetics). The term "target-site dose" is used here to mean the dose in the
target organ (liver) of the substance (DEHP or one or more of its metabolites or by-products of its
metabolism) that is ultimately responsible for causing (directly or indirectly) the observed increase in
the incidence of liver tumors. The term has been used in the past to refer to the dose of a genotoxic,
active metabolite of a proximate carcinogen. We use the term here in a more general sense, with no
implication of genotoxicity. A study of the dose-related pharmacokinetics of DEHP in rats has re-
cently been completed, and the results of the study appear to be useful for defining the likely dose-
response relation.
In addition to catalase, which breaks down' hydrogen peroxide, peroxisomes contain several en-
zymes that catalyze reactions that generate hydrogen peroxide. These include several oxidase en-
zymes, such as L-a-hydroxyacid oxidase, D-amino acid oxidase, and urate oxidase.1581 Recent studies
have demonstrated that peroxisomes contain a system of enzymes involved in the /3-oxidation of
fatty acids, which also generates hydrogen peroxide.(71)
The hypothesis that there is a relationship between peroxisome proliferation and liver carcinogen-
esis in rodents was proposed by Reddy et al.<"> on the basis of their findings with a structurally di-
verse group of chemicals, including some drugs used in the treatment of hyperlipidemia (clofibrate,
nafenopin, Wy-14,643, BR-931, and tibric acid). All 5 of these chemicals caused hypolipidemia (re-
duction in serum lipid levels, particularly triglyceride levels), liver enlargement (hepatomegaly) with-
out necrosis, proliferation of liver peroxisomes, and hepatocellular carcinoma in mice or rats, but
none were mutagenic in the Ames assay or caused DNA damage in the lymphocyte 3H-thymi.dine in-
corporation assay. Since then, additional chemicals have been found to display this same set of
effects.'72'
Although an association between peroxisome proliferation and development of hepatic tumors
seems clear for several chemicals, there may be some exceptions. Acetylsalicylic acid (aspirin) is a
weak inducer of peroxisome proliferation, but there is no evidence that it is carcinogenic.'72' Fenofi-
brate, a hypolipidemic drug related to clofibrate, caused hypolipidema, heptatomegaly and peroxi-
some proliferation in Sprague-Dawley rats at dose levels of 50-1000 mg/kg per day for 3 months,
but no significant increase in liver tumors was seen in Sprague-Dawley rats fed fenofibrate at 10 or
60 mg/kg per day for 117 weeks or in Swiss mice fed fenofibrate at 50 mg/kg per day for 92
weeks.(73) However-, the dose levels used in the carcinogenicity studies were fairly low and probably
122 .
-------
RISK ASSESSMENT OF DEHP
did not represent maximum tolerated doses as used by NTP with DEHP. It is, therefore, unclear
whether fenofibrate is a true exception to the correlation pattern or whether testing at higher dose
levels would reveal carcinogenic effects. This is highlighted by studies with bezafibrate, another hy-
polipidemic drug related to clofibrate. Although bezafibrate caused hypolipidemia, hepatomegaly,
and peroxisome proliferation in rats at dose levels of 10-500 mg/kg per day for 1 week, it caused no
increase in liver tumors in Sprague-Dawley rats when fed at 300, 750, or 1500 ppm for 24-26
months.(74) However, when the dose level was increased to 6000 ppm in the diet, liver tumors devel-
oped in the female rats.'"'
In considering how chemicals that cause peroxisome proliferation might induce cancer, one may
start by examining data on the physiological and biochemical effects of peroxisome proliferation
that might be relevant to carcinogenicity. Reddy and co-workers'28'75' have proposed that increased
production of hydrogen peroxide by the peroxisomes is responsible for the carcinogenic effects. Evi-
dence for this mechanism is incomplete, but it appears to be the best available explanation for the
carcinogenic activity of DEHP in rodents.
Although peroxisomes contain several enzymes that catalyze reactions in which hydrogen peroxide „
is produced,<58-60-7 most of these enzymes are little affected by. chemicals that cause peroxisome
proliferation.(7I) However, peroxisomal enzymes involved in /3-oxidation of fatty acids (which form
hydrogen peroxide as a by-product) are substantially increased in activity in liver cells from rats fed
hypolipidemic drugs that cause peroxisome proliferation. <71'77' Similarly, induction of peroxisomal
^-oxidation has since been demonstrated with DEHP.'78'82' The pathway for peroxisomal /3-oxida-
tion of fatty acids is illustrated in Figure 3. As indicated there, the reduced FAD produced by fatty
acyl-CoA oxidase is coupled directly to oxygen, producing hydrogen peroxide, rather than being
coupled to the respiratory chain, as it is in the mitochondria. (60-7U Because it is not coupled to the
respiratory chain, the peroxisomal /3-oxidation pathway is insensitive to cyanide; thus, an indepen-
dent measure of peroxisomal /3-oxidation can be made by using cyanide to inhibit selectively mito-
chondrial /3-oxidation.t78-79-82' Rats fed high dose levels of DEHP show substantial increases in per-
oxisomal /3-oxidation activity per mg protein or per gram of liver.(83"8S1 Canning et al.(82) found a
slightly lesser increase in j3-oxidation in rats fed the same concentration of MEHP but found no ef-
fect with phthalic acid or 2-EH. Similar results with- MEHP- and 2-EH were found by Morton.(80)
Recent work by Mitchell et al.(86) indicates that 2 metabolites of MEHP (metabolites VI and XI in
Figure 5) also induce /3-oxidation and may be the proximate inducers of peroxisome proliferation.
Of particular importance to our consideration of mechanism of carcinogenesis, these increases in
/3-oxidation activity were not accompanied by correspondingly large increases in catalase activ-
Ity^s.s^ai) -phis difference in extents of increase in /3-oxidation and catalase activity is important be-
cause, as mentioned before, catalase normally inactivates the potentially hazardous hydrogen per-
oxide generated by the /3-oxidation system, although glutathione peroxidase also plays a role in
disposing of hydrogen peroxide. Since the increase in /3-oxidation activity induced by DEHP is much
greater than the increase in catalase, it is possible that this creates an imbalance, resulting in excess
levels of hydrogen peroxide in the cell. The consequence of excess intracellular levels of hydrogen
peroxide could be detrimental to the cell. (61-87' Addition of hydrogen peroxide to mammalian cells ih
culture causes chromosome damage*64-"' and induces sister chromatid exchange.(66'67-69'70' Hydro-
gen peroxide can react with DNA, causing alteration and liberation of DNA bases and backbone
breakage.("' Also, recent Japanese studies, although not entirely conclusive, suggest that hydrogen
peroxide may have tumor-promoting or cocarcinogenic activity, <88-89) and it is mutagenic in a newly
developed strain of Salmonella bacteria.'90'
In addition to these effects of hydrogen peroxide itself, it can form the highly reactive hydroxyl
radical either by spontaneously splitting to form 2 hydroxyl radicals'"' or by reacting with ferrous
iron:'*7'
Fe II + HiO, - Fe HI + OH' + OH'
or by reacting in the Haber-Weiss reaction with superoxide ion, which may be generated by the
microsomes:'61'
HaO2 + O; - O» + OH" + OH'
123
-------
TURNBULL AND RODRICKS
R-CH -CH -COOH
Fatty acyl-CoA' synthetase
R-CH2-CH9-CO~S-CoA
FAD
Fatty acyl CoA
oxidase
FADH
H2°2
R - CH = CH-CO~S-CoA
Enoyl hvdratase
R-CHOH-CH-CO~S-CoA
.Catalase
NAD •
L-3-Hydroxy fatty acyl CoA dehydrogenase
\ADH + H+
R-CO-CH-CO~S-CoA
CoA-SH / \ Thiolase
,R-CO~S-CoA
FIG. 3. Pathway of peroxisomal 5-oxidation of fatty acids.
124 .
-------
RISK ASSESSMENT OF DEHP
The hydroxyl radical is highly reactive and can cause DNA damage directly, <61-621 or it can initiate
lipid peroxidation, which in turn yields several chemicals, such as fatty acid hydroperoxides, choles-
terol hydroperoxide. endoperoxides, fatty acid and cholesterol epoxides, enals and other aldehydes,
and alkoxy and hydroperoxy radicals that have been shown to exert rriutagenic, promoting, or carci-
nogenic effects."11
It is recognized that there are limitations to this hypothesis. The most important of these limita-
tions is that there is no evidence that excess levels of hydrogen peroxide are generated in the livers of
rats fed high levels of DEHP. Nor is there evidence of toxic effects, such as lipid peroxidation, that
might be expected to accompany excess levels of intracellular hydrogen peroxide. However, it is not
clear to what extent such effects have been looked for, and there is some evidence of increased
steady-state levels of hydrogen peroxide and increased levels ofjipofuscin (indicative of lipid peroxi-
dation) in the livers of rats treated with other peroxisome proliferators.'72'.
The proposed mechanism of carcinogenicity is summarized in Figure 4, which shows only the es-
sential outline of the peroxisome proliferation hypothesis. It is not intended as a comprehensive de-
scription of the fate and biological effects of DEHP.
DOSE-DEPENDENT PHARMACOKINETICS OF DEHP
Before discussing the available data on the pharmacokinetics of DEHP in rodents, it would be
useful to review briefly the major pathways of metabolism of DEHP. Our understanding of the me-
tabolism of DEHP is due largely to the work of Albro and co-workers, on whose work the following
summary is based. M7-"-92) The initial step in metabolism of orally administered DEHP in all species
studies is hydrolysis by a nonspecific lipase in the gut to yield MEHP and 2-EH. In the liver, MEHP
undergoes «- and (u-l)-oxidation to yield the metabolites shown in Figure 5. The metabolite num-
bering system is that of Albro et al.'47-"-92' Several other minor metabolites have recently been iden-
lified."11 In rats and mice, metabolite V, a product of co-oxidation, may then undergo /3-oxidation
with loss of a 2-carbon unit to yield metabolite I. This /3-oxidation is important because, as discussed
above, peroxisomes contain the enzymes involved in fatty acid /3-oxidation and may, thus, be in-
Enters Intermediary Metabolism
TCHP ""e"lntm-HFH11 <• l-Ethyliiexdiiul
motivation {Strand Breaks)
•Carcinomas 7
React ive
Nodular —
Hyperplasia
-Carcinomas
02 *• H2<3
FIG. 4. Schematic of the peroxisome proliferation hypothesis. At high doses of DEHP, it is pro-
posed that excess H2O2 or other oxygen species are produced in excessive amounts because catalase
production does not increase as rapidly as peroxide production.
125
-------
TURNBULL AND RODRICKS
:o
w-i oxidation
R-CH-CH_-CH,-CH-CH,
CH,
1 "•
CH,
R-CH-CH,-CH,-C-CH,
I - 'I '
CH, 0
! '
CH,
CH,
CH,
R-CH-CH,-CH,-CH,-CH.
R-CH-CH,-CH,-CK,-CH_
CH,
R-CH-CH,-CH,-CH,-CH,OH
CH,
R-CH-CH, -CH, -CH, -COOH
CH,
a-oxidacion
R-CH-CH,-CH,-CH,-CH3
COOH
8-oxidat ion
R-CH-CH -COOH
2
FIG. 5. Major pathways of metabolism of MEHP in rats. The pathways illustrated are inferred
from knowledge of the structures of the urinary metabolites shown, based on the work of Albro et
aj (47.83.92) Metabolites are numbered according to Albro's convention.
volved in this step of the metabolism of DEHP in rodents, generating hydrogen peroxide. However,
metabolite I has not been identified in humans and is only a minor metabolite in primates, (47-94> ap-
parently because humans achieve water solubility of DEHP metabolites by glucuronide formation
rather than by extensive oxidative metabolism. Peroxisome proliferation is much less extensive in
primates and presumably in humans than in rodents given DEHP.""" Hence, if peroxisomal /3-oxi-
dation is important in forming metabolite I in rodents, lower levels of metabolite I would be ex-
pected in primates, in agreement with the experimental findings.<47-*"
The influence of dose level and previous exposure on the metabolism of DEHP in rats has been in-
vestigated. <"' The most important finding in the context of this risk assessment was that in rats with
prior exposure to DEHP, the percentage of the dose excreted in urine as metabolite I increased dis-
proportionately with increased dose. The percentage of the dose excreted in the urine as metabolite I
doubled from about 1 l°7o of the dose at 1000 ppm to about 25% of the dose at 6000 ppm, and more
than doubled to 31 % of the dose at 12,000 ppm. This increase in the percentage of the dose excreted
in urine was offset by a decrease in the percentage of the dose excreted in feces, partially by a de-
crease in the percentage excreted as metabolite IX.
Metabolite I is believed to be formed by /3-oxidation of metabolite V,(92) and the urinary excretion
patterns observed for these metabolites were consistent with this^pathway. Therefore, repeated ad-
ministration of DEHP at 6000 and 12,000 ppm in the diet to rats apparently results in a nonlinear in-
crease in metabolism by /3-oxidatiori.
It thus appears that there is a nonlinear relationship between administered dose and the active
dose of DEHP metabolites. By "active," we refer to that which produces peroxisome proliferation,
an indirect measure of which is probably production of metabolite I by /3-oxidation of metabolite V.
126
-------
RISK ASSESSMENT OF DEHP
It is possible, however, that some of the enhanced conversion of metabolite V to metabolite I is at-
tributable to mitochrondrial oxidation as well as to peroxisome proliferation.
The 6000 ppm and 12,000 ppm exposure levels used in this study were identical to those used in the
NTP bioassay, which lead to the appearance of an excess of liver tumors. This study reveals that a
marked shift in the production of metabolite I (and in the pattern of other metabolites as well) oc-
curs between the 1000 ppm level and the higher levels. This observation strongly suggests that the
dose-response relation for peroxisome induction (and subsequent tumor induction) may decline in a
nonlinear fashion below the region of observed tumorigenic responses.
DOSE-RESPONSE EXTRAPOLATION OF RODENT DATA
In order to estimate the risk to humans of exposure to DEHP in the environment, it is necessary to
estimate the relationship between risk and exposure levels in the low range to which humans may be
exposed. As a first step, we need to extrapolate the observed relationship between DEHP dose and
tumor incidence in rodents to low dose levels. Several mathematical models have been developed and
proposed for extrapolating from observed data on tumor incidence at high experimental doses to
risk at low doses.t85-95-961 Which of these models is most appropriate to use in any particular case is
an important and controversial aspect of risk assessment methodology. Some of the controversy
arises because in some cases the various models predict risks at low exposure levels that differ by sev-
eral orders of magnitude,' even though they may all fit the experimental data in the high exposure
range almost equally well. For example, in the preamble to OSHAs Cancer Policy (45FR 5200), esti-
mates of lifetime risks to humans from vinyl chloride and saccharin at likely exposure levels are pre-
sented. These estimates vary 1 million-fold depending on which extrapolation model is used.
In selecting a model for carcinogenic risk assessment, the functioning of the model should be con-
sistent with the scientific information on the phenomenon being modeled. In the present case, it is
desirable that at least some of the models used should be consistent with our limited understanding
of the possible mechanism(s) of carcinogenicity of DEHP.
In conducting dose-response extrapolation, the dose in question is generally the applied dose of
the chemical, usually expressed in terms of mg/kg body weight per day. However, probably of more
importance in determining how the risk will change with dose is the dose of the ultimate carcinogen
at the target site. This latter dose will be a function of the pharmacokinetics of the chemical (its ab-
sorption, distribution, biotransformation, and elimination). As long as all of the pharmacokinetic
parameters of a chemical are linear with dose, any increase in applied dose will cause a proportional
increase in what we may call the "target-site dose," but at high doses processes such as absorption,
biotransformation, and excretion may become saturated, and there will no longer be a linear rela-
tionship between applied dose and target-site dose.'97-98' As an example of the application of this
phenomenon, Gehring et al.(99) examined data on the carcinogenicity and pharmacokinetics of vinyl
chloride. They demonstrated that metabolism of vinyl chloride was not linearly related to applied
dose but followed Michaelis-Menten kinetics. .
The situation with DEHP is different from that with vinyl chloride, which is a genotoxic carcino-
gen whose biotransformation to an active carcinogenic form is apparently saturated at high dose lev-
els. In the case of DEHP, our proposed mechanism involves peroxisome proliferation and, perhaps,
subsequent excess production of peroxide and other active oxygen species at high dose levels. If this
model is correct, the target-site dose in which we are interested is that of the various active oxygen
species. Unfortunately, no data are currently available to measure this target-site dose directly.
Instead, an indirect measure is proposed, recognizing the many uncertainties in the use of such a
measure.
This indirect measure is the level of urinary excretion of DEHP metabolite I. This is assumed to
provide a surrogate for the true target-site dose, which is a function of the effects of DEHP on the
peroxisomes,' perhaps involving a metabolic disturbance or the generation of active oxygen species.
There is circumstantial evidence that production of metabolite I may serve as an indicator of peroxi-
somal activity and perhaps of active oxygen generation. First, metabolite I is probably produced as a
f
127
-------
TURNBULL AND RODRICKS
result of /3-oxidation; (92> second, production of metabolite I is increased at high dose levels of
DEHP;'57' and third, peroxisomal /3-oxidation activity increases substantially at similar dose
levels.'78"82'
There is uncertainty in using metabolite I as a surrogate for target-site dose. We do not know
whether metabolite I is formed in the peroxisomes or in the mitochondria (or other cellular compart-
ment). Obviously, production of metabolite I in the mitochondria would not be an indicator of
peroxisomal activity and would not involve peroxide generation. Thus, the use of metabolite I
production would overestimate peroxisomal activity and peroxide generation to the extent that mito-
chondrial /3-oxidation is involved in metabolite I production. However, the involvement of the per-
oxisomes in the production of metabolite I is supported by the results of recent studies by Lhuguenot
et al.,<10°-102) who examined the effects of dose and time of the metabolism of MEHP using Wistar
rats in vivo and rat hepatocytes in vitro. They found that both in vivo and in vitro production of me-
. tabolite I increased in parallel with increases in cyanide-insensitive peroxisomal /3-oxidation activity.
Also the change in rate of production of metabolite I from metabolite V with time of the hepatocytes
in culture mirrored'the change in the activity of cyanide-insensitive palmitoyl Co A oxidation in the
hepatocytes.(102)
That mitochondrial /3-oxidation may also be involved in metabolite I production is suggested by
reports of increases in the activity of enzymes involved in ,3-oxidation in both peroxisomes and mito-
chondria after treatment of rats with DEHP.'79-80' However, provided the mitochondrial contribu-
tion to metabolite I production does not increase with dose of DEHP to a greater extent than the
peroxisomal contribution, the use of level of metabolite I production as a surrogate for peroxisomal
activity and, peroxide level will at least not underestimate the true values.
In the past, pharmacokinetic data have been used to estimate target-site doses of genotoxic carcin-
ogens, and these dose estimates have been used in place of measures of applied dose in risk extrapo-
lation.'99-103' In assessing the risks of DEHP, we may use our knowledge of the nonlinear relation-
ship between applied dose of DEHP and urinary excretion of metabolite I to provide a surrogate to
estimate the presumed nonlinear relationship between applied dose of DEHP and peroxisome activ-
ity, which is in turn an indicator of the target-site dose of peroxide or other active oxygen species. To
do this, it is necessary to examine the quantitative relationship between daily dose of DEHP and ex-
cretion of metabolite "I under conditions approximating those of the NTP bioassay. Data for this are
derived from the Little study of metabolism of DEHP in male rats fed DEHP at 1000, 6000, and
12,000 ppm in the diet for 20 days before receiving '4C-labeled DEHP at the same dietary levels.'571
These data are presented in Table 4.
The relationship between daily dose of DEHP and daily excretion of metabolite I was fit to a
power curve with equation:
I = 0.0187 D1-4287
where I = daily excretion of metabolite I, and D = daily dose of DEHP. This curve was chosen as
TABLE 4. RELATIONSHIP BETWEEN DAILY DOSE OF DEHP AND EXCRETION OF
METABOLITE I IN MALE RATS FED DEHP FOR 20 DAYS
Dietary
Concentration
of DEHP (ppm)
Equivalent
Daily Dose
(mg/kg per day)*
Fraction of Dose
Excreted in Urine
as Metabolite I
Amount of
•"' Metabolite I
Excreted per Day
(mg/kg per day)b
1,000
6,000
12,000
64.97
370.81
769.64
0.11
0.25
0.31
7.15
92.76
238.59
"Based on consumption of food containing '4C-labeled DEHP and terminal body weight.15"
bAmount excreted = daily dose of DEHP x fraction excretion as metabolite I.
128
-------
RISK ASSESSMENT OF DEHP
best fitting the data (r = 0.9996). This equation is used to adjust the doses in the NTP bioassay to
give doses that are surrogates for the target-site dose before fitting extrapolation models to the bio-
assay dose-response data. Because of the uncertainties outlined above, this adjustment must be con-
sidered only semiquantitative. However, the uncertainties are such that, the adjustment is likely to
overestimate risk at low dose levels and, hence, be conservative.
Choice of extrapolation methods
A review of the various extrapolation models that have been used for risk assessment reveal that
no single model is clearly the most appropriate for use with DEHP. We have, therefore, used several
procedures for low dose extrapolation:
1. Multistage model using applied doses of DEHP
2. Multistage model using a surrogate (metabolite I level) for target-site doses ,.,
3. Mantel-Bryan probit model using applied doses of DEHP
4. Mantel Bryan model using a surrogate for target-site doses
5. Threshold model
The multistage model is used because it is the model most widely used by regulatory agencies and
has a substantial basis in theories of cancer causation."04"107' This model has a biological basis in
that it is derived from a widely accepted view of the carcinogenic process. Under this view, cancer
arises in a single cell and is expressed after the cell or its progeny have passed through k transitions,
the rates of one or more of which are proportional to the concentration of the carcinogen at the tar-
get site."081 Some carcinogens may act at the first stage of the process (so-called initiators or early-
stage carcinogens), others may act only at later stages (promoters or late stage carcinogens), and still
others may act at both early and late stages of the process of tumor development (so-called complete
carcinogens). In the general case, where mechanisms are unknown, it has become the practice to
adopt the linearized multistage model.
The Mantel-Bryan probit model is another commonly used model.
-------
TURNBULL AND RODRICKS
Although we have not used them for this study, other extrapolation models, such as the Weibull
or multihit models, may also be appropriate for low dose extrapolation. <106-107)
Modeling of rodent carcinogenicity data
In estimating low dose risks to rodents, data derived from the NTP bioassay were used. To pro-
vide a conservative estimate of risk, the combined incidences of hepatocellular carcinoma and ade-
noma/neoplastic nodules were used. Also, to provide a better estimate of the background tumor in-
cidence,, pooled controls were used, combining data from the bioassays that were performed in the
same room of the same laboratory at the time as the DEHP bioassay. Thus, control data from bioas-
says of DEHP, butylbenzylphthalate, guar gum, and di(2-ethylhexyl) adipate were used. The dose
levels in mg/kg per day estimated by NTP<2) on the basis of food consumption were used. The 4 data
sets used for risk extrapolation are presented in Table 5. Included in Table 5 are the estimates of the
surrogate target-site doses based on the expected urinary levels of metabolite I for rats. These are de-
rived as described above using data from the multidose metabolism study.(S7) Since this metabolism
study involved only male rats, the adjustment, strictly, is applicable to male rats. However, we have
also conservatively applied the adjustment to the female rat data, since these data predict slightly
higher risks. Corresponding data are not available for mice. Therefore, no such adjustments are
made for the doses administered to mice; although a similar nonlinear relationship between applied
dose of DEHP and "target-site dose" probably also occurs in mice.
The computer program Global 82 written by Howe and Crump of Science Research Systems, Inc.,
Ruston, Louisiana, was used to perform low dose extrapolation using the multistage model. A com-
puter program was also used to perform low dose extrapolation by the Mantel-Bryan probit proce-
dure. As discussed earlier in the description of low dose extrapolation models, the Mantel-Bryan
probit procedure was not intended by its authors to provide valid estimates of low dose risk. Hence,
the program used generates estimates of "virtually safe dose" (VSD). The VSDs we report here corre-
spond to lifetime risks of less than lO'6 (1 in 1 million) and lO'8 (1 in 100 million).
The results of applying the multistage model to all four data sets in Table 5 are presented in Table
6, which presents the dose coefficients ^ and 2 from the multistage model that best fit the data
using the daily doses of DEHP. Also included are the values of g,*, the upper 95th percentile confi-
TABLE 5. DATA SETS USED FOR Low DOSE RISK .EXTRAPOLATION
DEPH (Metabolite I)
Species Sex Dose (mg/kg per day)
Incidence of Total
Hepatocellular Tumors
Rat
Rat
Rat
Rat
Rat
Rat
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Male
Male
Male
Female
Female
Female
Male
Male
Male
Female
Female
Female
0
322(71.6)a
674 (205.6)
0
394 (95.5)
774 (250.5)
0
672
1325
0
799
1821
8/149
6/49
12/49
3/197
6/49
.13/50
56/200
25/48
29/50
10/200
12/50
18/50
aNumbers in parentheses are estimated surrogate target-site dose levels calculated as de-
scribed in the text.
130
-------
RISK ASSESSMENT OF DEHP
TABLE 6.= MULTISTAGE MODEL DOSE COEFFICIENTS BASED ON DEHP DOSES IN NTP
qt (mg/kg per day)'1
q2 (mg/kg per day)-
qt* (mg/kg per day)'
Male rats
Female rats
Male mice
Female mice
1.42 x 10-4
2.13 x 10-4
4.59 x 10-"
2.35 x ID"1
2.86 x 10-'
2.07 x lO'7
0 ,
0
4.88 x 10-"
4.96 x 10-4
6.70 X 10-4
3.29 X 10-"
"The figures tabulated are the dose coefficients ql and q2 in the multistage model: P(d) = 1 - exp[-(?o +
qtd + qid1)] where P(d) is the probability of developing a tumor after lifetime exposure to a dose of d mg/kg •
per day. For small values of d, the excess risk of developing a tumor closely approximates (q,d + qd1)- Also
tabulated is ^i*, the upper 95th percentile confidence limit on ,.
dence limit on ,. At low dose levels, the excess risk above background to rodents is closely approxi-
mated by:
P(d) =• , + 2cP
where P(d) is the risk at dose level d and qv and q2 are the estimated dose coefficients. The upper
confidence limit on low dose risk is closely approximated by:
P(d) = qSd
Table 7 lists the corresponding values of „. q2, and qf derived using the surrogate target-site
dose values. To derive the risk estimate for a given dose (d) of DEHP using these values, it is first
necessary to convert the DEHP dose to the corresponding surrogate dose (/) using the equation:
/ = 0.0187J1-4287
The risk is then derived using the dose coefficients as described above.
Risk = qj + qj2
or
Upper 95th percentile confidence limit on risk = #,*/
TABLE 7. MULTISTAGE MODEL DOSE COEFFICIENTS BASED ON SURROGATE TARGET-SITE DOSES
(ESTIMATED LEVELS OF METABOLITE I) IN NTP
, (mg/kg per day)~
ql (mg/kg per day)-
q* (mg/kg per day)'
Male rats
Female ratsb
1.03 x 10-3
1.16 x 10-3
3.31
0
x 10-7
1.77 x 10-3
1.70 X 10'3
aThe figures tabulated are the dose coefficients , and 2 in the multistage model: P(d) = 1 - exp[- (q0 +
qvd + qid1)] where P(d) is the probability of developing a tumor after lifetime exposure to a dose of d mg/kg
per day. For small values of d, the excess risk of developing a tumor closely approximates (qtd + qd2). Also
tabulated is qt*, the upper 95th percentile confidence limit on
-------
TURNBULL AND RODRICKS
Table 8 lists the values of "virtually safe dose" corresponding to a risk of 10"6 and 10"8 predicted by
the Mantel-Bryan probit procedure when applied to the data on each species for the sex that showed
the higher risk (female rats and male mice).
TABLE 8. "VIRTUALLY SAFE DOSES" OF DEHP'CORRESPONDING TO RISKS OF 10'6 AND 10~8
FROM THE MANTEL-BRYAN MODEL FOR FEMALE RATS AND MALE MICE
BASED ON DEHP DOSE AND ON SURROGATE DOSE
Virtually Safe DEHP Dose fug/kg per day)
Based on DEHP Dose
Based on Surrogate Dosea
Risk = ID'6
Risk = IO-*
Risk =
Risk = 10-*
Female rats
Male mice
46.8
. H-9
6.5
1.6
791
198
aSee Footnote b to Table 7.
INTERSPECIES EXTRAPOLATION: RODENTS TO HUMANS
The risks per unit of exposure (or NOELs) derived by application of the various models listed in
the previous section pertain strictly to mice and rats. The next stage of analysis involves extrapola-
tion of these risks to humans. Extrapolation between species adds considerably to the uncertainty of
risk assessment, as has been discussed by many individuals from such groups as the National Acad-
emy of Sciences/3-11" the Interagency Regulatory Liaison Group,(85) and the Scientific Committee
of the Food Safety Council.(9S) In the preamble to its Cancer Policy (45 FR 5200), OSHA discussed
many of the complications of interspecies risk extrapolation and concluded:
Extrapolation from animal data to predict risks in humans introduces
many additional uncertainties. These include selection of appropriate scal-
ing factors for size, lifespan, and metabolic rate; differences in routes of
exposure, duration and schedule of exposure, absorption, "metabolism, and
pharmacokinetics; differences in intrinsic susceptibility and repair capabili-
ties; intra-population variation and susceptibility; and exposure to other
carcinogens and intrinsic and extrinsic modifying factors. At least theo-
retically, these factors can affect the relative response of humans and ani-
mals by many orders of magnitude. . '
There follows a discussion of some of these factors, as they apply to interspecies extrapolation for
DEHP, and the major approaches that have been proposed to take them into consideration.
Unit of dosage measurement
Consideration of an appropriate dosage unit encompasses consideration of animal size, lifespan,
and, to some extent, metabolic rate. In specifying a unit of dose there are generally three factors
involved: a measure of the amount of the substance administered (mg, ml, mmole, and so on), an
indication of the size of the organism (kg body weight, m2 body surface area, blood volume, and
so on), and some indicator of time (day, lifetime, and so on). Among the most commonly used units
of dose are mg/kg body weight per day, mg/kg body weight per lifetime, and mg/m2 body surface
132
-------
RISK ASSESSMENT OF DEHP
area per day. Debates over the choice of dosage unit have centered on the appropriate measure for
body size (kg body weight or m2 body surface area) and on the temporal descriptor (per day or per
lifetime).'9'-105-"2-"4'
Hoel et al.("2> proposed the use of dosage units in mg/m2 body surface area per day on the basis
of studies of the acute toxicity of anticancer drugs in humans and animals.<115) In these studies the
acutely toxic level was similar in mouse, rat, hamster, dog, monkey, and man when dosage was ex-
pressed as mg/m2 per day. This procedure has also been adopted by the Environmental Protection
Agency(IOS> and is the procedure used by the Consumer Products Safety Commission"071 in its risk
assessment of DEHP. However, the relationship between dose and body surface area, determined
on the basis of acute drug toxicity, a priori means very little when considering chronic effects, such
as cancer. Given the uncertainty about the factors contributing to carcinogenicity, the most appro-
priate basis for judging the suitability of a species-to-species conversion factor is empirical data on
the relative susceptibilities of different species.
What few relevant data exist have been analyzed by Crump et al.,'105) who concluded that the dos-
age measurement giving the closest correlation between species was mg/kg body weight per day. In
the absence of good evidence for the use of a more complex procedure, we propose the use of mg/kg
per day as a generally acceptable basis for interspecies dosage comparison. In addition to its relative
simplicity, this procedure appears to have the best empirical support.(10S)
All of the foregoing discussion and almost all of the discussion in the scientific literature concern
the relative merits of the possible measures of body size and temporal factors. Little attention has
been paid to the appropriateness of the measurement of the amount of substance applied (usually
measured in milligrams). In cases where the administered dose is linearly related to the dose of the
active carcinogen at the target site, a direct measure "of the amount applied is appropriate. However,
as alluded to earlier, in cases where nonlinear pharmacokinetics apply, measurement of the dosage in
terms of the material applied may be inappropriate, and, where available, pharmacokinetic data
should be used to modify the dosage measurement so that the units of dosage are an indication of the
level of the ultimate carcinogen at the target site. Alternatively, one might actually measure the
amount of an active metabolite of a precarcinogen or measure DNA interactions to estimate the tar-
get site dose. This would not be necessary if the pharmacokinetic parameters were the same in both
the experimental species and in humans. However, in many cases, the rates of production and inacti-
vation of active metabolites are likely to have species differences. For example, the rate of produc-
tion of active metabolites of chloroform, perchloroethylene, and 2-acetylaminofluorene (AAF) dif-
fers in different species and is correlated with the relative.carcinogenic potency of the particular
chemical in the different species.(98-116-"7'
Route, duration, and schedule of exposure
In the NTP Bioassay of DEHP, animals were fed DEHP at a uniform concentration in the diet
continuously from the age of about 5-6 weeks (i.e., after about 1/20 of their lifespan had elapsed)
for the subsequent 103 weeks. In contrast, although some dietary exposure may occur, human expo-
sure to DEHP is likely also to involve inhalation of low levels of DEHP vapor released from PVC
products, dermal contact with products containing DEHP, some of which may be absorbed through
the skin, and, in some cases, parenteral exposure from DEHP in medical devices, such as blood
bags. It is unclear if DEHP absorbed from these different routes is equivalent in carcinogenicity.
Certainly, although DEHP will be absorbed from blood bags intact, the action of nonspecific lipases
in the gut results in absorption of the hydrolysis products 2-EH and MEHP rather than DEHP itself.
For the purpose of this risk assessment, however,.we will assume that DEHP absorbed by all routes
is equivalent.
In addition to differences in routes of exposure, human exposure to DEHP differs from that of
the rodents in the NTP bioassay in its temporal pattern; humans are not exposed to DEHP at a con-
stant dietary concentration for their lifetime starting shortly after weaning. The actual temporal pat-
tern has not been evaluated in detail, but it is likely to be nonuniform. Infants may be exposed to
133
-------
TURNBULLANDRODRICKS .
higher than average amounts of DEHP as a result of chewing and sucking PVC teethers, pacifiers,
and toys and from skin contact with vinyl playpen liners, baby pants, and so on."07' Individuals re-
ceiving multiple transfusions of blood or blood products (e.g., hemophiliacs) are also likely to be ex-
posed to higher than average levels of DEHP, since the plasticizer is known to leach from the PVC
blood bags into the blood during storage. (118-119)
It is common practice in conducting carcinogenic risk assessments to equate risks in humans and
animals receiving the same average lifetime daily dose (mg/kg per day). However, if the multistage
model of carcinogenesis is correct, Day and Brown'I08) have demonstrated that for anyparticular
temporal pattern of exposure, the lifetime risk of cancer will depend on whether the carcinogen is an
early stage carcinogen (an initiator) or a late stage carcinogen (a promoter).
Unfortunately, we do not know at what stage or stages DEHP acts to produce carcinogenic ef-
fects. There is some indirect evidence that it may be a promoter. Reddy and Rao'120' have shown that
the hypolipidemic drugs Wy-16,634 and clofibrate promote the appearance of hepatocellular carci-
noma in the liver of rats given an initiating dose of the'liver carcinogen diethylnitrosamine (DEN).
,As discussed earlier, these drugs produce a similar spectrum of effects in the liver of rodents as does
DEHP. Hence, DEHP may also be a promoter. Additional support for this hypothesis comes from
the work of Ward et al.,(121> who found an increase in preneoplastic foci and adenomas in the liver in
mice given an initiating dose of DEN followed by treatment with DEHP compared to those receiving
DEN without DEHP. However, this effect has not been confirmed by others.(122) As discussed ear-
lier, Hirota and Yokoyama(88> and Ito et al.(89) have reported data suggesting that hydrogen perox-
ide is a tumor promoter in the gut. However, Levin et al.(90) have reported that hydrogen peroxide is
mutagenic in a specially constructed tester strain of Salmonella, and if the mechanism of action pro-
posed earlier involving peroxisome proliferation is correct, the DNA damage that would be caused
by the active oxygen species generated by the peroxisomes might initiate tumor development.(99) It is
possible that DEHP has both initiating and promoting activity.
Since we do not know whether DEHP acts as an initiator, a promoter, or both, we will use the
simple assumption that risk is a function of lifetime average daily dose.
Interspecies differences in target-site susceptibility
The liver in rodents appears to be particularly susceptible to the induction of tumors, as evidenced
by the high and variable incidence of spontaneous liver tumors in various strains of mice,'15-16' the
high spontaneous incidence in the livers of rats of preneoplastic cells that can be stimulated by pro-
moting agents to produce tumors, <17"19) and the high proportion of animal carcinogens whose site of
action is the rodent liver.'123' By contrast, few chemicals are known to cause liver tumors in hu-
mans,'20' and humans seem to be less susceptible to peroxisome proliferation than are rats.'72'
Hence, if the peroxisome proliferation hypothesis of carcinogenesis is correct, humans would be ex-
pected to be less susceptible than rodents. Unfortunately, no data are available that would allow this
difference in sensitivity to be taken into consideration in a quantitative way. We are limited to noting
qualitatively that this is another reason why use of the available animal data is likely to result in over-
estimation of the risk to humans of DEHP.
Interspecies differences in metabolism
Albro and co-workers have conducted and reported on numerous studies of the metabolism of
DEHP in various species of rodents, the Green monkey, and man. Data available from studies in
various species are summarized in Table 9,<47) which reveals striking differences in the pattern of uri-
nary metabolites between man and rat and the similarity between man and monkey. Another differ-
ence between the rat and man is, that,the major fraction of metabolites formed in man and the green
monkey is excreted in conjugated form, whereas little such conjugation takes place in the rat.'47'
Albro et al.'47' have suggested that rats compensate for not making glucuronides by carrying oxi-
134
-------
RISK ASSESSMENT OF DEHP
TABLE 9. DISTRIBUTION OF URINARY METABOLITES OF DEHP IN VARIOUS SPECIES
Percentage of Total Metabolites
Metabolite*
Residual DEHP
MEPH
I
II
III
IV
V
VI
VII
VIII
IX
X
A, B, C
Phthalic acid
Rat
Trace
17.2
2.0
1.2
3.3
51.3
2.6
2.6
13.3
0.6
4.1
1.8
Mouse
0.5
18.6
16.8
1.0
0.4
0.8
1.1
14.9
7.2s
. 12.3
2.2
8.1
12.4
Guinea
Pig
71.2
2.4'
0.4
0.5
0.4
6.9
1.1
0.8
3.4
1.3
6.2
5.4
Green
Monkey
2.2
28.9-
0.1
4.2
5.9
7.0
5.7
38.2 '
0.1
7.6
0.1
Man
18.3
1.8
1.2
5.3 <*,
12.1
11.9
8.1 .
36.2
0.1
4.9
0.1
Hamster
0.3
4.5
13.0
0.1
0.3
0.4
14.0
10.2
4.9
32.7
1.9
6.1
9.5
From Albro et al.14"
aSee Figure 5.
dative metabolism all the way to the highly water-soluble diacids. These authors also noted that if
formation of hydroxyl side chains involves by analogy with fatty acid u-oxidation, a mixed function
oxidase reaction, one would expect a net conversion of NAD(P)H to NAD(P). The additional steps
from alcohol to aldehyde (or ketone) and from aldehyde to acid, as well as the apparent a- and j3-ox-
idations needed to produce metabolites I, II, and III, would all be associated with net conversion of
NAD(P) to NAD(P)H. Thus, the overall demand on the oxidation potential of the liver when high
doses of DEHP are given would be in opposite directions for rat and primate. Albro et al.(47) con-
cluded that to the extent that metabolism of DEHP is involved in its biological activity, one must
question seriously whether rats are an appropriate model for man.
Although the metabolic differences between man and rats are striking, the differences are not so
marked when one compares mouse and man (Table 9). Because the NTP bioassay revealed that
DEHP is carcinogenic in mice, it may appear that metabolic differences are not important for carci-
nogenesis. However, Table 9 shows that rats and mice produce similar percentages of metabolite I,
and both differ from humans. If metabolite I is an indicator of peroxisome proliferation and en-
hanced /3-oxidation as has been hypothesized, then in keeping with the hypothesis relating peroxi-
some proliferation to hepatic tumors, the major differences between rodents and primates in metab-
olite I production imply equally major differences in susceptibility to cancer from DEHP.
As with all the other data we have described, there are gaps and uncertainties in the metabolism
data. Most important is the fact that the data shown in Table 9 were obtained under different experi-
mental conditions for the different animal species and are therefore not strictly comparable. To ad-
dress this uncertainty, a study has been conducted to compare the. metabolic fate of DEHP when
administered under identical conditions to monkeys, rats, and mice.(94) In this comparative metabo-
lism study, a single dose (100 mg/kg) of l4C-labeled DEHP in corn oil was administered by gastric in-
tubation to 3 male cynomolgous monkeys, 5 male F-344 rats, and 25 male B6C3F1 mice. All 3 spe-
cies excreted an average of 30-40% of the dose in the urine, all but 5% or less in the first 12 hours
after dosing in the mouse and rat and in the first 24 hours in the monkey. All'species excreted about
50% of the dose in the feces, all but 3% or less during the first 24 hours in the mouse and rat and
during the first 48 hours in the monkey.
The metabolites excreted in the urine were identified as shown in Table 10. In general, the pattern
135
-------
TURNBULL AND RODRICKS
TABLE 10. URINARY METABOLITES OF DEHP EXPRESSED
AS PERCENTAGES OF TOTAL URINARY RADIOACTIVITY
EXCRETED IN THE FIRST 24 HOURS AFTER DOSING
Metabolite*
Monkey
Rat
Mouse b
MEHP
Phthalic acid
I
II
III
IV
V
VI
VII
IX
X
XII
XIII
Uncertain
XIV
Uncertain
11
2
b.s
0.5
0.5
25
1
7d
18
9
2
6
15
2
1
_LC
2
11
0.9
4
,29
. 11
3
18
4
6
6
1
1
1
17
13
13
0.8
0.8
3
1
12
6
11
2
5
7
5
2
1
From Moran et ai.|94)
aMetabolites are numbered according to the convention of
Albro.
bThe mouse urine extract analyzed by HPLC contained only
79% of the radioactivity excreted in 0-24 hours. The remainder
of the radioactivity was eluted from the SAC-2 resin in the
acidic aqueous wash and probably contained some of the more
polar metabolites, perhaps including glucuronides.
c Radioactivity in the sample was less than twice background
for the system.
dThis fraction may include metabolite VIII, which was iden-
tified in monkey urine after IV administration of DEHP.1"'
of urinary metabolites identified in this study is similar to that described by Albro et al.(47) in rats,
mice, and monkeys. In particular, metabolite I constitutes a high proportion of the total urinary me-
tabolites in the mouse (13%) and the rat (11%) but only a very low proportion of the total urinary
metabolites in the monkey (0.5%). If the hypothesis described earlier regarding the mechanism of
carcinogenicity of DEHP is correct, and production of metabolite I is an indicator of target-site
dose, the carcinogenicity of DEHP in monkeys (and presumably in humans, who also excrete very
small amounts of metabolite I) is likely to be much lower than in rats or mice at the same dose level.
Since it is not known what proportion, if any, of the metabolite I produced in monkeys is formed by
^-oxidation in the peroxisomes, and hence would produce peroxide, it is not possible to quantify the
likely difference in susceptibility to DEHP-induced carcinogenicity between monkeys and rodents.
However, if comparative urinary excretion of metabolite I can be used as an indicator, monkeys
would be 20- to 25-fold less susceptible than rodents (Table 10).
If the metabolic and biological data are keys to carcinogenic activity, they strongly suggest that
risk predicted under all of our various models simply does not apply to man or, more likely, that the
magnitude of human risk is, at a given (low) level of exposure, very much less than that predicted for
rodents. Although direct evidence supporting this assumption is lacking, it is supported by the cur-
rently most reasonable hypothesis about the tumorigenic action of perbxisome proliferators in ro-
dents. The actual human risk per unit of exposure cannot be quantified but is likely to be less than
that predicted for rodents and may even be zero.
136
-------
RISK ASSESSMENT OF DEHP
In performing rodent to human extrapolation for this risk assessment we have used 2 procedures,
both of which probably overestimate the risk to humans. Both assume that humans are at equal risk
as rodents at the same dose level in mg/kg per day.(10S) The first, and most conservative, procedure
assumes that the relevant dose level is the applied dose of DEHP and, thus, takes into account none
of the information available on the likely mechanism of action of DEHP in rodents and the data sug-
gesting that the effects seen in rodents are not likely to occur in monkeys or humans.
The second procedure uses the dose adjustment discussed in the previous section. This adjustment
uses the relationship between applied dose of DEHP and urinary excretion of metabolite I in rats as
a surrogate for the likely target-site dose of active oxygen species generated by the peroxisomes. This
second procedure assumes the same relationship in rats and in humans between the applied dose of
DEHP and the dose of metabolite I, the latter being a surrogate for the dose of the ultimate carcin-
ogen. The dose of metabolite I at low doses of DEHP is calculated using the relationship determined
empirically in rats from data generated by Robinson et al.(57):
/ = O.OnSD1-4287
where / = dose of metabolite I, and D = daily dose of DEHP.
Under these assumptions, the data presented in Tables 6 and 7 may be applied directly to humans
to estimate the risk at low dose levels predicted by application of the multistage model without
(Table 6) or with (Table 7) the surrogate target-site dose adjustment described above. Similarly,
Table 8 shows the virtually safe doses predicted by the Mantel-Bryan model without .and with the
same adjustment.
NOEL/Safety factor approach to risk assessment
If the mechanism of carcinogenicity of DEHP that has been proposed above is correct, no in-
creased risk of cancer would occur at exposure levels that do not cause peroxisome proliferation and
subsequent excess peroxide production. Such pathological effects are of the type normally protected
against by the classic toxicological procedure of identification of a no-observed-effect level (NOEL)
and application of a safety factor to determine an acceptable daily intake (ADI). To use this proce-
dure, it is of course necessary to identify a NOEL. Unfortunately, it is not clear if a NOEL has been
identified. In the Phase I validation study of the CMA Voluntary TSCA testing program,(14) groups
of 12 male and 12 female F-344 rats were fed diets containing DEHP at 0, 1000, 6000, and 12,000
ppm for 3 weeks. The activity of the enzyme carnitine acetyltransferase, which occurs in the peroxi-
somes and the mitochondria, showed a dose-related increase in activity at all dose levels after as little
as 1 week of treatment. Dose-related effects were also noted on liver weight (increased 20, 66, and
98% in male rats at 3 weeks), serum triglycerides (decreased to 56, 31, and 18% of control in males
at 3 weeks), and a cytochemical test (dose-related increase in peroxisome proliferation). Males were
affected more than females. These effects were not evident, however, in the animals allowed 2 weeks
of recovery after DEHP treatment, indicating that the effects are reversible.
The European Chemical Industry Trade Association (CEFIC) sponsored a similar study in which
groups of male and female Alderly Park SPF-derived albino rats (number unspecified) were fed
DEHP at 0, 50, 200, and 1000 mg/kg per day (about 1000 ppm, 4000 ppm, and 20,000 ppm) for 28
days.'"4' Liver weight was increased in all treated groups in a dose-related manner. There was a
dose-related proliferation of peroxisomes starting at the lowest dose level and a similar proliferation
of smooth endoplasmic reticulum.
Morton'801 fed DEHP at various dose levels to groups of 5-12 male Sprague-Dawley rats for 7
days and measured such parameters as liver weight, serum triglyceride level, liver catalase, carnitine
acetyltransferase (CAT), carnitine palmitoyltransferase (CPT), and /3-oxidation activity. Liver
weight was significantly increased in a dose-related manner at DEHP dietary levels of iOOO, 2500,
and 5000 ppm but not at 50, loo, or 500 ppm. Serum triglyceride levels were significantly reduced in
137
-------
TURNBULL AND RODRICKS
f
a dose-related manner at all levels tested (50, 500, and 2500 ppm). Catalase activity was increased sig-
nificantly at 5000 ppm but not at 100 or 1000 ppm. CAT and CPT activities were significantly in-
creased in a dose-related manner at 100, 500, 1000, and 2500 ppm but not at 50 ppm.
Of perhaps most importance to the present discussion is the liver /3-oxidation activity. When total
liver /3-oxidation activity was measured, significant dose-related increases were seen at 500, 1000,
and 5000 ppm, and slight but not significant increases were seen at 50 and 100 ppm. Morton*80' also
examined /3-oxidation activity in isolated mitochrondria and peroxisomes after feeding DEHP, at 0,
100, 1000, and 5000 ppm. In peroxisomes, /3-oxidation activity was increased significantly only at
5000 ppm, but slight, nonsignificant increases were seen at 100 and 1000 ppm. In isolated mitochon-
dria, ^-oxidation activity was increased significantly at 5000 and 1000 ppm and slightly but not sig-
nificantly at 100 ppm.
In a study recently conducted by the British Industrial Biological Research Association,<125)
DEHP was fed to groups of 5 Fischer 344 rats of each sex at dietary levels of 0, 100, 1000, 6000,
12,000, and 25,000 ppm for 21 days. Males and females fed 6000 ppm or more showed significantly
increased liver weights and significantly increased peroxisomal (8-oxidation activity. Serum triglyc-
eride levels were significantly reduced at the same levels in males. Microsomal lauric acid 11- and 12-
hydroxylase activity was significantly increased in males at 1000 ppm and above, as was the number
of peroxisomes in the liver. These latter effects were seen in females only at 6000 ppm or more. No
significant changes in any of the parameters monitored occurred at 100 ppm, with the exception of
an increase in serum triglyceride level in males.
In attempting to identify a NOEL from these data, several choices are possible. Based on the dose
level causing a significant increase in peroxisomai jS-oxidation activity in the studies by Morton<80>
and BIBRA,11"' a NOEL of 1000 ppm could be identified (about 70 mg/kg per day based on Mor-
ton's food consumption and body weight data or 106 mg/kg per day based on BIBRA data). Based
on total liver /3-oxidation activity, the NOEL would be set lower, at 100 ppm (about 7 mg/kg per
day). A NOEL for all effects of 100 ppm (about 11 mg/kg per day) was identified in the BIBRA
study.<12S) However, a NOEL for all effects cannot be identified from the Morton study,<80) since a
significant reduction in serum triglyceride level was seen even at the lowest dose of 50 ppm (about 3.5
mg/kg per day), though such an effect was not seen in the BIBRA study at levels below 6000 ppm.
Since these possible NOELs are derived from studies of short-term exposure (7-21 days), estimation
of a chronic ADI for humans would typically involve application of a safety factor of 1000.(" " This
would lead to a chronic ADI of between 70 to less than 3.5 ^g/kg per day, with the most likely value
being 11 /ig/kg per day, which is derived from the NOEL for peroxisomal proliferation in the
BIBRA study.11"'
DISCUSSION
To provide a comparison of the implications for risk at low dose levels of these 5 models (multi-
stage and Mantel-Bryan models, both with and without target-site dose adjustment, plus threshold
model) their predictions of virtually safe dose (risk less than 10'6 on lifetime exposure) or ADI are
listed in Table 11. For the multistage* model, both maximum likelihood estimates and 95th percentile
upper confidence estimates are listed. For each model, results for the data set (species and sex) pre-
dicting the highest risk are presented.
Under the most conservative procedure (upper confidence limit on multistage model with applied
dose levels), exposure would need to be less than 1.5 ^g/kg per day to ensure a lifetime risk of less
than 1 in 1 million (10'6). At the other extreme, the Mantel-Bryan model using the surrogate target-
site dose adjustment predicts a risk of less than lO'6 at a daily dose of 791 pg/kg per day. The model
that is most consistent with our general understanding of cancer development (the multistage model)
when combined with our attempt to make the best use of data to provide inferences regarding the
likely shape of the dose-response curve at low doses (the surrogate target site dose adjustment) pre-
dicts a lifetime risk of 10"6 or less at DEHP4dose levels as high as 116 ^g/kg per day.
These estimates of "virtually safe dose" assume equal risk to humans and rodents at the same dose
138
-------
RISK ASSESSMENT OF DEHP
TABLE 11. ESTIMATES OF SAFE LEVELS OF EXPOSURE TO DEHPa
Virtually Safe Dose (risk < lO'6) or
ADI f/jLg/kg per day)
Extrapolation Model
Maximum
Likelihood Estimate .
Lower 95th
Percentile Estimate
Multistage with applied
doses
Multistage with surrogate
doses
Mantel-Bryan with applied
doses
Mantel-Bryan with surrogate
doses
Threshold
2.2
116
1.5
86.3
11.9
791
< 3.5-70
"The safe dose levels listed represent the dose levels associated with a lifetime risk of 10'6 or
less predicted by the multistage or Mantel-Bryan model or the ADI predicted by application of
a safety factor to the NOEL as described in the text. In each case, the value for the species and
sex predicting the lowest safe level (highest risk) is listed.
level of DEHP. However, as we have discussed above, there is information on interspecies differ-
ences in target organ susceptibility and in physiological and biochemical responses to DEHP that
suggest that humans are likely to be less susceptible to the carcinogenic effects of DEHP than
rodents. That is, all of the estimates in Table 11 probably overestimate the risk to humans from
DEHP exposure.
In conclusion, there is a substantial body of data to indicate that the simple application of a low
dose extrapolation model to the available data on the carcinogenicity of DEHP to.estimate human
risks likely overestimates these risks. Factors contributing to such overestimation are (1) the likely
mechanism of carcinogenicity of DEHP in rodents and the likely nonlinear relationship between the
administered dose of DEHP and 'the dose of the proximate carcinogenic species, (2) differences in
target-site sensitivity between humans and rodents for liver tumors in general, and (3) differences in
the response of monkeys and probably humans to DEHP, which indicate that the hypothesized
mechanism of carcinogenicity of DEHP in rodents does not occur or occurs to a lesser extent in hu-
mans than in rodents.
ACKNOWLEDGMENTS
We thank the members of the Chemical Manufacturers Association Phthalate Esters Panel, par-
ticularly Dr. Elizabeth J. Moran, for advice and support in the conduct of this study. We also thank
the members of CMA's Toxicology Research Test Group, particularly Drs. Bernard Astill, Arthur
W. Lington, and Bernard F. Schneider for useful discussions on the technical content of this paper.
We are indebted to Ms. Karen McCrary, Mr. James Woldahl, and Ms. Claudia Barber for typing
and Dr. Kenny S. Crump for providing computer runs of the multistage and Mantel-Bryan models.
REFERENCES
1. LITTLE, A.D. (1982). Phthalates'in Consumer Products. Contract No. CPSC-C-80-1001. Washington,
D.C.: Consumer Product Safety Commission. t
2. NATIONAL TOXICOLOGY PROGRAM (NTP). (1982). NTP Technical Report on the Carcinogenesis
139
-------
TURNBULL AND RODRICKS
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Bioassay of.Di(2-Ethylhexyl)Phthalate (CAS No. 117-81-7) in F344 Rats and B6C3F Mice (Feed Study).
NIH Pub. No. 82-1773.
NATIONAL ACADEMY OF SCIENCES (NAS). (1983). Risk Assessment in the Federal Government:
Managing the Process. Washington, D.C.: National Academy Press.
MILKOV, L.E., ALDYREVA, M.V., POPOVA, T.B., LOPUKHOVA, K.A., MAKARENKO, YU.L.,
MALYAR, L.M., and SHAKHOVA, T.K. (1973). Health status of workers exposed to phthalate plasticiz-
ers in the manufacture of artificial leather and films based on PVC resins. Environ. Health Perspect 3,
175-178.
GILIOLI, R., GULGHERONI, C., TERRANCE, T., FILLIPINI, G. MASSETO, N., and BOERI, R.
(1978). A neurological electromyographic and electroneurographic study in subjects working at the pro-
duction of phthalate plasticizers: preliminary results. Med. Lav. 69, 631.
THIESS, A.M., KORTE, A., and FLEIG, I. (1978). Study of morbidity in BASF workers exposed to di-
(2-ethylhexyl)phthlate (DOP). 18th Annual Meeting of the Human Society of Industrial Medicine DD
137-151. ' '
CARPENTER, C.P., WEIL, C.S., and SMYTH, H.F. (1953). Chronic oral toxicity of di(2-ethyl-
hexyl)phthlate for rats, guinea pigs and dogs. Arch. Ind. Hgy. Occup. Med. 8, 219-226.
HARRIS, R.S., HODGE, H.C., MAYNARD, E.A., and BLANCHET, H.J., JR. (1956). Chronic oral
toxicity of 2-ethylhexyl phthalate in rats and dogs. Arch. Ind. Health 13, 259-264.
LEFAUX, R. (1968). Practical Toxicology of Plastics. Cleveland: CRC Press.
W.R. GRACE AND CO. (1948). Dioctyl Phthalate. Technical Brochure. As cited in: Krauskopf, L.G.
(1973). Studies on the toxicity of phthalates via ingestion. Environ. Health Perspect. 3, 61-72.
SONTAG, J.M., PAGE, N.P., and SAFFIOTTI, V. (1976). Guidelines for Carcinogen Bioassay in Small
Rodents. NCI Technical Report Series No. 1. DHEW Publication No. (NIH) 76-801.
NORTHUP, S., MARTIS, L., ULBRICHT, R., GARBER, J., MIRIPOL, J., and SCHMITZ, T. (1982).
Comment on the carcinogenic potential of di(2-ethylhexyl)phthalate. J. Toxicol. Environ. Health 10
493-518. ' '
REDDY, J.K., MOODY, D.E., AZARNOFF, D.L., and RAO, M.S. (1976). Di-(2-ethylhexyl)phthalate:
An industrial plasticizer induces hypolipidemia and enhances hepatic catalase and carnitine acetyltrans-
ferase activities in rats and mice. Life Sci. 18, 941-946.
MIDWEST RESEARCH INSTITUTE (MRI). (1982). Toxicological effects of diethylhexyl phthalate. Fi-
nal Report. MRI Project No. 7343-B. Phthalate esters program panel, Voluntary Test Program, Health
Effects Testing, Phase I: Validation results. Volume II. Washington, D.C.: Chemical Manufacturers As-
sociation.
BUTLER, W.H., and NEWBERNE, P.M. (eds.). (1975). Mouse Hepatic Neoplasia. Amsterdam: Elsevier.
CLAYSON, D.B. (1981). International Commission for Protection against Environmental Mutagensand
Carcinogens. ICPEMC Working Paper 2/3: Carcinogens and Carcinogenesis Enhancers. Mutat Res 86
217-229.
OGAWA, K., ONOE, T., andTAKEUCHI, M. (1981). Spontaneous occurrence of q-glutamyl transpep-
tidase-positive hepatocytic foci in 105-week-old Wistar and 72-week-old Fischer 344 male rats JNCI 67
407-412. , . '
WARD, J.M. (1983). Increased susceptibility of livers of aged F344/NCr rats to the effects of phenobarbi-
tal on the incidence, morphology, arid histochemistry of hepatocellular foci and neoplasms. JNCI 71,
815-823.
SCHULTE-HERMANN, R., TIMMERMANN-TROSIENER, I., and SCHUPPLER, J. (1983). Promo-
tion of spontaneous preneoplastic cells in rat liver as a possible explanation of tumor production by non-
mutagenic compounds. Cancer Res. 43, 839-844.
INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). (1979). Chemicals and indus-
trial processes associated with cancer in humans. IARC Monographs, Volumes 1 to 20. Lyon, France: In-
ternational Agency for Research on Cancer.
GRIESEMER, R.A., and CUETO, C., JR. (1980). Toward a classification scheme for degrees of experi-
mental evidence for the carcinogenicity of chemicals for animals. IARC Scientific Publications No 27
259-281. ' '
SQUIRE, R.A. (1981). Ranking animal carcinogens: A proposed regulatory approach. Science 214,
877—880.
SIMMON, V.F., KAUHANEN, K., and TARDIFF, R.G. (1977). Mutagenic activity of chemicals identi-
fied m drinking water. In: Progress in Genetic Toxicology. D. Scott, B.A. Bridges, and F.H. Sobels (eds.).
Amsterdam: North-Holland, pp. 249-258.
140
-------
RISK ASSESSMENT OF DEHP
24. YAGI, Y., TUTIKAWA, K., and SHIMOI, W. (1976). Teratogenicity and mutagenicity of a phthalate es-
ter. Teratology 14, 259-260.
25. RUBIN, R.J., KOZUMBO, W., and KROLL, R. (1979). Ames mutagenic assay of a series of phthalic acid
esters: positive response of the dimethyl and diethyl esters in TA100. In: Abstracts of Papers, Eighteenth
Annual Meeting, Society of Toxicology, Inc., New Orleans, Louisiana. New York: Academic Press.
26. KIRBY, P.E., PIZARELLO, R.F., LAWLOR, T.E., HAWORTH, S.R., and HODGSON, LR. (1982).
Evaluation of di-(2-ethylhexyl)phthalate and its major metabolites in the Ames test and L5178Y mouse
lymphoma mutagenicity assay. Environ. Mutagen. 5, 657-663. •
27. ZEIGER, E., HAWORTH, S., SPECK, W., and MORTELMANS, K. (1982). Phthalate estertesting in
the National Toxicology Program's Environmental Mutagenesis Test Development Program. Environ.
Health Perspect. 45, 99-101.
28. WARREN, J.R., LALWANI, N.D., and REDDY, J.K. (1982). Phthalate esters as peroxisome prolifera-
tor carcinogens. Environ. Health Perspect. 45, 35-40.
29. BARBER, E.D. MLJLHOLLAND, A., JAGANNATH, D.R., CIFONE, M., CIMINO, M., MYHR, B.,
and RUNDELL, J. (1985). The testing of di(2-ethylhexyl)phthalate (DEHP), mono(2-ethylhexyi)phthalate
(MEHP), di(2-ethylexyl)adipate (DEHA), and 2-ethylhexanol (2EH) in a battery of genotoxicity assays.
Paper presented at Society of Toxicology meeting, March, 1985.
30. TOMITA, I., NAKAMURA, Y., AOKI, N., and INUI, N. (1982). Mutagenic/carcinogenic potential of
DEHP and MEHP. Environ. Health Perspect. 45, 119-125.
31. YOSHIKAWA, K., TANAKA, A., YAMAHA, T., and KURATA, H. (1983). Mutagenicity study of nine
monoalkyl phthalates and a dialkyl phthalate using Salmonella typhimurium and Escherichia coii. Food
Chem. Toxicol. 21, 221-223.
32. DIVINCENZO, G.D., DONISH, W.H., MUELLER, K.R., HAMILTON, M.L., and BARBER, E.D.
(1983). Mutagenicity testing of urine from rats dosed with 2-ethylhexanol derived plasticizers (abstf.). En-
viron. Mutagen. 5, 471.
33. MALCOLM, A.R. (1982). Mutagenicity and tumor-promoting potential of di-(2-ethylhexyl)phthalate in
Chinese hamster cells. United States Environmental Protection Agency. Environmental Research Labora-
tory. Narragansett, R.I. Draft.
34. HODGSON, J.R., MYHR, B.C., McKEON, M., and BRUSICK, D.J. (1982). Evaluation of di-(2-ethyl-
hexyl)phthalate and its major metabolites in the primary rat hepatocyte unscheduled DNA synthesis assay
(abstr.). Environ. Mutagen. 4, 388.
35. BUTTERWORTH, B. (1984). The genetic toxicology of di(2-ethylhexyl)phthalate (DEHP). CUT Activi-
ties 4, 1-8.
36. TURNER, J.H., PETRICCIANI, J.C., CROUCH, M.L., and WENGER, S. (1974). An evaluation of the
effects of diethylhexyl phthalate (DEHP) on mitotically capable cells in blood packs. Transfusion 14,
560-566.
37. STENCHEVER, M.A., ALLEN, M.A., JEROMINSKI, L., and PETERSEN, R.V. (1976). Effects of
bis(2-ethylhexyl)phthalate on chromosomes of human leukocytes and human fetal lung cells. J. Pharma-
ceut. Sci. 65, 1648-1651.
38. TSUCHIYA, K., and HATTORI, K. (1976). A chromosomal study of cultured human leukocytes treated
with phthalic acid esters. Rep. Hokkaido Inst. Public Health 26, 114.
39. ABE, S., and SASAKI, M. (1977). Chromosome aberrations'and sister chromatid exchanges in Chinese
hamster cells exposed to various chemicals. JNCI 58, 1635-1641.
40. PHILLIPS, B.J., JAMES, T.E.B., and GANGOLLI, S.D. (1982). Genotoxicity studies of di(2-ethyl-
hexyl)phthalate and its metabolites in CHO cells. Mutat. Res. 102, 297-304.
41. ISHIDATE, M., JR., and ODASHIMA, S. (1977). Chromosome tests with 134 compounds on Chinese
hamster cells in vitro—A screening for chemical carcinogens. Mutat. Res. 48, 337-354.
42. THIESS, A.M., and FLEIG, I. (1978). Chromosomal studies of employees after exposure to di-2-ethyl-
hexyl phthalate (OOP) Zentralbl. Arbietsmed. Arbietsschutz Prophyl. 28, 351-355.
43. PUTMAN, D.L., MOORE, W.A., SCHECHTMAN, L.M., and HODGSON, J.R. (1983). Cytogenetic
evaluation of di-(2-ethylhexyl)phthalate and its major metabolites in Fischer 344 rats. Environ. Mutagen.
5,,227-231.
44. SINGH, A.R., LAWRENCE, W.H., and AUTIAN, J. (1974). Mutagenic and antifertility sensitivities of
mice to di-2-ethylhexyl phthalate (DEHP) and dimethoxy-ethyl phthalate (DMEP). Toxicol. Appl. Phar-
macoi. 29, 35-46.
45. AUTIAN, J. (1982). Antifertility effects and dominant lethal assays for mutagenic effects of DEHP. En-
viron. Health Perspect. 45, 115-118.
141'
-------
TURNBULL AND RODRICKS
46
47
48,
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
RUSHBROOK, C.J., JORGENSON, T.A., and HODGSON, J.R. (1982). Dominant lethal study of di-(2-
ethylhexyl phthalate and its major metabolites in ICR/SIM mice (abstr.). Environ. Mutagen 5, 387
ALBRO, P.W., CORBETT, J.T., SCHROEDER, J.L., JORDAN, S., and MATTHEWS, H.B. (1982).
Pharmacokinetics, interactions with macromolecules and species differences in metabolism of DEHP En-
viron. Health Perspect. 45, 19-25.
VON DANIKEN, A., LUTZ, W.K., JACKH, R., and SCHLATTER, C. (1984). Investigation of the po-
tential for binding of di(2-ethylhexyl)phthaiate (DEHP) and di(2-ethyihexyl)adipate (DEHA) to liver DNA
in vivo. Toxicol. Appl. Pharmacol. 73, 373-387.
SEED, J.L. (1982). Mutagenic activity of phthalate esters in bacterial liquid suspension assays Environ
Health Perspect. 45, 111-114.
OISHI, S., and HIRAGA, K. (1979). Effect of phthalic acid esters on gonadal function in male rats Bull
Environ. Contain. Toxicol. 21, 65-67.
OISHI, S., and HIRAGA, K. (1980). Effect of phthalic acid esters on mouse testes. Toxicol Lett 5
413-416.
ALBRO, P.W., CORBETT, J.T., SCHROEDER, J.L., and JORDAN, S.T. (1983). Incorporation of ra-
dioactivity from labeled di-(2-ethylhexyl)phthalate into DNA of rat liver in vivo. Chem. Biol. Interact. 44,
1—16.
WEISBURGER, J.H., and WILLIAMS, G.M. (1980). Chemical carcinogens. In: Casarett and Doull's
Toxicology, 2nd ed. J. Doull, C.D. Klaassen, and M.O. Amdur (eds.). New York: Macmillan, pp. 84-138.
MILLER, B.C. (1978). Some current perspectives on chemical carcinogenesis in humans and experimental
animals. Cancer Res. 38, 1479-1496. .
MILLER, E.G., and MILLER, J.A. (1966). Mechanisms of chemical carcinogenesis: Nature of proximate
carcinogens and interactions with macromolecules. Pharmacol. Rev. 18, 805-838.
REDDY, J.K., AZARNOFF, D.L., and HIGNITE, C.E. (1980). Hypolipidaemic hepatic peroxisome pro-
liferators form a novel class of chemical carcinogens. Nature 283, 397-398
ROBINSON, E., LINGTON, A., DIVINCENZO, G., CHADWICK, M., BRANFMAN, A., SILVEIRA
D., and McCOMISH, M. (1985). Nonlinearity of metabolism of DEHP with dose and prior exposure!
Paper presented at Society of Toxicology Meeting, March 1985.
MASTERS, C., and HOLMES, R. (1977). Peroxisomes: New aspects of cell physiology and biochemistry
Physiol. Rev. 57, 816-882.
LORD, J.M. (1980). Biogenesis of peroxisomes and glyoxysomes. In: Subcellular Biochemistry D B
Roodyn (ed.). New York: Plenum Press, Vol. 7, pp. 171-211.
TOLBERT, N.E. (1981). Metabolic pathways in peroxisomes and glyoxisomes. Annu. Rev. Biochem. 50,
CHANCE, B., SIES, H., and BOVERIS, A. (1979). Hydroperoxide metabolism in mammalian organs
Physiol. Rev. 59, 527-605.
FREESE, E. (1971). Molecular mechanisms, of mutation. In: ChemicalMutagens. Principles andMethods
for Their Detection. A. Hollaender (ed.). New York: Plenum Press, Vol. 1, pp 1-56
WANG, R.J., ANANTHASWAMY, H.N., NIXON, B.T., HARTMAN, P.S., and EISENSTARK, A.
(1980). Induction of single-strand DNA breaks in human cells by H2O, formed in near UV (black light)-
irradiated medium. Radial. Res. 82, 269-276.
SCHONEICH, J. (1967). The induction of chromosomal aberrations by hydrogen peroxide in strains of
ascites tumors in mice. Mutat. Res. 4, 384-388.-
STICK, H.F., WEI, L., and LAM, P. (1978). The need for a mammalian test system for mutagens: Action
of some reducing agents. Cancer Lett. 5, 199-204.
BRADLEY, M.O., HSU, I.C., and HARRIS, C.C. (1979). Relationship between sister chromatid ex-
change and mutagenicity, toxicity and DNA damage. Nature 282, 318-320.
MACRAE, W.D., and STICH, H.F. (1979). Induction of sister chromatid exchanges in Chinese hamster
ovary cells by thibl and hydrazine compounds. Mutat. Res. 68, 351-365
PARSHAD, R., TAYLOR, W.G., SANFORD, K.K., CAMALIER, R.F., GANTT, R., and TARONE,
R.E. (1980). Fluorescent light-induced chromosome damage in human IMR-90 fibroblasts. Role of hydro-
gen peroxide and related free radicals. Mutat. Res. 73, 115-124.
SPEIT, G., and VOGEL, W. (1982). The effect of sulfhydryl compounds on sister chromatid exchanges.
II. The question of cell specificity and the role of H2O2. Mutat. Res. 93, 175-183.
SPEIT, G., VOGEL, W., and WOLF, M. (1982). Characterization of sister chromatid exchange induced
by hydrogen peroxide. Environ. Mutagen. 4, 135-142.
LAZAROW, P.B., and DeDUVE, C. (1976). A fatty acyl-CoA oxidizing system in rat liver peroxisomes;
enhancement by clofibrate, a hypolipidemic drug. Proc. Natl. Acad. Sci. USA 73, 2043-2046.
142
-------
RISK ASSESSMENT OF DEHP
72. REDDY, J.K., and LALWANI, N.D. (1983). Carcinogenesis by hepatic peroxisome proliferators: Evalu-
ation of the risk of hypolipidemic drugs and industrial plasticizers to humans. CRC Crit. Rev. Toxicol. 12,
1-58.
73. BLANE, G.F., and PINAROLI, F. (1980). Fenofibrate: animal toxicology in relation to side effects in
man. Nouv. Presse Med. 9, 3737-3746. .
74. FAHIMI, H.D., REINICKE, A., SUJATTA, M., YOKATA, S., OZEL M., HARTIG, F., and STEG-
MEIER, K. (1982). The short- and long-term effects of bezafibrate in the rat. Ann NY Acad. Sci. 386,
111-135.
75. REDDY, J.K., WARREN, J.R., REDDY, M.K., and LALWANI, N.D. (1982). Hepatic and renal effects
of peroxisome proliferators. Ann. NY Acad. Sci. 386, 81-110.
76. LAZAROW, P.B. (1978). Rat liver peroxisomes catalyze the 0-oxidation of fatty acids. J. Biol. Chem.
253, 1522-1528.
77. LAZAROW, P.B. (1977). Three hypolipidemic drugs increase hepatic palmitoylcoenzyme A oxidation in
the rat. Science 197, 580-581.
78. OSUMI, T., and HASHIMOTO, T. (1978). Enhancement of fatty acyl-CoA oxidizing activity in rat liver
peroxisomes by di-(2-ethylhexyl)phthalate. J. Biochem. 83, 1361-1365.
79. OSUMI, T., and HASHIMOTO, T. (1979). Subcellular distribution of the enzymes of the fatty acyl-CoA
(3-oxidation system and their induction by di(2-ethylhexyl)phthalate in rat liver. J, Biochem. 85, 131-139.
80. MORTON, S.J. (1979). The hepatic effects of dietary di-2-ethylhexyl phthalate. Ph.D. Thesis, The Johns
Hopkins University, Baltimore., MD.
81. CANNING, A.E., and DALLNER, G. (1981). Induction of peroxisomes and mitochondria by di(2-.ethyl-
hexyl)phthalate. FEES Lett. 130, 77-79.
' 82. CANNING, A.E., KLASSON, E., BERGMAN, A., BRUNK, U., and DALLNER, G. (1982). Effect of
phthalate ester metabolites on rat liver. Acta Chem. Scand. Sec. B. B36, 563-565.
83. ALBRO, P.W., HAAS, J.R., PECK, C.C., ODAM, D.G., CORBETT, J.T., BAILEY, F.J., BLATT,
H.E., and BARRETT, B.B. (1981). Identification of the metabolites of di-(2-.ethylhexyl)phthalate in urine
from the African green monkey. Drug Metab. Dispos. 9, 223-225.
84. NIXON, J., and JACKSON, S.J. (1982). Bis(2-Ethylhexyl)Phthalate: A Comparative Subacute Toxicity
Study in the Rat and Marmoset. Report No. CTL/P/690. Prepared for CEFIC (Council European of the
Federations of the Industry Chemical).
85. INTERAGENCY REGULATORY LIAISON GROUP (IRLG). (1979). Scientific bases for identification
of potential carcinogens and estimation of risks. JNCI 63, 244-268.
86. MITCHELL, A.M., LHUGUENOT, J.C., BRIDGES, J.W., and ELCOMBE, C.R. (1984). Identifica-
tion of the proximate peroxisome proliferator(s) derived from di-(2-ethylhexyl)phthalate. Toxicol. Appl.
Pharmacol. In press.
87. HILL, H.A.O. (1979). The chemistry of dioxygen and its reduction products. In: Oxygen Free Radicals
and Tissue Damage. Ciba Foundation Symposium 65, 5-17.
88. HIROTA, N., and YOKOYAMA, T. (1981). Enhancing effect of hydrogen peroxide upon duodenal and
upper jejunal carcinogensis in rats. Gann 72, 811-812.
89. ITO, A., NAITO, M., NAITO, Y., and WATANABE, H. (1982). Induction and characterization of gas-
tro-duodenal lesions in mice given continuous oral administration of hydrogen peroxide. Gann 73, 315-
322.
90. LEVIN, D.E., HOLLSTEIN, M., CHRISTMAN, M.F., SCHWIERS, E.A., and AMES, B.N. (1982). A
new Salmonella tester strain (TA 102) with A"T base pairs at the site of mutation detects oxidative muta-
gens. Proc. Natl. Acad. Sci. USA 79, 7445-7449.
91. AMES, B.N. (1983). Dietary carcinogens and anticarcinogens. Science 221, 1256-1264.
92. ALBRO, P.W., THOMAS, R., and FISHBEIN, L. (1973). Metabolism of diethylhexyl phthalate by rats.
Isolation and characterization of the urinary metabolites. J. Chromatography 76, 321-330.
93. ALBRO, P.W., TONDEUR, I., MARBURY, D., JORDAN, S., SCHROEDER, J., and CORBETT, J.T.
(1983). Polar metabolites of di-(2-ethylhexyl)phthalate in the rat. Biochem. Biophys. Acta 760, 283-292.
94. MORAN, E., LINGTON, A., DIVINCENZO, G., ROBINSON, E., CHADWICK, M., BRANFMAN,
A., McCOMISH, M., and SILVEIRA, D. (1985). Species differences in the metabolism of DEHP in mon-
keys, rats, and mice. Paper presented at Society of Toxicology meeting, March 1985.
95. FOOD SAFETY COUNCIL. (1980). Proposed System for Food Safety Assessment. Final Report of the
Scientific Committee of the Food Safety Council, Washington, D.C., June, 1980.
96. BROWN, C.C. (1984). H5gh-to low-dose extrapolation in animals. In: Assessment and Management of
Chemical Risks. J.V. Rodricks and R.G. Tardiff (eds.). ACS Symposium Series 293. Washington, D.C.:
American Chemical Society, pp. 57-79.
143
-------
TURNBULL AND RODRIGKS
97. GEHRING, P.J., and BLAU, G.E. (1977). Mechanisms of carcinogenesis: Dose response. J. Environ.
Pathol. Toxicoi. 1, 163-179.
98. REITZ, R.H., GEHRING, P.J., and PARK, C.N. (1978). Carcinogenic risk estimation for chloroform:
An alternative to EPA's procedures. Food Cosmet. Toxicoi. 16, 511-514.
99. GEHRING, P.J., WATANABE, P.O., and PARK, C.N. (1978). Resolution of dose-response toxicity
data for chemicals requiring metabolic activation: Example—vinyl chloride. Toxicoi. Appl. Pharmacol.
44, 581-59L
100. LHUGUENOT, J.C., MITCHELL, A.M., and ELCOMBE, C.R. (1984). Dose- and time-dependency of,
mono(2-ethylhexyl)phthalate (MEHP) metabolism in primary rat hepatocyte cultures. Presented at Society
of Toxicology Conference, Atlanta, GA, March 1984. Abstract No. 383.
101. LHUGUENOT, J.C., LOCK, E.A., and ELCOMBE, C.R. (1984). Dose- and time-dependency of mono-
and di-(2-ethylhexyl)phthalate (MEHP and DEHP) metabolism in rats. Presented at Society of Toxicology
Conference, Atlanta, GA, March 1984. Abstract No. 382.
102. LHUGUENOT, J.C., MITCHELL, A.M., and ELCOMBE, C.R. (1984). Dose- and time-dependent me-
tabolism of mono(2-ethylhexyl)phthalate and its major metabolites in rat hepatocyte cultures. Toxicoi.
Appl. Pharmacol. In press.
103. ANDERSON, M.W., HOEL, D.G., and KAPLAN, H.L. (1980). A general scheme for the incorporation
, of pharmacokinetics in low dose risk estimation for chemical carcinogenesis: Example —vinyl chloride.
Toxicoi. Appl. Pharmacol. 55, 154-161.
104. CRUMP, K.S., HOEL, D.G., and LANGLEY, C.H. (1976). Fundamental carcinogenic processes and
their implications for low dose risk assessment. Cancer Res. 36, 2973-2979.
105. CRUMP, K.S., HOWE, R., and FIERING, M.B. (1980). Approaches to carcinogenic, mutagenic and
teratogenic risk assessment. Task A, Subtask No. 5, Summary Report, contract No. 68-01-5975, U.S. En-
vironmental Protection Agency.
106. U.S. ENVIRONMENTAL PROTECTION AGENCY (EPA). (1980). Water Quality Criteria Document;
Availability. Appendix C—Guidelines and Methodology Used in the Preparation of Health Effect Assess-
ment Chapters of the Consent Decree Water Critera Documents. Fed. Reg. 45, 79347-79357.
107. CONSUMER PRODUCTS SAFETY COMMISSION (CPSC). (1983). Children's Chemical Hazards. Risk
Assessment on Di-(2-Ethylhexyl)Phthalate in Children's Products. Chemical Hazards Program, Director-
ate for Health Sciences, CPSC. August 1983.
108. DAY, N.E., and BROWN, C.C. (1980). Multistage models and primary prevention of cancer. JNCI 64,
977-989.
109. MANTEL, N., and BRYAN, W. (1961). "Safety" testing of carcinogenic agents. JNCI 27, 455-470.
110. MANTEL, N., BOHIDAR, N.R., BROWN, C.C., CIMINERA, J.L., and TUKEY, J.W. (1975). An im-
proved Mantel-Bryan procedure for "safety" testing of carcinogens. Cancer Res. 35, 865-872.
111. NATIONAL ACADEMY OF SCIENCES. (1977). Drinking Water and Health. Washington, D.C.: NAS.
112. HOEL, D.G., GAYLOR, D., KIRSCHSTEIN, R., SAFFIOTTI, U., and SCHNEIDERMAN, M. (1975).
Estimation of risks of irreversible delayed toxicity. J. Toxicoi. Environ. Health 1, 133-151.
113. NATIONAL ACADEMY OF SCIENCES. (1975). Pest Control: An Assessment of Present and Alterna-
tive Technologies. Vol. 1. Contemporary Pest Control Practices and Prospects: The Report of the Execu-
tive Committee. Washington, D.C.: NAS.
114. CROUCH, E., and WILSON, R. (1979). Interspecies comparison of carcinogenic potency. J. Toxicoi. En-
viron. Health 5, 1095-1118.
115. FREIREICH, E.J., GEHAN, E.A., RALL, D.P., SCHMIDT, L.H., and SKIPPER, H.E. (1966). Quan-
titative comparison of toxicity of anticancer agents in mouse, rat, hamster, dog, monkey, and man. Can-
cer Chemother. Rep. 50, 219-244.
116. SCHUMANN, A.M., QUAST, J.E., and WATANABE, P.O. (1980). The pharmacokinetics and macro-
molecular interactions of perchloroethylene in mice and rats as related to oncogenicity. Toxicoi. Appl.
Pharmacol. 55, 207-219.
117. MILLER, E.G., MILLER, J.A., AND ENOMOTO, M. (1964). The comparative carcinogenicities of 2-
acetylaminofluorene and its N-hydroxy metabolite in mice, hamsters, and guinea pigs. Cancer Res. 24,
2018-2031.
118. JAEGER, R.J., and RUBIN, R.J. (1970). Plasticizers from plastic devices. Extraction, metabolism, and
accumulation by biological systems. Science 170, 460-462.
119. JAEGER, R.J., and RUBIN, R.J. (1973). Extraction, localization, and metabolism of di-2-ethylhexyl
phthalate from PVC plastic medical devices. Environ. Health Perspect. 3, 95-102.
120. REDDY, J.K., and RAO, M.S. (1978). Enhancement by Wy-14,643, a hepatic peroxisome proliferator, of
diethylnitrosamine-initiated hepatic tumorigenesis in the rat. Br. J. Cancer 38, 537-563.
144
-------
RISK ASSESSMENT OF DEHP
121. WARD, J.M., RICE, J.M., CREASIA, D., LYNCH, P., andRIGGS, C. (1983). Dissimilar patterns of
promotion by di(2-ethylhexyl)phthalate and phenobarbital of hepatocellular neoplasia initiated by diethyl-
nitrosamine in B6C3F1 mice. Carcinogenesis 4,, 1022-1029.
122. DEANGELO, A.B., and GARRETT, C.T. (1983). Inhibition of development of preneoplastic lesions in
the livers of rats fed a weakly carcinogenic environmental contaminant. Cancer Lett. 20, 199-205.
123. PURCHASE, I.F.H. (1980). Interspecies comparisons of carcinogenicity. Br. J. Cancer 41, 454-468.
124. HINTON, R.H., PRICE, S.C., HALL, D., GRASSO, P., and MITCHELL, F. (1982). Report to CEFIC
on a 28 Day Dose and Time Response Study on Di-ethylhexyl Phthalate in Rats. Study 5/81/TX. Robens
Institute of Industrial and Environmental Health and Safety, University of Surrey, Guildford, Surrey,
England.
125. BRITISH INDUSTRIAL BIOLOGICAL RESEARCH ASSOCIATION (BIBRA). (1984). A 21-day dose-
response study of di(2-ethylhexyl)phthalate in rats. Report No. 0512/1/84. Draft. July 1984.
Address reprint requests to:
Duncan Turnbull
Environ Corporation
WOO Potomac Street, N. W.
Washington, DC 20007
Submitted July 26, 1984
Accepted February 11, 1985
145
-------
A Time-to-Response Perspective on
Ethylene Oxide's Carcinogenicity
Robert LSielken Jr.
President
Sielken, Inc
Suite 410,3833 Texas Avenue
Bryan,Texas 77802
September 1987
-------
(b) the secondary data set, equaling the primary data set plus the histologically
examined female rats which were sacrificed at the end of the experiment,
(c) the EPA data set, corresponding to the secondary data set minus the rats
which died or were sacrificed prior to 18 months plus the 18-month
scheduled sacrifices in the 0 ppm and 100 ppm groups but not in the 10 ppm
and 33 ppm groups. (49)
More detailed time-to-response data was also available. Time was measured in
terms of months from the onset of exposure. The time of death was recorded for all
rats which died. No further classification of the cause of death for these animals
was reported.
The carcinogenic response analyzed in this chapter is defined as death with MCL
in a female rat.
Any female rat which had MCL at the time it died or was sacrificed was
considered to have had the carcinogenic response, death with MCL
Risk Characterizations Emphasizing Time-to-Response but Not Requiring Dose-
Response Modeling
It is appropriate to characterize risk in terms of the time that a response occurs or
the amount of time proceeding a particular consequence (29,36,39). Some such
characterizations can be made without having to do dose-response modeling. As
shown below, these largely assumption free, straightforward representations of the
experimental data suggest a high degree of similarity between the real risks at
0 ppm and 10 ppm.
Table 2 indicates the number of female rats which died with mononuclear cell
leukemia (MCL) during each of the exposure months, the number which died
without MCL, and the number which died but were lost to follow-up (that is, not
-------
histologically examined for MCI). There were 120 female rats in each of the two
identical control groups (denoted as Control I and Control II or Cl and Cll herein) and
in each of the 10 ppm, 33 ppm, and 100 ppm exposure groups. The control level is
referred to herein as the 0 dose level.
There were, 1,10,10, and 20 rats randomly selected at 3,6,12, and 18 months
respectively for interim sacrifice at each dose level (0,10.33, and 100 ppm). There
was also a terminal sacrifice at approximately 24 months for all female rats. All
rats are accounted for in Table 2.' The animals which were randomly selected for
sacrifice at scheduled times can be distinquished in Table 2 from the other rats; in
that, a table entry is a ratio like x/y if there were y > 0 rats which were scheduled
sacrifices. For example. Table 2 implies the following:
(a) there were no female rats which died with MCL (among those not lost
to follow-up) before 18 months of exposure;
1$
(b) in Control II and at 33 and 100 ppm one. rat died with MCL during the
18-th exposure month;
(c) the one rat which died with MCL at 100 ppm, during the 18-th
exposure month, was a scheduled sacrifice; and,
(d) at 100 ppm there were a total of 16 female rats which died with MCL in
scheduled sacrifices while 12 died with* MCL by deaths that were not
*
scheduled sacrifices.
The life table analysis estimates of the probabilities of a female rat dying with
MCL following various numbers of exposure months are shown in Table 3. These
estimates are straightforward nonparametric estimates in that they do not presume
any model or mathematical form for how these probabilities change with dose or
time. The estimates are computed using the fairly standard technique described in
(47). This method incorporates all of the rats and the length of time they were at
risk.
-------
11
There are at least two conclusions which can be drawn from Tables 2 and 3. The
first is that, if a response occurs at all, it occurs very late relative to an average rat
lifetime -- mostly in the 24-th month and no sooner than the 18-th month in a 24-
month study. The second conclusion is that the number and timing of the responses
atO ppm and 10 ppm are very similar. The latter conclusion is also evident in Figure
1 which indicates the plots of the life table analysis estimates versus time for the 10
ppm exposure group and the two control groups. The plot for the 10 ppm group
lies between those of the two control groups except for the very last month.
There are several other ways to characterize
(a) the lateness of the MCL response in female rats,
(b) the similarity between the response times for the rats exposed to 10
ppm and the response times for the control rats, and
(c) the changes in the response times at 33 ppm and 100 ppm relative to
those at 0 ppm and 10 ppm.
For example, Table 4 indicates the number of exposure months until the proportion
of animals which die with MCL reaches a prescribed percentage (1%, 5%, 10%, or
20%). In particular, it took the same number of months (23 months) for both the 10
ppm group and the combined controls to reach a 5% incidence rate whereas it took
a noticeably shorter time attKe 33 ppm and 100 ppm levels (namely, 20 and 21
months respectively). Throughout Table 3 the rats at 0 ppm and 10 ppm took similar
times to reach the same incidence rate (risk level); in fact, it never took less time to
reach a specified incidence rate at 10 ppm than it did for the combined controls.
The average amount of time until a response occurs cannot be calculated when
some individuals do not develop the response. However, the average amount of
time in an experiment during which an individual has not had the response can be
calculated. This latter average, called the mean response free period, is not the same
as the average time until the response occurs. For instance, if no rats died with MCL
-------
during a 24 month experiment, the average response free period would be 24
months; however, the average time to death with MCL would certainly be greater
than 24 months but how much greater is unknown..
Table 5 shows the average number of exposure months in the experiment that
the female rats were free from the specified response, death with MCL. The
maximum possible response free period is 24 months (the length of the
experiment). The female rats in the 10 ppm group averaged 23.55 months of
response free time (98.1 % of the maximum response free period). The two control
groups averaged 23.45 months (97.7%) and 23.70 months (98.7%) without
mortality with MCL. Hence, the average for the 10 ppm group is between the
averages for the two control groups. The average for the 33 ppm and 100 ppm
groups were slightly lower (namely, 23.10 months (96.2%) and 22.88 months
(95.3%) respectively) but noticeably different from that for the 10 ppm group. The
similarity between the control groups and the 10 ppm group is clear. The
dissimilarity between the 10 ppm group and both the 33 ppm and TOO ppm groups
is also clear.
Simple Statistical Analysis of Time-to-Response Data
Several two-sided 95% confidence intervals on the lifetime probability of a
female rat dying with MCL are shown in Table 6. Regardless of which subset of the
a
experimental data is being considered, there is substantial overlap between the
confidence intervals for the MCL response rate at 0 ppm and at 10 ppm (see also
Figure 2). There is no statistically significant difference between the lifetime MCL
response rate for the 10 ppm group and the combined controls (even at the 20%
significance level). This suggests that there is no statistically detectable increase in
-------
13
the lifetime probability of dying with MCL from being exposed to 10ppmEO
instead of 0 ppm.
Not only is the lifetime probability of dying with MCL similar at 0 ppm and 10
ppm, but also there is no difference between 0 ppm and 10 ppm in when an
individual is likely to die with MCL In particular, being at 10 ppm instead of 0 ppm
does not appear to increase the likelihood of dying with MCL earlier. The
hypothesis that there is no shifting of the likelihood of dying with MCL from one
time to another (especially to an earlier time) is the hypothesis that the distribution
of the time to death with MCL is the same for different dose levels. This hypothesis
i
was tested using the available information on the length of time without a female
rat dying with MCL. The statistical procedures decribed in (48) are appropriate for
this analysis and were applied to the data on the female rats which were
histologically examined for MCL and which died or were sacrificed at the end of the
experiment. (The procedures considered the frequency of death with MCL in three
intervals: 0-18,19-21, and 22-24exposure months.) The hypothesis that the 0,10,
33, and 100 ppm groups had the same time-to-response distribution was rejected at
the 5% significance level. The hypothesis that the 0,10, and 33 ppm groups"had the
same time-to-response distribution was also rejected. However, the hypothesis that
the controls and the 10 ppm group had the same time-to-response distribution was
not rejected. These tests, which do not require any specific dose-response modeling
assumptions, imply that the time-to-response distributions for the controls and the
rats at 10 ppm are similar, and that the time-to-response distributions at 33 ppm
and 100 ppm are different from those at 0 ppm and 10 ppm. These results are
consistent with the implications of Table 5 where the estimated mean times without
a female rat dying with MCL are nearly identical at 0 ppm and 10 ppm but smaller at
33 ppm and 100 ppm. In short, even though the same cannot be said for the 33 ppm
-------
u
and 100 ppm groups, the 10 ppm group is similar to the controls with respect to
when and how often death with MCL occurs.
EPA's Position on Time-to-Response Information
The final report (49) published by the U.S. EPA in June, 1985, contains a summary
of their quantitative risk assessment for EO. Their approach to estimating the risk at
very low doses like many of the past attempts by regulatory agencies, fails to reflect
the available time-to-response information in either its dose-response modeling or
its risk characterizations. The methodology to incorporate this additional
information has been available for almost 10 years and, as discussed previously, it
should be a part of most assessments if the objectives of the EPA's Cancer Risk
Guidelines (46) and those of the NAS(1) are to be met.
There have frequently been several methodogical shortcomings in the past
quantitative risk assessments prepared by regulatory agencies for ethylene oxide,
and other chemicals, in addition to the failure to reflect the available time-to-
response information. Many of these shortcomings are discussed briefly in
Appendix B. *
EPA's Interpretation of the Ethvlene Oxide Bioassav Data
The EPA's quantitative risk assessment failed to incorporate a consideration of
the time-to-response information in its summary risk characterizations (49). Since
1982, the EPA has summarized the results of their risk modeling efforts in terms of
what EPA calls "unit risk estimates" (although these are really bounds on risk and
not estimates). These risk characterizations ignore the time-to-response
information and are based solely on the proportions of animals developing the
-------
15
specified response at any time. Such tumor/no tumor data or response/no response
data are quantal response data and do not reflect the time at which the response of
concern occurs.
In EPA's description of the BRRC study, the cumulative mortality percentages
versus exposure months were tabulated, and the agency noted that
"At no time were significant increases in mortality observed in the 10 ppm
exposure group of either sex." (49, page 9-100)
Regretably they did not present the corresponding tables and implications for death
with MCL or other potential carcinogenic responses. Had they done so, they would
have discovered that the similarity between 0 ppm and 10 ppm observed for
mortality from all causes combined extends to the specific carcinogenic responses
they eventually focused on (MCL in particular). Furthermore, they would have
realized that there was more to the risk characterization of EO than that suggested
by the quantal response data alone.
The only time-to-response information for the carcinogenic responses which the
EPA reported was a table of "Time in months to First Tumor" and "Time in months
to Median Tumor". EPA summarized their interpretation of this data in the
following two sentences (49, page 9-105):
"The time to first tumor for some neoplasms (but not for mononuclear cell
leukemias) was decreased in the high-dose group as compared to controls,
as shown in Table 9-23. Median time-to-tumor was not reduced."
This information deserves more emphasis and, since they pursued no further time-
to-response analyses, more discussion and follow-up. The similarity of the time-to-
response behavior at 10 ppm to that at the control level was not explicitly noted by
the EPA, although, as discussed previously, it woulcl appear to be very relevant data
to be considered by risk managers. Further, the omission of most of the time-to-
-------
16
response information and implications should be included in the EPA's discussion of
the uncertainties in their risk assessment.
Effect of Evaluating Only a Particular Portion of the Quantal Response Data
The inclusion or exclusion of subgroups of female rats on the basis of different
factors such as sacrificed versus non-sacrificed or gross examination versus
histological examination and inconsistency across the dose levels in the
determination of the number of animals at risk caused the EPA risk characterization
to differ markedly from the risk characterizations corresponding to alternative (and
probably more reasonable) representations of the quanta! response data. In most
experiments there are several ways to detemine which animals are included in the
quantal response data and which animals are excluded (not counted). The different
determinations may not be equally appropriate or equally reflective of the
underlying dose-response relationship. Furthermore, the same risk characteristic
can differ by several orders of magitude for different determinations of the quantal
response data. The most informative determination may depend on the
experimental protocol and the chemical's time-to-response behavior. The
sensitivity of the risk characterization to the determination of the representation of
the quantal response data should be evaluated and made clear to the risk manager.
In the primary data set the observed proportion dying with MCL did not increase
as the dose increased. As indicated in Table 7, the observed percentages were
30.6%, 13.6%, 32.3%, and 25.5% for the combined controls, 10 ppm, 33 ppm, and
100 ppm groups respectively. In fact, the percentages at 10 ppm, 33 ppm and 100
ppm are between the 22.2% and 38.9% observed in the two individual control
groups. Among the female rats which died there was no increased risk of dying
-------
17
with mononuclear cell leukemia from a lifetime of inhalation exposure to ethylene
oxide at or below 100 ppm.
In their analysis (49), the EPA evaluated a mixture of sacrificed and non-sacrificed
animals. The usefulness of this analysis is questionable since there was not an '
increasing dose-response relationship observed among the non-sacrificed animals,
but there was one observed in the female rats sacrificed at 24 months.
The observed percentages of female rats with MCL in the 24-month sacrificed
group were 9.5%, 20.4%, 29.2%, and 57.7% in the combined controls, 10 ppm, 33
ppm, and 100 ppm groups respectively. The percentage for the 24-month sacrificed
group at 100 ppm is significantly greater than the percentage for the non-sacrificed
group at 100 ppm (at the 1% significance level). The corresponding differences in
the percentages for the 10 ppm and 33 ppm are not significant (even at the 20%
significance level). However, the percentage for the 24-month sacrificed controls is
significantly decreased relative to the percentage for the non-sacrificed controls (at
the 1% significance level).
The reasons for the observed differences between the non-sacrificed and
24-month sacrificed rats are not evident from the numbers themselves. However,
the numbers are consistent with the hypothesis that the dose-related occurrences of
MCL were "incidental" as opposed to "lethal"; that is, consistent with the dose-
related occurrences not leading to an increased threat to life. The numbers are also
consistent with the claim that the dose-related occurrences of MCL happen very late
in the lifetime of female rats. In either case, the EPA's risk characterizations do not
reflect health impact since they either are reflecting an incidental carcinogenic
response and/or they do not reflect the amount of a lifetime adversely affected by
exposure.
The specific criteria used by EPA to select their single representation of the
quantal response data did not have a consistent impact on the different dose
-------
18
groups. In fact, the criterion of including only histologically examined rats appears
reasonable at first glance; however, it meant that the 20 female rats sacrificed at 18
months in each of the 10 ppm and 33 ppm groups (i.e., about 17% of these dose
groups) were excluded from the quantal response data (because they were only
grossly examined) while the 40 female control rats and the 20 female rats at 100
ppm which were sacrificed at 18 months were included (because they were
histologically examined). It is true that the rats in the 10 ppm and 33 ppm groups
were only grossly examined for MCLand not histologically examined; however,
none of the 101 female rats which were histologically examined before 18 months
had MCL, and only 1 of the 20 histologically examined female rats which were
exposed to 100 ppm and sacrificed at 18 months had MCL Thus, the female rats
which were at 10 ppm and 33 ppm and sacrificed at 18 months probably did not
have MCL Therefore, by including the 40 control rats sacrificed at 18 months and
excluding the 20 rats at each of 10 ppm and 33 ppm which were sacrificed at 18
months, EPA's choice decreased the proportion with MCL among the controls and,
most likely, increased the proportion with MCL at 10 ppm and 33 ppm. This choice
exaggerates any increase in the proportion with MCL at 10 ppm and 33 ppm relative
to the controls. In short, the EPA's non-uniform treatment of the female rats
sacrificed at 18 months results in biased dose-response information which inflates
the risk characterization -- in fact, more than doubled the implied risk.
Several alternatives to the EPA's representation of the quantal response.data on
female rats with MCL are indicated in Table 7. They include
(a) the primary data set,
(b) the secondary data set -- the primary data set plus the 24 month
sacrifices,
(c) EPA's choice plus the rest of the 18-month sacrifices,
-------
19
(d) all histologically examined female rats which survived the first year of
the experiment, and
(e) the female rats in (d) plus the rest of the 18-month sacrifices.
In all of these alternatives, the apparent increase in the percentage of female rats
with MCI between the controls and the 10 ppm group is less than the implied.
increase under EPA's choice. For alternative (a), the primary data set, there is no
implied increase. For alternatives (b), (c), and (e) the implied increases are only
roughly one-half of the implied increase under the EPA's choice. Alternative (d)
suffers from the same disparate treatment of the 18-month sacrifice data as EPA's
choice and implies risks twice those implied by (e).
All of the representations of the quanta! response data on MCL which combine
the sacrificed and non-sacrificed female rats hide the fact that there was no
observed increasing dose-response relationship among the non-sacrificed rats. Of
course, these representations also failed to indicate the very long period of time
without evidence of a carcinogenic effect.
Risk Characterizations can Emphasize Time-to-Response
As suggested previously, a practical way to characterize the effects of a particular
exposure is to describe the corresponding average amount of time in a specified
observation period during which the subject is free from a specified response. This
characterization has been called the mean response free period or mean free period
(36,39). If the observation period is 24 months (as it was in the BRRCstudy), then
the mean response free period could be as long as is 24 months. As indicated in
Table 5, when the response is death with MCL, the observed mean response free
periods (observed mean numbers of months without dying with MCL) are 23.58,
23.55,23.10, and 22.88 months for a female rat at 0,10,33, and 100 ppm
-------
20
respectively. Thus, the risk for female rats at 10 ppm can be characterized as a
decrease of 0.03 months in the mean response free period (0.03 = 23.58 - 23.55).
For female rats exposed between 0 ppm and 10 ppm, the overall observed rate
of decrease in the observed mean respqnse free periods for dying with MCL was
23.58 months - 23.55 months = 0.003 months/ppm
10 ppm-0 ppm
If a linear interpolation between 0 ppm and 10 ppm is applied, this rate of decrease
implies that a female rat's timespan without dying with MCL during a 24 month
exposure decreases approximately 0.003 months (approximately 0.09 days, 2.2
hours, or 0.0125% of a 24 month lifetime) with each 1 ppm increase in exposure
from 0 ppm to 10 ppm. The linearity assumption is almost certain to overstate the
actual risks. If the mean response free period decreased linearly with the dose level
at a rate of 0.003 months/ppm, then dose levels of 2.2 ppm, 0.31 ppm, and 0.013
ppm would correspond to average reductions of approximately 0.00391%,
0.00016%, 0.0000027%, of a 24 month expected lifetime. These percentages
correspond to approximately one week, one day, and one hour in a 70 year
expected lifetime.
The observed mean response free periods and estimates of dose levels
corresponding to specified reductions in the mean response free period described in
the two preceding paragraphs do not take advantage of the experimental
information on the shape of the dose-response relationship obtained by combining
the animal information from several different dose levels together. Dose-response
modeling allows the information from different experimental dose levels to be
combined. In addition, instead of arbitrarily assuming how risk characteristics
should be interpolated between dose levels, dose-response modeling allows the
-------
21
experimental data themselves to imply how interpolations or extrapolations over
doses and over times are to be done.
The family of multistage-Weibull time-to-response models can be used to
reasonably model the dose-response relationships for ethylene oxide inhalation in
rats. These models are generalizations of the quantal multistage model typically
used by the EPA. The multistage-Weibull models can be used to extrapolate over
time as well as over dose. The multistage-Weibull models fit to the available dose-
response information has the probability of a specified response occurring by time t
at dose d in the absence of competing risks equal to
P(t;d) = 1 -exp{-[oo + aid + a2d2 + a3d3] x [t-pilp2>
where ao. ai, 02, as, pi. and 02 are the unknown parameters whose values are to be
estimated from the experimental data (12).
In such analyses, the dose scale used for modeling the dose-response relationship
should be the same as the dose scale that is going to be used for species
extrapolation. Table 8 indicates the dose levels to which the rats in the BRRC study
were exposed on the three dose scales corresponding to EO concentration in air
(ppm), intake relative to body weight (mg/kg/day), and intake relative to surface
area (mg/kg2/3/day).
The rats in the BRRC study were exposed for 6 hours/day, 5/days/week for 2 years
which is equivalent to approximately 18% of a complete 2-year lifetime. Hence, the
lifetime average exposure concentrations are 18% of the experimental
concentrations (0,10,33, and 100 ppm). Tyler and McKelvey (50) reported that the
average cumulative daily intake for male Fischer 344 rats inhaling ethylene oxide
vapor for 6 hours/day was
* " • ' -
(a) 2.7 mg/kg/day when the concentration of ethylene oxide was 10 ppm,
and
-------
22
(b) 20.24 mg/kg/day when the concentration of ethylene oxide was 100
ppm.
Since the BRRC rats were exposed 5 days/week, the corresponding lifetime average
*'
exposures were 5/7 times these numbers; that is 1.93 mg/kg/day at 10 ppm and
14.46 mg/kg/day at 100 ppm. Linear interpolation between these two numbers
gives 5.13 mg/kg/day for 33 ppm. These numbers were also assumed to apply to
female rats. The exposures to male and female rats can be converted from a body
weight scale (mg/kg/day) to a body surface area scale (mg/kg2/3/day) by multiplying
the mg/kg/day exposures by (0.42)1/3 and (0.22)1/3 for males and females
respectively where 0.42 kg and 0.22 kg were the average weights of male and
female rats respectively in the BRRC study (42,43).
The estimated dose-response relationships for female raits dying with MCL were
affected by the dose scale used during the model fitting. Such affects were due
solely to the differences in dose scales and occur even without extrapolating across
species. These effects are illustrated in Table 9 using the experimental data for the
controls, the 10 ppm group, and the 33 ppm group. As shown, for a female rat
f
inhaling EO at 1 ppm for 6 hours/day, 5 days/week for a lifetime, the fitted model
values for the added probability of dying with MCL were 0.0056,0.0028, and 0.0028
when the modeling dose scales were lifetime average concentration (ppm), intake
relative to body weight (mg/kg/day), and intake relative to surface area
(mg/kg2/3/day). The corresponding decreases in the mean response free period
during the 24 month experiment were 0.0065,0.0033, and 0.0033 respectively.
Thus, for this data set the choice of the dose scale used during the model fitting
made about a two-fold difference in the estimated risks for female rats
extrapolated from the experimental levels of 10 ppm and 33 ppm to 1 ppm.
-------
23
The reason for these differences in estimated risk is that, when the dose scale is
changed from ppm to mg/kg/day or mg/kg2/3/day, the relative distance between the
doses corresponding to 0,10, and 33 ppm changes. On the lifetime average ppm
scale the value corresponding to 33 ppm is 3.3 times greater than that for 10 ppm
JP
whereas on either the mg/kg/day or the mg/kg2/3/day scales the value
corresponding to 33 ppm is only about 2.7 times greater than that for 10 ppm. Thus,
the increases in the proportion dying with MCL observed between 0 ppm and 33
ppm are shallower (more linear and more rapidly increasing) on the ppm scale than
the other scales, and consequently the low dose risks are estimated to be greater
when the modeling is done on the ppm scale. There is no difference between the
estimates for female rats obtained using the mg/kg/day and the mg/kg2/3/day scales
since these two scales are linearly related - here one scale is a simple constant
multiple of the other.
The Best Available Risk Characterization
The multistage-Weibull time-to-response model, when applied to the primary
*
data on female rats dying with MCL, indicated that risks do not increase as the dose
level increases. That is, based on the primary data set, the fitted time-to-response
model implied that, as the dose increases between 0 ppm and 100 ppm, the
probability of a female rat dying with MCL does not increase and the mean response
free period does not decrease! These implications based on the time-to-response
information are consistent with the implication of no increases in risk with increases
in dose based on the quanta! response data for female rats which died.
The best available risk characterization of EO based on the female rats dying
with MCL is that there is no increased risk at low levels of exposure, particularly
below 10 ppm. The remainder of this section considers what the risk characteristics
-------
B.3
DOSE-RESPONSE
EVALUATION
Attribute 2 The presentation of dose-response evaluation includes both an
, upper and lower bound of potency estimates and, wherever
possible,- some measure of the central tendency.
SOURCE Case Study D. Formaldehyde (Pages 1-29 to 1-31).
Note This excerpt shows a tabulation of upper bound estimates of risk for
several exposure scenarios and MLE estimates as a central tendency.
A lower bound of zero is stated. Uncertainty is also discussed.
SOURCE Case Study A. DEHP (Pages 138439),
Note See Dose-Response Attribute 1 in this Appendix. This report displays
upper and lower estimates for a "virtually, safe dose" or ADI derived
from several models and data sets. Uncertainty- is discussed, but no
central tendency is indicated.
-------
Assessment, of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
1.4,3. Numerical Risk Estimates
The risk estimates for the linearized multistage procedure,
upper bound (UB) and maximum likelihood estimates (MLE)7 at
various exposure levels are presented in Table 1-2. Risks at any
exposure level range from the upper bound to zero. An
established procedure does not yet exist for making "most likely"
or "best",estimates of risk within the 'range of uncertainty
defined by the upper bound and zero. The upper bound estimate
for excess lifetime risk of developing cancer is 3 x 10~"4
[Group Bl]a for apparel workers exposed to formaldehyde at the
0.17 ppm level, 2 x 10"4 [Group Bl] for residents of mobile homes
who are exposed for 10 years to an average level of 0.10 ppm; and
1 x 10" [Group Bl] for residents of some conventional homes who
are exposed for 10 years to an average level of 0.07-ppm. The
upper bound unit risk estimate for an ambient exposure of 1 ug/m3
The shapes of most models' upper bound estimates .tend to
parallel the shapes of the models themselves, unless a procedure
has been devised to provide otherwise. This is the case for the
linearized multistage procedure, which provides a linear upper
bound estimate at low dose. The maximum likelihood estimate
(MLE), which is the estimate given by a fitted model, takes only
the experiment to which the model has been fitted into account.
The upper bound estimate, on the other hand, is intended to
account for experiment to -experiment variability as well as
extrapolation uncertainties. •
g •
EPA's Guidelines for Carcinogen Risk Assessment recommend
categorizing chemicals in Group B (Probable Human Carcinogen)
when "the evidence of human carcinogenicity from epidemiologic
studies ranges from almost 'sufficient' to 'inadequate.' To
reflect this range, the category is divided into higher and lower
degrees of evidence. • Usually, category Bl is reserved for agents
for which there is at least limited evidence of carcinogenicity
to humans from epidemiologic studies."
1-29
-------
(0.00082 ppra) for 70 years -is 1.3 x 10"5 [Group Bl]. The fitted
model gives the maximum likelihood estimate curve and, specific
**-
to the CUT study/ it has a pronounced S-shape. By contrast/ as
the linearized multistage procedure's upper bound estimate is
traced toward lower doses, its linear nature accomodates
increasing variability and extrapolation uncertainty. Both
estimates are shown in Table 1-2 to illustrate how the
perspectives they give on risk differ. Thus at 3 ppm (which is
in the experimental range), the difference between the MLE and
the UB is ten-fold, whereas at about one-tenth of that exposure,
a 100,000 fold difference is generated.
The lower bound on risk is always recognized to be as low as
zero. The upper bound estimate is ordinarily shown to allow for
extrapolation uncertainty. It is for this reason, along with
adherence to EPA's Guidelines for Carcinogen Risk Assessment,
that the upper bound was selected,to represent potential human
risk. While some of the existing information on formaldehyde is
3
consistent with non-linear interpretations, some support for a
*
linearized upper bound comes from the epidemiologic studies. The
excess cancer incidences observed in the epidemiologic studies
are about the same as the upper bound on lifetime risk based on
the rat nasal carcinoma data.
_ in
-------
TABLE 1-2
SUMMARY OP CANCER RISKS ASSOCIATED WITH FORMALDEHYDE EXPOSURE
Population Segement
{Exposure Level)
Lifetime
Individual Risk
Curre.nt OS'HA std. (3 ppm)
UBb 6 x 10~;j [B1J
MLEC 6 x 10"4 [Bl]
Garment Workers
NIOSH •
(0.17 ppm)
UB 3 x 10~4 [Bl]
MLE 4 x 10
-9
[Bl]
Mobile Home
Residents
(0.10 ppm 10-yr average)
UB- 2 x 10~/J [Bi:
MLE 2 x 10"10 (Bl]
Conventional Home*
Residents
(0.07 10-year average)
UB 1 X 10"4 [Bl]
MLE 6 X IQ"IL [Bl]
Home/Environment
Background Upper Limit
(0.0 5 ppm)
10.yr.
70 yr,
UB 7.0 x 10"5 [Bl
MLE 1.0 X .10"11 [B
[Bl]
,-4
UB 5.0 x 1.0"?n[Bl] •
MLE 1.0 x 10"iu [Bl]
* For hom«s containing substantial amounts of urea-formaldehyde
pressed wood (e.g./ floor underlayment and/or paneling)
upp«r Bound
c Maximum Likelihood Estimate
d 'Airborne Unit Risk, 1 ug/m3
UB - 1.3 x 10"5 [Bl].
- 70 yrs; Lifetime individual risk,
1-31
-------
-------
B.3
DOSE-RESPONSE
EVALUATION
Attributed
The report offers an explicit rationale for any preferred data
set(s) and model(s) used in dose-response evaluation;
strengths and weaknesses of the preferred data sets and
models are discussed, and scientific consensus or lack thereof
is indicated for critical issues or assumptions.
SOURCE Case Study H. Methylene Chloride (Pages 71-87).
Note See General Attribute 2 in this Appendix.
SOURCE Case Study A. DEHP (Pages 111 - 145) .
Note See Dose-Response Attribute 1 in this Appendix.
SOURCE Case Study D. Formaldehyde (Pages 1-23 to 1-28).
Note See Dose-Response Attribute 1 in this Appendix. This excerpt presents
the rationale for selection of data sets and models. Scientific consen-
sus that supports use of generic assumptions is discussed.
SOURCE Case-Study J. Red Dye No. 3 (Page? 73-86).
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS/NIH
Dr. Stan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA
Dr. Theodore Meinhardt, NIOSH/CDC
•Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries - ,
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NGTR/FDA
July, 1987
-------
July, 1987
CHAPTER 7 - ISSUES IN QUANTITATIVE RISK ESTIMATION
A. Introduction
Since the thyroid tumors in rats produced by R-3 appear likely to be
the result of a secondary mechanism, the question of the appropriate method
of low-dose extrapolation arises. Because thyroid tumors occur in the
control animals, it can not be ruled out that R-3 may be affecting a
tumorigenic process already operating in control animals in the absence of
R-3. It has been shown by several authors (147-149) that if a background
tumor risk exists which is not totally independent of the mechanism of the
production of tumors by administration of an agent, regardless of the mech-
anism, that linearity is the appropriate model at low dose levels of the
actual oncogen. If an agent produces tumors by the same biological mechan-
ism that produces some spontaneous tumors in the controls, then those
a
effects are additive. If tumors are produced in the control animals due to
the presence of an endogenous or environmental agent, then the effective
dose in the control has already surpassed the threshold dose, if one
exists. If the addition of a small amount of an agent increases the level
of the active oncogen, either directly or indirectly, this will result in a
proportionately small increase in tumors. From a standpoint of not under-
estimating the risk, these arguments support the use of linear extrapola-
tion in the low-dose region.
Since thyroid tumors in rats produced by R-3 appear likely to be the
result of a secondary mechanism, arguments could be presented for the
existence of a threshold dose of R-3 below which no increase in thyroid
tumors occur. For example, a threshold dose would occur if there is a dose
of R-3 for which no increase of the active oncogen at the target site
occurs in any individual.
Two approaches to risk assessment are considered in- this chapter. One
approach is based upon a no-observed-effect-level (NOEL) of R-3 in humans
which does not appear to produce an increase in TSH. Safety factors are
applied to this NOEL to arrive at an acceptable daily intake (ADI). The
second approach is based upon estimates not likely to underestimate risk
using linear low dose extrapolation of thyrdid tumor incidence from animal
bioassay data. The animal"thyroid tumor data are examined below.
73
-------
July, 1987
B. Animal Tumor Data
Detailed statistical analyses of age-adjusted thyroid tumor rates were
performed by the Panel for the two IRDC chronic bioassays (72,74) conducted
in rats, described in Chapter 4. Age-adjusted analyses (150) were used to
correct tumor rates for differences in mortality across dose groups. Anal-
yses were conducted for follicular cell and C-cell 'adenomas and carcinomas
of the thyroid. It was assumed that these tumors did not contribute to the
death of any animal. A summary of the number of animals with thyroid
tumors is given in Tables 13 and 14 for males and females, respectively,
derived from the detailed reports by the various pathologists. Differences
were considered significant if they existed at the P<0.05 level.
1. Follicular Cell Tumors .
Statistical tests for dose-response trends on age-adjusted tumor preva-
lence rates were conducted according to procedures given by Kodell et_ al.
(151) based upon Peto et al. (151). Separate analyses were conducted <• on
each study. Where one group provided diagnoses for both studies, the trend
test statistics were combined to give an over-all significance level. The
results of the trend tests are given in Table 15.
The numbers of animals developing follicular adenomas in females were
lower in the 4.0% dose group in Study 410-011 (74) than in the 1.0% dose
group from Study 410-002 (72). Since these tumors tend to develop late in
life, the lower rates in Study 410-011 may be due, in part, to the shorter
length of 122 weeks compared to the length of 128 weeks for Study 4LO-002.
It is noted from Table'15 for males that there is a highly statisti-
cally significant increase in follicular adenomas at the 4.0% dose level in
Study 410-011. For the diagnoses of follicular adenoma in males by IRDC,
there is a marginally significant dose-response trend in Study 410-002.
Based on the FDA diagnoses, there is a marginally statistically signi-
ficant (P<0.048) increase in follicular carcinomas in male rats.
For adenomas and carcinomas combined in male rats, the results are
dominated by. the larger prevalence of adenomas. For all pathology groups,
a highly statistically significant increase in follicular tumors was found
in male rats at the 4.0% dose level in Study 410-011.
For female rats, there were not enough follicular carcinomas to perform
statistical tests *for dose-response trends. For follicular adenomas or
follicular tumors in female rats, there were statistically significant
74
-------
July, 1987
TABLE 13
Numbers of male rats with thyroid tumors
Cla C2 0.1% 0.5% 1.0% , C3B 4.0%
°F I F I F IF IF IRFG I R F G
Number
Diagnosed •
Follicular
Adenoma
Follicular
Carcinoma
C-Cell
Adenoma
C-Cell
Carcinoma
69 64 69 61-64 - 66 69 57
00 1. 0-3 -8 31
00 01-3 -0 03
05 07-7 -2 03
00 00-0 -0 00
-•
70'
0
. -
1
3
1
67 68
1 1
2 1
3 4
1. 1
69 70 68 69
0 16d 15d 14d
0 3d 3d 5d
88 6
244
69
8
2
Cl - 0, C2 - 0, 0.1, 0.5, and 1.0% dose levels in Study 410-002 (72).
C3 - 0 and 4.0% dose levels in Study 410-011 (74).
Group: I- IRDC (81), F- FDA (82), R- CCMA consultant (83), G- A Canadian group (84)
One animal had both a follicular adenoma and carcinoma.
75
-------
July, 1987
TABLE 14
Numbers of female rats with thyroid tumors
cia
Ifc Rf
Number 70 69
Diagnosed
Follicular 0 1
Adenoma
Follicular 1 0
Carcinoma
C-Cell 0 2
Adenoma
C-Cell 0 1
Carcinoma
C2 0.1% 0.5% 1.0% C3b 4.0%
If Rf ' If Rf • If Rf If Rf If Rf If Rf
'70 68 - 67 - 69 68 68 67 66 70 65
00 -1 -3 65 0022
00 -0 -0 01 00 10
-
03 -10-7 01 24 35
01 -1 -3 0102 01
Cl - 0, C2 - 0, 0.1, 0.5, and 1.0% dose levels in Study 410-002 (72).
C3 - 0 and 4.0% dose levels in Study 410-011 (74).
If- IRDC (86) and R- consultant for CCMA (87).
76
-------
TABLE 15
Significance Levels (P-values) of Dose-Response Trend Tests
of Age-adjusted Prevalence of Follicular Tumors
Sex Type
Male Follicular
Adenoma
™
Follicular
Carcinoma
Combined
Female Follicular
Adenoma
Combined
Group
I
R
F
6
I
R
F
G
I
R
F
G
I
R
I.
R
410-002
0.043
0.108
—
a
-
0.154
—
0.043
—
0.054
^
0.0002
0.0022
0.0017
0.0006
410-011
0.00005
0 .00040
0.00100
0.00390
0.188
0.380
0.068
0.094
0.00005
0.00089
0.00029
0.00118
0 . 120
0.109
0.085
0.109
Combined
0.00002
0.00054
. -
0.048
-
0.00002
_
0.00011
-
0.0075
0.012
0.011
0.007
No carcinoma reported.
77
-------
July, 1987
dose-response trends in Study 410-002. ' Failure to achieve significance in
Study 410-011 may be due, in part, to the shorter length of 122 weeks of
that study compared to 128 weeks for Study 410-002.
2. C-Cell Tumors
There was a marginally statistically significant increase in C-cell
tumors (adenomas and/or carcinoma) in male rats administered 4.0% R-3 in
Study 410-011. The statistical significance levels for age adjusted
prevalence of C-cell tumors in male rats were 0.056* 0.025, and 0.120 for
diagnoses provided by groups I,R, and F, respectively.
3.
There was a highly statistically significant increase in follicular
adenomas in male rats administered 4.0% R-3. Female rats showed a statist-
ically significant increase, in follicular adenomas in Study 410-002, but
not in the shorter Study 410-011.
C. Dose-Response Models
Let P » g(d) represent the relationship between the proportion of ani-
mals above background with tumors, P, and the daily dose, d, of the active
oncogen. Let d - f (D) represent the relationship between the effective
dose and the daily dose of the administered agent, D.- This relationship,
f(D), represents the combined effects of absorption, distribution, physio-
logic, activation, and detoxification processes which produce the effective.
oncogenic dose. Then, P - g[f(D)] - h(D). That is, the chronic bioassay
data provide a measure of the relationship between risk and administered
dose without knowledge of the relationship between the effective dose and
the administered dose. If the dose response, P - h(D), is sublinear
(curving upward) in the low dose region, then low-do"se linear extrapolation
will overestimate risk;, see, e.g.. Gay lor and Kodell (152). If an oncogen
at least partially augments an oncogenic process already in progress in
control animals, then low-dose linearity is expected. Since this possibil-
ity cannot be ruled out, an approach which overestimates risk is to assume
low dose linearity with the effective dose, P - bd> b*f(D), where b is the
slope. Whittemore jet al. (153) proposed a similar approach for incorporat-
ing pharmacokinetic data into the risk assessment process to utilize the
effective dose. Whether or not this model can be extrapolated below the
experimental dose range for estimating ' risk depends upon the validity of
78
-------
July, 1987
low-dose linearity and the accuracy of the form of the model used for d -
f(D). If the true model for effective dose is sublinear (curving upward),
low-dose linear extrapolation overestimates risk. Even if tumors ^are
produced by a secondary mechanism, this does not necessarily imply that a
threshold dose for tumors exists. If the effective dose function, d »
f(D), is a threshold function of the administered dose, then doses below
this dose threshold would not result in an increase in tumors .
D. No-Observed-Effect-Level (NOEL) and Acceptable Daily Intake (ADI)
Because the Panel has concluded that the " mechanism in R-3 oncogenesis
is probably indirect, an important question is: what is the mathematical
relationship between the administered dose of R-3 and the active oncogen?
If a particular secondary mechanism is assumed (e.g., "TSH elaboration) , the
effective agent becomes TSH, not R-3. The background tumor incidence in-
dicates that it is possible that normal TSH levels in some animals are
above any tumor threshold, if one exists. For TSH the relationship of
effective dose to administered dose of R-3, d » f(D), cannot be established
from the available animal data. Thus, an estimate of the tumor risk based
on the dose-response in animals for this secondary mechanism cannot be
computed directly by P - g(d).
Without this information, one is forced to use assumptions. One may
assume a threshold function for the dose-response of R-3 and TSH based on
the premise that these reactions have thresholds. Thus, one could esta-
blish what appears to be an experimental dose level with little or no
biological effect, and divide this level by a safety factor to account fpr
the uncertainty in the data in order to arrive at an ADI
It appears, from human studies (Chapter 6) that an exposure of 20-60
mg/d of R-3 may be a NOEL for humans. Using a traditional adult male body
weight of 70 kg, an exposure level of 20 mg/d is equivalent to an exposure
of 20/70 - 0.29 mg/kg-d or 290 Ug/kg-body weight-day. Since a NOEL depends
on the -number of subjects examined, the biological endpoints measured, and
the variability in the measurements, every study is limited in its ability
to detect small effects. Therefore, a NOEL is not necessarily an exposure
level without adverse biological effects, particularly when short-term
tests are used to model lifetime chronic exposure.
In using a NOEL to establish an ADI, a key question is, what are the
79
-------
July, 1987
appropriate safety factors to use? How some safety factors can be used is
illustrated below. For instance, since only 10 males in the study of in-
terest were used, employing a safety factor of 10 would be standard
practice to account for variable sensitivities to chemicals among indivi-
duals. Since only a 14-day exposure was used to model a lifetime chronic
exposure, an additional safety factor would likely be used. For example,
if 10 is used for the additional safety factor, the ADI would be the NOEL/
100 - 290/100 - 2.9 Ug/kg-d. Also, if 60 mg/d is used as the NOEL, the ADI
would be 8.6 Hg/kg-d. Alternatively, one might also use a higher dose
level, e.g., 200 mg/d, and use an additional safety factor to account for
the apparent effect at this level. Other safety factors could be used
resulting in different ADI's. The choice of safety factors is a risk man-
agement function arid they are only employed here to illustrate the
procedure.
E. R-3 Risk Estimates
Since the data are insufficent to establish a threshold function of R-3
and TSH, either on experimental or mechanistic grounds, there may be a
tumor risk at the ADI. Hence, it is necessary to estimate the potential
tumor risk at low doses. If one does not make the assumption of a thres-
hold, the estimate of the risk using the multistage model can be calculated
using GLOBAL82 (154) using the 'data from the two IRDC chronic bioassays
discussed in Section B for rats. The procedure of Kodell jsit al. (155) was
used to calculate lifetime tumor incidence for each dose group standardized
to the mortality experience of the control animals. These standardized
lifetime tumor incidence rates were then used in the multistage model. The
point estimates of risk are dependent upon the choice of the model. How-
ever, the estimates of upper limits on low-dose risks obtained from linear
low-dose extrapolation are rather insensitive to the choice of the model
used in the experimental dose range. Choices of other models in the exper-
imental data range used in conjunction with linear low-dose extrapolation
would have little influence on the upper limits of risk. In the case of
R-3, the upper confidence limits are not far above the point estimates
because of the linear component exhibited in the'chronic bioassay data for
thyroid follicular cell tumors in rats. The nearly linear relationship
between TSH and R-3 reported in humans (134), although in a limited study,
80
-------
July, 198,
tends to support use of a low-dose linear extrapolation to estimate upper
limits on tumor risk at low doses. This does not necessarily mean that a
straight line is fit to the data. Rather, the curvilinear multistage model
is fit to the data and linear extrapolation is used below the experimental
data range to estimate upper limits on low-dose risk.
Low-dose risk estimates were based on the results of the FDA patholo-
gists for male rats and by the results of the CCMA consultant for the
female rats because these were the most complete pathology data sets for
each study. In Chapter 6, the average lifetime exposure to R-3 was calc-
ulated to be 1.41 mg/day, based upon production for the past 10 .years.
Using the average body weight of the U.S. population for ages 1-74 of 61 kg
(145), the average exposure to R-3 is 23 ng/kg-day. Point estimates of the
risk for lifetime exposure to 23 ug/kg-day of R-3 based upon the multistage
model are shown in Tables 16 and 17 for male and female rats, respectively,
as are the point estimates of the exposure corresponding to a lifetime risk
of one in a million. Upper limits of risk and lower limits on the exposure
corresponding to a risk of one in a million based upon the linearized mul-
tistage formula are also given.
If, as is suggested in the exposure information, the exposure to R-3 is
higher at young ages, and if R-3 effects the early stages in carcinogene-
sis, the multistage model would predict that the risk estimates based on
extrapolation of an average adult dose will be too low.
A plot of the standardized lifetime thyroid follicular cell tumor
(adenoma and/or carcinoma) incidence for male and females are given in
Figures 3 and 4, respectively. The fitted multistage models are also
plotted. Considerable variability of the data from the fitted multistage
model is noted. If the high dose level (4%) is omitted, the slopes become
somewhat steeper for males and considerably steeper for females. Hence,
discarding the 4.0% dose group tends to result in higher estimates of risk
(Tables 16 and 17). It appears that the absorption of R-3 is similar for
rats and humans. Further, it is assumed in. the -absence of data that the
lifetime incidence of tumors in humans is equivalent to the incidence of
tumors in rats for equal daily doses on a body weight basis. Even if the
same mechanism produces tumors in rats and humans, exposure to equal doses
of R-3, could have different quantitative tumor rate effects because the
dose-response slopes may be different at the different spontaneous tumor
81
-------
. . July, 1987
rates for rats and humans.. Generally, it is expected that the slope and
hence the additional risk would be lower for exposure above the human back-
ground level than at the higher background level in rats. If the dose
conversion between rats and humans is made on a surface area basis, the
estimated risks would be approximately a factor of five higher. Alterna-
tively, if R-3 produces tumors by a different mechanism than spontaneously.
occurring tumors and if a threshold of R-3 exists for each individual in
the population which is above his/her exposure level (i.e., no active on-
cogen is produced by R-3 for any individual), the tumor risk would be zero.
F. Discussion
The ADI and risk estimation are two different approaches used by regul-
atory agencies in the management of health hazards. For R-3 the ADI is
baaed on no observed short-term changes in TSH in adult males and risk es-
timation is based on tumor incidence in rats. Two different kinds of
safety considerations are utilized in the two approaches. The ADI is
derived from applying safety factors to a dose level considered to be a
NOEL. An upper limit on the risk estimate is derived from a mathematical
extrapolation model, which is believed to overestimate the risk. Since
there may not be a threshold dose of the oncogen or since the ADI might be
c
above a threshold dose, if one exists, there is a potential for tumors at
the ADI. Hence, it is useful to estimate the magnitude of the risk at the
ADI. Applying the multistage model, and using the information in Tables 16
or 17, the risk at an ADI of 0.0029 mg/kg-d is 0.0029/0.007 x lo"6 - 0.4 x
10" . Similarly, the risk at an ADI of 0.0089 mg/kg-d is 0.0086/0.007 x
10~6 - 1.2 x 10"6.
The multistage model may be used to estimate the potential risk from
oral exposure to R-3 through food (or food and drugs). " This requires a
choice be made from the options for selecting the tumor data base for the
extrapolation model. The options have been listed and discussed in Chapter
6.1. The slope of the extrapolation model, and thus the point estimate of
the risk, depends heavily on the chosen data base. With the information in
Tables 16 and 17, risks can be calculated for different estimates of daily
dose levels by direct proportionality. For example, reduction of the daily
exposure to 1.2 mg/d - 0.020 mg/kg-d (based upon the. annual certified
poundage of R— 3 averaged over the most recent, five years rather than the
82
-------
July, 1987
past ten years) would result in a risk of 20/23 times the risk listed in
Tables 16 and 17.
For comparison, a recent effort by a joint FAO/WHO expert committee
defined a temporary ADI for R-3 at 0-0.6 mg/kg body weight, or 0-600 ug/kg-
d (155). If finalized, this would be approximately one to two orders of
magnitude greater than the possible ADI's used for illustration here. .How-
ever, it appeared from the discussion in the summary document that the
expert committee did not have the advantage of some of the studies in rat
and man that the Panel had. WHO considered its estimate only tentative.
Low—dose estimates of oncogenic risk could be improved if the active
oncogenic dose were estimated as a function of the administered dose of
R-3, i.e., establish d - f(D). Or, if it were possible to at least iden-
tify an intermediate agent whose dose level is proportional to the tumor
incidence, then the dose level of that intermediate agent could be used as
a surrogate for the oncogenic dose." Under these circumstances, it would
only be necessary to establish a relationship between the dose of the
."
intermediate agent and the dose of R-3 administered.
83
-------
July, 1987
. TABLE 16
Estimates of Lifetime Risk of Thyroid Follicular Cell Tumors in Male Rats
Based on Results of FDA Pathology Analysis (See Table 13)
Tumor Type Data
Sets*
Estimates of Risk at
23 ug/kg-d
Point Upper
Estimates of Dose for
Risk of 10~6
Point Lower
Adenoma
All
2.5 X 10~6 4.0 X 10~6
9.1 ug/kg-d 5.8 ug/kg-d
alone -4% 5.4 X 10~6 10.4 X 10~6
4.2 ug/kg-d 2.2 ug/kg-d
Carcinoma All
0.7 X 10~6 1.6 X 10~6 31.8 ug/kg-d 14.5 ug/kg-d
alone
ca. 0
4.3 X 10
-6
5.4 ug/kg-d
Combined All 3.2 X 10~6 4.8 X 10~6 7.2 ug/kg-d 4.8 ug/kg-d
- 4Z 6.4 X 10~6 12.3 X 10~6 3.6 ug/kg-d 1.9 ug/kg-d
- 4% means that the data used exclude the 4% group.
x - The point estimate gives practically no risk, therefore, the dose for a
risk of 10~ is high, approximately 14.0 mg/kg-d.
84
-------
July, 1987
TABLE 17
Estimates of Lifetime Risk of Thyroid Follicular Cell Tumors In Female Rats
Based on Results of CCMA Consultant's Analysis (See Table 14)
Tumor Type Data Estimates of Risk at
Sets8 23 ug/kg-d
Point . Upper
Estimates of Dose for
Risk of l(f6
Point Lower
Adenoma
All 0.5 X 10~6 1.3 X 10~6
48.1 lig/kg-d 17.5 Hg/kg-d
alone
4% 3.1.X 10"6 5.5 X 10"6
7.4 ug/kg-d 4.2 ng/kg-d
Combined
All 0.6 X 10~6 1.4 X 10"6 41.7 ng/kg-d 15.9 ug/kg-d
-4% 3.6 X 10"6 6.1 X 10~6 6.5 ug/kg-d 3.8 Ug/kg-d
- 4% means that the data used exclude the 4% group.
85
-------
-------
B.3
DOSE-RESPONSE
EVALUATION
At mDUte «l T^e report reveals how dose-response relationships change
with alternate data sets, assumptions, and models.
SOURCE Case Study D. Formaldehyde (Pages 7-12 to 7-13).
Note Excerpts from the EPA report on formaldehyde illustrate an analysis
of benign tumor incidence and a resultant unit risk estimate. This
analysis was subsequently compared to the results from analysis of
malignant tumor incidence.
SOURCE Case Study F. Lead (Pages iii, iv, ix, 1, 49, 50).
Note This illustration is taken from a very comprehensive analysis of the
effects of blood lead levels on young children. The authors elicited
judgments from experts on the effects at various blood levels of lead as
a basis for their estimates. Sensitivity analysis was used to determine
the effects of two critical aspects of the study. The authors' presen-
tation is closely tied to the technical context of the discussion. The
illustration is included in the appendix to highlight how a sensitivity
analysis can be clearly presented. To assist the reader in understand-
ing the context of the sensitivity analysis, the authors' abstract, the
Table of Contents, and the list of acronyms preface the discussion of
* sensitivity analysis.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
7.2. Riate Batiaatea Baaed on Polypoid Adenoma Data
There appears to be little credible evidence that polypoid
adenomas progress to any of the malignant tumors seen in the Kern
et al. (L983) study. However, while the adenomas should not be
combined statistically with the squamous carcinomas for hazard
identification purposes/ they represent an endpoint that can be
quantified separately for analysis pruposes.
Because it is beyond the capability of•the various
extrapolation models to fit data with a negative slope, an
alternative extrapolation procedure is to drop the two highest-
doses and use the data from the 2.0 ppm rat exposure,grouo
(straight line to zero). However, since the true slope of the
dose-response curve is unknown below 2.0 ppm, this approach may
vastly overestimate the true risk if the curve is convex, and
underestimate it if it- -is concave. .- The reason the occurrence of
. polypoid adenomas has a negative slope probably lies with the
fact that the cell type in the respiratory epithelium from which '
these tumors arise is lost sooner and to a greater extent with
increasing dose due,, to squamous metaplasia. The less, respiratory
epithelium available the smaller the chance for adenomas to
develop. 0-ther explanations are also possible as discussed in
section 7.4.1.
Risk estimates using polypoid adenomas appear, in Table
7-3. For polypoid adenoma as the endpoint instead of squaraous
cell carcinoma there is no difference between the two procedures
described earlier to adjus,t for animals at risk. The first
7-12
-------
observation of a polypoid adenoma ,was in a rat sacrificed at 10
months. Eliminating all rats dead of any cause prior to, that
time and applying the method used for the carcinoma data.leads to
7/159 for the response at 2 ppm with 1/156 at control, the same
as if all rats dead prior to an including the 18 month sacrifice
were excluded.'
Table 7-3.
RISK EXTIMATES USING POLYPOID ADENOMA DATA
Category
tobile Hone
Residents
Based on HUD
Target Level
Dose
0.15
(112 hrs/wk
for 10 yrs)
Manufacturers
of
-------
ARGONNE NATIONAL LABORATORY
9700 South Cass Avenue, Argonne, Illinois 60439
ANL/AA-32
ASSESSING THE RISKS TO YOUNG CHILDREN
OF THREE EFFECTS ASSOCIATED WITH
ELEVATED BLOOD-LEAD LEVELS
by
Thomas S. Wallsten* and Ronald G. Whitfield
Energy and Environmental Systems Division
Decision Analysis and Systems Evaluation Section
December 1986
work sponsored by
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
*L.L. Thurstone Psychometric Laboratory, University of North Carolina, Chapel Hill
-------
CONTENTS
ACRONYMS lx
SYMBOLS x
ACKNOWLEDGMENTS xi
ABSTRACT • • ' 1
1 INTRODUCTION l
1.1 Report Organization ? 2
1.2 Motivation «
1.3 Judgmental Probability Encoding 5
1.4 Dose-Response Uncertainty • 6
1.5 Risk Assessment Strategy 6
2 PROBABILISTIC DOSE-RESPONSE FUNCTIONS FOR LEAD-INDUCED
ELEVATED EP LEVELS 7
3 PROBABILISTIC DOSE-RESPONSE FUNCTIONS FOR LEAD-INDUCED
Hb DECREMENTS • • n
3.1 Protocol Development • 11
3.2 Protocol Outline • *2
3.3 Conduct of the Sessions 13
3.4 Encoding the Judgments - 13
3.5 Representing the Judgments : ijj
3.6 The Experts • • • JJ
3.7 Results "
3.7.1 Hb Level < 11 g/dL, Ages 0-3 • 18
3.7.2 Hb Level < 11 g/dL, Ages 4-6 20
3.7.3 Hb Level < 9.5 g/dL, Ages 0-3 21
3.7.4 Hb Level < 9.5 g/dL, Ages 4-6 22
3.8 Discussion 23
4 PROBABILISTIC DOSE-EFFECT AND DOSE-RESPONSE FUNCTIONS FOR
LEAD-INDUCED IQ DECREMENTS • • • 27
4.1 Protocol Development • • • • 28
4.2 Protocol Outline 30
4.3 Conduct of the Sessions • • • • • 30
4.4 Encoding the Judgments 31
4.5 Representing the Judgments 32
4.6 The Experts • 32
4.7 Results • 33
4.7.1 Control-Group Mean IQ .. * • • • 34
4.7.2 Within-Group IQ Standard Deviation • 35
4.7.3 Mean IQ Decrements for the Low SES Group 36
4.7.4 Mean IQ Decrements for the High SES Group 36
-------
CONTENTS (Cont'd)
4.7.5 Change in Percentage of Low SES Group with IQ < 85 .............. 37
4.7.6 Change in Percentage of High SES Group with IQ < 85 ., ..... !!!!!.'! 40
4.8 Discussion ............................... . ................. « • • • •
5 ESTIMATED RISKS OF ADVERSE HEALTH EFFECTS VERSUS GEOMETRIC
MEAN PbB LEVEL ...... ,. ...... . ......... ............................... 43
5.1 Estimated PbB Distributions ____ .......................... . 43
5.2 Overview of the Risk Results f or EP ........ .'.'.".*!.'!.'.'."!!!.'!!"""""" 44
5.3 Overview of the Risk Results for Hb ............ ...'.......... .......... 45
5.4 Overview of the Risk Results for IQ . ........... •.' 1 !.'.*.'!!!.'!.'-!!.*!.'!!!!!! 45
5,4.1 Risk Distributions over Mean IQ Decrement ..... . . ...... !!!!...! 47
5.4.2 Increased Probability of Lead-Induced IQ Levels Seine ......
^IQ* ....... .................................. ;...: ....... ... 48
5.5 Sensitivity Analysis ................................ 4 ...... 4q
6 CONCLUDING REMARKS ....... . ..... ... .............. . ____ ............. 51
REFERENCES ......................... co
****•******•"*••,*••••••»••••••••*••••.»•» O /
APPENDIX A: Fitting Functions to Data on Lead-Induced Elevated EP
Levels..... ............................................ ...... 55
APPENDIX B: Fitting Functions to Encoded Judgments Relating to
Lead-Induced Hb Decrement ............ ... ...... . ............. gj
APPENDIX C: Fitting Functions to Encoded Judgments Relating to
Lead-Induced IQ Effects ...................................... 101
APPENDIX D: Risk Distributions ..... .......... ..... ...... . . .. ............... 141
TABLES
1 Probability of Suffering a Specified Health Effect under Alternative
NAAQS, Given Complete Information ...................... ............... 4
2 Probabilities of Suffering a Specified Health Effect under Alternative
NAAQS, Given Incomplete Information ......... , ....... . .................. 5
3 Sample Sizes for the EP Data ........ . .................. .......... 9
4 Consultants for the Hb Protocol ........................ .. ........... 12
5 Experts Participating in the Hb Encodings ......... ...... ........ '. ......... 17
6 Consultants for the IQ Protocol .. ..... . ..... .......... ............ . 29
7 Experts Participating in the IQ Encodings .................................. 33
-------
ACRONYMS
ALAD 6-aminolevulinic acid dehydrase
ALAS 6-aminolevulinic acid synthase
CD criteria document
CDF cumulative distribution function ,
CI credible interval
CNS central nervous system
ECAO Environmental Criteria Assessment Office
EDTA ethylenediaminetetraacetate
EEG electroencephalogram
EP erythrocyte protoporphyrin
EPA U.S. Environmental Protection Agency
FEP free erythrocyte protoporphyrin
GM geometric mean
GSD geometric standard deviation
Hb hemoglobin
HERL Health and Environmental Research Laboratory
IQ intelligence quotient
NAAQS National Ambient Air Quality Standard(s)
NHANES II second National Health and Nutrition Survey
NOLO normal-on-log-odds
OAQPS Office of Air Quality Planning and Standards
PbB blood lead
PDF probability density function
PMF probability mass function
SES socioeconomic status
ZPP zinc protoporphyrin
iz
-------
ASSESSING THE RISKS TO YOUNG CHILDREN
OF THREE EFFECTS ASSOCIATED WITH
ELEVATED BLOOD-LEAD LEVELS
by
Thomas S. Wallsten and Ronald G. Whitfield
ABSTRACT
- Formal risk assessments were conducted as part of the U.S.
Environmental Protection Agency's current review of the primary
National Ambient Air Quality Standard for lead. The assessments
focused on three potentially adverse effects of exposure to lead in
children from birth through the seventh birthday: erythrocyte
protoporphyrin (EP) elevation, hemoglobin (Hb) decrement, and
intelligence quotient (IQ) effect. The same general strategy was
followed in all three cases: for two levels of each effect, probability
distributions over population response rate were estimated at a series
of blood-lead (PbB) levels. These distributions were estimated from
data in the case of EP elevation and from expert judgments in the
cases of Hb decrement and IQ effect. Although of interest in their
own right, these estimates were combined with PbB distributions to
yield probability distributions over the estimated percentages of
children experiencing the particular health effects.
1 INTRODUCTION
The Clean Air Act charges the U.S. Environmental Protection Agency (EPA)*
wjth setting and reviewing both primary and secondary National Ambient Air Quality
Standards (NAAQS) for selected pollutants. Each primary standard must be set at a level
sufficient to protect public health with an adequate margin of safety. This report
presents the results of a risk assessment performed to assist in the review of the primary
NAAQS for lead.
**
For each review, the scientific basis for revising the primary lead NAAQS is
presented in an updated document entitled Air Quality Criteria for Lead (EPA, 1986a),
hereafter referred to as the criteria document (CD). It summarizes and analyzes
available scientific evidence about the adverse health effects of lead. After evaluating
and interpreting the information in the CD, a draft EPA staff paper (EPA, 1986b)
identifies the critical elements that EPA staff believe should be considered in the review
and possible revision of the lead NAAQS. Particular attention is paid to those subject
*A11 acronyms used in this report are listed alphabetically on pp. ix and x.
-------
49
among low SES children sheltered from lead to be quite low (around 85). No estimates
are given for Expert I because that individual did not provide the needed judgments about
mean IQ levels for children sheltered from lead exposure or for IQ standard deviation.
5.5 SENSITIVITY ANALYSIS
Two sensitivity analyses were conducted to study .the effects of (1) having dose-
response distributions on intervals smaller than 10 ug/dL and (2) changing the GSD value
assumed for the PbB distributions. We considered the Hb risk assessment for the first
sensitivity analysis and the EP, Hb, and IQ risk assessments for the second sensitivity
analysis.
We chose the Hb judgments for the first sensitivity' analysis because the log-odds
transformation resulted in functions with approximately equal slopes. This "common
slope" allowed interpolation with a high degree of confidence between the probability
distributions encoded for Experts C, D, and E. The IQ judgments were less suitable for
this type of analysis because the slopes of the transformed distributions were not equal.
Furthermore, we chose to consider children aged 0-3 because they are the most sensitive
to lead exposure and their dose-response distributions display the largest variations,
which tends to accentuate any sensitivities that may be present.
Twenty-one dose-response distributions were specified for Experts C and E (six
distributions on 10-yg/dL intervals were encoded) by plotting the median values of the
transformed distributions versus the six PbB levels and drawing a smooth curve through
the points. The smooth curve allowed estimation of median values in steps of 2.5 pg/dL
from 5 yg/dL to 55 pg/dL. (Eighteen distributions were specified for Expert E because
only five distributions, beginning at PbB = 15 pg/dL, were encoded in his case.) These
values, along with the common slope value, completely specified the dose-response
distributions. These distributions were then combined with PbB distributions in a fashion
identical to that used to produce the results described in Sees. 5.2-5.4.
For Experts C, D, and E, the risk distributions based on the larger number of PbB
levels are virtually identical to those based on the smaller number of PbB levels. The
differences between the two sets of ..calculations are very small, but systematic, for each
expert. For Experts C and E, the risk distributions based on fewer PbBlevels are about
0.1% closer to the origin than are the corresponding risk distributions based on more PbB
levels. In other words, the risk estimates we reported are slightly smaller than those
that would have been produced by a finer-grained analysis. The opposite is the case for
Expert D: the risk distributions based on more PbB levels are about 0.2% closer to the
origin. These results strongly indicate that encoding at only five or six PbB levels was
adequate, at least for Hb effects.
The second sensitivity analysis was simpler than the first. We repeated the risk
calculations for EP, Hb, and IQ effects (performed assuming a GSD of 1.42 pg/dL for the
PbB distribution) for two additional GSD values: 1.3 pg/dL and 1.5 yg/dL. The chosen
values bound those reported in the literature and summarized in the CD for lead (EPA,
1986a). Geometric standard deviation values for PbB distributions in populations of
children are extensively discussed in the EPA staff paper that reviews NAAQS for lead
(EPA, 1986b). The 1.42-g/dL value is reported in NHANES II (Annest et al., 1982).
-------
50
For EP level > 53 yg/dL, risk results are virtually unchanged at low (< 7.5 yg/dL)
GM PbB values. Differences are greatest at GM = 15 yg/dL: results at GSD = 1.3 yg/dL
are about 28% lower (i.e., a response rate of 4% versus 5.5%), and results at GSD =
1.5 yg/dL are about 22% higher than those* assuming GSD = 1.42 yg/dL. At GM =
27.5 yg/dL, results for GSD values of 1.3 yg/dL and 1.5 yg/dL are within 11% of those
assuming a GSD of 1.42 yg/dL. The threshold for lead-induced EP effects, which was
calculated by Piomelli et al. (1982) to be at a PbB level of about 16.5 yg/dL, probably
explains the observation that results are most sensitive to GM PbB values around
15 yg/dL. '
Results for Experts C and E are virtually unaffected by the GSD value for Hb
level < 9.5 yg/dL and children aged 0-3. For Expert D, differences in mean values of the
risk distributions are about 10% for GM values > 15 yg/dL (e.g., median response rates at
GM = 27.5 yg/dL are 4.6%, 5.3%, and 5.7% for GSD = 1.3, 1.42, and 1.5 yg/dL,
respectively). The difference beginning at 15 yg/dL can probably be attributed to Expert
D's judgment that a threshold exists for lead-induced Hb effects in the 15-25 yg/dL PbB
range.
For IQ decrement among low SES children, the different GSD values have
essentially no effect on the risk distributions for any of the six IQ experts (Experts F
through K). The differences are generally less than 0.1 mean IQ points. The same is true
for the IQ response rate at IQ* = 70 and low SES children. Differences between, GSD =
1.42 yg/dL and the other two GSD levels are less than 0.2% (in terms of response rate)
for Experts H, J, and K, and less than 0.1% for Expert G. Risk distributions for Expert F
are unaffected.
-------
-------
B.4
EXPOSURE ASSESSMENT
Attribute 1 The purpose and scope of the exposure assessment and the
underlying methodologies are clearly described.
SOURCE Case Study K. Tetrachloroethylene (Pages 1-77).
-------
UCRL-15831
Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
K. T. Bogeh, L. C Hall, T. E. McKone,
D. W. Layton, and S. E. Fatten
Environmental Sciences Division
Lawrence Livennore National Laboratory
University of California
P.O. Box 5507
Livennore, CA 94550
April 10, 1987
Prepared for
California Public Health Foundation
P.O. Box 520
Berkeley, CA 94701
-------
1. INTRODUCTION
This document presents an assessment of the potential health risks
associated with exposure to tetrachloroethylene (also known as
perchloroethylene or PCE) dissolved in California drinking waters. .This
assessment is being provided to the California Department of Health Services
(CDHS) for the development of drinking-water standards to manage the health
risks of PCE exposures. Other assessments required in the risk-management
process include analyses of the technical and economic feasibilities of
treating water supplies contaminated with PCE. A primary goal of this
health-risk assessment is to evaluate dose-response relationships for observed
and potential toxic end points of PCE in order to define dose rates that can
be used to establish standards that will protect members of the general public
from .adverse health effects resulting solely from water-based exposures to
PCE. We also analyze the extent of human exposures attributable to
PCE-contaminated ground water in California.
The document consists of eight sections, plus supporting appendices.
Each section provides information that risk managers at the CDHS can use to
develop PCE drinking-water standards that will safeguard human health. Our
assessment begins in Section 2 with a review of the uses of PCE, its
environmental chemistry, and concentrations measured in different
environmental media. The next section provides an overview of published
studies on the absorption, distribution, metabolism, and elimination of PCE,
emphasizing those studies that have defined the rate and extent of these
.processes in rodents and humans. In Section 4, we review studies of the
acute, subchronic, and chronic toxicity of PCE to animals, including a summary
of data from bioassays conducted to evaluate PCE carcinogenicity. In
Section 5, we provide an overview of PCE health effects in humans, review
epidemiological studies involving PCE, and examine human data on PCE's toxic
effects on specific organs and systems.
In Section 6, we describe our procedure for calculating human PCE
exposures attributable to contaminated groundwater supplies. This section
takes an integrated approach to the exposure assessment. A household
consisting of two adults and two children uses approximately 1000 L/d of water
from wells or surface water. Our approach considers how PCE contained in this
-------
amount of water can result in human exposure through ingestion,"inhalation,
and dermal absorption. For each pathway we develop pathway-dose factors that
translate a unit concentration in mg/L in tap water into a lifetime equivalent
dose rate in mg/kg-d. We use the pathway-dose factors and data from A81803
surveys (CDHS, 1986) toadetermine the magnitude and distribution of
human-lifetime dose rates attributable to PCE in California groundwater
supplies.
We do a quantitative dose-response assessment for PCE cardnogenicity in
Section 7, using a "linearized" multistage dose-response extrapolation model
along with four sets of animal cancer-bioassay data as input to that model.
In this quantitative carcinogenic potency assessment, a relationship between
the doses applied in the animal bioassays and the corresponding effective or
metabolized doses is derived using a simple pharmacokinetic model and
available data on PCE metabolism in rodents. The results of our metabolic
analysis are compared to and shown to be consistent with results based on
other analytic methods; the method we use is also shown to provide a good
prediction of available data on human PCE metabolism. Our calculated
carcinogenic potencies of PCE to animals based on different sets of bioassay
data are then extrapolated to humans using two different methods of
inter-species extrapolation. This yields a set of 112 alternative potency
values based on different assumptions that might be applied to humans exposed
to PCE in the context,of regulatory risk assessment. Finally, we discuss
methods applicable to calculating PCE concentrations in water associated with
given, predicted cancer risk levels using information provided in this section
and in Section 6.
The last section addresses some of the key uncertainties associated with
the health-risk assessment and also presents some research recommendations for
reducing those uncertainties.
-------
. 2. CHEMICAL AND PHYSICAL PROPERTIES AND ENVIRONMENTAL
TRANSPORT AND TRANSFORMATION
Tetrachloroethylene is commonly referred to as PCE, or perchloroethylene,
This compound is a volatile, chlorinated hydrocarbon that is widely used as a
degreasing solvent. In this section we provide an overview of its ,use in the
U.S. and, importantly, its transport and fate in the environment.
CHEMICAL AND PHYSICAL* PROPERTIES
PCE is a colorless, nonflammable liquid with a chloroform-like odor. .It
is slightly soluble in water and has a vapor pressure of 15.8 mm Hg at 209C
and'a boiling point at 12TC. Table 2-1 lists the chemical structure,
alternative names, and identifiers of PCE. Its physical and chemical
properties are listed in Table 2-2.
PRODUCTION AND USES : ,
In the U.S., PCE is primarily produced via chlorination of hydrocarbons
(based on ethane or propane) and from processes based on ethylene d1chloride
(i.e., oxychlorination of 1,2-dichloroethane). Additives, amines or esters,
are added in small amounts to stabilize the product. PCE is the most stable
of the chlorinated ethanes and ethylenes, requiring only small amounts of
stabilizers (Keil, 1979). PCE production plants vary in size from 20 million
to 90 million kg annual production (Lowenheim and Moran, 1975). In 1983, the
United States production totaled 308,076 metric tons (308 million kg) (CARB,
1984). The only producer of PCE in California is Dow Chemical Company,
Plttsburg, with an annual capacity of 22.7 million kg (Chemical Marketing
Reporter, 1983). There are no known natural sources of tetrachloroethylene.
PCE is a widely used solvent with applications as a dry-cleaning agent, a
metal degreaser, and a chemical intermediate in the manufacturing of
fluorocarbons. It 1s also used as a fumlgant, in the extraction of caffeine
from coffee, in removal of soot 'from industrial boilers, and as a heat-transfer
medium. Of the total PCE used 1n 1983, 59! was as a dry-cleaning agent and in,
textile processing, 2U as a metal degreaser, 111 for export, 6t as a chemical
intermediate, and 31 in miscellaneous uses (Chemical Marketing Reporter, 'l983).
-------
Table 2-1. Chemical structure of tetrach'loroethylene,-alternative names,
identification numbers, empirical formula, and molecular weight.
Chemical structure:
Cl - C - C - Cl
Cl Cl
Empirical formula: CaCU Molecular weight: 165.85
Chemical Abstracts Service registry number: 127-18-1
NIOSH Registry of Toxic Effects of Chemical Substances number: KX3850000
«
Alternative names: PCE, Perc, tetrachloroethene, perchloroethylene, ethylene
tetrachlorlde, carbon dichloride
Common trade names: Antisol, Dee Solv, Per Sec, and Texranec.
ENVIRONMENTAL TRANSPORT AND TRANSFORMATION
Tetrachloroethylene tends to partition primarily to the atmosphere. It
has been estimated that 85 to 901 of the PCE produced is eventually released
to the atmosphere (U.S. EPA, 1985a; WHO, 1984). The key properties of PCE
that affect its movement in the environment are its high vapor pressure and
low solubility in water.
GEOTOX (McKone and Layton, 1986) was used to estimate the equilibrium
distribution of PCE in air, soil, and water. GEOTOX is a multimedia
compartment model that simulates the environmental transport and transformation
of a chemical, based on its physical and chemical characteristics and the
properties of the landscape into which it 1s released. A simulation*of the
environmental partitioning of PCE was run using California landscape data,
properties of PCE (see Table 2-2), and PCE source-emission data from the
California Air Resources Board (CARS). The PCE source term was represented by
an annual release of 1.83 x 107 kg/y over an area of 411,000 km (CARS,
1984); of this source, 101 1s assumed released into soil, 11 to surface water,
and the remainder directly to *he atmosphere. The simulated equilibrium
distribution of PCE is shown in Fig. 2-1. Most of the PCE released to the
environment is found in the atmosphere. However, the equilibrium
distributions, 861 in the air and 111 in surface water, reflect both the
-------
Atmospheric
86i
Biomeu
0.02%
Upper toil
0.4%
i
Lower soil
0.5%
i
Atmospheric per tides
4 x 10"8%
I
Ground weter
8
1%
t
Surface wctar
11%
Sediments
1%
8
Flgurs 2-1. Environmental distribution of PCE undtr steady-state conditions,
Partitioning between compartments Is predicted by the computer model GEOTOX
(McKone an4 Layton, 1986).
-------
Table 2-2. Chemical, physical, and organoleptic properties of
tetrachloroethylene.
Property
Boiling point at 760 mm Hg
Freezing/melting point
Density at 20aC
Vapor pressure at 20aC
Henry's law constant at 20aC
Conversion factor
Units .
ac
9C
g/cm3
mm Hg
atm-nH/mol
mg/m^-ppmv
Value
121
-22.4
1.65
15.8
0.0227
6.89
Reference
Hawley (1981)
Hawley (1981)'
Hawley (1981)
Sittig (1985)
Mackay and Shiu (1981)
Verschueren (1983) .
m
2/s
Diffusion constants
at 1 atm, 20aC
Air
Water
Solubility in water at 25'C mg/L
Log octanol/water
partition coefficient
Odor threshold in water
Unitless
mg/L
7.4 x 10~6
7.6 x 10-10
150
3.14
2.46
Lyman et aU (1982).
MacMnson e_t aj..
(1981)
Leo (1983)
Callahan et al. (1979)
3.0 x 10-1 Zoeteman et a_L (1974)
relative magnitude of the source (891 to air and 11 to water) and the
effective residence times. The loss rate of PCE in air is an order of
magnitude greater than that in surface water. This accounts for the apparent
"enrichment" of PCE in surface water.
Air
PCE in the atmosphere is subject to relatively rapid chemical or
photochemical degradation. In the troposphere, it photodegrades, ultimately
leading to the formation of hydrochloric acid, trichloroacetic acid, and carbor
dioxide in the presence of atmospheric water (U.S. EPA, 1985a). PCE can also
be removed by scavenging mechanisms, primarily through hydroxyl radicals
(Dimltriades et a].., 1983). Estimates of its atmospheric residence time are
on the order of one year or less (see U.S. EPA, 1985a).
-------
Singh et a_L (1981) compiled monitoring data for the concentrations of
several volatile organics in ambient air and found that for the. western half
•3
of the U.S. the average PCE concentration was 4.3 ng/m and the overall range
was 0.23 to 51.6 pg/m . The U.S. EPA (1985a) reported ambient air PCE
concentrations in California (1972-1980) ranging from 0.2 to 19.0 ng/m3. The
California Air Resources Board (Nystrom, 1986), based on preliminary data,
found average ambient air PCE concentrations for several California locations:
Los Angeles 1180 ± 900 parts-per-trillion by volume (pptv) (8.1 ± 1.2 jig/m3);
San lose 490 ± 330 pptv (3.4 ± 2.3 iag/m3); Long. Beach 1030 ± 560 pptv
(7.1 ± 3.9 yg/m3); Stockton 450 ± 170 pptv (3.1 ±1.2 ng/m3), and Simi
Valley 330 ± 250 pptv (2.3 ± 1.7 jjg/m3). These data indicate that PCE
concentrations in the ambient air of urban areas are higher than those in
rural areas (or less densely populated areas).
Water
In surface waters, PCE rapidly volatilizes into the atmosphere. Wind
speed,, agitation of the water, and water and air temperatures affect
evaporation rates. Photodegradation, in contrast, is a slow decay process and
does not appear to be an important transformation mechanism in water. The
half-life of PCE in shallow water due to volatilization has been estimated at
24 to 28 min in laboratory experiments (Oilling e_t a_L, 1975). Zoeteman
et a].. (1980) measured PCE persistence in surface waters of the Netherlands
from 3 to 30 days (half-life), while in lakes and ground waters, the half-life
was estimated to be 10-fold higher.
In ground water, PCE is relatively persistent, with degradation occurring
through hydrolysis and biotransformation. It is denser than water as an
undissolved liquid, consequently it tends to sink in ground water. Vogel and
McCarty (1985) have shown that PCE biotransforms to trichloroethylene (TCE),
dichloroethylene, and vinyl chloride via reductive dehalogenation under
anaerobic conditions. They further suggest that the potential exists for the
complete mineralization of PCE to carbon dioxide in aquifer systems. The
half-life of PCE due to aqueous hydrolysis in natural waters can be on the
order of months (Dilling et a_L, 1975) to several years (Pearson and
McConnell, 1975).
-------
The U.S. Environmental Protection Agency (1985a) reported a mean PCE
concentration of 1 vg/L from 1102 surface water measurements in 45 states
(from August 1975 to September 1984). An important source of data on the
concentrations of PCE in drinking water supplies is a survey of large water
utilities in California (i.e., utilities with more than 200 service
connections) that was conducted by the California Department of Health Services
(1986). From January 1984 through December 1985, the wells in 819 water
systems were sampled for contamination by organic chemicals. The water systems
considered included a total of 5650 wells, 2947 of which were sampled. The
wells sampled were selected based on the likelihood of contamination. PCE was
found in 199 wells in concentrations up to 166 jig/L, with a median
concentration of 1.9 n9/L. Generally, the highest fraction of contaminated
wells and the wells with the highest concentrations were found in the heavily
urbanized areas of the state. Contamination was state-wide. Los Angeles
County registered the greatest number of contaminated weMs (i.e., 140).
Soil
There is limited information on the behavior of PCE in soil. The solvent
can be adsorbed to soil or leached through soil when dissolved in water or as
a separate organic phase (as in large spills). PCE associated with soil air
or soil water is more mobile than the absorbed portion (Schwarzenbach and
Westall, 1981).
The adsorption of PCE to soils appears to be correlated to its octanol/
water partition coefficient, the organic carbon content-of the soil, and the
concentration of PCE in the liquid phase. PCE appears to leach rapidly
through soils of low (<0.1%) organic carbon content (U.S. EPA, 1985a;
Schwarzenbach and Westall, 1981). .
Several studies have documented the mobility of tetrachloroethylene in
soi 1/groundwater systems (Piet et a].., 1981; Schneider et aj.., 1981;
Schwarzenbach and Westall, 1981). Wilson et a_]_. (1981) showed that most of
the chemical was lost from the soil via leaching or volatilization to the
atmosphere. Persistence in soil ranges from months to years.
-------
3. PHARMACOLOGY AND METABOLISM
PCE 1s readily absorbed through the lungs and gastrointestinal tract and
may, to a lesser extent, be absorbed through the skin. Once in the body, PCE
distributes into all tissues. Steady-state tissue concentrations are a
function of the absorbed dose, partitioning factors, and pharmacokinetic
properties, such as rate of metabolic conversion and elimination.
8 The primary metabolic pathway of PCE 1s thought to, involve oxidation to
an epoxide as the first step, although this epoxide intermediate has never been
isolated in vivo (Bonse et aj.. 1975; Greimeta_[., 1975). The epoxide
undergoes rearrangement to form trichloroacetyl chloride, and ultimately
trichloroacetic acid (Yllner, 1961; Daniel, 1963; Moslen et aT., 1977; Costa
and Ivanetich, 1980). Studies In which radiolabeled PCE was administered to
anjmals have occasionally recovered oxalic acid as a significant urinary
metabolite miner, 1961; Dimltrieva, 1967;' Pegg et aj.., 1979)., Carbon
dioxide is also commonly produced (Pegg et a].., 1979; Schumann et aj.., 1980).
In this section, we present an overview of published studies on the
^absorption, distribution, metabolism, and elimination of PCE. Our emphasis is
on studies that have defined the rate and extent of each of these processes in
humans and tn rodents. Proposed metabolic pathways are discussed in some depth
because metabolism is responsible for the transformation of PCE to one or more
reactive species.
ABSORPTION
In the following paragraphs we review relevant data on PCE uptake through
ingestlon, dermal absorption, and inhalation.
Inqestion
Absorption of PCE from the gastrointestinal tract has been measured
indirectly as percent of dose recovered. The percentage of dose recovered
after administration of PCE is similar in mice and rats, varying between 80 to
1001. PCE is absorbed rapidly; peak blood concentrations were measured in
rats one hour after a 500 mg/kg dose (Pegg et aj.., 1979),
-------
Little information exists regarding amount or rate of oral absorption of
PCE in humans. Koppel et a_]_. (1985) reported that the blood concentration of
PCE following an oral dose of 400 mg was described by a two-compartment model
with half-lives of 160 min and 33 h, respectively (unpublished data cited in
Koppel et aj... 1985). The same authors (Koppel et a_L , 1985) measured the
concentration of PCE in blood at 21.5 yg/mL within one hour of ingestion of
12 to 16 g. Although far from definitive, these reports suggest fairly rapid
and complete oral absorption.
Dermal absorption
Jakobsen et a].. (1982) measured the absorption of PCE through guinea pig
skin. Animals were in contact with liquid PCE for 6 h; during the exposure,
blood concentrations rose rapidly, and peaked within 30 min. Tsuruta (1975)"
estimated the rate of absorption of PCE through mouse skin to be
2
24 nmol/min-cm of skin.
Percutaneous penetration of PCE vapor in humans exposed to ambient air %
concentrations of 600 ppmv is approximately one percent of pulmonary absorption
(Riihimaki and Pfaffli, 1978). In studies in which volunteers immersed their
thumbs in liquid PCE, a peak concentration of 0.3 ppmv in expired air within
40 min was measured, which decreased thereafter. Four other chlorinated
solvents were tested by this method; trichloroethylene, carbon tetrachloride,
rnethylene chloride, and 1,1,1-trichloroethane. Peak alveolar concentrations
were 0.5, 0.6, 3.0, and 0.7 ppmv, respectively. PCE produced the lowest
maximum concentration in alveolar air, and it had the slowest rate of
elimination in breath. The relatively low concentration of PCE in exhaled air
suggests that dermal absorption is Mmi ted (Hake and Stewart, 1977; Stewart
and Oodd, 1964). However, the work of Brown, et a].. (1984) suggests that
dermal absorption of dilute aqueous solutions may contribute a significant
amount to the overall absorption of PCE (see Section 6).
Pulmonary Uptake
During inhalation, PCE diffuses across the lungs and dissolves into the
bloodstream. The rate of transfer is dependent on the blood/gas partition
coefficient for PCE. One estimate (Monster et a].., 1979) placed the human.
10
-------
blood/gas coefficient of PCE at 16, while Gargas et a_K (1986) reports a value
of 10.3. Both values reflect the'fact that PCE is lipophilic and readily
diffuses into the blood. The uptake of PCE by the lungs is also determined by
the alveolar ventilation rate (i.e., that fraction of total respiratory
ventilation from which volatile organic compounds, such as PCE, may be cleared
by absorption into alveolar capillary blood), ambient concentration, exposure
duration, and metabolism.
At steady state, the amount of PCE taken up or retained will be equal to
the amount of PCE metabolized. However, as pointed out by Guberan and
Fernandez (1974), the time it takes for humans to approximate a steady-state
equilibrium upon inhalation of a constant concentration of PCE is one to two
weeks. Since no human studies for PCE involve constant exposure for this
length of time, steady-state conditions were not approximated in any of these
studies. Therefore, the uptake or retention rate varies as a function of time
for all of these" studies, and no generalizations can be made that are
independent of the duration of exposure. The value of these observations for
purposes of risk assessment are therefore limited, since generalizations
relating uptake to metabolism cannot be confirmed by data from any of these
reports (see Section 7).
Several experimental human studies have quantified PCE absorption in
terms related to what we here refer to as "percent absorption" or "percent
uptake" (operationally defined as one minus the ratio of alveolar to ambient
air concentration, ^multiplied by 100X) and we review such values here.
However, as noted above these values are of limited use in predicting the net
quantity of PCE retained over extended periods of time following environmental
exposure.
Yllner (1961) reported the average pulmonary absorption of mice exposed
to 1.3 mg/g of PCE in .air to be 701. In these animals, absorption varied from
42 to 87%. In a study by Pegg et aj.. (1979), the peak blood concentration of
PCE in rats during a 6-h exposure to 600 ppmv was approximately 10 ng/mL.
Monster et aJL (1979) observed an inverse relationship between uptake and
exposure duration in humans over the course of a 4-h inhalation exposure to 72
or 144 ppmv of PCE. The net uptake at the end of 4 h was approximately 60t of
that during the first hour. This observation indicates that net uptake
decreases as blood and tissue concentrations of PCE equilibrate with PCE in
the air space of the lungs. Net uptake is affected by differences in
11.
-------
ventilation rate. When volunteers were exposed to 142 ppmv while under a work
load (i.e., an increased ventilation rate), the uptake of PCE increased to
over two times what it was at rest (Monster et a_L, 1979).
Fernandez et aJL (1976) exposed humans to 100 ppmv PCE for 8 h and
measured the concentration in alveolar air. Alveolar air concentration rose
rapidly in the first half hour and then continued to increase throughout the
experiment, although at a slower rate, reflecting a sustained decline in
percent uptake observed.
The percentage of PCE absorbed in humans through the lungs has been
estimated at 50t (Ohtsuki et a].., 1983). After a 6-hr exposure of five
volunteers to 0.39 mg/L PCE, retention of PCE became stabilized after 1.5 h to
an average of 62% of respired PCE (Bolanowska and Golacka, 1972). The report
of Bolanowska and Golacka (1972) is contradict, by the data of Fernandez
et ah (1976) and Monster et a].. (1979). These investigators found that
retention of PCE showed no signs of approaching equilibrium after exposures of
6 to 8 h.
Table 3-1 summarizes the absorption and recovery of experimentally
administered PCE in animals. Table 3-2 lists parameters of absorption,
metabolism, and disposition of PCE in humans.
DISTRIBUTION AND BIOACOJMULATION
PCE diffuses into the bloodstream and distributes to tissues, primarily
organs and fat. The binding of metabolites of PCE to hepatic protein has been
measured by Pegg et aj.. (1980), and Mitoma et aj.. (1985). Savolainen et a_K
(1977) has also observed substantial levels of PCE metabolites in the
perirenal fat of rats 17 h after the end of exposure (200 ppmv, 6 h/d for
4 d). Seventy-two hours after oral or inhalation exposure to
tetrachloroC14C]ethylene, measurable radioactivity was found in the liver,
kidneys, fat, lungs, heart, and adrenals of rats. The major part of the
radioactivity was concentrated in the liver, kidneys, and fat (Pegg et a]..,
1979). A similar distribution was observed by Frantz and Watanabe (1983)
after PCE was administered to rats in a saturated drinking-water solution
(containing approximately 150 ppmv).
12
-------
,
VI
f**
anima
e
41
e
V
-C
2
o
u '
13
L.
41
O
U
41
0
U
41
t.
-o
S
f»
e
o
.u
^L !
a
VI
J3
41
41
U
C
41
41
ec
4i e
1- 0
• 3 •<-
VI **
o /a
a. i.
X 3
uj -a
c
a
3 41
IV VI
S. 0
e
41 U
U O
O
41
.4-1
3
VI 4rf
— en
, u —
4) 41
41
3
'"ifl
>
Parameter
^—
**Q
en
u
41
C
>
.C
CM
en
"X
CO
*
0
13
*9
^»
Ji
e
*•»
41
VI
i
_
CO
o
CM
01
o en
• e
g £
•^
41
' J3
U
O
VI
Percent at
(average)
' ' 00
*™
*
^j
' jW
3
x
-
x
•o .
in
en
en
s
1
0
^
41
VI
i
m
\o
an
41 .
3
VI
41 0
VI Q.
O X
•o.a 4»
o -a ' vi
4i a
*j u a.
C 41
41 > JZ
u o
<- u oo
41 -C
u a
i. U CM
41 41 r-
a. u ^
i
~
^
^™f
1^1
«i-
,5
a
'£
^
•a
1/1
en
jt
i1
- ^p
S3
0
1
•*J
ec
•w
CO
41
U
3
VI
41 O
vi a.
O X
•o JS
u o
t. u oo
OL 2 3
en
^™'
„
•
'"^
•a .
en
en
41
^
41 3
r— VI
f |
> 41
"
^S.
-
2
o
t*^
en
- S
a
CO
o
41
k. .
3
VI
31 §.
O X
-a 4>
a -a «i
41 O
— u a.
e 4i
41 > £
u o
U. U CM
4) 4i r-
Q. U ~.
en.
!«•
^
^J
43
at
en
-£
41 3
a, S
e a.
— x
t/t 4)
9
•s.
at
0
o
2
o
a
*j in
<« CM
OK —>
CM
cn
•J
L.
3
VI
4* O
VI CL
a x
•Q (I)
•^ *4
o -a vi
4i a
« v. a.
C 41
41 > j;
u «
<- U CM
4» 41 I-.
a. t. ^
v
e
*^
3
j
•a
e
'Q
N
_ fO
u oo
t. en
f
CM
en — .
J£ •
GA O
5 ••
— a.
oo -2.
41
S.
e en
•*• c
•'^
•— c
O -a
"en
03
CM
O "
m
m CM
S3
O
, . - O
£
w a
U u CN
41 41 r-
a. i. —
13
-------
i
8
01
3
C
0
O
PI
i
4)
(J
C
*•
VM
QC
4) e
1. 0
VI ^J
a 19
a. u
X 3
uj -a
e
o
^j 4)
^ O
_• -a
S u
u a
0
41
O
cc
VI «J
.2 f
ftl 41
ex 2
10 —
§
2
Parameter
1
^
ti
"3
Ol
Ol
41
Q.
.c
VO
Q.
a.
o
Q
•5
i
• Ol
•** vrt
<9 04
oc •—
en
£
3
VI
4> a
vi a.
a x
•a 4i
^B ^t
O T3 VI
« 0
** U Q.
C 09
41 > .C
w a
u u oj
cu 4) PH
a. u ~~
' «r>
1
J
y iq
m
Ol
Ol
41
a.
^
vO
i.
a.
o
VO
a
"S
.c >
c
"S
o
* O4
OC — •
0
0
a
41
3
1 Of O
0 X
•0 41
^^ ^J
o -a vi
41 a
.u u a.
C 4)
vj O
U U CM
« 41 r*»
a. u —
' §
. 2
3
tl
e
e
1 .
i/i
.c
vO
a.
a.
o
a
19
j=
e
41
i
o
§
£
'3
VI
4i a
vi a.
e x
•a 4i
*^ " ^
o -a vi
4i a
*j u a.
e «
41 > -C
u a
u u oj
4i 4) r»
a. u «—
to
2
_
19
L.
3
VI
- 1—
_
i
4)
a
41
VI .
i
e
Q
O
•x
c
CM
41
19
VI 1-
3
o e
Percutane
absorptio
CM
01 ~
~ ^
J 3
C
^^1 4)
OJ( VI
Ol O
Ol -*
41 - - 15
a. o
41
U
41 3
r~ VI
Ol O
CO. JS
••• X
01
s
1
-
2 1
p a
Ol
a.
01 ^
41
o e
,19 04 3
a ~- o
3
X
o
^ a. v/)
^- <»»• a
•o -a
1 J
"^ i ^ §'
19 •— ifl —
41 *> 41 *J
a.
-------
•a
41
3
e
• «•
^ .
e
a
i^
,
^B
|
4)
^3
1—
4)
U
e
4)
L.
41
41
oc
4i e
i. a
VI -*
a fl
o» ^
X 3
LbJ •^
e
a
Z 4)
41 -C
— O)
U •—
41 41
& S
(SI ^
V;
41
3
U
41
4)
g
U
2.
*•
&
*
*
^•J
I9
"Si
0)
o>
41
O.
^g
*O
1
i
e
a
%
"^
^^
e
^^
^
oe
^^
1 _
'c "3
* *
O) 9)
IT) O
K C3
U
fQ
e
1
3
a.
4)
^ ^j
a a.
h- 3
VI
41
><-
t~
O
^3
(^
4J ~
•
• -c
41
i.
A
a
VI
*J
e
i
4)
1.
3
VI
tQ
i
>,
jg
a
•o
41
e
3
a
u
u
UJ
°- .
a
41
VI
a
•o
4)
41
VI
'c
1
<4-
a
e
a
^^
u
a
o.
41
a
VI
u
41
2
•a
41
41
a
«
u
41 41
VI 3
a vi
•a vi
*4B ^
a'
•a
•^ e
4)
u
u iq
41 -w
& 41
1.
|Q U
X
41
N
3
a
_
o
^m*
^~ •
U
e
41
•a
•W
e
41
1.
ifl
VI
U
41
1.
Q.
a •
•"
41
^
41
41
a
.c
e
e
o.
i.
a
VI
13-
^
e
41
U
U
4)
a*
a
4)
'£'
U
^g :
e
41
J3
a
^
41
41
.^
VI
u
VI •
"* "n
T9 1^1
4t go
41 <—
a
U <£
41 Q.
I. UJ
e vi
41
u 3
a. iJ-T
4» "
J5 *••
^^
^3 •
"SI
15.
-------
fvl
r»
en
u
c
4)
U
41
St.
41
ae
e
o
*j
2
a
o
'U 41
c u a
§ 4) U
-c u , o
c-
C 0
— VJ
UJ
*•«
o
s
•*» A)
'M 0
a.
V)
'•5
•0
C
B* 41
VI 3
**• f"»
•"• ^
o >
*
1
e
a
^J
a.
u
o
VI
<
•
CM I.
1 41
4)
41 §
r- g
fQ fQ
H- a.
on
CO
ai
•
<3
•S3
vt
••»
Jrf
3
VI
*J
;§
e
a
m
*fl
•§
*"*
ei
vn
f
£
^j
*J
CJ
JQ
U
O
W1
^3
r^
^j
e
41 VI
u en
u e
41 3
a. <—
vO *fl ^r • • • •
ececk.
41 e 414) e u /o i) ra "o 41
VI e VIM |Q J 41 41 41 4) in
e ^ ce *— 4>.jQ.o.a.dc
a 41 55 O •••'3333Q
x u. xx 09 O *> *J U1
U1 Irt Irt , •
(\4 {V) t/3*T U) I/1COCOCOCN4O
vo <— v _ cn vn »• •—
e
a
f4l 41 41 —
"3 "3 "3 • •«
g«> *j w i.
Q. a. Q. -u
j: 3 33 e u 4)
-u eu — 3 >.
>> >i >i e u 19 * •&
•o u u u a evia
41 fd l«IQ — O •« •— J3
u a a o e — 41 c TS
a s as 41 j« ifl cn — fl o
I/I F— r— r— *•> r—
<9 0. Q. O. U a. 41 — J 0 ^>- j=
^| flj 3J QJ ^J ^ *
e en encn e *j i i i i i i
4) VI nj > > 41 —
cv -- < < < a. t-
16
-------
.
^
41
3
e
••••
e
Q
LJ
s
|S
& o
ON
m ~
e
o
ro
2
' C
• '
m
CM
VI
•o 01
•<-> 3
1 .2
41 "O
41
^i O)
e c
41 ro
u j:
u u
41 C
a. 3 ,
>T
•^
N
41
•a
e
ra
c
u
41
14.
•a
c
e
41
3
U
-e
00
a.
•a.
0
*^;
C
*•
<•
c
Ol
E
o
o
0
c
41
• rs
3
.a
• |
T
••^
41
•a
e
e
41
• U.
•o
s
e
u
41
J3
3 *
^
CO
a.
a.
o
^~
c
o
2
_g
c
OJ
Ol ifl
E .S
o x
o o
m u
. a.
c '
41
•a
3
41
VI
O
a.
3
r*
^-*
N
41
C
C
1.
41
U.
•a
c
rO
e
u
41
3
0
-c
00
a.
o
^
c
o
*
J?
c
-C
•
r^
CNi
—
*^
'JZ
^
ai
Q
o
-O
•59
41
(j
1
a.
•^
C ^
ON
*"
rO 4J
U U
I 1
1,2
•a e
C 0
•a
(fl Q)
•O VI
41 ro
1—4 •«!*
41
e? "iT
•^ t
VI 3
— VI
-c a.
r^ 4)
a.
a.
0
*"~
e
a
"3
2
c
^^
in
M3
CM
-IT
a
rO
^
f— „
a.
VI
41
ae
2
^_
u
3
i
js
C
•a
.41
"
•a
Lrt
, m
•^^
' J=
00
s
a.
0
o
o
*"~
c
o .
rO
•5
C
•JC
1
2
eg
>s»
^**
*^
-------
•o
01
3
'*•
C
0
o
B
CM
41
J3
>—
41
u
e
41
0)
"S
ee
e
o
IS
u
°
e
0
'H 41
||
4) U
U O
e
o
u
^j
3
O
ce
at
3
IS
41
1
1
IS
Q.
f".
2
j
ta
u
41
VI
c
5
JC
CO
1 -
o
a
e
a
'£
is
fQ
•C
c
^
uv
CM
41 41
IS O
*^ OU
VI ••-
1 -9
T3
IS •
51
VI •-
a .a
«J — 41
— 3
4) — VI
g 3 VI
.S er —
t— 4) **
i
j
ta
^
3
VI,
jC
0
f
CO
i
a.
1
e
a
'2
IS
IS
e
u
8
.12
o
J3
1
a
e
o
i.
3
IS
v>
in
CO
en
-
^j
/.•
41
a.
o
4)
k.
3
VI
e
a.
X
41
41
'01
. e
1/1
2
|
^**
/Q
u
O
e
'i
JS
o
va n
.• ro
CM
~
Q
e
J
PMB
UJ
c
'I
41
IS
a.
3
41
«J
" .1
VI
41
41
a>
41
§
U
**"
•a
41
*
3
U
Is
u
VI
41
3
>
i
41
<*.
*«-
T 4)
>— J3
IS IS
JS U
I/I 1—
, 4i a.
w a.
O IS
e
41 ->J
•a o
e
-- i'
J3 U
18
-------
Information on the distribution of PCE in humans comes largely from
reports of accidental exposures (Stewart et a_L, 1961a; Stewart, 1969; Hake
and Stewart, 1*977; Koppel et a].., 1985). Overexposure can result in central
nervous system (CNS) depression, cardiac arrhythmias, alteration of kidney
function, and liver injury. These observations provide indirect evidence that
PCE distributes to the nervous system, liver, and'kidneys. Additionally,
tissue concentrations in the liver and brain have been measured at levels 10
to 50 times greater than those in blood following fatal intoxication
(Lukaszewski, .1979). , "
PCE is a relatively stable molecule and is metabolized slowly. PCE is
soluble in lipids, and this factor and its slow rate of metabolism lead to its
accumulation in tissue following repeated exposure (Filser and Bolt, 1979;
Loew et ah, 1983).
On the basis of a model' of PCE uptake, distribution, and elimination,
Guberan and Fernandez (1974) predicted that the solvent woulS distribute
primarily to three body compartments; adipose tissue, muscle, and tissues rich
in blood vessels (some PCE will also enter poorly perfused tissues such as
bone and cartilage). This pattern of uptake and distribution is dependent on
the blood flow to a given tissue, the volume of each tissue, and on the
solubility of PCE. The model of Guberan and Fernandez (1974) shows that the
accumulation of PCE in all body compartments will increase rapidly during the
course of an 8 h exposure. At the end of this time, the greatest
concentration of PCE will be in muscle and fat (these tissues wiH have
approximately equal amounts of PCE).
Because the blood supply to adipose tissue is less than to other .tissues,
and because PCE is more soluble 1n fat than in blood, levels of PCE will
continue to increase in fat for many hours after the end of exposure. (This
is in contrast to muscle and other tissues, where concentrations of PCE begin
to decrease as soon as exposure is terminated.)
Once whole-body, steady-state concentrations are reached in relation to
concentrations in air, the amount of PCE in each tissue depends on'the tissue/
blood partition coefficient. PCE is lipophilic and the adipose/blood partition
coefficient is the highest of any tissue type (at 37°C the adipose/blood
partition coefficient is about 107) (Guberan and Fernandez', 1974). The model
of Guberan and Fernandez (1974) predicts that dai.ly occupational exposure to
-------
PCE at TOO ppmv, 8 h/d would lead to accumulation of PCE in fat .(at
equilibrium, more than 90% of the body burden of PCE will be in the fat).
Savolainen et a_K (1977) noted accumulation of PCE in the blood, liver,
fat, kidneys, and brain of rats following 5 d of inhalation exposure at
200 ppmv. PCE levels in perirenal fat, brain, and lungs rose continuously
during the experiment. Concentrations in blood and liver also increased, but
the rate of accumulation slowed by the third day.
Evidence of PCE accumulation is also available from examinations of the
concentration of PCE in exhaled air. Nhen volunteers were exposed to
100 ppmv, 7 h/d for 5 d, the concentration of PCE in expired air increased
with each exposure (Hake and Stewart, 1977). This suggests that the uptake .
capacity of tissues had not been reached by 5 d of 7-h exposures and that
consequently, body burden was still increasing with each exposure.
METABOLISM AND ELIMINATION
Among the most important enzyme systems for metabolism of toxic substances
are the mixed-function oxygenases (MFO). These enzymes are concentrated in
the liver, kidneys, lungs, and skin, and are present in other tissues as well.
Mixed-function oxygenases catalyze the addition of oxygen to compounds, which
facilitates their excretion from the body. Oxidation of a compound can
function as a mechanism of detoxification or can transform it to a reactive
(toxic) substance.
Evidence that MFO's are directly involved in the metabolism of PCE comes
from the work of several investigators. Moslen et a].. (1977) demonstrated tha-
a number of substances, including phenobarbital (PBT) and Aroclor 1254, induce'
hepatic MFO. Pretreatment of rats with either of these compounds, followed by
administration of PCE, increased the metabolism of PCE five to seven times ove.
controls. Costa and Ivanetich (1980) showed that substances that inhibit
cytochrome P450, a component of MFO's, also inhibit metabolism of PCE in rats.
Induction of cytochrome P450 by PBT or pregneno1one-16a-carbonitrile
increased the metabolism of PCE. PCE was also shown to bind to the active
site of P450 in rat hepatic microsomes.
The first step in metabolism of PCE is thought to be transformation to ar
epoxide by the MFO, although this epoxide has never been isolated in vivo
(Bonse et §1., 1975; Greim et a].., 1975). The epoxide of PCE rearranges
20
-------
spontaneously with migration of a chlorine to form trichlorqacetyl chloride
and, ultimately, trichloroacetic acid (Moslen et a_L, 1977; Leibman and Ortiz,
1977; Reichert, 1983). Trichloroethanol has also been reported as a
metabolite; however, a pathway has not been proposed that explains its
formation (Ikeda and Ohtsuji, 1972; Ikeda et a_L, 1972; Monster et a_L, 1933;
Koppel et a]... 1985).
As yet, it is not clear if conjugation (phase II) reactions are involved
in the metabolism of PCE (Pegg et §1., 1979; Lafuente and Mallol, 1986).
Although not consistently reported, the production of oxalic acid and CO as
metabolites of PCE has been documented in rodents (Yllner, 1961; Pegg et a_L,
1979; Schumann et §1., 1980). Their formation argues for a second and ~~ '
possibly minor pathway of oxidative metabolism.
It has been proposed that oxalic acid and C02 are formed as end products
of a metabolic path that also includes ep.oxide formation as the first step
(Pegg et a].. 1979). In this scheme, chloroethylene glycol is formed from the
epoxide by the action of epoxide hydrolase/ This reaction is followed by
hydrolysis to oxalic acid and/or decarboxylation to CO- and possibly formic
acid.
Chloride, dichloroacetic acid,and ethylene glycol have also been reported
as urinary metabolites of PCE in rodents (Daniel, 1963; Yllner, 1961;
Oimitrieva, 1967). It appears that these substances are minor metabolites,
and little is known about their formation. Figure 3-1 shows the structure of
metabolites as well as the proposed metabolic pathways.
There are some similarities in the urinary metabolites produced by humans
.and rodents following exposure to PCE. Trichloroacetic acid appears to be the
principal metabolite formed, regardless of species. Tables 3-3 and 3-4 .list the
metabolites identified in humans, mice, and rats, as well as the conditions of
exposure.
Yllner (196V) was the first to analyze urinary metabolites of PCE in mice.
Although 18% of the radiolabeled compound was not accounted for, 52% of the
portion metabolized was recovered as trichloroacetic acid (TCA), 11% as oxalic
acid, and a trace amount as dichloroacetic acid. Daniel fed labeled PCE to
rats and found TCA (0.6%) and inorganic chloride as the only metabolites
(Daniel, 1963). Total urinary metabolite production in rats was measured by
Moslen et a].. (1977). They did not identify the actual metabolites,' with the
exception of TCA (which was the major metabolite produced). Pegg et ah (1979)
21
-------
Cl
c=c.
ci
Tetrach loroethy lena
Mixed-function
oxidases
(02, NADPH)
CL
Cl
Cl
Tatrachloroethylerw oxide
Epoxide
hydrolu*
Hydrolysis
Hydrolysis
CO, -r-
Carbon
dioxide
H-
OH
HO
OH
Formic Kid
Oxalic acid
Chloride
migration
Cl
Cl
\
Cl
Cl
Trichlofoaeetyl chloride
Hydrolysis
Cl
Cl
c—c
OH
Cl
Trichloroacatic acid
Figure 3-1. Metabolic pathways of PCE (brackets denote compounds that have not
been Isolated In vivo) (Daniel, 1963; Pegg et'alL , 1979; Costa and Ivanetich,
1980). Although trlchloroethanol, dlchloroacetic add, ethylene glycol, and a
thioether have been Identified as urinary metabolites, the route(s) by which
each Is formed has not been characterized. ,
22
-------
VI
C
CD
O
^
C
"~
C
o
4-1
u
•a
•o
^
0>
4->
•f~
*™-
O
.0
ra
CD
UJ
a.
•
i
m
OJ
2
re
t—
a;
u
c .
CD
w
OJ
CD
Q£
O
CU
c u
O 3
re a.
U X
3 CU
a
<*- a>
0 u
3
CU vi
-M O
3 a.
0 x
c
0
4n» CU
rO VI
i- O
4-» -a
c
CD !_
U O
o •
CJ
OJ Q i—
*•* ^rf •
..- re
U • £
CU O •«-
a. s c
oo --/ /a
0)
4-1
•^
»••
O
ra
cu
3E
_
WO
Ol
*—
t
c
»—
^•>
*~
-C
CM
c
o
-M
ITS
*—
ITS
J=
c
01
1
*—
in
3)
VI
3
i
u
£
cu
u
re
0
o
,—
u -a
•j_ u
I— re
WO
CJ1
m
WO
O^ re
1— >
CD
• •*-•
CU 4-1
•^ »f-»
C £
^ •*••
O O
4>
^
3 T3
VI
O rn
a.
x u
4-1
<£ 2
CM
2
— "
3
-M
i—
O
•o
c
ITS
re
•a
CU
^^
, 1— 1
^^
co
c
o
4->
rO
*—
re
fm
• C
f
^
£
a.
a.
o
o
CO
4-1
a:
<^j
2
— "
3
^J
O
•a
c
ra
ra
•o
CU
^£
1—1
cu
i.
3
vt^
o
0.
X
cu
CU
"01
c
•^*
00
Q.
Ol
^
f^
CO
r—
CM
p—
4^
•a
CM
2
— "
3
• -M
r-
O
•o
c
ra
re
•o
cu
J^
*—4
^^
CO
c
o
4-1
O ra
z ce
-
-M
cu
re
—
_
o
"O
c
re
c
cu in
J2 OO
3 cy>
03 •—
'"^3
vO
o
^
4!
•a
"i
cu
Ol
ITS
re
Ol
•^^
Ol
os
-MO
o
o o
CM CM
O cu
4-> VI
o
T -a
w <-»
O) CU •—
vi a. cu
3 >
O ^- cu
Z CM —
CM
2
•*•••
•f"^
3
VI
^.
O
•b
c
rO
•a
cu
^*
^~"
.c
CO
e
.2
•^
fV>
!•"»
rtS
.e
c
^
£
Q.
Q.
O
o
CM
CO
>~'
4-1
re
cu
'o
c
re
JS
cu
o
o
.£:
u
•f^
H—
CM
2
.,-*
3
VI
.C
0
•^
c
re
re
•a
cu
•^y
H-l
^
03
e
o
4-1
ITS
ITS
C
*
^
£
a.
a.
o
o
CM
^
*^" '
CU
(/I
3
i
CM CM
P** P^
O^ Crt
"~ "~
•r- — "
3 3
VI VI
4^ 4-*
UZ <~
0 0
•O T3
C C
re re
re re
CD CU
I— 1 1— 1
CU CD
3 3
? VI
0 0
a. a.
X X
cu cu
CU CD
Ol Ol
c c
oo oo
o. .a. •
1-1 .NH.
O) O)
^ ^
^•B ^_,,
§*
gs
i
oo co
CM CM
P"» CM
>*x >— '
CU
VI
4-> 3
rO O
23
-------
s"%
•a
ai
c
c
o
0
•
en
i
en
at
«— \
itj
H-
(*-
O
C
O
Id
u
3
a
o
ai
4-1
3
O
OS
o
4-i
rd
+j
C
ai
u
e
t/i U-
O) O
°u •
0) O
a. c
CO >-»
Ol
u
c
ai
i^
01
<+-
0)
ce
ai
t-
23
i/i
a
X
Ol
u
3
VI
O
a.
X
•^•J
•f—
^"
o
.a
rd
4*1
d)
3E
,_
VO
cn
»••—
«k
U
0)
c
^H»
*"
f
CM
c
!J3
rd
i^—
fa
_£J
C
I—I
en
C3J
£
en
^
ai
v>
3
•a
'u
u
P»
rd
X
O
VO
o»
^^
•
rd
>
OJ
*^>
•M
•f"
E
5
•a
en
u
0
<*_
•a
^2
in
c
O
^
rd
^^
rd
j^
C
_J
*«*,,
i1
0
^^
4^
ra
ae
S S
•— <—
. .
, •
f«>l P^
rdl rd
4->| 4-1
ail at
O) CT
O) O)
03
C fl
« (J
> en
t ^
<=L f
8 o
vo o
in
^
^D I*
^3
o
en en
4-1 4->
rd rd
r-
•~
.
•
*—
rd
4-1
Ol
en
en
Ol
a.
^^
VO
c
o
'^3
rd
i—
rd
JZ
c
>
t
a.
o
O
VO
j
O
o
en
^j
a:
Ol
•a
X
O
•5
c
o
J3
t»
rd
O
O^
•~
.
•
^— I
rdl
4>j I
0)|
en
en
o>
a.
Ol
=
i/l
0
a.
X
OI
, ai
en
c
CO
Ol
en
rd
rd
en
i1
O
o
o
en'
^
4-1
rd
ae
0 0
so co
O"l OA
-
"rdl "rd
4J| 4J
Oil O)
C C
c c
i i
.C -C
u u
01
e 3
I/i
O
a.
x
Ol
0)
J= C
vo J5
c
o
4^
rd 01
rd rd
JC >
C rd
N»l (^
> en
E ^
a. -^.
a. ' en
8 o
VO O
IT)
^
0 s-
o
en en
^ «-*
O) Ol
> >
3 3
_
VO
O^
»—
•
s_
Ol
c
im
*" : ;
^gj
CM
=
o
*4j
rd
^M
r«
^^
c
1— 1
en
. ,^^
m
P5
en
•
IT)
^ .
01
en
3
• .2
u
rd
O
Lm
Q
f^
.s -a
• U .t-
1 *^ CJ
O rd
VO
cn
«
rd
^
Ol
•^
u
•f«
E
a
•a
en
VJB
0
*^»
•a
iG
m
e
0
4-1
rd
fmm
• rd
^c
C
MM
.J
tk ^^
O)
ID
^^
4^
rd
cc
^
o
u
,_*
en
01
c
O)
1^—
>•)
^^
4-1
LU
en
vo
•CTl
— —
*
Ol
C
rd
O
-.*
3
t/1
o
o.
X
Ol
Ol
en
c
•^
00
*••» .
rt3
^
o
en
^
en
E
0
2
vo
***
4-1
rd
ae
Ol
T3
•i^
^
O
»— "
jC»
O
24
-------
t
c
cu
ai
^4i—
cu
ce.
(4-
O
0)
C i-
O 3
4J O
IQ a.
S_ X
3 ai
a
'o-u
3
O) vi
•M O
3 ex
0 x
Q£ . .,—
•<-• o >
"o Z;o
ra • .r- -
«u
LU
<_>'
Q. v» C C CU
CU 3 (9 t9 ^3
O C
•a 19 «M -o ~o c
-^ O en _^ ^ ai
>— I CO •— l— i I— I U.
•
Si>*
O)
^" co ^o co
.
CN CS4
r^s ^
wi C^
§ao GO ao in
^^ c^ ^n GO ^j * .,
£ - ._ _ ^ ^ ._
3 ...
o ...
X — j — —
a
S- V. i_ U
cu cu a> cu •—
2 ? '"
* * M aC
; « ° S
J • * c 1
cu «
» '
4J 4J4-I4JCU c c c a. cucu _i _ ,_,
cu
^^ ^^ ' '-• *
• — cu cu cu
•a . — « 4-> 4-1
vo _= cu .- 'i ^
v» 3 i_ i. GO
j= ^o.^c ^c4->
GO **O QJ C3 ^i— ' O s**1' <~~
c c c c e
o o o o o
4-1 4-1 4-1 4-1 4J
J2 ^2 ^2 j2 ^2
^ rtS /4 rti /ti
•C -C .C J= .C
= C C C C
•— ' •— • . H- 1 P— 1 1— 1
i > i i-
a. E a. a.
1 ^ 'I i J
o .S o 2 2
_ .0 o o
O en o O O
>— m «M csj —
m to ^ vo «g-
QO (43 CM
U
^.4
cu .
o
O
o
(—
u -a
•^ »^™
^ ^ ^: 3
-^ 4-1 j_i
a, ^ ^ -J S ' ^ = =
— j=^!ca. "" * *
Cjl<-s - x ij **
C CU ^ *J* ^ CU *O ^, ^
"- u to en un s ' ' 's •
v» 3 « CU . £ £
s_»v» o O O— "O >>eu >»cu
O -M 4-< 4-> CT "^ —4-1 —4-J
-C ex u c -C •*- e «— cz
^j P^^ -^ • _l| ^^ ^^ "^ ?^ •• •
^rcu m m m > o
O a. ex ex o
0. 4->
O r» m
r^ r^. ^ . (D oo
vo M en en -
c , c c
o o o
'•*J 4J 4J
^S ^0 ~ ^fl
^^ ^i» ^_
ro n3 TO
c c c
S > CL
Si. . °-
o * 8
5 S ^
o o ' 5
4-> 4-1
00 0
— CM CM
IT) «5J- VO
<^ (43
—
O
c
ra
£
cu
O -
0
•—
u
b_
25
-------
•o
O)
3
C
C
O
u
>
^»
1
m
O)
.0
* flj
O 1-
3
CD v>
4-» O
3 a.
O X
cr ai
c
O
•*••
4>j ^
id v>
U 0
4-> -a
c
oj v.
u o
1
vt
"id
<*- 3
o ~o
O *>
•5
C
•*••
O)
4-1
'""
f^
J=]
•a 3 »
>t OJ ^— ' t/1 O
~- 4-> O 4J
••- C -CO.
o >-* ^ d> fi
e c c
o O O
4V 4-> , 4-1
(& ^O ^0
^MB ^H» *H*
^9 ^S ^9
-C J= -C
c c c
1— 1 1— 1 I-H
e >
Q. m
a. , a.
a.
0
0 O
4J i_ e
o a
8 o »
«T <43 ~ LO
CM
^
mm
03 CO —
O1O% O
^d
^l •—! ^
rdl id| "O
c
4J| 4->| id
^1 w|
O)
W ^ 4^
O) 3 VO
e c «- ao
3E X — 1 •—
22 2
4= , 4= -a
^T ^T W^
un un
*
O O T3
4-> 4-> •»«
4=
m m 03
e c c
O O O
4-14-1 4-1
id id id
^" f«^ ^"»
• & rG . ^
^W ^M -C
c e c
1— 1 1— 1 1—4
>
g
a.
a.
rd id O
> > m
I. i. o
a. a. 4-1
r-» m LO
^r un -—
01 01 ^
a>
^
•»-
id
^
•»i—
V.
O)
"^
u
ai
4=
^•rf
CD
id
u
O)
^
id
I
•o
O)
4-1
^**
•£•*
OJ
2
1
f-
4-1
4-»
C
0)
at
^
a.
a;
^
VI
3)
3
^™
^a
>
rd
26
-------
and QimHrieva (1967) Identified oxalic acid as the primary product of PCS
metabolism in rats. Carbon dioxide has been recovered as a metabolite of PCE
in mice (Schumann et ah (1980) and in rats (Pegg et ah, 1979).
In contrast to rodents, TCA and trichloroethanol are the only metabolites
of PCE that have repeatedly been Identified In humans. In some instances,
identification'of trichloroethanol has been by the Fujiwara reaction (Ikeda
et a_L, 1972; Ikfda and Ohtsujl, 1972). The results of this test are
qualitative and the accuracy is questionable. However, Monster et a_K (1983)
and Koppel et a],. (1985) determined the presence of trichloroethanol by gas
chromatograpny. Still other studies have failed to detect trichloroethanol;
therefore, it 1s far from clear under what circumstances this substance is
formed (Monster el: aK, 1979; Fernandez et a_K, 1976). Measurements of human-
metabolite production are confounded by the fact that carbon dioxide, oxalic
acid, and chlorine are normal products of metabolism. Their presence as
products of PCE metabolism could be quantified only by the administration of
radiolabeled PCE—a procedure which has not been undertaken with humans.
Lafuente and Mallol (1986) have reported the presence of a thioether
derivative of PCE in the urine of women occupationally exposed to PCE. A
gradual Increase in thioether production was observed over the course of a
week. This Increase appears to have been associated with continued exposure
to PCE (with concomitant accumulation) over the work week. However, measured
amounts of thioether in exposed women were reported not to be statistically
different from levels found in non-exposed individuals.
The recent Identification of thioether derivatives in humans exposed to
PCE, coupled with .problems associated with accurate and complete Identification
of metabolites, indicates that not all human metabolites of PCE have been
accounted for. It is possible that characterization of PCE metabolites in
rodents is incomplete as well, since no complete mass-balance studies have been
conducted. Although the major metabolites recovered in all species are
presumably produced by the same enzymes, many uncertainties remain. Based on
present Information, humans and rodents appear to metabolize PCE in
qualitatively similar ways.
PCE is eliminated from the body by two major processes; metabolism
followed by excretion of urinary metabolites, and pulmonary elimination of
unchanged PCE. Although some PCE may be eliminated through the skin,.
preliminary measurements indicate that this is a minor route, at least in
27
-------
humans (Bolanowska and Golacka, 1972). Most of the PCE systemically absorbed
under experimental conditions (e.g., in mice, rats and humans) was eliminated
unchanged In expired air, so that metabolic degradation and subsequent
elimination are thought to account for only a fraction of absorbed PCE.
Experimental data reviewed below indicate that metabolism is dose-dependent
and saturable, and that the amount metabolized appears to be species-dependent
as wel1.
When the production of urinary metabolites was measured 1n mice given
1.3 mg/g of PCE for two hours, Yllner (1961) found that only 21 of an Inhaled
dose was excreted by this route. Seventy, percent of the parent compound was
recovered in expired air. Pegg et a_K (1979) compared the metabolism of PCE
in Sprague-Dawley rats at different doses and routes of exposure. PCE was
administered by gavage (1 or 500 mg/kg) or by inhalation (10 or 600 ppmv). The
primary route of elimination of PCE was through the lungs as the unmetabolized
parent compound. Urinary excretion of metabolites accounted for the majority
of the remaining PCE. The percentage of PCE metabolized was dose dependent:
elimination of unmetabolized PCE in expired air following oral exposure
Increased from 721 after 1 mg/kg, to 901 after a dose of 500 mg/kg. there was
a corresponding decrease in metabolites from 28 to 101. An analogous pattern
was seen after inhalation exposure. At 10 ppmv, 681 of the dose was
eliminated unchanged through the lungs and 321 was recovered as metabolites.
Treatment at 600 ppmv caused an increase in pulmonary elimination to 881 of
the dose, with a concomitant decrease in metabolites to 121. Pulmonary
elimination of PCE was linear and had a half-life of approximately 7 h. The
half-life was Independent of'dose or route of administration. The half-life
of PCE in blood was 6 or 7 h, after oral or inhalation exposure, respectively.
The pharmacpkinetlcs of PCE metabolism in Sprague-Oawley rats and B6C3F1
mice were studied by Schumann et aj.. (1980). Rats were given a single oral
dose of 500 mg/kg or were exposed to 10 ppmv 14C-tetrachloroethylene by
inhalation. The authors reported that at 10 ppmv the major route of
elimination was excretion of unmetabolized PCE 1n expired air, although
supporting data were not provided. Some PCE was metabolized, and metabolites
were recovered in the urine. After an oral dose of 500 mg/kg of labeled^PCE,
radioactivity was measured in the expired air, urine, feces, and carcass. The
relative importance of each route in the elmination of PCE was not discussed.
28
-------
The elimination of PCE by B6C3F1 mice differed, depending on the dose, and
possibly on the route of administration. Metabolism, with urinary excretion
of PCE, was the primary route of elimination after inhalation exposure at
10 ppm; 62.51 of the dose was recovered as urinary metabolites and only 121
was excreted through the lungs. Eighty-three percent of a single oral dose
(500 ing/kg) was eliminated through the lungs, while 10.31 appeared as urinary
metabolites (Schumann et aj.., 1980).
The fate of PCE 1n Sprague-Dawley rats fed PCE in their drinking water
was reported by Frantz and Watanabe (1983). Animals were given PCE in a
saturated solution over a 12-h period, resulting 1n an average dose of
8.1 mg/kg. Treatment was followed by a 72-h observation period prior to
sacrifice. Most (87.91) of the PCE was eliminated unmetabolized via the lungs.
Although urinary excretion was the second largest route of elimination, the
amount metabolized was .relatively small (7.21). The half-life of pulmonary
elimination was 7.1 h; this 1s nearly identical to the value determined by
Pegg et a].. (1979) after oral doses of PCE were given to rats.
Mitoma et aj.. (1985) studied the metabolic disposition of PCE in Osborne-
Mendel rats and 86C3F1 mice using a follow-up period of onl^y 48 h.
Substantial differences between species in the amount of PCE eliminated
unchanged in expired air were noted. There were also differences in the
percentage-of dose metabolized. Rats eliminated 791 through
»the1r lungs and metabolized 51 of a 1000 mg/kg dose. Pulmonary elimination by
mice (of a 900 mg/kg dose) was 57.51, while 221 was metabolized. The actual
amount of PCE metabolized (measured as mmol/kg) also differed by species. Mice
metabolized approximately four times as much of the total dose of PCE as rats.
However, in both species, as the dose was quadrupled, the amount metabolized
only increased about 2.5-fold. These data suggest that metabolism approaches
saturation at high doses in both species.
Buben and 0*Flaherty (1985) also demonstrated that oxidative metabolism
of PCE in Swiss mice is a saturable process. Animals received 0, 20, 100, 200,
500, 1000, 1500, or 2000 mg/kg-d of PCE by gavage and were followed for 72 h
post exposure. TCA was measured as an index of PCE-metabol1te production; the
estimated maximum rate of urinary metabolite excretion was 136 mg/kg-d. As
the dose Increased, the percentage of the dose metabolized decreased. At the
lowest doses, approximately 251 was metabolized; this decreased to 51 at
29
-------
the highest doses. Tables.3-5 and 3-6 summarize data on the metabolism of PCE
in rats and mice, respectively. Table 3-7 lists some pharmacologlc constants
of PCE 1n rodents.
The data reviewed above Indicate that metabolism of PCE in both rats and
mice displays saturable kinetics. The percentage of dose metabolized decreases
as the dose 1s Increased until the amount metabolized is no longer a function
of the dose (a zero-order reaction). The extent of metabolism of PCE is
apparently species-dependent. Rats consistently metabolize a relatively small
amount of PCE regardless of the route of administration. Mice typically
metabolize a greater percentage of a dose than rats.
Measurements of PCE metabolism 1n humans have many uncertainties
associated with them. Total tr1chlorinated metabolites in urine have usually
been determined colorimetrically and therefore are difficult to evaluate
quantitatively. By this method, production of chlorinated metabolites is
equated with the amount metabolized. This approach does not account for the
possibility that some metabolites may be produced that are not chlorinated,
such as C02, the thioether, and oxalic acid.
Studies qf PCE metabolism 1n humans have consistently identified TCA as
the principal metabolite. TCA production has been measured to estimate the
extent of metabolism of PCE and also to characterize the kinetics of urinary
elimination. Volunteers exposed to 87 ppmv of PCE for 3 h excreted about
0.40 mg/h of TCA (Ogata et ajk, 1971). Monster et a].. (1979) analyzed the TCA
content of blood and urine from volunteers exposed to 72 or 144 ppmv of PCE for
4 h. The mean production of TCA (over a 70-h period) was 6.0 and 11.0 mg.
Blood levels of TCA increased over the course of the experiment and continued
to rise until about 20 h after the end of exposure. TCA was eliminated from
blood by first-order processes; the half-life of elimination was 65 to 90 h.
Urinary elimination of TCA followed the disappearance of TCA from blood.
Fernandez et a].. (1976) also measured excretion of TCA from individuals exposec
to 150 ppmv for 8 h. Over a 72-h collection period, the average amount of TCA
produced was 25 mg. This 1s equivalent to about 0.34 mg of TCA per hour.
Ikeda (1977) and Ikeda and Imamura (1973) qualitativelymeasured the half-life
of urinary trichloro compounds and calculated thejnean half-life to be 144 h
(range of 123 to 190 h). The lengthy half-life may be due to the continued
formation of metabolites,from PCE mobilized from tissues where it has
accumulated.
30
-------
e
41
u
41
H-
41,
at
VI
Ol
«
k
41
C
41
4)
O
a
44
41
a
a
VI
I/I
1
41
U
« e
44 3
i:-
4i -a
44 §,
c e
41 14
U A
U U
4i e
a. 3
u
VI
41
44 a
,23
VI
41
£
I/I
2
a. «
x
41
£ i
3 ••-
S «
II
44
14 -a
W 41
44 VI
e a
4i -a
u
S £
u
ex
3
£
in
m
>*3
2
41 3
r- VI
01 a
e a.
£ S
Ol
s
^ •»
-
•s
td
s
41
CO
CM
PO
^
CO
(SJ
^^
_g
^D
^^
Ol
JC
^^
1
CM
2 3
a
44 , .
•2
45
1
rtl • ' ^ft ^ft
2 2 2
5 - -
1 ^ '. *
m rtj (jj
4ta*
C CO O) O)
rtj CD - O) O)
k ty\ 4u - 4V
u. — a. a.
§. CO 0
CO ^^
CM CM • O
""""""
CM \O
r. vo *T
F'
CM . CM CM
r» . p»'-. r»
41
« 3
«— VI
? S.
CM — X
^
01
-* Ol
j*> §. s" "oS
0» CM
• o ci a
co vo — • in
^^
u
41
- s
O>
— e
.x a
e —
•a —
"- o
1
-
•si
m
2
z
r>
' in .
•T
CM
.S
CM
Iff
x •
^y
Jt
X
•a
in
O)
i1
a
o
o
2
o
-
^
41
•*•
•3
44
4l
g'
VI
X
44
.C
44
41
VI
a
•o
41
w
41
a
2
, *a
^J
e
41
U
U
41
0.
41
e
4)
VI
41
O.
41
U
I/I
41
3
"?
. 41
VI
41
fi-
(0
Ol
S
CM
a
*£
Ol
41
X
41
Ol
U
41
C
e
•a"
"^
"a
CO
0
o
41
44
U
01
!c
41
U
.a
- *
Ol
• C
'i
3
VI
VI
TS
.
i1
c
41
I/I
a
•o
41
g*
44
VI
41
U
3 .
VI
e
a.
X
41
e
a
^
2
.S "5
- k •*«
a e
U. 19
* i
a.
•
e
o
44
41
U
X
41
14*
a
VI
41
3
O
f*m
(Q
U
^B
^
41
O
U
41
U
41
44
^M
a
flj
i
44
e
41
VI
41
U
a.
41
VI
41
3
H
41
VI
41
J=
U
31
-------
gj
u
n
e
41
C
«
*><
At
41
2
0
u
2
4)
«J
V4-
o
C
O
£
O
0.
VI
•o
I
r»
.5
PMB
o
(U
•
^3
f*i
4)
i
41
U
C
01
u
01
**•
oc
VI
T3 OI
oi e
At 3
s -
"* 2
•5 '>
*« tj
01
g 1
U -C
u u
a
N O J3
•"- AI
5 2
i»
1 *
VI
e x AI
41 U u-
u m •—
U CO
4i ••- -a
a. i. *
=> |
qj
i. -a
1 3 0
At I/I —
M a v.
O a. 0)
a. x a.
4i e
U 0
3 —
VI At
i.2
X 3
ui -a
§
*£*
2%,
At M
e a
4i -a
I»
O
*«
i
et
f 1 •'•*?•
• I
"5 • «
Ite ^M ^^
O H VO O
4» O\
"c c c
1
1 * 1 1 1 ->
3 ^^ U ^™ 3 Gft
CO - «« > OB -
S ^ 23
Z ^* n» Z
0
^ S i 2
V
in m o in
CM CM •
i W >«,
•a At ^ -o
M V0 CM in
j
^^
a> jf 9 Jt
I |y s* >
o a CM • a
CM i— •-• •"• CM
e e
o o
Al At
1— F-
1 ' 1 11
>>
Al
U
01
i
lA.
O
•Q
e
tj
e
oi m
JS CO
CO —
2
$
VO
CM
A
0
-5
^o
Jtf
>»
•o
m
at
JC
>»
2
e
o
CO
<*V in
•— CO
•~
*fll •
«l *
G ^(
e
u —
W X
vo in
CM r«
co in
^" CM
^ CM
— CM
<*» rr
O »T
*•• ^
' Si J=
CM CO
J<
^y
01
U
,41 3 JC
^» 0 "X
C CV ^
-------
3
a
e
•^
e
a
•
^D
d,
41
^
^
^
_
•Q
4)
1
"5
p^
41
^J
C
41
U
W
41
a.
•a *«
41 •<•>
N 0
•» I—
f^
e
.a
nj
1
e >i
41 U
U V»
vi e
o a.
a. x
41
£
3
VI
a
X
UJ
§
4
i.
•
U
4t
*
e
13
•o
e
•
•o
-s
WB
IT)
5
o
o
(4.
a
.2
nj
u
o»
!c
19
41
' U
A
13
g>
i
VI
v>
•o
41
3 ~
U
U
VI
M
X
Ol
-N
?
e
VI
a
•o
5
4^
•
^^
VI
"v
u.
3
I/I
X
41
e
.2 •
^rf ^*
•i— S
ns —
^ e
«^
^ I
u.
91
10
O
CM
.
^^
e
o
VI
41
.«•
a
^
^^
n
«
c
.^
3
a
u
41
a
u
41
U
e
41
VI
£
S"
u
VI
41
a
^
41
i/l
41
ja
41
e
41
"5,
^j
4rf
4)
a
a
2
u
u
41
•-
^3
41
. U
•^
X
4>
a
e
a
4J
Q.
41
U
X
41
41
^
4^
X
•
4)
U
a
VI
a
Q.
X
41
a
vi
4)
a
a
. w
^•B
/Q
§
*•
t.
41
a
u
41
e
41
VI
£
a.
41
V
VI
41
3
^
4)
VI
41
I—
U
41
jg
ifl
131
(3
«
^J
a
c
•'i
Q
33
-------
,
VI
e
ei
•a
o
k
e
41
e
*>•
*j
o
JZ
u
1.
*l
<*»
o
VI
1
VI
o
u
u
'en
a
o
u
1
m
£
r»
A
ti
1
41
U
2
41
41
«> e
k o
3 •-
VI *J
O 19
a. k
X 3
Ul Tl
I
3 *
2 S
«j -a
41 k
U 0
a
«
3
a
oc
*^
VI **
.2 *
a. s
3
>
U
41
I
^ft ^ft 1J1
^* ^^ ^^
o o^ o
*"" *~
•a 3 3
tJ "S3 "Si
en en en
en en en
41 4) 4)
a. a- a.
41 , 41
u k
41 3 41 3
•—VI f— VI
en o en o
CQ. j= e e. .e
— X •«• X
* 41 O l/l 41 ^
en >
•x S.
€ , I =
en • o \e
8 S. in
^W «
k O
a o o -a
o o co
in « *.(»*.
I , 1
4J e *»
— fl i— <•
>a -c m .c
k c k e
a N* o —
.. ._•
en en en
o o o
*> in *• in -w in
* CM It CM 19 CM
S
0.
f * * *
vo r» r>. -~
19 19
^ ^ *
f- f- Q
S § S
5 '5 3
C C r-
•5 i2 9^
1 "« !s
1*
a a i CM
•2 ^ "si
S S a. *T
Cft ^^
^ft ^ft
•d -a
44 ^M
en en
en en
d! 2
41
k
41 3
r— VI
en a
CO. J=
•^ X
VI 41 V0
en
1 1
S S
lO vO
S
JJ
iq
^**' iQ
<9 -C
C e
0 «
en en
o o
•M m *• in
fl CM « CM
a* •— ex —
"L "L
CM O
e d
*? C
e e
3 3
VI VI
s ' e
a e
w w
1 I
cr'£ cr'i
J» 8 -"8
s 3 * 3
s " s *
03'— CO v
34
ON
2
j
'•a •
s
41
a.
41
k
41 3
1— VI
en o
e a.
— X
> «4
k *
<9 e
||
41
<«
U.
O
•a
S
e
.0 03
3 91
CO —
X
vfl
• •
jf
X
•a
in
1
e '
a
o
CM
O
a
CM
^»
19
k
0
4>
VI
' i
13
£
SP
CO
Vto
O
fM -O
41 41
1 = 3
ill
• -
2 >» * 2
i k •«
(O 41 V
x e • CM
3— C
> s 5--.S
.
2
j
^
03
•a
e
19
k
41
VI
^J
CM
"c
e
i-s S
a. ••- c
*> 41
o — u
ace
o ••• e
.— ~- u
S
'£
19
/t)
J=
e'
^ .
en
CM
e
U CM
ae ~-
en
1
^
i
p*
V
u
'o
3rm +* -^
8 e
'x ••-
3 21 *
•i .- e
> > U
1
J
'o
•o
e
i>
k
41
VI
'• iZ1'
^J
CM
i
a
o
a
o
=
'3
«
4
&
e
en
a
CM
O
« 0
oe CM
»
U1
a
S S
1 1--
.^
3 *"
cr c
41 »
VI
er e
41 O
^ U
-------
u
e
41
b
Vto
2 S
3 —
Is
x a
ui -a
e -
'a
* vt
i> a
^j ^9
e
41 k
u a
§
4)
at
VI *»
.2 i,
u ••-
,*i
S
i
4)
*>
i
I.
2
C ^
^" ^n *
•3
i -s ^
a 73 §
33 1
«l 41 U
A
• ^™
^9 O \O
* .
| 1 i
2 1 2
S S §
'-S '^ *^
* * «
•s -<= -2
£ 5 '- -5
0 0 S
*• vrt *• ui a
<« PH <« CM O
OC — 06 — £
o> e> S
||f
* 2 "
a? f> >—
! ! 1
J J J
•S1 -S" -o*
s s i
e
a
e
'i
41
1
Y
JZ
41
5
VI
/H
35
-------
•Measurements of TCA production suggest that human metabolism of PCS is
limited. For example, the values published by Monster et a_[. (1979) (6.0 and
11.0 mg TCA after a 4-h exposure to 72 or 144 ppmv PCE) represent only about
1 to 2% of the estimated absorbed dose. Ogata et a].. (1971) reported that TCA
excretion was equivalent to 1.8% of the retained PCE, and that total excretion,
of organic chloride accounted for only 2.8X of the retained dose. These data
are in agreement with those of Fernandez et a_L (1976), who estimated that
13SO mg of PCE would be absorbed after" exposure to 150 ppmv PCE for 8 h. The
25 mg of TCA produced corresponds to metabolism of 1.85% of the absorbed dose.
Ikeda.et aj.. (1972) and Ohtsuki et a]..- (1983) have estimated that only about
2% of an 8-h exposure to 50 ppmv PCE would be metabolized and that 381 would
be eliminated through the lungs unchanged .by the end of the exposure period,
the remaining 60% of the inhaled dose was hypothesized to be stored in the
body and available for subsequent metabolism and/or pulmonary elimination.
Ohtsuki et a]..-(1983) found that human urinary-metabolite production-did no
appear to be linearly related to exposure concentration. A graph of total
trichloro compounds (from urine) against PCE concentrations in air showed tr :
metabolite production appeared to be dose-dependent, leveling off a't
approximately 400 ppmv PCE -(8-h exposure), suggesting metabolic saturation;
however, no statistical test of departure from linearity was performed in this
study, and a questionable nonlinear model was assumed.
In contrast to the findings just reviewed, Bolanowska and Golacka (1972)
proposed that at steady state approximately 62% of respired PCE is metabolized
(based on their measurements of respiratory PCE retention discussed above).
This conclusion is contradicted by other studies of PCE uptake and metabolism
in humans (Ogata et aj... 1971; Ikeda et aj... 1972; Fernandez et aj.., 1976;
Ikeda, 1977; Monster et al.., 1979; Ohtsuki et aj.., 1983).
A long period of time is necessary for pulmonary elimination of
unmetabolized PCE. Stewart and co-workers (Stewart et aj.. 1970; Hake and
Stewart, 1977) have analyzed the pulmonary excretion of PCE following
experimental human exposures and found that it is biphasic. Initially,
elimination is rapid but the second phase is prolonged, with a half-life of
approximately 65 h. Monster et aj.. (1979) determined that human pulmonary
elimination of PCE has three different phases, with half-lives of 12 to 16 h,
30 to 40 h, and 55 to 50 h, respectively. Fernandez et aj.. (1976) note that
humans exposed to 100 ppmv for 8 h will require about 2 wk to eliminate PCE.
36
-------
4. TOXIC EFFECTS IN ANIMALS
Estimates of human-health risks resulting from exposure to a toxic
substance are frequently based on an assessment of animal dose-response data
because specific human data are often Inadequate for this purpose. In this
section, we review studies of the toxldty of PCE to animals, Including data
from bioassays conducted to evaluate the cardnogenlcity of PCE. Bloassay
results are also used as the basis of the quantitative assessment of.
carcinogenic potency In Section 7. The toxldty of PCE has also been reviewed
by IARC (1979), Re1chert (1983), HHO (1984), and the U.S. EPA (1980, 1982
1984b, 1985a, 1985b).
We begin our discussion of PCE toxldty with analyses of toxic effects to
major body organs and systems. For these effects/Appendix A presents a
review and summary of dose-response Information for different routes and
periods of exposure. The Information 1n Appendix A would be relevant to the
development of safety limits for PCE concentration In drinking water to
protect against acute, subchronlc, and certain (specifically, noncardnogenlc
and nonmutagenlc) chronic toxlcologlcal end points. In the absence of adequate
human-toxldty data. We then examine studies dealing with the teratogenidty
of PCE. The section concludes with a review of the mutagenlc potential of PCE
and Its metabolites and a summary of the results of animal cardnogenicity
bioassays.
TOXIC EFFECTS ON ORGANS AND SYSTEMS
Liver
Cornish et a].. (1973) administered 0.3 to 2.0 mL/kg (0.33 to 4.95 mg/kg)
of PCE IntraperUoneally (IP) to rats. Liver damage was measured by an
increase In serum glutamlc oxalacetlc transamlnase levels (SCOT), and was
observed at all doses (Cornish et a].., 1973). Ogata et ah (1968) observed a
decrease 1n the adenoslne tHphosphate (ATP) content of liver as well as an
Increase 1n the content of Uplds and trlglycerldes after mice were exposed to
800 ppmv for 3 he. Elevation of serum glutamlc pyruvate transamlnase (SGPT)
levels In mice was elicited by exposure to 3700 ppmv for 9 to 12 h, as well as
by IP administration of 3900 mg/kg (GehHng, 1968). Klaassen and P1aa.<1966>
37
-------
also measured increased levels of SGPT. in mice that received single
intraperitoneal doses of 2.9 ml/kg. These animals had enlarged hepatocytes
and slight liver necrosis. A single intraperitoneal dose of 1.23 ml/kg
elevated SGPT levels in dogs.
Cornish and Adefuin (1966) studied the effects of PCE administered in
combination with .ethanol. A single dose of ethanol was administered by
stomach tube (5 g/kg). Rats were then exposed to 4000, 5000, or 10,000 ppmv
PCE for 6, 4, or 2 hours, respectively. None of these treatments had a
statistically significant effect on levels of SCOT, SGPT, or SICO (serum
isocitric dehydrogenase).
Carpenter (1937) studied the subchronic and chronic inhalation toxicity
of PCE in rats. Although no effects were observed in animals treated with
70 ppmv (8 h/d, 5 d/wk for 7 months), rats that received 150 exposures of
230 ppmv had less glycogen storage than unexposed animals. Exposure to
470 ppmv (150 d) caused liver congestion and swelling. Rowe et a!- (1952)
exposed guinea pigs to 100 ppmv 7 h/d over a period of 17 to 185 d. No
effects were apparent in animals that received 13 exposures in 17 d. However,
when the number of exposures was increased to 132 over 185 days, females had a
significant increase in liver weight (p - 0.01) and animals of both sexes
exhibited Hpid accumulation in the liver.
Schumann et a].. (1980) dosed mice and rats orally with 100, 200, 500, or
1000 mg/kg of PCE daily for 11 d. In mice, all dose levels produced
hepatocellular swelling, a significant increase in absolute liver weight
(p<0.05), and a significant decrease in hepatic DNA content (p<0.05). All of
these changes are indicative of hypertrophy (the enlargement of an organ due
to an increase in size of its constituent cells). Mice that received 100 mg/kg
displayed an increase in hepatic DNA synthesis. In contrast to the pathologies
that developed in mice, only the highest dose, 1000 mg/kg, induced hepatic
toxicity in rats. These animals had a statistically significant increase in
relative liver weight (p<0.05). A change in the staining affinity of
hepatocytes (for hematoxylin and eosin) was also observed. The significance
of the latter observation is not known.
Buben and 0'Flaherty (1985) treated mice by gavage with dosages ranging
from 20 to 2000 mg/kg, 5 d/wk for 6 wk. Liver weights and liver triglycerides
were significantly greater than controls at doses of 100 mg/kg and above
(p<0.001). A Juje-dependent increase in liver degeneration and karyorrhexis
38
-------
was also observed at TOO mg/kg and greater. Activity of glucose-6-phosphatase
(G6P) was Inhibited, and there was a significant Increase 1n SGPT at 500,
1000, 1500, and 2000 mg/kg (p<0.001). Hepatic DNA content was measured in
mice treated with 200 pr 1000 mg/kg; animals that received 1000 mg/kg had
significantly lower levels of DNA (p<0.01).
In an NTP-sponsored study (NTP, 1986) of the effects of PCE on mice and
rats, animals were exposed to PCE by Inhalation 6 h/d, 5 d/wk for 103 wk. Mice
were exposed to 100 or 200 ppmv and rats to 200 or 400 ppmv. Male and female
mice of both exposure groups developed liver degeneration and necrosis.
Development of these pathologies appeared to be dose related. The incidence
of liver degeneration in male mice was as follows: controls, 2/49; low dose,
8/49; and high dose, 14/50. The observed incidence 1n female mice 1n the
control group was 1/49; in the low dose group, 2/50; and in the high dose
group, 13/50. The number of male mice exhibiting necrosis in the treatment
groups increased with increasing exposure concentrations (I.e., controls,
1/49; low dose, 6/49; and high dose, 15/50). For female mice the incidence of
necrosis was: controls, 3/48; low dose, 5/50; and high dose, 9/50. Treated
male mice (but not females) had a greater incidence of hepatic nuclear
inclusion than controls (i .e., for controls, 2/49; low dose, 5/49; and high
dose, 9/50). The statistical significance of these data was not evaluated.
Under the conditions of this study, rats did not develop hepatic lesions in
response to exposure to PCE.
Kidney • ,
«.
Klaassen and Plaa (1967) administered PCE to dogs intraperitoneally and
measured excretion of phenolsulforiephthalein (PSP), a substance used to test
for renal function. Control dogs excreted 561 of the PSP within 30 minutes;
excretion of less than 39X was .considered to be an Indicator of kidney
dysfunction. Kidney function was significantly affected after a single IP dose
of 1.4 ml/kg (statistical significance was not given). Plaa and Larson (1965)
reported that all mice given a single IP dose of 2.5 ml/kg exhibited swelling
of the proximal convoluted tubule, and one animal (of six treated) developed
necrosis of the proximal convoluted tubule. Mice that received a single dose
of 2.5 or 5.0 ml/kg excreted protein in their urine. The statistical
significance of these responses was not reported. Carpenter (1937) found that
39
-------
rats given 230 ppmv (8 h/d, 5 d/wk) for 21 d developed swelling and congestion
of the kidneys. This response was exacerbated when the concentration was
increased.to 470 ppmv. Subchronic exposure of mice aand guinea pigs to
400 ppmv, 7 h/d for 169 times in 236 d caused swelling of the tubular
epithelium along with an increase in kidney weight (Rowe et §1-, 1952).
The NCI (1977) cancer bioassay of PCE documented a high incidence of toxic
nephropathy in both species of rodents and in all dose groups (toxic
nephropathy was defined as degenerative changes in the proximal convoluted
tubule, fatty degeneration, and necrosis.'of the tubular epithelium). PCE was
administered by gavage. In this study, mice received time-weighted-average
(TWA) daily doses of 386 to 1972 mg/kg; toxic nephropathy was observed in 82
to 100% of the animals. Rats received TWA daily doses of 471 to 949 mg/kg; 58
to 941 of these animals developed toxic nephropathy (see'Table A-7 for
specific data).
A bioassay sponsored by the NTP (NTP, 1986) documented kidney casts,
nephrosis, and tubular cell karyomegaly in mice (animals received 100 or
200 ppmv of PCE 6 h/d, 5 d/wk for 103 weeks). Casts occurred more .frequently
in treated male mice than in controls (incidence in controls, 3/49; low dose,
9/49; and high dose, 15/50). The trend in female mice was not clearly dose
related (incidence in controls, 4/48; low dose, 4/49; and high dose, 15/50).
Nephrosis developed at a greater incidence in treated female mice (control,
5/48; low dose, 14/49; high dose, 25/50). For male mice, the incidence of
nephrosis in the controls was 22/49; low dose, 24/49; and high dose, 28/50.
Karyomegaly of tubular cells was treatment-related. The Incidence of this
pathology in male mice was control, 4/49; low dose, 17/49; high dose, 46/50.
In female mice, the incidence of nephrosis was 0/48, 16/49, and 38/50 in the
controls, low-dose, and high-dose groups, respectively.
The same study (NTP, 1986) reported a dose-related increase of renal
tubular cell karyomegaly in rats of both sexes. Low-dose animals received
200 ppmv of PCE; high-dose animals received 400 ppmv (the exposure regime was
the same as listed above for mice). The incidence of karyomegaly in male rats
for the corresponding ccntrol group was 1/49; low dose, 37/49; and high dose,
47/50. In female rats, the incidence'was 0/50, 8/49, and 20/50, respectively.
Male rats exhibited a dose-related increase in renal tubular cell hyperplasia
(controls, 0/49; low dose, 3/49; and high dose, 5/50). Only one high-dose
female rat had renal tubular cell hyperplasia.
40
-------
Pancreas
Hamada and Peterson (1977) studied the effects of PCE on the electrolyte
and protein concentration in bile duct-pancreatic fluid (BOPF). The actual
source of this fluid (bile duct and/or pancreas) is not known. Rats were
given a single IP dose of PCE (1.3 ml/kg in corn oil). Animals were then
fasted for 24 h, at which time BOPF was collected and analyzed.
PCE caused a significant increase in BOPF flow, a decrease in
concentration of protein in the BOPF, and an increase in the concentration of
chloride and potassium (p«3.05 for all parameters). The mechanism of
enhanced BDPF is not known. Although Hamada and Peterson (1977) discussed
possible mechanisms that may be analogous to secretion or cholinergic
stimulation, they concluded that PCE (and other chlorinated aliphatic
hydrocarbons) altered BDPF by an unknown mechanism, and that the toxicological
significance of the reported effects is not known.
Lungs. Skin, and Eyes
PCE is an eye and skin irritant. Application of PCE to the eye of
rabbits caused abrasion of the epithelium and conjunctivitis. PCE was also
extremely irritating when applied topically to the skin of rabbits (Duprat '
e£ a_L, 1976). However, Jakobsen et aj.. (1982) saw no visible sign of skin
irritation when guinea pigs were exposed to liquid PCE.
The NCI (1977) reported a high incidence of pneumonia in animals used in
the bioassay of PCE. Sixty-two to 791 of treated rats, and 29 to 661 of
treated mice developed pneumonia. However, 95 to 1001 of control rats, and 28
to 3SX of control mice also developed pneumonia. Because of the relatively
high incidence of pneumonia observed in control animals, PCE probably did not
contribute directly to tHe infection. In a separate study, chronic inhalation
of PCE caused a dose-related incidence of passive congestion of the lungs in
mice (NTP, 1986).
Reproductive System
The only indication that PCE has any effect on the reproductive system
comes from the work of Rowe et a!. (1952). Seven male guinea pigs were
41
-------
exposed to 1600 ppmv, 7 h/d for 8 exposures within 10 d. Microscopic
examination of tissues revealed slight degenerative changes in the germinal
epithelium of the testes. The Implication of this observation was not
discussed and subsequent studies have not confirmed-the finding. Therefore,
1t is difficult to evaluate the significance.of this report.
Cardiovascular System . ' -
PCE has been associated with sudden death from cardiac failure (Rowe
et aJL., 1952; Reinhardt et a].., 1973). It has been suggested that PCE (as
well as a number of other solvents) may sensitize the heart to the effects of
. endogenous 1y produced epinephrine (Price and Dripps, 1970; Reinhardt et a]..,
1973). If sensltization occurs, epinephrine-induced stimulation can lead to
tachycardia and cardiac failure. This response is apparently precipitated by
physical exertion and exposure to high concentrations.
Kobayashi et aj.. (1982) Investigated the action of intravenously
administered PCE on cardiac rhythm. Rabbits were anesthetized with urethane,
while cats and dogs were anesthetized with pentobarbital. A mean dose of
10 mg/kg of PCE administered with 0.7 yg/kg of epinephrine produced
tachycardia In rabbits (although the most sensitive animals were effected by
5 mg/kg of PCE). Tachycardia also occurred in dogs given a mean dose of
13 mg/kg PCE with 4.2 yg/kg of epinephrine. while doses of 30 to 40 mg/kg
decreased left intraventricular pressure. Cats exhibited ventricular
arrhythmias after 24 mg/kg of PCE was administered in conjunction with 13 to
14 jig/kg of epinephrine.
Rowe et a!. (1952) speculated that death in some rats exposed to
concentrations of 3000 ppmv PCE or more In air was caused by cardiac failure;
however, it is possible that cardiac failure occurred as a result of extreme
CNS depression. Reinhardt et aj.. (1973) studied the cardlotoxicity of PCE by
exposing unanesthetized dogs to 5000 or 10,000 ppmv. Arrhythmias and cardiac
failure were not observed, and there was no evidence of'sensltization.•
The dose levels used in the aforementioned studies are not representative
of typical human exposure levels. Furthermore, the work of Kobayaahi et a]..
(1982) utilized anesthetized animals, administered PCE Intravenously, and used
42
-------
e ,
relatively large amounts of epinephrine, none of which readily facilitates
extrapolation of results to humans. The cardiotoxic potential of PCE requires
additional experimental work before any conclusions can be drawn.
Central Nervous System
Acute exposure to PCE typically induces CNS depression. Initial
depression can progress to loss of consciousness, anesthesia, and respiratory
failure with prolonged or massive exposure. Single oral doses of 1623 mg/kg
produced reversible CNS effects in cats (Maplestone and Chopra, 1933) while a
single dose of 6492 mg/kg caused lethal CNS depression (Lamson et a_L, 1929).
Death from CNS depression resulted from single doses of 4700 mg/kg (rat) and
6492 mg/kg (dog) (Smyth et aj.., 1969; Lamson et a].., 1929). Over a 4-h period,
2300 ppmv caused an impairment of muscular coordination in female rats, which
contributed to a loss of 80 percent of avoidance and escape responses. Animals
apparently developed some tolerance to PCE, because this effect did not persist
when dosing was continued over a two-week period (Goldberg et aK, 1964).
Rats exposed to 6000 ppmv lost consciousness within a few minutes;
decreasing the concentration to 3000 ppmv required several hours to elicit the
same effect (Rowe et aj.., 1952). The NTP (1986) study found that exposure of
mice to 2917 ppmv for 4 h was lethal to all animals. Rats appear to be less
sensitive to PCE, because a 4-h exposure to 5163 ppmv was required to produce
100 percent mortality (NTP, 1986).
Carpenter (1937) studied the effects on rats of chronic exposure to 70,
230, 470, or 7000 ppmv of PCE. Although various pathological changes were
»
observed at 230 ppmv and above, no CNS effects were reported. Savolainen
et a_K (1977) observed a slight decrease in brain RNA content and an increase
in nonspecific cholinesterase in rats exposed to 200 ppmv (6 h/d for 4 d). A
one-month study conducted by Honma and co-workers (1980) documented a dose-
dependent decrease in dopamine content of the striatum in rats exposed to 200,
400, or 800 ppmv 12 h/d. The decrease was not statistically significant.
Norepinephrine content of the hypothalamus and serotonin levels of the cortex
and hippocampus increased after exposure to PCE.(all three concentrations).
None of the increases were statistically significant. A significant decrease
in acetylcholine (ACh) levels in the striatum was measured after exposure to
800 ppmv (p<0.05). Drowsiness and other symptoms indicative of CNS depression
43
-------
were reported by Rowe et a].. (1952) after rats were exposed to 1600 ppmv 7 h,
5 d/wk over a 25-d period. At 2500 ppmv, 13 exposures within 18 d caused the
death of most rats and guinea pigs from CNS depression. Rabbits that received
the same treatment displayed signs of CNS depression but did not lose
consciousness. ' ,
TERATOGENICITY
The teratog'enlc activity of PCE has been studied in rats (Nelson et aJL,
1980; Schwetz et a_K, 1974; Schwetz et a_L, 1975) and mice (Schwetz et a]..,
1975). Maternal exposure levels ranged from 100 ppmv to 1800 ppmv. Although
some minor effects were seen, in the progeny, PCE is not considered to be a
teratogen. A summary of these studies is provided in Table 4-1.
Investigations of the teratogenic and/or developmental effects of PCE have
most commonly shown evidence of maternal toxldty, rather than adverse effects
on the progeny. Toxicity was evident at 300 ppmv (Schwetz et a].. 1974; Schwetz
et ah, 1975) and 900 ppmv (Nelson et aj.. 1980), while maternal death occurred
at 1800 ppmv (Nelson et aj.., 1980). Dams exposed to 300 ppmv 7 h/d on days 6
to 15 of gestation had reduced body weights (rats) or an increase in liver
weight (mice). The pups of these mice had lower body weights, and there was a
slight increase 1n the number of runts. Some fetuses had subcutaneous edema
or delayed ossification of the skull and sternebrae, as well as splits in the
sternebrae. These pathologies probably reflect developmental delays, and as
such, are considered to be reversible. Developmental delays are believed to
result from maternal toxlcity rather than from any direct teratogenic activity
of PCE (Schwetz et §1., 1975). The mechanism of maternal toxicity is unknown,
but is thought to Involve CNS depression (weight loss) and cytotoxicity
(change in liver morphology). Because fetal health is often a reflection of
the health of the mother, maternal toxlcity is a significant concern. The
loss of maternal weight, most probably due to decreased feed consumption from
sufaclinical.effects (ataxla and anesthesia), can have a great impact on the
growth and maturation of the fetus. Maternal malnutrition can cause
developmental retardation of the fetus (Doull et aj.., 1980). The
hepatotoxicity observed 1n mice could also have a profound effect on the
44
-------
VI
"fl
g
'e
^
o
41
e
41
^S
ol
O
e
4=
U
JM
U
41
*J
i^
9
0)
9
2
fl
u •
41 '
1—
*_,
^.
41
ii—
"fl
41
fl V)
U O
e
41 1-
U 0
e
o
VI
41
U
41
m , m
o r*» f** o • o
en ••• •" ^ ° • ^
• => . • - • •
"fli "^ _ -^i -^l
^ji ^ji
*J( 44 41 ^4 *J(
4l| Qjl jjj
N N
c •* -u c e
O 41 41 9 9
£ X J JJ VI
41 U U 41 41
Z .W CO Z Z
-^ -» c ' -e t- " c
"fl e >* 15 vi . . — u • —
C O U1 •»•> U 4! C VI . 4> •>-> O 41 *J
41 -^93*^-^ fl Cvl flO)> flOlv)
«< U^>ja«l 41 41 94 4) .— T3 fl 41 .— -o m
fl 3 C •<- U VI JJ 41 U 41 41 J= U 41 41 *J
S -a T *J •— .— u fl u U2JS4) w2j=
41-^^jC C*>>4li-»W 4fVIXl 4IVI-W
u u ai4ii»>, —^w-a^-c -a-o— -a -o — . c
9 «J .— vl — O) U 41 3 •— C C 41 C C 4i
• • .C ,— fl •• •— 41 ^ J< fl •• fl •— E . . fl .— u
i— >. Ol VI 41 *. >i 41 "O fl VI • C >» SO >> Svl
fl -J — U O ^iZi— 4lS *i C •- VI *J C — fl
*> .^ 4i .. u .— . . 41 i*. fl -a — o -a •— o ~9
4i u j >, e a\ u u >, -a 9 u 41 u — e u— e
.1^ .«• » •_ _ .— 4) «> ^ ..- «J .. — •— *J •• O
x >> .— x > •— • c 4i vi x a. >< xo.>>
•a . O -Q U —J VI 9 — • U «l -9 e 3 9K«'4I O S ••" 41
41 -ui O •- C C **•«.• j: •— U O *J 3 •— U jJ 3 •— u
> J3XIQ9 X O» ** 41 41 VI U C VI U C
U >» ^- O u •— i— 41 9 .^ fl *J C — C •— fl — - C •— fl
4) *J fl C -^ — *J fl > «J 11 U VI fl fl O X C fl 9 X E
vi— c fl vi. a. e— X— •* c u 9 C c u o C
0 — 4) S fl C 9 4) * « X — — U 4)13 <*-4rf 41-0 *.
X J-> ^ CD VI *< ^ *J -o vl i— J3 «J4IO>^VI -J 41 Q. U
99 fl C 41 •— 41 fl 41 '41 O VI i 3 fl 41 3 41 41 fl 41 3 41
Z •«-> Z •— U. VI 1. X I- U. ^ 9 VI VI Z'^a.&w Z'^&O.
^
O ' O
CMC me me me CM
9 •— 9 ^9 —9
9 — — — — 9
«>«! O *J ' O *> 9 *J *J
fl ' JJ fl *J fl *J fl
^ VI VO VI \O VI 1^ VI •—
41 41 41 41
VI O) VIS) VIS) VIS) VI
X >, >. >. >.
fl Vk. fl •«- fl >*- fl V>- fl
o o Q o a 9 09 a -
-Q -^ ^ • « -a ~o
»x -V >» «s. ^' .
fl .c j: fl fl
(X [*% f^ f^ f^.
--- — -- - .__-—-— — "--
.1 11 1 1
a. a. a. a. a.
o o o o o
O O O O 0
^» r^ PO ' ^^ ^^
II 1 1
41 41 — • 41 4)
O)">i O) ">« ,VI4I Oi"^ Sl>>
fl 41 fl 4) VI i-> fl 41 fl 41
U — U -- 41— VI U>- 1- —
o.x a.x vix^a o,x Q. i
.tolt/Yfl
fl i^ O fl ^^ O ' • 9 ^^ 2 fl *"^ O fl *<^ O
ae ce X ae at
45
-------
4>
U
C
41
4)
oc
u
41
e
o
e
at u
u o
e
o
o
.<*
<« >» -e u
c c a. «s c c
>M ••• V> tj 1Q •••
s
ui
01 O
w ••
58
46
-------
growth of the fetus. Maternal toxicity is also suspected of causing a small
but significant (p<0.05) increase in the number of resorptions in treated rats
(300 ppmv, 7h/d) (Schwetz et aK, 1975).
Nelson et aJL (1980) performed a series of behavioral and biochemical
tests on the offspring of exposed rats. There were no adverse effects to
mothers or their pups following exposure of the mothers to 100 ppmv of PCE
(7 h/d) on days 14 to 20 of gestation. Exposure at 900 ppmv for 7 h/d during
days 7 to 13 of gestation produced significant differences in neuromuscular
coordination (p<0.02) and wire mesh ascent e(p<0.05). Exposure at 900 ppmv for
7 h/d during days 14 to 20 of gestation caused diminished performance in the
wire-mesh ascent test, but increased the performance in the neuromuscular
coordination test. A neurochemical analysis of whole brain (minus cerebellum)
was performed on newborn and 21-day-old pups. Twenty-one-day-old pups from
dams exposed during either period had a significant decrease in acetylcholine
(p<0.05). A significant decrease in dopamine (p<0.05) was measured in 21-day-
old pups from dams exposed during days 7 to 13 of gestation.
Elovaara et a].. (1979) injected 5 to 100 nmol of PCE into the air space
of chicken eggs and studied the gross effects on the embryo. Malformations
observed were exteriorization of viscera, as well as skeletal and eye
abnormalities. These deformities occurred in six embryos (out of 61 examined).
Tests conducted in rodents have not clearly demonstrated that PCE is a
teratogen. However, there is some evidence that inhalation exposure of
pregnant rodents to PCE can induce developmental delays and altered
performance in behavioral tests of the offspring.
f.
MUTAGENIC EFFECTS • •
Short-term assays have been conducted to evaluate the ability of PCE to
permanently alter genetic information. Most of the tests of genetic activity
have been microbial assays that measured forward or reverse mutations.
Chromosomal effects have been studied In cultured mammalian eel Is.
The ability of short-term assays' to.detect mutagens is compromised by lack
of knowledge of the mechanisms involved, different sensitivity and predictive
ability of each test, and by variations in protocols used by separate labs.
Despite these problems, short-term assays provide supportive evidence in the
evaluation of a compound's carcinogenic potential.
^ 47
-------
H1crob1al Assays
Greim et a_K (1975) evaluated tne mutagenic activity of PCE (purity
>99.91) in Escherichia coli K12 with and without metabolic activation (S-9).
The results were negative at four loci tested. Three of these loci are back
mutation systems (gal*, arg*. and nad*), while one measures a forward mutation
that produces resistance to 5-methyl-OL-tryptophan. Only one concentration was
used (0.9 mM), and detailed data were not reported. This study was the only
one cited by Fishbein (1976) in his review of the mutagenicity of halogenated
aliphatics.
Cerna and Kypenova (1977) reported in an abstract that PCE of unspecified
purity produced base-pair substitutions and frameshift mutations in Salmonella
typhimurium without S-9 (Ames test). Concentrations of 0.01, 0.1, and
1.0 mg/mL of PCE produced a dose-dependent increase in the number of
revertants. This response was significant only in TA100 (51 level of
significance), a strain sensitive to base-pair substitutions. In a host
mediated assay that used ICR mice and S. tvphimurium strains TA1950, 1951, and
1952, PCE reportedly caused a significant increase in the number of revertants
(level of significance was not given). The doses used were lifted as LD5Q
and l/2LDgo, but exact quantities were not specified. Because no information
was provided on the purity of PCE used, revertant counts, or controls, the
significance of these data cannot be evaluated.
The results of Bartsch et aj.. (1979) conflict with those of Cerna and
Kypenova (1977). PCE (99.71 pure) was studied in the Ames test with strain
TA100. Concentrations up to 4 x 10"3 M were not mutagenic in the presence
of an S-9 liver fraction from mice pretreated with phenobarbital. Toxicity
-4
was observed at concentrations greater than 5 x 10 M.
Kringstad et ii- (1981) evaluated the mutagenic activity of PCE in the
Ames test. PCE (99.01 pure) was tested at a single concentration
(0.1 mg/plate) 1n S. tvphlmurium. strain TA 1535. No source of exogenous
metabolic activation (S-9) was used. A slight increase in the number of
revertants was observed (31/piate after PCE compared to 19/plate in the ether
controls); however, the response was considered negative.
The NTP (1986) reported the results of a series of Ames tests on PCE
conducted by the Environmental Mutagen Test Development Program. Salmonella
tvphimurium (strains TA 98, 100, 1535, and 1537),was incubated with technical-
48 ,
-------
grade PCE in covered test tubes for 20 min. The test was conducted both with
and without S-9 (S-9 fractions were prepared from the livers of male Sprague-
Dawley rats and Syrian hamsters pretreated with Aroclor 1254). The greatest
number of revertants was observed in strain TA100 (all doses, both with and
without S-9); high doses (333 yg/plate) were toxic to TA1535 and 1537 in the
absence of S-9. However, PCE was judged not to be mutagenic in any of the
strains, regardless of the concentration tested.
Callen et a_K (1980) used Saccharomvces cerevisiae. strain D7, to study
the frequency of gene conversion (trp5 and ilvl loci) and mitotic recombination
(ade2 locus). PCE (purity not given) was added directly to a cell suspension
(log phase) and incubated in a closed vial for one hour. Samples were
centrifuged, resuspended in buffer, and plated on a medium with glucose."
Survival decreased as dose increased from 0 to 4.9, 6.6, or 8.2 mM. At 8.2 mM
survival was greatly reduced and assessment of* genetic activity was precluded.
Exposure to 6.6 mM elicited substantial increas.es in mltotic recombination
(52.6 recombinants per 10 survivors vs 3.3 recombinants per 104 survivors
in controls). The number of gene conversions (trpS) also increased at 6.6 mM
(8.3 convertants per 10 survivors vs 1.4 x 105 convertants per 105 survivors
in controls); the number of revertants at the ilvl locus was not measured at
this exposure concentration. Fabre (1978) has proposed that mitotic
recombination and gene conversion may.be induced by the same mechanism. If so,
the response may have been inaccurately evaluated, as ade2 recombinants were
estimated from plates that had been previously used to determine the number of
trp5 convertants. Consideration of this possibility as well as a lack, of
statistical analysis of results limit the strength of evidence presented by
Call en et §1. (1980). However, mitotic recombination and gene conversion are
indicative of interaction of a substance with DMA. Since the increase in
frequency of mitotic recombination was pronounced, this response may warrant
additional study. . .
Bronzetti et a].. (1983) also studied the effect of PCE on the trpS, ade2,
and ilvl loci in S. cerevisiae. Cells in the stationary phase were exposed for
two hours to 5, 10, 20, 60, or 85 mM PCE. All results were negative. The PCE
used was 99.51 pure, whereas the purity of that used by Callen |t a_L (1980)
was not reported. When comparing the results of these two studies,
consideration must be given to the possibility that cells in the stationary
phase may be re'sistant to the mutagenic or toxic action of xenobiotics.
49
-------
In addition, Bronzetti et a_[. (1983) conducted an intrasanguineous host-
mediated assay that tested the ability of PCE to induce genetic effects in S.
cerevisiae. Yeast were exposed to PCE and its metabolites in the liver, lungs,
and kidneys of mice. PCE did not induce point mutations, mitotic
recombination, or mitotic gene conversion.
Drosophila ,
The NTP (1986)" reported the results of an assay that tested the ability
of PCE to induce sex-linked recessive lethal mutations in Drosophila. Males
were exposed by injection or by feeding and were mated to a series of untreated
females. Neither route of administration produced a statistically significant
increase in sex-linked recessive letnal mutations.
Mouse Cells
*
PCE was not mutagenic in L5178Y/TK*'" mouse lymphoma cells, with or
without metabolic activation. Cells were treated for four hours with 6.25,
12.5, 25.0, 50.0, and 100.0 nL/mL of PCE. Following incubation for 48 h, celi,
were plated in medium supplemented with trifluorothymidine for selection of
cells mutant at the thyraidlne kinase (TK) locus. No statistically significant
increase in mutation frequency was observed at any dose level (NTP, 1986).
Somatic mutations are thought to occur when a substance or one of its
metabolites interacts with DNA. Alkylation of DNA is therefore one possible
indicator of genotoxlcity. Schumann et aj.. (1980) measured binding to hepatic
macromolecules of mice after administration of radiolabeled PCE. No
radioactivity was detected bound to hepatic DNA. The specific activity of the
tetrachloroC14C]ethylene used in this study was too low to permit detection ;
of low levels of DNA binding. Therefore, these results do not rule out the
possibility of DNA alkylation following exposure to PCE.
Genetoxic Activity of Metabolites
Tetrachloroethylene oxide (PCE oxide) is believed to be the first
intermediate formed by mlcrosomal oxidation of PCE. A concentration-dependent
mutagenic response was produced by PCE oxide (0.5, 1.3, 2.5,* 5.0, and 25.0 mM)
50
-------
in S. typhimurium TA1535 (without metabolic, activation) but PCE oxide was not
mutagenic to £. coli WP2 uvrA. The mutagenicity of this epoxide was also
evaluated,in a DNA-repair-deficient strain of E. coli (E. coli pol A 1").
In the latter assay, genotoxicity is measured by comparing .differential growth
inhibition of the DNA-polymerase-deficient strain, pol A V", with a polymerase-
proficient strain, pol A 1*. PCE oxide gave positive results in this test;
growth of the pol A 1" strain was inhibited at all dose levels used (Kline
et aj... 1982).
Trichloroacetic acid (TCA) is the principal metabolite of PCE excreted in
the urine of rodents and humans; trichloroethanol has also been identified as
a metabolite of PCE. In an Ames test conducted with metaboTic activation, TCA
CO.45 mg/plate) and trichloroethanol (7.5 mg/plate) were not mutagenic to
strains TA^98 and TA100. There is some indication, however, that
trichloroethanol can induce sister-chromatid exchange in cultured human
lymphocytes (Gue_ta_L, 1981).
CHROMOSOMAL DAMAGE
• Cerna and Kypenova (1977) did not observe any cytogenetic effects in mouse
bone-marrow cell's following single or repeated IP Injections (daily injections
for five days). Cells were recovered for analysis 6, 24, or 48 hours'after the
last injection of PCE. No details were provided regarding doses used or the
specific end points that were studied (the study was published only as an
abstract). The NTP (1986) published the results of assays for PCE induced
chromosomal aberrations and sister-chromatid exchanges in Chinese Hamster Ovary
(CHO) cells, both with and without S-9. The S-9 fractions were obtained from
livers of male rats pretreated with Aroclor 1254. Data were reported in tables
only, with no supportive discussion (such as the number of cells scored for
each dose level). However, at least three dose levels were used (with and
without S-9), as were both positive and negative controls. PCE had little, if
any, cytogenetic effect in.either assay.
CARCINOGENICITY IN ANIMALS
*
Two lifetime bioassays have been completed on PCE (NCI, 1977; NTP, 1986).
Additionally, there are three other studies that have addressed the question
51
-------
of PCE card nogeni city (Rampy et a_L, 1978; Thelss et a_L, 1977; Van Ouuren
it aK, 1979). In Table 4-2 we have summarized results of the bioassay,
studies that resulted in significant increases in malignant neoplasms among
the exposed animals.
The National Cancer Institute (NCI, 1977) conducted a study in which
86C3F1 mice and Osborne Mendel rats were administered PCE in corn oil by
gavage, 5 d/wk for 78 wk. Animals were then observed for 32 wk (rats) or 12 wk
(mice). Mice were 25 days old at initial treatment; rats were 35 days of age.
The time-weighted average daily doses ofjPCE were 536 and 1072 mg/kg for male
mice, 386 and 722 mg/kg for female mice, 471 and 941 mg/kg for male rats, and
474 and 949 mg/kg for female rats.
PCE causeQ a statistically significant increase (p<0.001) in the incidence
of hepatocellular carcinoma in mice of both sexes and both dosage sroups
(Table 4.2). The time to first tumor development was considerably shorter in
treated mice than in controls. Hepatocellular carcinomas were first detected
at weeks 91 and 90 in untreated and vehicle controls. However, in male mice,
hepatocellular carcinomas were detected after 27 weeks (low dose) and 40 weeks
(high dose). The first hepatocellular carcinomas were observed in female mice
at week 41 (low dose) an'd week 50 (high dose).
Exposed mice also exhibited a high incidence of toxic nephropathy, a
conditibn that was not observed in controls.- Median survival times of mice
were inversely related to dose. Control males had median survival times of
» •
more than 90 weeks; this decreased to 78 weeks in low-dose males and 43 weeks
in high-dose males. The median survival time of control females was also
greater than 90 weeks. Median survival times of low- and high-dose females
were 62 and 50 weeks, respectively (NCI, 1977).
Early mortality occurred in all groups of rats dosed with PCE. Half of
the high-dose males had died by week 44; half of the high-dose females died by-
week 66. The median survival time of control animals was 88 to 102 weeks,
depending on sex. The National Cancer Institute (NCI, 1977) determined that
there was a statistically significant association (p<0.001) between increased
dosage of PCE and increased mortality. The early mortality obser /ed in rats
and its statistical association with dose of PCE indicate that the doses given
to rats in this bioassay were inappropriately high*. Because the optimum dosagi
was not used, and because significant early mortality occurred, these results
preclude any conclusions regarding the carcinogenicity of PCE in rats.
52
-------
V)
CU
•5
^j
^**
c/l
ITS
(/I
vt
tj
o
3
1
'c
s
£
ItJ
•o
CU
u
CU
t3
U
e
•^
i
w
•M
0
1_
E
1
1
"
CU
3
.•ts
H-
cu
"<*»
1
CL
CU
u
e
•ft CU
CU •Q
e *u
O c
cu
u
§
h- eu
st
P""
0
+* CU
Concentra
or dos
X
CU
OO
CU ~-
•#• ^0
U i-
CU -4-1
CL >
>» .
a
oo \
J3 J3 . .Q J3
^•* P—1* «^ V"^
< o o vo CM
CO CO OO ^^
o 10 i— o en I**
^
^C Lw
0
VO
CO
u
s
o*
*
'X .
JS JD JS JS
< O O O (_)
%%% ^== ^^^^$=
•
ill ill llllll
CL CL CL CL
CLO O CLO O . CLO O CLO O
0 — CM O°-CM O 2 CM 0 — CM
^» ^_ .^
^" ^b UB
U.
g
CD
U
•^
OO
at . ,
^~
a."
z
' 53
-------
•a
3
*#"•
C
8
CM
^*
41
JS
IQ
41
3
fd
>
I
a.
4)
O
e
v» O>
««—
^3
vi •«-
e u
o e
VI
41
V-
u
cu
>»
o
4-> 4>
(0 VI
J- 0
c
41 U
U O
O
0
X
4)
oo
vi e
**** rtJ
u u
O. VI
~ "O
4-1
00
u u u u
i^ ^* f^ CO
SO CM LO
5 O O
< d d < o o
z z
II .11
0.0. 0.0.
000 OO O
LO to in in tr> 10
co r^^ ^* oo o ^ft
CM n n •^ ro CM
zzz zzz
^ — > >
• E s e
> o. o. > o. a.
S 0. 0. E 0. 0.
0. 0.
O» C3 ^*? CC ^3 ^3
^j ^^ ^^ ^3
O CM ^ O CM ^
Z *J-
s
^
fl
u.
V)
4-1
VO
03
CT»
a.
H-
Z
*
•a
c
c
3
u
5
o.
5
1
(J
VI
S —
^^ ^9
4J <
U
f 41
U. _J
J" _•
49 49
41 41
^™ ^*»
4^ £
f~ ^"*
H^ *^
j3 J3
^9 ^9
"o "o
u u
a. a.
.a u
•
c
*
u
' «^
<*-
gj
en
•*"
>•
• "*
41 U
!a 4J
4 (/)
pM» H^
2 2
<3 V)
^J <^B!
O ' O
e e
i N
< 00
z z
.54
-------
Questions have been raised about the purity of PCE used in the NCI mouse
and rat bioassays. The PCE was produced by Aldrich Chemical Co. and had a
purity of 99%. However, epichlorohydrin (ECH) was apparently used as a
stabilizer. It has been suggested that the presence of this contaminant may
have directly contributed to tumor induction. ECH is a direct-acting
alkylating agent and-is mutagenic (Kucerova et a_L, 1977; Bridges, 1978). Van
Duuren et aT. (1974) demonstrated that ECH was carcinogenic in mice when
injected subcutaneously. A subsequent study by Laskin et aj_. (1980) showed
that ECH induced neoplastic lesions of the nasal cavity of rats. Most of these
tumors were carcinomas of "the squamous epithelium. Interestingly, 30-d
exposures to 100 ppmv produced a much greater incidence of cancer than lifetime
exposures of 30 ppmv (exposures were 6 h/d, 5 d/wk). A study by Konishi et
a_l. (1980) and Kawabata (1981) also showed that ECH fed discontinuously to
rats in drinking water at a concentration of 1500 ppm (and at a lifetime TWA
dose of approximately 40.2 mg/kg-d) induced a significantly increased
incidence of papillomas and squamous cell carcinomas of the forestomach above
that of control animals.
The exact quantity of ECH present in the PCE used in the'NCI study is not
known, but it has been estimated that high-dose male mice received 0.42 mg/kg-d
(U.S. EPA, 1985a). This represents a small fraction of the dose that elicited
squamous cell carcinomas in rats. Furthermore, ECH appears to initiate tumors
by a localized tumorigemic reaction at sites where it is in direct contact with
tissue, such as nasal or forestomach squamous-cell epithelium (U.S. EPA,
1984c). No animal in the NCI bioassay developed tumors at these sites. ECH
is,among the weakest of the more than 50 suspect carcinogens evaluated by the
U.S. EPA Carcinogen Assessment Group, having an estimated upper-bound
carcinogenic potency, or effect per unit dose at low doses; to humans of
9.9 x 10~ (mg/kg-d)~ , based on data indicating increased nasal cavity tumor
incidence in rats exposed to ECH via drinking water "(U.S. EPA, 1984c). Using
the methodology of U.S. EPA (1984c), the equivalent potency to mice would be
9.9 x 10~3 x (fm/fh), where fm and fh are the fractions of body weight
consumed as water per day, equal to 0.17 and 0.029 for mice and humans
respectively. The potency for ECH to mice is therefore estimated to be
0.058 (mg/kg-d)" . Using this potency estimate, the highest dosed animals
(high-dose male mice) in the NCI (1977) bioassays*would be expected to incur
55
-------
an increased cancer risk of (0.42) x (0.058) - 0.024, or less than 2.5%.
Therefore, it is unlikely that ECH contributed significantly to the observed
increased tumor incidence in PCE-exposed mice in the NCI (1977) bioassay.
Rampy et al_. (1978) exposed male and female Sprague-Dawley rats to PCE by
inhalation (300 or 600 ppmv) 6 h/d, 5 d/wk for 12 months. Animals were
subsequently observed for 18 months. High-dose males had slightly greater
mortality than did controls, but neither sex exhibited an increased incidence
of tumors, regardless of dose. Interpretation of this study is limited by the
duration of the exposure and by the fact that it was reported only as an
abstract.
Theiss et a].. .(1977) studied the ability of PCE to induce lung adenomas
in A/St male mice. Animals six to eight weeks old were given 80, 200, or
400 mg/kg of PCE in tricapryliri (intraperitoneally) three times a week. Each
group received 14, 24, or 48 injections. Animals were sacrificed 24 weeks
after the first injection and were examined historically for the presence of
pulmonary tumors. Treated animals did not exhibit a significant increase in
the average number of lung tumors,when compared to controls. The relevance and
validity of these test results are of questionable significance, though, as
this test has not produced positive results with several known animal
carcinogens.
The ability of PCE to initiate and/or promote skin tumors in ICR/Ha Swiss
mice was investigated by Van Duuren et a].. (1979). In one group, 163 mg of PCE
was applied once tq surface skin. Fourteen days after this, phorbol myristate
acetate, a promoter, was applied to the same area three times a week for 428
to 576 d. A second group,received 54 mg of PCE by topical application three
times a week for.440 to 594 d. PCE did not show any initiating activity. It
also gave negative results in the portion "of the experiment that tested its
ability to act as a complete carcinogen. It is difficult to interpret these
data in relation to the carcinogenic action of PCE .because the significance
and sensitivity of skin application tests are not thoroughly understood.
The most definitive study of the carcinogenic potential of PCE was
conducted by Battelle Pacific Northwest Laboratories for the National
Toxicology Program (NTP, 1986). B6C3F1 mice and F344/N rats were exposed to
99.9% pure PCE by inhalatfon, 6' h/d, 5 d/wk for 103 wk. Mice were exposed to
concentrations of 0, 100, or 200 ppmv; rats were exposed to concentrations of
0, 200, or 400 ppmv. .
56 .
-------
Treated male rats had lower survival rates than control animals (controls,
23/50; 200 ppmf, 20/50; 400 ppmv, 12/50). Survival rates among female rats
showed little variation (controls, 23/50; 200 ppmv, 21/50; 400 ppmv, 24/50).
Both exposure concentrations produced significant increases in mononuclear cell
leukemia in female rats (incidence in controls, 18/50; in rats receiving
200 ppmv, 30/50; and in rats receiving 400 ppmv, 29/50). Life Table analysis
showed the significance of these increases to be p = 0.023 (200 ppmv) and
p = 0.053 (400 ppmv). Treated male rats also developed mononuclear eel 1
leukemia in greater numbers than controls (controls, 28/50; 200 ppmv, 37/50;
400 ppmv, 37/50). Levels of significance (evaluated by Life Table analysis)
are p = 0.046 (200 ppmv) and p » 0.004 (400 ppmv) (Table 4-2).
Renal tubular-cell adenomas are rare neoplasms with a historical
occurrence at Battelle Laboratories of less than one percent (Appendix F, NTP,
1986). Renal tubular-cell adenocarcinomas are even less common, and have not
been documented in any NTP studies (NTP, 1986). Male rats (at the 200 and
400 ppmv exposure levels) exhibited an increased incidence of both of these
neoplasms (see Table 4-3). Although the increases were not statistically
significant, they appeared to be dose-related. Tubular-cell hyperplasia was
observed in eight treated males but only in one treated female rat. Tubular-
cell karyomegaly developed in a majority of male rats but was less common in
females.
Brain glioma is a rare tumor of neuroglial cells (the cells that compose
the supporting structure of nervous tissue). Brain gliomas were observed in
one male control rat and in four male rats that were exposed to 400 ppmv PCE
(NTP, 1986). This increase was not statistically significant. However,
because the historical incidence of these tumors is quite low (0.2% at Battelle
Laboratories), the increased incidence in treated animals is noteworthy.
In the NTP study (NTP, 1986), the survival of low-dose male mice (after
week 74) and of high-dose male mice (after week 75) was significantly lower
than controls
-------
Table 4-3. Incidence of renal tubular cell adenomas and adenocarcinomas in
male rats exposed to PCE by inhalation.4
Neoplasm
Tubular cell adenoma
Tubular cell adenocarcinoma
Control
1/49
0/49
Treatment
200 ppmv
3/49
0/49
400 ppmv
2/50
2/50
Tubular cell adenoma or
adenocarcinoma
1/49
3/49
(p = 0.259b)
4/50
(p . 0.070b)
a Data are from NTP, 1986.
b P-values are based on Life Table Tests (Appendix E, NTP, 1986).
• * , ' • '
Hepatocellular adenomas occurred in both sexes of mice, and at both
concentrations of PCE (Table 4-2). The incidence of adenomas was not
statistically significant. However, the combined incidence of hepatocellular
adenomas and hepatocellular carcinomas was significant. In males, the combined
incidence was: controls, 16/49; low dose, 31/49; (p « 0.002); and high dose,
40/50 (p<0.001). In females, the incidence of hepatocellular adenomas and
carcinomas was: controls, 4/48; low dose, 17/50 (p = 0.001); and high dose,
38/50 (p<0.001).
Summary of Evidence of Carcinoqenicitv in Animals
The NCI (1977) bioassay of PCE found that administration of PCE by gavage
was associated with a statistically significant increased incidence (p<0.001)
of hepatocellular carcinoma. This increase was documented in low- and high-
dose B6C3F1 mice of both sexes. A decrease in the time to first tumor
development was also observed in treated mice of both sexes and both dose
groups. Early mortality in rats prevented an analysis of PCE's carcinogenic
potential in this species. The NCI (1977) concluded that under the conditions
of this study, PCE was a liver carcinogen to B6C3F1 mice of both sexel.
In 1979, IARC reviewed the NCI study on PCE as well as the animal
carcinogenicity studies of Rampy et a].. (1978) and -Theiss et a].. (1977). Only
two short-term assays were evaluated (Cerna and Kypenova, 1977; Greim et aj...
1975). IARC (1979) determined that there was "limited evidence" that PCE is
carcinogenic in mice.
58
-------
IARC re-evaluated the evidence of carcinogenocity of PCE to animals in a
1982 publication (IARC, 1982). Although this review was extended to include
an epidemiologic study as well as recent information on PCE's activity in
short-term tests, no additional animal data were available. As in 1979, IARC
concluded that there was only "limited evidence" that PCE is carcinogenic to
animals (IARC, 1982).
The final report of the NTP inhalation bioassay on PCE was released in
1986 (NTP, 1986). The NTP determined that under the conditions of this study,
there was "clear evidence of carcinogenicty" of PCE for male F344/N rats, "some
evidence of carcinogenicity" of PCE for female F344/N rats, and "clear evidence
of carcinogenicity" of PCE for both sexes of B6C3F1 mice. In rats, these
conclusions were based on an increased incidence of mononuclear cell leukemia
in males and females. Male rats also developed renal tubular cell neoplasms
(a rare type of tumor). The evaluation of carcinogenicity in mice was based
on an increased incidence of hepatocellular adenoma and hepatocellular
carcinoma in males, and an increased .incidence of hepatocellular carcinoma in
females. '
59
-------
5. TOXIC EFFECTS IN HUMANS
To assess correctly the health risks from a chemical, consideration of
human toxidty data is essential. Unfortunately, information on human toxicity
for many substances is limited or is anecdotal in nature. For PCE, however,
there have been some controlled inhalation exposures.for the purpose of
defining occupational limits. In addition, epidemiological studies have be.en
conducted to explore the relationship between exposures to PCE vapors and .
potential health effects. We begin this section with a brief overview of the
health effects of PCE exposures. That discussion is followed by a review of
different epidemiological studies dealing with PCE. Finally, we examine human
data on the toxic effects of PCE on specific organs and systems.
GENERAL TOXICITY ,
Acute exposure to PCE can produce skin irritation and burns, as well as
irritation of the eyes and respiratory tract. Central nervous system (CNS)
depression is the most immediate effect of exposure, but high concentrations
can also cause loss of consciousness and respiratory failure. Liver and kidney
toxicity can result from single exposures (Stewart et aj... 1961a; Stewart,
1969; Hake and Stewart, 1977), but the concentration and duration are typically
greater than those that cause transient CNS effects. Chronic occupational
exposure to PCE has caused headache, dizziness, hangover, intoxication,
diminished cognitive abilities, and a decreased performance in the Romberg and
Flanagan Coordination Tests (Stewart "et ah, 1961b; Stewart et aJL, 1970;
Stewart et a].., 1974; Stewart et a].., 1976). Extended exposure to PCE (2-1/2
months to several years) has also caused changes in kidney and liver function,
cirrhosis, and toxic hepatitis (Coler and Rossmiller, 1953; Meckler and
Phelps, 1966; Hake and Stewart, 1977).
In experimental studies human volunteers have been exposed to PCE by
inhalation at various concentrations and for various durations. Because
subjects were allowed to leave exposure chambers .when they felt discomfort,
observed adverse effects have been restricted to the respiratory tract and CNS
(Carpenter, 1937; Rowe et al_., 1952; Stewart et aj.., 1961b; Stewart et a_L,
1970; Stewart et al_., 1974; Hake and Stewart, 1977). Accidental exposure to
PCE has occurred primarily as a result of its use as an industrial solvent.
60
-------
Although specific exposure levels have not always been determined,
concentrations have been high enough to cause liver and kidney dysfunction
(Coler and Rossmiller, 1953; Hake and Stewart, 1977; Koppel e_t aj.., 1985).
EPIDEMIOLOGIC EVIDENCE FOR CARCINOGENICITY IN HUMANS
Epidemiologic studies of PCE exposure have been reviewed by Reichert
(1983) and by the U.S. EPA (EPA, 1985a). Blair et aj. (1979) analyzed the
death certificates of 330 union laundry and dry-cleaning workers (out of a
cohort of 10,000). Of 330 decedents, 279 had worked solely in dry-cleaning
establishments (while union members). The solvent(s) used by the dry cleaners
were not identified. Length of union membership ranged from one to 25 years,
with a mean of 13 years. The number of expected deaths from cancer was 67.9
(based on proportionate mortality of the U.S. population) while 87 deaths from
cancer were observed. The authors also compared the number of years of union
membership with cause of death. With the exception of nonwhite males, length
of union membership was nearly identical for cancer and noncancer deaths.
Increased mortality (p<0.05) from cancers of the respiratory tract, cervix,
and skin, was documented. When all malignancies were evaluated together, the
number of observed deaths was also significantly greater than expected
(p<0.05). The authors noted that the excess of cervical cancer may be related
to the typically low wages and socioeconomic class of this occupational
group. Although an excess of liver cancer and leukemia was also observed,
these increases were not statistically significant.
The increases in cancer deaths among the study group probably contributed
to a lower than expected relative frequency of deaths from other causes. It
is noteworthy that death from circulatory disease was significantly lower than
expected (p<0.005). The factors contributing to this phenomenon are not known.
Causes of death were not determined separately for laundry and dry-cleaning
workers in this study. The actual solvent(s) used were not identified, and
smoking history was not documented. The lack of control for smoking was a
significant deficiency of the study because lung cancer was a major
contributor to the total number of cancers. Although this study identifies a
potential occupational hazard,, data provided are not adequate to evaluate the
carcinogenic potential of PCE.
61
-------
Katz and Jowett (1981) analyzed the mortality patterns of 671 white female
laundry and dry-cleaning workers. Data were obtained from the death
certificates of. individuals who died in the period 1966 to 1977. Occupational
code's listed on the certificates did not distinguish between the two types of
work. Data on the duration of employment were not available, nor were the
investigators able to determine to which solvent(s) the individuals were
exposed. Smoking history was not known. Cause-specific proportionate
mortality"ratios were calculated for 25 causes of death. A significant
increase in risk of death from cancer of the kidneys (p<0.05) and genitals
(p<0.01) was documented. An excess risk from skin and bladder cancer was also
found; however, neither increase was statistically significant. Individuals
in the group under study had a greater risk of death from cancer of the cervix,
ischemic (obstructive) heart disease, and diabetes mellitus. However, when,
the effect(s) of low-wage occupations were accounted for, only the risk for
diabetes mellitus remained statistically, significant (p<0.05).
Other studies of laundry and dry-cleaning workers have also reported an
increased risk of death from cervical cancer (Blair et a_L, 1979; Kaplan,
1980); however, these investigators have not compared mortality data by
low-wage occupation. Although not definitive, the findings of Katz and Jowett
(1981) suggest that factor(s) other than (or in addition to) solvent exposure
are important contributors to cervical cancer.
Kaplan (1980) completed a retrospective mortality study of 1597 dry-
cleaning workers exposed to PCE for at least one year (prior to 1960). By the
end of the study period, 1028 of the cohqrt were alive, 285 had died, and the
status of the remaining 254 was not known. Although a considerable effort was
made to determine the history of solvent exposure, the solvent history of
approximately half of the dry-cleaning establishments was unknown. Of those
shops with known solvent history, none had used trichloroethylene; individuals
who had worked in shops that had used carbon tetrachloride were eliminated from
the study. However, prior to 1960 (the period of interest in this study), the
majority of dry cleaners used petroleum solvents (NIOSH, 1980). In keeping
with this information, Kaplan decided that employment in a shop with unknown
solvent history probably involved exposure to petroleum solvents. Similarities
in the physical properties and physiological effects of gasoline (which has
been associated with kidney cancer in rats) (Kitchen, 1983), and petroleum
solvents suggest that use of these solvents may contribute to an increased
62
-------
risk of cancer. The inability of Kaplan (1980) to quantify solvent exposure
adds further uncertainty to the interpretation of this study. The mean
exposure concentration of individuals to PCE was calculated to be 22 ppmv for
^
dry-cleaning machine operators and 3.3 ppmv for all other jobs. These values
are based on a NIOSH survey (NIOSH, 1980) cited by Kaplan (1980).
A Standardized Mortality Ratio (SMR)* was used to compare the number of
observed deaths to the number of expected deaths, by cause. Kaplan (1980)
found an elevated SMR (182) for malignant neoplasms of the colon (11 observed
deaths, 6.77 to 6.98 expected deaths). In discussing this observation, Kaplan
(1980) points out that those individuals of high socioeconomic status are at
greater risk for cancer of the colon than individuals of low socioeconomic
status. Because dry-cleaning workers generally receive low wages, the study
cohort may have overrepresented individuals of low socioeconomic status, and
therefore, included a disproportionate number of individuals who are at low
risk for colon cancer. If this risk trend is valid, the elevated SMR reported
for colon cancer may actually be an underestimate.
In addition to colon cancer, SMR's for malignant neoplasms of the rectum
(158), pancreas (152), respiratory system (140), urinary organs (198), an9
"other and unspecified sites (major)" (156) were observed. Although Kaplan
did not evaluate SMR's for statistical significance, a review of this study by
the EPA (U.S. EPA, 1985a) included an evaluation of significance of these
malignant neoplasms. The SMR's for cancer of the rectum, pancreas,
respiratory system, urinary organs, and "other and unspecified sites (major)"
were not significant at the p<0.05 level. However, cancers of the respiratory
system, urinary organs, and "other and unspecified sites" were of borderline
significance (0.10
63
-------
Although the relatively small cohort 1n this study limits conclusions
about the carcinogenic potential of PCE, the study results suggest a
relationship between colon cancer and solvent exposure. Because the NTP study
of PCE (NTP, 1986) found hyperplastlc and neoplastlc changes In the kidneys of
treated rats, the Increase In the SMR from malignant neoplasms of the urinary
organs raises the possibility that occupational exposure to solvents may
Increase the risk of cancer In these organs. Since petroleum products have
also been linked to kidney cancer in rats (Kitchen, 1983), and because it is
probable that some of the cohort were exposed to petroleum solvents (as well
as to PCE), the possible contribution of PCE to an increased risk of urinary
organ cancer cannot be ascertained. An additional problem in this study was
the inability of investigators to collect data on (and thus control for)
smoking history. Since smoking is associated with an elevated risk of many
types of cancers (including lung and kidney), i.ts contribution to the elevated
SMR's reported by Kaplan needs to be evaluated (U.S. DHEW, 1979).
Duh and Asal (1984) studied the cause(s) of mortality among 440 laundry
and dry-cleaning workers from Oklahoma who died during 1975 to 1981. This
study had the same problem as the studies of Blair et aj[. (1979) and Katz and
Jowett (1981)—smoking histories were not available and separation of the two
groups by occupation was not possible. Therefore, duration or characterization
of individual exposure was not reported. However, Duh and Asal noted that the
two groups of workers probably experienced substantially different solvent
exposure. (NIOSH (1980) reported that, although 75% of dry-cleaning
establishments in the U.S. use PCE, Oklahoma may be unique in that petroleum
solvents account for more than 50% of total solvents used.) A Standardized
Mortality Odds Ratio (SMOR)* revealed elevated SMOR's for all digestive
diseases (1.5), cirrhosis of the liver (1.3), and homicide (3.8). SMOR's less
than 1.0 were reported for diabetes mellitus (0.7), ischemic (obstructive)
heart disease (0.8), emphysema (0.8), and suicide (0.2). A SMOR less than 1.0
suggests that laundry and dry-cleaning workers may be at low risk for these
diseases.
*SMOR was defined by Duh and Asal (1984) as a method that compares the
number of deaths by specific cause to the number of deaths due to other causes
in the exposed population (the odds) to the expected odds derived from a
, comparison population.
64
-------
Analysis of deaths due to cancer showed an increase in the SMOR for
cancers of the respiratory system (1.8), lung (1.7), and kidney (3.8). Deaths
from breast cancer were considerably less than expected (SMOR = 0.1).
Brown and Kaplan (1987) conducted a retrospective, cohort-mortality study
of workers employed in the dry-cleaning industry to evaluate the carcinogenic
potential from occupational exposure to PCE. The study cohort consisted of
1,690 members of four labor unions (located in Oakland, Detroit, Chicago, and
New York City). Individuals selected for the study had been employed for at
least one year prior to 1960 in dry-cleaning shops using PCE as the primary
solvent. Complete solvent-use histories were not known for about half of the
shops included in the study. Because petroleum solvents were widely used by
dry cleaners prior to 1960, most of the cohort had known or potential exposure
to solvents other than PCE (primarily, various types of Stoddard solvents).
The investigators also identified a subcohort of 615 workers who had been
employed only in establishments where PCE was the primary solvent.
PCE exposure in shops included in the study was evaluated independently
(Ludwig et al- .1983). The geometric mean of time-weighted-average exposures
was 22 ppmv for machine operators, and approximately 3 ppmv for other workers.
Brown and Kaplan (1987) calculated person-years-at-risk (PYAR) for each
worker. The PYAR were then .combined into five-year calendar periods and
five-year age groups by the life-table-analysis-system (Waxweiler et al..,
1983). PYAR values were also evaluated by length of employment and by time
lapsed since first employment in a shop that used PCE. The expected number of
deaths was calculated by multiplying PYAR (by age and calendar period*) by the
U.S. mortality rates. Risk of death due to a specific cause was calculated by
means of a SMR.
Among the,(main) cohort, the number of observed deaths from all causes
(considered together) was less than expected (493 observed, 575.5 expected;
SMR = 86). No deaths occurred from liver cancer, although 3.5 were expected.
There were also fewer deaths due to diseases of the'circulatory system and
nervous system (SMR -i 70 and 73, respectively).
However, observed deaths from all types of neoplasms were higher than
expected (142 observed, 122.9 expected; SMR = 116). Elevated SMR's from
malignant neoplasms of the intestine (136) and pancreas (172) were documented.
Malignant neoplasms of the urinary tract caused a significant excess of deaths
(12 observed, 4.7 expected; SMR = 255). Of these urinary tract cancers,
65
-------
kidney cancer caused four deaths (2.0 expected; SMR - 200), while eight deaths
from bladder cancer were observed (2.7 expected; SMR = 296). Mortality from
calculi of the urinary system (a nonmalignant disease) was also greater than
expected (2.0 observed deaths, 0.3 expected; SMR = 667). Although Brown and
Kaplan note that there may be an association between calculi and malignant
disease of the urinary tract, the association is speculative. An elevated SMR
for cancer of the cervix (196) and a decreased SMR due to cancer of the breast
(87) were attributed to factors associated with the low socioeconomic status
of the cohort (Hoover et al_., V975). The subcohort (workers employed only in
shops where PCE was the primary solvent) had only one death from urinary tract
cancer (1.3 deaths expected). All deaths from urinary calculi (2) occurred in
this group. •
In summary, a statistically significant excess of deaths from urinary
tract cancer was observed in those workers potentially exposed to both PCE and
petroleum solvents. Individuals employed in shops where PCE was the primary
solvent did not have an increased risk of mortality from kidney or bladder
cancer. Although these findings do not rule out PCE as the causative agent of
urinary tract cancer, the data suggest that other factors or agents.may have
contributed to the development of neoplastic disease.
The possible relationship between exposure to petroleum solvents and
kidney cancer has already been noted (kitchen, 1983). An excess risk of
bladder cancer has been associated with cigarette smoking (Matanowski and
Elliot, 1981). Brown and Kaplan (1987) were not able to document smoking
history; however, they calculated the possible effects of smoking on the risk
for bladder cancer (based on Axelson, 1978). They concluded that the
three-fold increase in bladder cancer among the cohort could not be attributed
to smoking.
The epidemiologic studies that are currently available add limited
information to our understanding of the health hazards associated with
occupational exposure to PCE. Although there is some indication that use of
dry-cleaning solvents poses a health risk, the contribution of individual
solvents to the overall problem is far from clear. Until studies are
completed that include a thorough analysis and quantification of PCE exposures,
. the usefulness of epidemiological studies in the assessment of the human
nealth risks of PCE will be limited. :
66
-------
TOXICITY TO MAJOR ORGANS AND.SYSTEMS
Information on human toxicity following exposure to PCE has been obtained
from case reports of accidental exposures, as well as from a limited number of
experimental studies. Human-health effects from short- or long-term exposures
are similar to those observed in animals. PCE initially affects the.CNS, and
larger doses cause various degrees of hepatic and renal damage. Table 5-1
summarizes human-health effects resulting from experimental inhalation
exposures to PCE.
Liver
The earliest reports of PCE-induced liver damage are associated with its
use as an anti-helminth in the 1920's and 1930's. Hundreds of thousands of
individuals were treated with PCE for hookworm infestations. PCE was typically
given as a single oral dose of 0.12 ml/kg (maximum of 5 ml) (Reichert, 1983).
PCE reportedly produced hepatic necrosis on some occasions. Damage was
transient, and recovery took place within 1 to 2 wk (Hall and Schillinger,
1925; Lambert, 1933). Meckler and Phelps (1966) documented a case in which an
individual developed hepatitis after a massive inhalation exposure to PCE (of
undetermined concentration). The individual's liver remained enlarged six
months after the exposure. Interestingly, in a separate case of oral
overexposure reported by Koppel et a_L (1985), .an individual did not have any
measurable liver or kidney damage after ingestion of 8 to 10 ml of relatively
pure PCE. Clinical measurements of organ function that were within normal
limits included SCOT, SGPT, alkaline phosphatase, red and white Wood-cell
counts, and serum creatinine. However, the individual was hospitalized for
treatment within 1 h of ingestion, which probably averted organ damage (Koppel
et a].., 1985).
Stewart and co-workers have investigated several cases of acute
overexposure to PCE that caused liver damage. The earliest report documented
the clinical effects of occupational exposure to the vapor of a petroleum-based
solvent mixture that contained approximately 50% PCE. The individual was
exposed to this mixture for about 3.5 h, which caused a loss of consciousness.
Simulation of exposure conditions gave,an estimated concentration of 250 ppmv,
with levels that reached 1000 ppmv for the last 30 min. Nine days after the
67
-------
x>
LU
Du
O
CU
5-
3
to
O
Q.
cu
c
o
4-»
rd
x:
c
•r-
Id
c
p-
•r*
t-
CL
X
CU
et
s
u
co
U
eu
<*-
cu
x:
4-*
rd
cu
i
c
3
«••
•Jv.
*
1
LO
cu
X)
rd
eu
u
c
cu
cu
eu
cu
4-"
u
cu
<*-
<*-
LU
cu
3 O
CO *f~
O i-
cx cu
X CL
LU
CU
cu E
CO «r-
acn
CD
V-
,
O
4-1
U *>
1 « c3
C CL
CU CU
O *-^
s
•* •—
en en
* *
• •
'rdl "id
•Ml -M
CU1 CU
•M | •
V- U
^3 ^3
3 3
cu cu
4-> 4->
CO CO
C
0
•r—
rd rd
LU T—
LU S-
«*%
o -t ' •—
J
^^" r
1*** • C
3 *"*
—• r»~
O CO
^y\ ^^ o~t o^ ^^ crt
r— - *•— r—
* : » ' **
* • •
15| ^| Id
«LJ| ^^1 luf
cu| cu| cu
1 . 4J 4J
V- J- S-
CM •— r
LO
en
»— •
i— »•
id
r— 1 4->
rd| CU
-Ml 4-> -i
eu| u
— CM •—
UO
» cr* *
• r^ . •
rd Idl
-* r« 4»i|
a> id eu|
P -Ml ' 4->
u cu| s-
3 3 X O) 32CU2
cu eu cu
1 « • 4^ 4~*
CO CO CO
en
T3 J-
c cu
, id xi eu c
E > O
co » O id «i—
co CU OS 3 4->
co co • - i , id
c o i= -a rd 4->
•r- C O CU -Mrd «r-
Q. .r- .1- r— (J V_
CO -4-" <4- CU LU S-
> • -+-> s- t_ rd
en cu eu e M- •— o ' eu >> •—
C x: CU rd U co -M rd
rd [ *
«— f—
U •»— -t-> E co rd «i— CL
rd co id c_ CU >
o en -o e o O -M i-.r- 4->
O i — eu U x: x) cu cu o
Z CO
^
'i
x: **
0
LO -M
cu
j
—
c/;
O
s- a
o x
C|_ CU
OC4-»-l->rd4-» i-Hid CO
.^
•o
x:
o
x: LO -M
^^ ^^ ^^
*
eu cu
v. v_
3 3
CO CO
3 eu cu 3 cu
O -M 4
M O -M
Oi CO CO OS . CO
• M
c <=
O rd
.r- en
. 4-> < — rd
rd id C
4-> -M rd -
•r- C •—
S- O LL.
c- S-
•i— <4- C
-M en
r^ W
3 CU
c ox>
O ••- E
co T- **_ o
3 -M <*_ Q£
M C id T-
10 . •!— «r- T3 • —
1> co c_ rd
• • CU co c_
•i- en c c co o
cu M—
J_ o "- CU C
>, O co eu -r- c:
cu cu x> -M • • xu en
c u E id c cu c .
4->O O O -MOT3"-
C .r- U OS • .r- «i- Id C
O) 4-> co CO -Q V--M eu-r-
•r- CO CU . '
CO CU CO C_ O
c en 3 cu -M
id c c 3 co
s- O ••- o cu
i— u co _i -M :
,
*
— CO V* CO X. rd
id CU -i- CO 4-> -4->X)
E s- en x: c -M
u O eu c cn.r- co
O a >> O •*- re) cu
Z CO LU O — 1 E -M
x: c
| CM 'i
o r-
O • -M CO
-o co •—
x:
x: LO
cu
s_
3
V)
O O O
CL i- a. a. >-
X O X X O
eu M— eu eu <4_
^/ QJ co \y cu cu -^£
2 r— r— • 3 *"" l~m f
•aaecc -o3cc TSS
•r- O
O S -M
4->
LO O
xi "3- . co
CU CO . CU
c_ c_ t_
3 33
CO CO CO
O O O
CL CL Q.
XXX
cu cu cu
cu cu cu
^yt c?) en
c c c
Loiotvotr) LO'LOCOCO in 10 - O) •).)
0 t
CVJ
o —
00 o
^T
CM
o
o •*•* . ••-
O O O co . O
O LO O O O O LO
O co O
o •— •—
CM CM CM
68
-------
^_^
T3
O)
3
CI
•r™
4-*
C
o
o
x_s
•
^-»
1
LO
Ol
1—
O)
u
(U
cu
"oi
Q£
o
O)
(4-
LU
Ol
i- -0
3 O
in .i—
0 i-
Q. O)
X Q.
m
C Q.
ai a.
Q.
Q.
O
oo
CM
congestion; transient
nausea
•
m
u
CU
c
CU
Q.
O
Ol
c
>f~
*~~
01 O)
3 <4-
c
oo JZ
Ol
O 01
4-" • -
rd C C
4-> O O
S- 4-> -M
I- Ol rd
•r- CU •—
Ol CU
0)' C
>> O <*-
UJ u o
£
0
m
*""" '
CU
3
0
Q.
X
cu
Ol
c
CO
^
CL
O.
LO
CM
LO
—
r
rd
"cu
CU
O
* «h
C
O
4->
S_
• r-*
CU
00
O
c:
T3
C
rd
UJ
,-
C
'i
o
^""
CU
3
01
O
CL
X
CU
CU
Ol
CO
ex
CL
o
0
VO
c
O O
o
E 01
01
•o o
O) r—
01 CU
= O
•I— 00
E
•a c"
o
• -.r- C
oo 4-> O
01 rd •!—
CU C 4->
C T- -r-
"- T3 J3
N i- -f-
N 0 J=
•r- O C
•a u .1—
« «
"
m
"" -
**
CU
C
cu
ex
S-
•
OJ
3
Ol
X
OJ
at
Ol
c
CO
Ch
CM
LO
2
*
"id
4->
O)
O>
• O
>,
s_
O
rd
s_
Q.
Ol
CU
i_
O)
O.
3
C
o>
UJ
c
'i
CM
' 0
—
«
• 3
01
O
0.
X
CU
O)
Ol
c
CO
o
0
rn
^™
jj"
4->
C rd <—
O c CU 4->
•.— 4-> U
S- (U
I- CU -r-3
••- 1- J3
3 3
4-> 01 Ol
U O
S- X r—
•+-> LU rd
C
E
LO
^
-
O)
3
01
O
Q.
X
cu
01
Ol
<=
•f—
CO
"
o
o
0
CM
.'
£
s_
Ol
o
"id
.C
Q.
CU
U
C
0)
o
4-)
O
CU
»— •
ID
UJ
UJ
.
O
c
rd ~^y
•i- S-
X O
fO ' O
rd
i °
4->
-t-> CZ
C QJ
cu E
E 0)
O) 1-
S- 3
3 01
01 re
rd OJ
OJ E
E
• • i •
4-> 01
01 O)
OJ 4->
4->
C
Ol rd
i- Ol
O) rd
E ri
5 L7
rd 43 U
69
-------
Incident, urinary urobilinogen and serum bilirubin levels were elevated, an
indication of liver impairment. On the 18th day after exposure, a slight
elevation of SGPT was measured (Stewart,et a_L, 1961 a). In a separate incident
reported by Stewart (1969), a worker was overcome by PCE vapors (of unknown
concentration). The exposure lasted about ten minutes and produced hepatic
dysfunction. Clinical measurements of liver function were normal shortly after
exposure. A slight increase in SCOT levels was measured on the third and
fourth days following exposure, and urinary urobilinogen levels were elevated
on the ninth day. Elevated serum enzymes were also measured by Hake and
Stewart (1977) after a massive overexposure to PCE (estimates of the exposure
concentration were not made). 'The individual was found lying in a pool of PCE,
and some was probably absorbed dermally as well as through the lung. Recovery
was complete within 21 days.
Eight of nine firemen exposed to PCE vapors for three minutes (unknown
concentration) had elevated SCOT levels. Hepatomegaly and splenomegaly were
also found in one individual. Normal function was regained within 22 to
63 days (Saland, 1967). ,
The effects of chronic occupational exposure to PCE vapor were studied in
seven individuals exposed for 2 toa6.y (Coler and Rossmiller, 1953). Short-
term tests indicated that exposure levels ranged from 232 to 385 ppmv. The
authors assumed that these measurements were representative of levels that
workers ^were exposed to 8 h/d, 5 d/wk. Of the seven individuals, three had
abnormal liver-function tests, and one was diagnosed as having cirrhosis.
Kidneys
In animals, kidney damage generally occurs at exposure levels greater.
than those that cause liver toxicity. This trend seems to hold true for
humans as well. Several cases of overexposure great enough to cause loss of
consciousness have not produced measurable kidney damage (Stewart et aj..,
1961a; Stewart, 1969; Patel et a].., 1973; Patel et a].., 1977). In one case of
overexposure (where an individual was unconscious in a pool of PCE for 12 h),
kidney damage was measured by proteinuria and hematuria. These effects lasted
for 20 d and for 8 d, respectively (Hake and Stewart, 1977)..
70
-------
Lungs
• Pulmonary edema has been documented in only one instance. Exposure to
PCE was probably massive (>1500 ppmv) because the individual was comatose and
required mechanical ventilation upon admission to the hospital. Recovery was
complete within 4 d (Patel e_t al_., 1977).
Skin and Eyes
Dermal contact with PCE causes localized irritation; prolonged exposure
can cause erythema, first- and second-degree burns, and blistering (.Gold,
1969; Stewart et a]_., 1961b; Hake and Stewart, 1977). PCE vapor is also
irritating to the eyes (Carpenter, 1937; Rowe et al_., 1952). Exposure of human
volunteers to 106 ppmv produced transient eye irritation; this became more
pronounced when the concentration was Increased to 216 ppmv (Rowe et aJL,
1952). Stewart et a].. (1970) has also reported eye irritation when humans
were exposed to 100 ppmv for a period of 7 h.
Connective Tissue
Sparrow (1977) described a patient who had a connective tissue disorder
with similarities to a syndrome observed in vinyl chloride workers. The
individual in question was exposed to PCE vapor during his work at a dry
cleaners. No measurements of concentration were reported, but at least once a
week (over a four-year period), exposure was high enough to cause dizziness and
sleepiness. The individual displayed pathological changes in the skin of the
hands, acrocyanosis, and polymyopathy. Abnormalities in the immune system,
and in hepatic function (mild hepatitis) were also documenfed.
The patient.,may have been abnormally sensitive to PCE, perhaps related to
an existing abnormality in the immune system. Indications that the individual
did have an abnormal immune system are suggested by intermittent alopecia
areata (since childhood), vitiligo (an apparently autoimmune condition
characterized by destruction "of melanocytes), and an absence of immunoglobulin
A (IgA). This app'ears to be a unique'case report, whose information is
71
-------
confounded by the abnormal medical history of the individual. Although PCE
exposure may have contributed to the disease, it is not clear if PCE was the
sole causative agent.
Central Nervous System
'Acute exposure to PCE generally causes temporary CNS effects such as
dizziness, headache, and confusion. However, massive single exposures can
cause loss of consciousness (Patel et a_L, 1973; Patel et a].., 1977; Hake and
Stewart, 1977), and have been fatal in at least two instances (Lukaszewski,
1979; Levine et al.., 1981). Protracted exposure produces symptoms similar to
those observed after short-term exposure, although the effects apparently
persist, even after exposure is terminated.
Rowe et a].. (1952) did not observe any CNS effects in humans exposed to
106 ppmv (single exposure of unspecified duration). Minor CNS effects were
produced by 216 ppmv (45 min to 2 h), and a 10-min exposure to 600 ppmv
significantly affected motor coordination. Eleven volunteers exposed for a
single 7-h peri'od to 100 ppmv experienced headache, dizziness, and somnolence
(Stewart et al.., 1970). Three of the eleven had abnormal scores on the Romberg
test, which measures ataxia. Tests of coordination, visual inspection, visual
acuity, and depth perception were normal. Carpenter (1937) noted only minor
and transient CNS effects in individuals exposed to 500 ppmv for 2 h. When
the same subjects were exposed to 911 ppmv, they complained of lassitude,
exhilaration, and inebriation.
In a study conducted by Stewart et al.. (1974), individuals were exposed
to 20 to 150 ppmv of PCE, 7.5 h/d over a 5-wk period. No alterations in the
EEG were noted at 20 ppmv, but some aberrant EEC tracings were seen after
100.ppmv. A decrease in Flanagan Coordination Test scores was observed
following exposure to 150 ppmv. In a subsequent experiment (Stewart et al..,
1977), individuals were repeatedly exposed to 0, 25, or 100 ppmv of PCE,
5.5 h/d over 11 wk. CNS effects were observed by measuring the subjects'
response on the Romberg, Michigan Hand-Eye Coordination, and Flanagan
Coordination Tests, and on the EEG. In contrast to the 1970 study, exposure
to 100 ppmv did not produce abnormal scores on the Romberg Test. PCE at
100 ppmv did cause a significant decrease in Flanagan Coordination Test scores.
•72
-------
Coler and RossmiHer (1953) documented subjective compla.ints of malai.se,
dizziness, headache, light-headedness, and intoxication in individuals
regularly exposed to concentrations of 232 to 385 ppmv PCE. Gold (1969)
described a case history of an individual exposed to PCE vapors, six to seven
days a week for three years. The individual was hospitalized after exhibiting
confusion, disorientation, agitation, and an inability to concentrate. A
neurological examination revealed a normal EEC. However, his performance on
psychological tests that required concentration was "poor", and he showed
"marked confusion." These problems persisted over a 12-month follow-up period,
although there was no further exposure to PCE (the individual was lost to
follow up at this time). Gold (1969) concluded that there was suggestive
evidence of both cerebral and cortical damage, and basal ganglia involvement.
However, it was not possible to obtain conclusive evidence of any neurological
damage. Gregersen et al_. (1984) examined 65 Danish workers exposed to various
organic solvents for neurotoxic effects. Although only 15% of the cohort were
exposed to PCE (approximately 100 ppmv), the results are intriguing.
Individuals were examined for intelligence, given a neuropsychological exam,
and were subjected to a number of neujropsychological tests. The authors
concluded that solvent exposure was correlated with acute neurotoxic symptoms,
as well as longer-lasting symptoms of - intellectual impairment. A relationship
between exposure and signs of peripheral neuropathy was also observed.
McMullen (1976) reported on a case in which an individual was exposed to
PCE vapor (500 ppmv) for an undetermined period of time. CNS depression was
observed; the effects were described as resembling alcohol intoxication. No
clinical measurements of sensory or organ function were made, but the
individual apparently recovered within six hours.
Reproductive System
No studies have addressed the question of whether exposure to PCE affects
the human reproductive system. At present, there is no evidence of human
reproductive toxicity from PCE exposure. .
£
Cardiovascular System
Some organic solvents have been associated with cardiac arrest due to
ventricular fibrillation. It is thought that these compounds sensitize the
73
-------
heart td epinephrine-induced arrhythmias. There is suggestive evidence that
PCE has this effect in animals (see Section 4), and there is one report on
PCE-induced cardiac arrhythmias in humans. Abedin et al_. (1980) observed that
occupational exposure to PCE probably caused dizziness and premature
ventricular contractions in one case. Although PCE may not have been the only
factor in this response, removal from exposure to PCE alleviated the symptoms.
TERATOGENIC EFFECTS
\
There is no published information on the teratogenicity of PCE in humans.
MUTAGENIC EFFECTS
Ikeda et al_. (1980) examined lymphocytes from individuals exposed to PCE
for 3 months t
-------
for evaluating evidence on the carcinogenic activity of a substance. The
International Agency for Research on Cancer (IARC, 1982) separates strength of
evidence of carcinogenic activity into four groups: sufficient evidence,
limited evidence, inadequate evidence, and no evidence of carcinogenicity.
Inclusion in any of these categories is based on data from short-term assays,
as well as animal and human studies (if available).
The U.S. Environmental Protection Agency uses the same groupings, but
places a substance in one or the other category solely on the basis of animal-
bioassay data (U.S. EPA, 1984a). To assess overall evidence of carcinogenic
potential to humans, six additional categories are used: Group A - Human
Carcinogen, Group B - Probable Human Carcinogen (further separated into Bl and
B2), Group C - Possible Human Carcinogen, Group 0 - Not Classified (due to
inadequate animal evidence), and Group E - No Evidence of Carcinogenicity for
Humans (U.S. EPA, 1984a).
In the absence of sound epidemiological data, the greatest weight of
evidence in a carcinogen assessment is typically given to the results of
lifetime animal bioassays. The criteria employed in analysis of bioassay data
include an increase in the incidence of tumors in treated animals over those
noted in controls, a decrease in latency (time to tumor development),
development of rare tumors, and an increase in the number of tumors in
individual animals.
In 1982, the IARC evaluated available information on PCE and determined
that there was "inadequate" evidence to conclude that PCE is carcinogenic to
humans (IARC, 1982). This assessment was based primarily on the epidemiologic
study of Blair et aj.. (1979). (An analysis of animal data gave "limited
evidence" of carcinogenicity; evidence for any activity of PCE in short-term
tests was "inadequate" to judge its carcinogenic potential) (IARC, 1982). The
U.S. EPA Health Assessment Document for Tetrachloroethylene (Perchloroethylene)
(EPA, 1985a) has also analyzed the evidence of carcinogenicity of PCE. Thts
evaluation included an extensive review of short-term test results, data from
animal tests, and several epidemiological studies. EPA concluded that the
evidence for the carcinogenicity of PCE in animals is "limited," and that the
epidemiological data were inconclusive. PCE was placed in Group C, a possible
human carcinogen (EPA, 1985a);
75
-------
It should be noted however, that since these two analyses were published,
other epidemiologic and animal study results have become available. It is
possible that a subsequent evaluation by either the U.S. EPA or IARC would
result in a different conclusion.
76
-------
6. HUMAN EXPOSURES TO PCE FROM CONTAMINATED NATER SUPPLIES
In this section, we describe the procedures we use to assess human
exposures attributable to contaminated water supplies. Efforts to assess
human exposure to contaminated drinking water have revealed that significant
exposures to volatile organic compounds (VOC's) can occur from pathways other
than the direct ingestion of water. Several researchers have investigated the
relative importance of a variety of VOC exposure routes in the home from use
of contaminated water supplies (Cothern et a_L, 1986; Shehata, 1985; Andelman,
1985). In addition, there have been studies of the contribution to indoor
exposures of waterborne radon-222, another highly volatile substance (Hess
et al_., 1982; Prichard and Gesell, 1981). These studies indicate that exposure
to volatile chemicals from routes other than direct ingestion of fluids may be
as large as or larger than exposure from ingestion alone. These other routes
include inhalation from indoor air of contaminants mobilized by showers, baths,
toilets, dishwashers, washing machines, and cooking; ingestion of contaminants
in food; and dermal absorption of contaminants while washing, bathing, and
showering. We divide exposures attributable to contaminated ground water into
ingestion, inhalation, and dermal absorption pathways. Our discussion of human
exposure to PCE is divided into three subsections. In the first subsection,
we provide a background on the general approach we use for assessing human
exposure to VOC's. This approach addresses relative contributions from
.ingestion, inhalation, and dermal absorption. Our focus in this subsection is
on the type of information that is needed for risk assessments. The second
subsection covers the method used to estimate *ingestion, inhalation, and
dermal-absorption dose factors. These factors convert water concentrations in
mg/L into human-population exposures in mg/kg-d. The third subsection presents
our calculations of the magnitude and distribution of human exposures
attributable to PCE contamination in California groundwater supplies.
+
BACKGROUND ON HUMAN-EXPOSURE ESTIMATES
The purpose of the human-exposure estimate is to provide a distribution
of population dose of a chemicalfrom the various phases of the environment.
Three primary pathways must be addressed — inhalation, ingestion, and dermal
77
-------
-------
B.4
EXPOSURE ASSESSMENT
AtmDUTe 2 The specific populations and subpopulations that are the
subjects of the assessment are clearly identified,,and the
reasons for their selection and any exclusions are given.
SOURCE Case Study D. Formaldehyde (Pages' 6-2 to 6-4).
Note References to previous studies are given as background and as a basis
for selecting two.population groups for study.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
derivatives are irreversibly Corned and usually contain only
residual levels of unreacted HCHO. Under extreme conditions,
such as very high temperatures or highly acidic conditions,
of the derivatives may degrade and release HCHO.
HCHO's major nonconsumptive uses are (1) disinfectant, (2)
preservative, (3) deodorant, and (4) .textile' and oaoer uses.
The major" pseudo-consumptive uses are (I) urea-HCHO resins
which are used: in fiberboard, particleboard, plywood, laminates,
% •
urea-HCHO foams, molding compounds, and paper, textiles, and
protective coatings; (2) urea-HCHO concentrates which are used to
produce time-release fertilizers, and (3) hexanethylenetetramine
i
which is used as a special anhydrous form of HCHO to cure resins
and to treat textiles and-rubber.
The major consumptive uses are (1) melanine-HCHO resins
which are used for molding compounds, fiberboard, particleboard,
plywood, laminates, paper and textiles, (2) phenol-HCHO resins
which are used in fiberboard, particleboard, olywood moldina
compounds, and insulation; (3) nentaerythritol which is used to
produce alkyd resins. (4) 1,4-butanediol which is used to produce
tetrahydrofuran, (5) acetal resins which are used in the
manufacture of engineering plastics, and (6) trimethyloloropane
which is used in the production of urethanes.
6.2. Batiaates of Current Human Exposure
To obtain estimates of human exposure to HCHO, the Agency
commissioned a contractor study (Versar, 1982). This studv
integrated the existing monitoring data, engineering or -nodelir.^
•»,
6-2
-------
estimates, use data, population estimates, and assessment of the
likelihood of exposure from HCHO-related activities into an
exposure assessment detailing those activities having a high HCHO
exposure potential. EPA updated some portions of this assessment
to reflect new data received in response to the FEDERAL REGISTER
notice of November 18, 1983 and other data gathered by EPA. The
combined data were used as the basis for the May 1935 draft risk
assessment.
#
Subsequent to the draft risk assessment, the Agency
commissioned additional contractor studies to assess garment
worker (PEI, 1985) and residential (Versar, 1986a,b,c) exposure
to HCHO in more depth. The, exposure estimate's from these reports
were used as the primary basis for this risk assessment. The
conclusions of these contractor reports are summarized in this .
document; more detailed information regarding exposure can be
obtained by referring to the contractor reports.
6.3. Populations at Risk
The two populations at risk examined here are certain home
residents and garment workers.
6.3.1. Home Residents
Based on a projection of manufactured housing starts by
Schweer (1987), it is estimated that 7,800,000 persons may occupy
new manufactured homes during the next ten years. This figure
assumes 295,000 starts per year and 2.64 persons per home.
Similarly, an estimated 214,000 new conventional homes
containing significant quantities of pressed wood products as
«"• .
construction materials will be started each year for the next ten
6-3
-------
years with an occupancy rate of 2.95 persons for a total of
6,310,000 persons.
6-3.2. Garment Workers
The number of potentially exposed garment workers is
estimated to be 777,000 (Versar, 1982) out of 1,100,000 workers
employed in the U.S. apparel industry (Ward, 1984). This figure
may drop in the future due to-increased foreign competition and
the introduction of labor saving equipment.
6.3.3. Summary
«
Table 6-1 presents population estimates ,for the two housing
segments. Assuming that the number of potentially exposed
garment workers remains steady at 777,000, then a total of almost
15,000,000 persons over the next ten years may have the potential
to be exposed to elevated levels of HCHO.
Table 6-1.
POPULATIONS AT RISK
Category
Manufactured homes
Conventional homes
* Schweer (1987)
per yr
779,000
631,000
Population
Estimates
10 yraT
7,790-, 000
6,310,000
6-4
-------
-------
B.4
EXPOSURE ASSESSMENT
Attribute 3
Available data are considered and critically evaluated, and the
degree of confidence in the data expressed. (Reasons for any
data exclusion are presented.)
SOURCE Case Study I. Para-dichlorobenzene (Pages B-3. to B-5).
Note Limitations in drawing conclusions from the data in a large multi-
chemical monitoring study are pointed out.
-------
ASSESSMENT OP HUMAN CANCER RISKS
FROM PARA-DICHLOROBENZENE
FEBRUARY 13- 1987
Prepared by Barbara Mandula, Project Manager Karl Baetcke
Diane Beal, C.J. Nelson, Vanessa Rodriguez, Gary Thorn
Risk Analysis Branch Office of Toxic Substances EPA.
-------
a - 3
l)
Measurements
a) TEAM Study
Extensive monitoring of PDCB exposure
was carried out as part of.EPA's Total Exposure Assessment
Methodology (TEAM) study (Pellizzari jt .al_. , 1985). The TEAM
study used personal air monitoring devices to measure the
concentration of 20 organic compounds in daytime and overnight
air samples among 600 persons selected in four geographical
regions. The samples were taken in two consecutive 12-hour
intervals. In addition, exhaled breath samples were taken at the
end of the 24 hours.
Although meta- and para-dichlorobenzene
were measured together, the values until now have been assumed to
represent PDCB, since there were no known common sources of the
meta- compound. However, a recent experiment found that a
liquid deodorizer contained approximately equal amounts of meta
and para isomer (Wallace, 1986c). Therefore, the TEAM
measurements may in some cases reflect meta- plus para- isomers
of dichlorobenzene, and therefore represent an upper limit on
possible PDCB levels. It is likely that most .of the: measured m-
»p-DCB in the homes sampled represented PDCB since production of
the meta- compound is approximately an order of magnitude lower
than that of the para- compound (U.S. EPA, 1985).
Tables B-2 and B-3 show the TEAM results
for daytime and nighttime personal air samples. Daytime and
overnight exposures are similar. It is unclear to what extent
these findings reflect differences in exposure,at home and away.
-------
0)
M
0)
£>
O
u
O
o
•r*
Q
I
a
IU
o
4)
*-4
a
s
10
V)
a
o
(0
— 4
O
CN
3
1
0)
C
3
in
r-t
CN
z
o
in
i
CN
CN
O
O
CN
CN
0
CN
i-4
O
CN
"*
vO
•
en
in
00
03
ON
— i
_Q
0)
EC4
C
"
ON
en'
§
0 0
en CN . fn
CN f» T '
1 1 1
ON eN .en
Tf ~4 O
0 0 O
< < O
Z Z O
— 1
000
CN in en
in r» r»
o o o
en ON vO
-4 CN CN
vO ON
, J3
0 (0 0)
cu u <
Q O J
O -i
in in
CN ^4
i i
in in
0 0
o o
o
•
O ^0
f^ ^
^4
^5 , ^^
^^ ^\
CN CN
t en
03 in
0 O
.'
CN
»* I~"
i-4
03
ON
i—4
^*
0) 03 .
C ON
3 -l
1 0)
>i C
(0 3
^4 CO
in o
i i
o \o
ir> ^
en
0
U
*0
S-l
CN -P
C
< o
3 o
in
00
ON
0
XI
(0
fl
rtj
o
z
0)
cu
-------
•o
3
•U
CO
s
3 s
_ C
5 5
u "•
U-l
*"£
^ <5) of
$ -. in
•H CIN
c *i
.5 c
« 8
c !*
s *-
i £
0)
O
u
o
-•§ *
S £
? 1
• a s
0
« Z
0> •|~l
-U *•
1 i S
to ' * ^
•^
^ <
-^
I «
2 o|
f fil
e
(0
•
ro
i
03
J O
51-1
0
.a
U} CO
^ c
'-I CN -H <1) -H CN
o
2 ^ i-j
-------
Ficjure
POMHXnOM tXCttOHM CONCJWTRATION SHOWN
iifjDi t»sat iu» u»
1000
1.000
i m,p-DICHLOROBENZENE
100
NWMT
<.ICINO
> MMONM. AUI j
•IO.MM !
-. ourooe* urn
N-.M1
i.ooi
100
I
N
.NIGHT
/
~0* 90% «% W*
12JOO **OOD 115JOO 127.000
KJWUkTION MUOW* CONCtNTHATKJN SHOWN
nonj of per»n»l exposure*.
collected in the vic.mtv of tne participant*
Samples taken Fall, 1981
Source: Wallace et,al., 1985
-------
B - 4
although-overnight samples would be expected to reflect primarily
indoor horn* air. In some cases, daytime measurements may also
reflect home air to a great extent, depending on an individual's
daily routine. -The populations sample'd were chosen statistically
to reflect a much larger population. Thus, for example, the
sample included children (older than 7 years old), and one-half
of the persons sampled were not in the workforce.
The distribution of concentrations
provides some indication of range of exposures. Thus, the median
value is-close to the outdoor measurements, which are usually
below 2 ug/m3, whereas the 75th and 95th percentile values are
£ • '
higher. Approximately a 10,000-fold range was found for measured
values of this compound. Figure B-l shows the distribution-in
New Jersey.
For the samples in New Jersey, if. we
take 50 ug/m as an arithmetic mean concentration of daily
personal exposure, and assume an -inhalation rate of 1 m^/hour, we
get a daily intake of about 1.2 mg. The nighttime median is about
3.5 ug/m3 and the 75th percentile about 13 ug/m3. Mean 24-hour
values in the other sampled areas are 6 to 18 ug/m3, giving a 24-
hour daily intake of 0.14 to 0.42 mg/day. The 12-hour 75th
percentile concentration values excluding New Jersey range from-
2.5 to 9.4 ug/m3, and the median ranges from 0.5 to 2.9 ug/m3.
The TEAM measurements presumably include
some persons not using any PDCB products at home, as-well as some
using PDCB-products in several locations. The 10,000-fold range
in nighttime personal exposures is consistent with this
-------
3 ,- 5
explanation (Wallace, 1985). There is essentially no information
on whether the persons with high exposure were using PDCB
products. Follow-up studies are planned in the near future to
determine the sources of PDCB in homes and their overall
contribution to PDCB loading. Such follow-up studies might
provide an explanation for the higher levels found in New Jersey
as compared with the other sampled areas.
b) Home with Known Source
In a recent experiment related to the TEAM study, Wallace
(1986c) reported measurements of PDCB in a home with a known
source. Two consecutive 8-hour measurements established a
baseline indoor PDCB concentration of about 1 ug/m3. A room air
freshener containing PDCB was then opened and PDCB levels
measured in the home, and detected by personal monitors, averaged
300 ug/m3 over the next six 8-hour measurement periods. Breath
concentrations of the residents increased approximately linearly
over the same two-day period to 50 ug/m3.
c) Concentration of PDCB in Public Restrooms
In another short-term experiment,
Midwest Research Institute (MRI 1986) determined PDCB
concentrations and related variables in two public bathrooms.
Figures B—2 and B-3 provide information about the bathrooms.
Bathroom 1 had one urinal block in the open air on the floor, .
whereas bathroom 2 had two blocks, one in a urinal and one in( a
toilet bowl. These experiments were carried out in large well-
ventilated bathrooms under conditions that do not necessarily
reflect the conditions that would be encountered in a home. In
-------
B.4
EXPOSURE ASSESSMENT
Attribute 4
If models are used, their bases are described, along with their
validation status.
SOURCE Case Study I. Para-dichlorobenzene (Pages B-8 to B-ll).
Note The limitations of the model are discussed. None of the case studies
treated model validation status extensively.
-------
ASSESSMENT OF HUMAN CANCER RISKS
FROM PARA-DICHLOROBENZENE
FEBRUARY 13- 1987
Prepared by Barbara Mandula, Project Manager Karl Baetcke
Diane Beal, C.J. Nelson, Vanessa Rodriguez, Gary Thorn
Risk Analysis Branch- Office of Toxic Substances- EPA.
-------
B - 8
For Table B-4, worst case and be'st
estimate values are given for adults, and for children
approximately one year old. The calculations are based on the
modeled .concentration in a given room, an estimate of time spent
per day in that room (1-14 hours) and the indicated inhalation
rate. In Table B-5 best estimates for exposure are obtained by
considering the concentration throughout a house, the residence
time in the house (1-14 hours/day) and the indicated inhalation
rate.
Even using the house concentrations, the
values obtained are in the highest range of those found in the
TEAM study. Therefore, it seem likely that these models do not
adequately reflect the actual exposures that people experience
from PDCB.
b) Simplified Model .
Wallace (1986a) has suggested a
simplified equation that may be applied to PDCB sources in a
home. The equation
cin =S/V(a+k)
where C^n is the indoor concentration in ug/m3
S is th« source generation fate in ug/hour
V ia the volume of the home in m3
a is the air exchange rate/hour
and k is a decay constant due to absorption, chemical
reaction etc in h~^ , between 0 and 1.
-------
B - 9
This is the basic equation that can be-
used to estimate PDCB concentrations in homes with sources. The
source generation rate may be estimated from knowledge of the
Loss rate as provided by the manufacturer. The volume of the
home and the air exchange rate may be estimated or measured.
However, little is known about the magnitude of the decay term k,
representing the loss of the vapor through adsorption or other
chemical processes. Presumably some absorption on fibers occurs,
since clothes that have been recently stored near moth crystals
retain odors. Absorption on surfaces such as paint or wood is
also possible.
A second major problem in employing this
equation is our lack of knowledge of air flow in homes. The
above equation assumes perfect mixing within the home, whereas in.
fact the mixing is affected by many factors including the type of
ventilation system (central forced air provides better mixing
«L
than other systems), use of fans, presence of closed doors
between interior rooms, etc. A "mixing factor" jn can be defined
that acts as a multiplier to the air exchange rate. The value of
m may range between 0.1 and 0.5 in practical situations.
Alternatively, we may simply use somewhat smaller values of the
air exchange rate in the above equation to,take into account
imperfect mixing. Similar concerns apply to any model.
Without better knowledge of absorption
processes and mixing factors, we cannot hope to use models to
provide accurate predictions of observed concentrations.
However, by assuming realistic values for the various model
-------
B - 10
parameters./ we may establish a range of concentrations that can
then be compared to measured values.
We can now estimate concentrations for
various scenarios, using suitable values from Table B-7.
Worst Case
A small tight apartment with strong sources
of PDCB and no materials to absorb the PDCB vapors. V = 100;
k = 0.0; a = 0.1; S = 100 g/month. We convert S to ug/h by
dividing by about 720 h/month.
C = 140,000 ug/h 14,000 ug/m3
100 m3 x 0.1 h'1
Typical Case
An average-size home without special
insulation. V = 400; k = 0.5; a = 0.5; S = 10 g/month.-
14,000 ug/h
400 m3 X (0.5 •)- 0.5) h"1
= 35 ug/m
This calculation yields 35 ug/m3, or about 840 ug/day. This value
is in the range of the observed arithmetic mean values of from 5
to 78 ug/ra3 found by the TEAM study. Furthermore, this value is
1/3 to 1/12 of that estimated by the house method in Tables 3-5
and B-6.
The simplified equation was also applied to
the MRI bathroom in which a value for S of 102 g/month had been
measured. The concentration was calculated as 400 ug/m3; the
measured value was 600 ug/m3. It appears that the simplified
-------
B-7. Parameters for Simplified Indoor Air Model
Parameter
Volume (m3)
Air exchange rate (h"1)
Source generation
rate (g/ month)
Decay rate (h"1)
Symbol
V
a
S
k
Low
100a
0.1
1
0.0
Medium
,400b
0.5
10
0.5
High
700
1
100
1
.0
.0
a Small apartment (about 400 sq. ft.)
b Average new home (about 1700 sq. ft.)
From Wallace, 1986c.
-------
B - 11
model yields concentrations at least in some cases, that agree
with measurements.
3) Number of Households Using PDCB Products
The number of households using PDCB as space
deodorants, toilet deodorants, or as moth repellent is not -
«*
known. We can get some estimate of the numbers of households
using these products by assuming 4 million kg/year total used for
each of the three uses, 80% of use by consumers, and reasonable
,/•
estimates for numbers of grams/month/household. Also, to the
extent that the exposures in the TEAM study indicate household
use, we can presume that approximately one-half the households
may be exposed to close to ambient levels.
Simmons (1982) reports that about 25 % of
households, or about 20 million households, use "in-bowl toilet
fresheners." Two-thirds of households use air fresheners and
room deodorizers; approximately 25%, or 20 million, households,
use a solid product. It is unclear to what extent these products
represent PDCB use, but it seems likely that much of the use of
these solid products reflects PDCB use.
Using the data on total consumption, if we
assume 10 g per month/household for toilet deodorant, we get.25
million households using the product; assuming 16 g per month for
space deodorant gives 16' million households using space
deodorant, and 37 g per month (a one-pound package per year) of
moth repellent gives 7 million households. If the
use/month/household is overestimated, then more households are
using the products; if the use/month/household, is
-------
-------
EXPOSURE ASSESSMENT
Attribute 5
Potential sources, pathways, and routes of human exposure
are identified and quantified; the reasons why any are not,
included in the assessment are presented.
SOURCE Case Study K. Tetrachlproethylene (Pages 4-8; 73-93).
-------
UCRL-15831
Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
K. T. Bogen, L. C. Hall, T. E. McKone,
D. W. Layton, and S. E. Patton
Environmental Sciences Division
Lawrence Livermore National Laboratory
University of California
RO. Box 5507
Livermore, CA 94550
April 10, 1987
Prepared for
California Public Health Foundation
P.O. Box 520
Berkeley, CA 94701
-------
structyr? °f tetrachloroethylene. alternative names
numbers, empirical formula, and molecular weight.
Chemical structure:
Cl - C - C - Cl
Cl Cl
Empirical formula: C2C14 Molecular weight: 165.85
Chemical Abstracts Service registry number: 127-18-1
NIOSH Registry of Toxic Effects of Chemical Substances number: KX3850000
Alternative names: PCE, Perc, tetrachloroethene, perchloroethylene, ethylene
tetrachloride, carbon dichloride
Common trade names: Antisol, Dee Solv, Per Sec, and Texranec.
ENVIRONMENTAL TRANSPORT AND TRANSFORMATION
Tetrachloroethylene tends to partition primarily to the atmosphere. It
has been estimated that 85 to 90% of the PCE produced is eventually released
to the atmosphere (U.S. EPA, 1985a; HHO, 1984). The key properties of PCE
that affect its movement in the environment are its high vapor pressure and
low solubility in water.
GEOTOX (McKone and Layton, 1986) was used to estimate the equilibrium
distribution of PCE in air, soil, and water. GEOTOX is %a multimedia
compartment model that simulates the environmental transport and transformation
of a chemical, based on its physical and chemical characteristics and the
properties of the landscape into which it is released. A simulation of the
environmental partitioning of PCE was run using California landscape data,
properties of PCE (see Table 2-2), and PCE source-emission data from the
California Air Resources Board (CARB). The PCE source term was represented by
an annual release of 1.83 x 107 kg/y over an area of 411,000 km2 (CARB,
1984); of this source, 10% is assumed released into soil, 1% to surface water,
and the remainder directly to the atmosphere. The simulated equilibrium
distribution of PCE is shown in Fig. 2-1. Most of the PCE released to the
environment is found in the atmosphere. However, the equilibrium
distributions, 86% in the air and 11% in surface water, reflect both the
-------
Atmospheric gas
86%
Atmospheric particles
4 x 10~8%
Bio miss
0.02%
Upper soil
0.4%
4
t
Lower soil
0.5%
•
|
Ground water
6
1%
Surface water
11%
Sediments
1%
8
Figure 2-1. Environmental distribution of PCE under steady-state conditions.
Partitioning between compartments Is predicted by the computer model GEOTOX
(McKone and-Layton, 1986).
.5
-------
Table 2-2. Chemical, physical, and organoleptic properties of
tetrachloroethylene.
Property
Units
Value
Reference
Boiling point at 760 mm Hg 8C
Freezing/melting point °C
Density at 20°C g/cm3
Vapor pressure at 20°C mm Hg
Henry's law constant at 20°C atm-m3/mo1
Conversion factor mg/m3-ppmv
Diffusion constants
at 1 atm, 20°C
Air m2/s
Water m2/s
121
-22.4
1.65
15.8
0.0227
6.89
Solubility in water at 258C mg/L
7.4 x 10-6
7.6 x 10-10
150
Log octanol/water
partition coefficient
Odor threshold in water
Unitless
mg/L 3.0 x 1Q-1
3.14
2.46
Hawley (1981)
Hawley (1981)
Hawley (1981)
Sittig (1985)
Mackay and Shiu (1981)
Verschueren* (1983)
Lyman et al. (1982) *
Mackinson et al.
(1981)
Leo (1983)
Callahan et al.. (1979)
Zoeteman et a_K (1974)
relative magnitude of the source (89% to air and 1% to water) and the
effective residence times. The loss rate of PCE in air is an order of
magnitude greater than that in surface water. This accounts for the apparent
"enrichment" of PCE in surface water.
Air
PCE in the atmosphere is subject to relatively rapid chemical or
photochemical degradation. In the troposphere, it photodegrades, ultimately
leading to the formation of hydrochloric acid, trichloroacetic acid, and carbon
dioxide in the presence of atmospheric water (U.S. EPA, 1985a). PCE can also
be removed by scavenging mechanisms, primarily through hydroxyl radicals
(Dimltriades et a!., 1983). Estimates of Its atmospheric residence time are
on the order of one year or less (see U.S. EPA, 1985a).
-------
Singh et al- <1981> compiled monitoring data for the concentrations of
several volatile organlcs 1n ambient air and found that for the western half
of the U.S. the average PCE concentration was 4.3 jig/m3 and the overall range
was 0.23 to 51.6 ng/m3. The U.S. EPA (1985a) reported ambient air PCE^
concentrations 1n California (1972-1980) ranging from 0,2 to 19.0 jig/m . The
California Air Resources Board (Nystrom, 1986), based on preliminary data,
found average ambient air PCE concentrations for several California locations:
Los Angeles 1180 * 900 parts-per-trillion by volume (pptv) (8.1 ± 1.2 Hg/m );
San lose 490 ± 330 pptv (3.4 ± 2.3>g/m3); Long Beach 1030 * 560 pptv
(7.1 ±3.9 vg/m3); Stockton 450 ± 170 pptv (3.1 ± 1.2 yg/m ), and Si mi
Valley 330'* 250 pptv (2.3 ± 1.7 iig/m3). These data indicate that PCE
concentrations in the ambient air of urban areas are higher than those in
rural areas (or less densely populated.areas).
Hater
In surface waters, PCE rapidly volatilizes into the atmosphere. Wind
speed, agitation of the water, and water and air temperatures affect
evaporation rates. Photodegradation, in contrast, is a slow decay process and
does not appear to be an important transformation mechanism in water. The
half-life of PCE in shallow water due to volatilization has been estimated at
24 to 28 rain in laboratory experiments (Dilling et al-, 1975). Zoeteman
et al. (1980) measured PCE persistence in surface waters of the Netherlands
from 3 to 30 days (half-life), while in lakes and ground waters, the half-life
was estimated to be 10-fold higher.
In ground water, PCE is relatively persistent, with degradation occurring
through hydrolysis and biotransformation. It is denser than water as an
undissolved liquid, consequently it tends to sink in-ground water. Vogel and
McCarty (1985) have shown that PCE biotransforms to trichloroethylene (TCE),
dichloroethylene, and vinyl chloride via reductive dehalogenation under
anaerobic conditions. They further suggest that the potential exists for the
complete mineralization of PCE to carbon dioxide in aquifer systems. The
half-life of PCE due to aqueous hydrolysis in natural waters can be on the
order of months (Oilling et alv 1975) to several years (Pearson and
McConnell, 1975).
-------
The U.S. Environmental Protection Agency (1985a) reported a mean PCE
concentration of 1 yg/L from 1102 surface water measurements in 45 states
(from August 1975 to September 1984). An important source-of data on the
concentrations of PCE in drinking water supplies is a survey of large water
utilities in California (i .e., utilities with more than 200 service
connections) that was conducted by the California Department of Health Services
(1986). From January 1984 through^ December 1985, the Dwells in 819 water
systems were sampled for contamination by organic chemicals. The water systems
considered included a total of 5650 wells, 2947 of which were sampled. The
wells sampled were selected based on the likelihood of contamination. PCE was
found in 199 wells in concentrations up to 166 jjig/L, with a-median
concentration of 1.9 jig/L. Generally, the highest fraction of contaminated"
wells and the wells with the highest concentrations were found in the heavily
urbanized areas of the state.. Contamination was state-wide. Los Angeles
County registered the greatest number of contaminated wells (i.e., 140).
Soil
There is limited information .on the behavior of PCE in soil. The solvent
can be adsorbed to soil or leached through soil when dissolved in water c-r as
a separate organic phase (as in large spills). PCE associated with soil air
or soil water is more mobile than the absorbed portion (Schwarzenbach and
Westall, 1981). *
* The adsorption of PCE to soils appears to be correlated to its octanol/
water partition coefficient, the organic carbon content of the soi.l, and the
concentration of PCE in the liquid phase. PCE appears to leach rapidly
through soils of low (<0.1%) organic carbon content (U.S. EPA, 1985a;
Schwarzenbach and Westall, 1981).
Several studies have documented the mobility of tetrachloroethylene in
soil/groundwater systems (Piet et §_[., 1981; Schneider et a].., 1981;
Schwarzenbach and Hestall, 1981). Wilson et aK (1981) showed that most of
the chemical was lost from the soil via leaching or volatilization to the
atmosphere. Persistence in soil ranges from months to years.
-------
Coler and Rossmlller (1953) documented subjective complaints of malaise,
dizziness, headache, light-headedness, and intoxication in individuals
regularly exposed to concentrations of 232 to 385 ppmv PCE. Gold (1969)
described a case history of an individual exposed to PCE vapors, six to seven
days a week for three years. The individual was hospitalized after exhibiting
confusion, disorientation, agitation, and an inability to concentrate. A
neurological examination revealed a normal EEC. However, his performance on
psychological tests that required concentration was "poor", and he showed
"marked confusion." These problems persisted over a 12-month follow-up period,
although there was no further exposure to PCE Cthe individual was lost to
follow up at this time). Gold (1969) concluded that there was suggestive
evidence of both cerebral and cortical damage, and basal ganglia involvement.
However, it was not possible to obtain conclusive evidence of any neurological
damage. Gregersen et- al_. (1984) examined 65 Danish workers exposed to various
organic solvents for neurotoxic effects. Although only 15% of the cohort were
exposed to PCE (approximately 100 ppmv), the results are intriguing.
Individuals were examined for intelligence, given a neuropsychological exam,
and were subjected to a number of neuropsychological tests. The authors
concluded that solvent exposure was correlated with acute neurotoxic symptoms,
ft.
as well as longer-lasting symptoms of intellectual impairment. A relationship
between exposure and signs of peripheral neuropathy was also observed.
McHullen (1976) reported on a case in which an individual was exposed to
PCE vapor (500 ppmv) for an undetermined period of time. CMS depression was*
observed; the effects were described as resembling alcohol intoxication. No
clinical measurements of sensory or organ function were made, but the
individual apparently recovered within six hours.
Reproductive System
No studies have addressed the question of whether exposure to PCE affects
the human reproductive system. At present, there is no evidence of human
reproductive toxicity from PCE exposure.
Cardiovascular System .
Some organic solvents have been associated with cardiac arrest due to
ventricular fibrillation. It is thought that these compounds sensitize the
73
-------
heart to epinephrine-induced arrhythmias. There is suggestive evidence that
PCE has this effect in animals (see Section 4), and there is one report on
PCE-induced cardiac arrhythmias in humans, Abedin et aK (1980) observed that
occupational exposure to PCE probably caused dizziness and premature
ventricular contractions in one case. Although PCE may not have been the only
factor in this response, removal from exposure to PCE alleviated the symptoms.
TERATOGENIC EFFECTS
There is no published information on the teratogenicity of PCE in humans.
MUTAGENIC EFFECTS
Ikeda et a_L (1980) examined lymphocytes from individuals exposed to PCE
for 3 months to 18 y. Chromosomal aberrations, sister-chromatid exchanges, and
alterations of the mitotic index were the cytogenetic effects studied. The
exposure lev.el in one group of workers was 92 ppmv (geometric mean), while a
second group was exposed to 10 to 40 ppmv (the authors did not give TWA
exposure concentrations). Although a control group was included, the criteria
usedJ:o select and/or match controls was not reported. Exposed individuals did
not have a significantly greater frequency of chromosomal aberrations or
sister-chromatid exchanges, nor were there any substantial differences in the
mitotic index.
Trichloroethanol is a metabolite of PCE isolated from the urine of humans.
#
Gu et §1. (1981) reported a slight increase in the number of sister-chromatid
exchanges per cell in human lymphocytes exposed to 178 mg/L of
trichloroethanol. Neither of these studies is adequate to assess the
mutagenic potential of PCE and its metabolites in humans.
SUMMARY OF EVIDENCE OF HUMAN CARCINOGENICITY
Evaluation of the carcinogenic potential of a chemical is based on the
results of short-term assays of mutagenesis, pharmacological data (e.g.,
distribution and metabolism), lifetime animal bioassays, and epidemiological
evidence. Several agencies and groups have developed systems of classification
74
-------
for'evaluating evidence on the carcinogenic activity of a substance. The
International Agency for Research on Cancer (IARC, 1982) separates strength of
evidence of carcinogenic activity into four groups: sufficient evidence,
limited evidence, inadequate evidence, and no evidence of carcinogenicity.
Inclusion in any of these categories is based on data from short-term assays,
as well as animal and human studies (if available).
The U.S. Environmental Protection Agency uses the same groupings, but
places a substance in one or the other category solely on the basis of animal-
bioassay data (U.S. EPA, 1984a). To assess overall evidence of carcinogenic
potential to humans, six additional categories are used: Group A - Human
Carcinogen, Group B - Probable Human Carcinogen (further separated into Bl and
B2), Group C - Possible Human Carcinogen, Group D - Not Classified (due to
inadequate animal evidence), and Group E - No Evidence of Carcinogenicity for
Humans (U.S. EPA, 1984a).
In the absence of sound epidemiological data, the greatest weight of
evidence in a carcinogen assessment is typically given to the results of
lifetime animal bioassays. The criteria employed in analysis of bioassay data
include an increase in the incidence of tumors in treated animals over those
noted in controls, a decrease in latency (time to tumor development),
development of rare tumors, and an increase in the number of tumors in
individual animals.
In 1982, the IARC evaluated available information on PCE and determined
that there was "inadequate" evidence to conclude that PCE is carcinogenic to
humans (IARC, 1982). This assessment was based primarily on the epidemiologic
<*
study of Blair et aj_. (1979). (An analysis of animal data gave "limited
evidence" of carcinogenicity; evi'dence for any activity of PCE in short-term
tests was "inadequate" to judge its carcinogenic potential) (IARC, 1982). The
U.S. EPA Health Assessment Document for Tetrachloroethylene (Perchloroethylene)
(EPA, 1985a) has also analyzed the evidence of carcinogenicity of PCE. this
evaluation included an extensive review of short-term test results, data from
animal tests, and several epidemiological studies. EPA concluded that the
evidence for the carcinogenicity of PCE in animals is "limited," and that the
epidemiological data were inconclusive. PCE was placed in Group C, a possible
human carcinogen (EPA, 1985a).
75
-------
It should be noted however, that since these two analyses were published,
other epidemiologic and animal study results have become available. It is
possible that a subsequent evaluation by either the U.S. EPA or IARC would
result in a different conclusion.
76
-------
6. HUMAN EXPOSURES TO PCE FROM CONTAMINATED WATER SUPPLIES
In this section, we describe the procedures we use to assess human
exposures attributable to contaminated water supplies. Efforts to assess
human exposure to contaminated drinking water have revealed that significant
exposures to volatile organic compounds (VOC's) can occur from pathways other
than the direct ingestion of water. Several researchers have investigated the
relative importance of a variety of VOC exposure routes in the home from use
of contaminated water supplies (Cothern et a_L, 1986; Shehata, 1985; Andelman,
1985). In addition, there have been studies of the contribution to indoor
exposures of waterborne radon-222, another highly volatile substance (Hess
et a].., 1982; Prichard and Gesell, 198T). .These studies indicate that exposure
to volatile chemicals from routes other than direct ingestion of fluids may be
as large as or larger than exposure from ingestion alone. These other routes
include inhalation from indoor air of contaminants mobilized by showers, baths,
toilets, dishwashers, washing machines, and cooking; ingestion of contaminants
in food; and dermal absorption of contaminants while washing, bathing, and
showering. We divide exposures attributable to contaminated ground water into
ingestion, inhalation, and dermal absorption pathways. Our discussion of human
exposure to PCE is divided into three subsections. In the first subsection,
we provide a background on the general approach we use for assessing human
exposure to VOC's. This approach addresses relative contributions from
ingestion, inhalation, and dermal absorption. Our focus in this subsection is
on the type of information that is needed for risk assessments. The second
subsection covers the method used to estimate ingestion, inhalation, and
dermal-absorption dose factors. These factors convert .water concentrations in
mg/L into human-population exposures in mg/kg-d. The third subsection presents
our calculations of the magnitude and distribution of human exposures
attributable to PCE contamination in California groundwater supplies.
BACKGROUND ON HUMAN-EXPOSURE ESTIMATES .
The purpose of J:he human-exposure estimate is to provide a distribution
of population dose of a chemical from the various phases of the environment.
Three primary pathways must be addressed — inhalation, ingestion, and dermal
77 ".
-------
absorption. The exposure estimates form the basis for determining the absorbed
doses, which are expressed as the amount of chemical (in mg) absorbed or
metabolized per unit body weight (in kg).
* - * •*
Exposure. Dose, and Risk '
Exposure refers to human contact with a chemical or physical agent.
Exposure can be expressed in terms of a concentration, such as the airborne
level of PCE in mg/m3, or in terms of the quantity that comes in contact with
the human system through lung, gut wall, or skin, expressed in mg/d or mg/kg-d.
An individual breathing 20 m3/d of air containing 1 mg/m3 is exposed to
20 mg/d. Dose, or dose rate, expresses the amount of chemical actually
absorbed into the body where it can be subsequently metabolized and/or
transported to other tissues. Our risk estimates are based on the equivalent
lifetime dose rate expressed in mg/kg-d absorbed througlf the lung, skin, or
gut.
For a risk-based approach, in which there is uncertainty about the level
of exposure and the dose-response function, it is appropriate to employ a
stochastic approach for estimating the incidence of health effects within a
population. As recommended by the United Nations study on population
exposures to ionizing radiation (UNSCEAR, 1977; UNSCEAR, 1982), a general
approach for estimating the incidence of health effects within an exposed
population is obtained from an expression of the form:
J* dD / dR(D)n(D) p[R(D)]
o o
(6-1)
where
H = expected number of health effects within the population;
D * absorbed dose, mg/kg-d;
R(D) - risk factor that expresses the lifetime probability of health
effects at a dose level in the range D to D + dD, kg-d/mg;
n(D) = number of people receiving a dose level in the range D to
D + dD; and
pCR(D)] m probability density function that expresses, the probability
that the dose/response function at dose level D has a value
between R(D) and R(D) + dR(D).
78
-------
If exposure levels are low, such that the risk factor R(D) can be approximated
by a linear function that is independent of dose rate D, then R(D) ~ q1,
where q^ is the cancer potency in kg-d/mg. (The derivation of q] is taken
up in Section 7.) Nith a linear dose-response function, Eq. 6-1 becomes
H - q S dD n(D)
o
(6-2)
Our goal in this section is to estimate the distribution of n(D) for water-
borne PCE exposures within the population of California. For the three .
pathways we consider—ingestion, inhalation, and dermal absorption—the total
dose rate is given by the expression
(6-3)
where
D - total lifetime average absorbed dose rate, mg/kg-d;
i m index referring to pathway (l=ingestion, 2=inhalation, 3-dermal
absorption); and
a, - absorption factor for the ith pathway, unitless; and
E. - daily average lifetime exposure by the ith pathway, mg/kg-d.
Anatomical and Dietary Parameters for Humans
Our human-exposure estimates are structured to provide input to a risk
assessment. To achieve this we must determine the lifetime average exposure
within the human population in terms of the daily intake per unit body weight,
mg/kg-d. In this subsection we review the information required for these
estimates and provide tables of representative values for infants, children,
and adult males and females.
Table 6-1 lists values of human-body mass and surface area as a function
of age. The lower portion of this table lists the values used in this study
to characterize each of the major age categories — adult, child, and infant.
The values listed include arithmetic means and standard deviations. The
surface area as a function of body weight is calculated from a formula taken
from ICRP (1975):
79
-------
Table 6-1. Human body weight and surface area by age and sex (from ICRP,
I .7 / D / • '- " .
Age
(y)
Newborn
1
2
4
8
12
16
20
40
Infant
Newborn to 2 y
Child
2 to 16 y
Adult
16 to 70 y
Sex
male/female
male/female
male/female
male/female
male/female
male/female
male
female
male
female
male
f emal e
male/female
male/female
male
female
Arithmetic mean + one standard
SA . 4W + 7
iM W + 90
in which
SA - surface
W - body we
2
area, m , and
ight, kg.
Mass (kg)*
3.5 ±
10 ±
12 ±
18 ±
26 ±
41 ±
62 +
55 ±
70 +
58 ±
75 ±
62 ±
8.5 ±
32 ±
73 +
60 ±
deviation
0.6
2
2
2
5
8
8
8
10
9
10
10
3'
16
10
9
•
Surface areaa
(m2)
0.22 ±
0.47 ±
0.54 ±
0.73 ±
0.96 ±
1.3 +
1.7 ±
1.6 +
1.8 ±
1.6 ±
1.8 +
1.7 +
' 0.42 + 0
1 .1 ± 0
1.8" + 0
1.6+0
0.02
0.07
0.06
0.06
0.1 .
0.2
0.1
0,1
0.1
0.1
0.1
0.2
.1
.4
.1
.1
(6-4)
The standard deviation in surface area is calculated as the product of the
derivative of surface area with respect to body weight (W) and the standard
deviation in body weight:
80
-------
7SA
1 ^
90
4H + 7
(H + 90)'
'w
(6-5)
In Table 6-2 we present values of the hourly breathing rate by age and
activity level. Also provided are the daily average breathing rates based on
the time spent at' rest or awake and the daily average breathing rate per unit
body weight. These values represent the volume of air that enters and leaves
the lungs within a one-hour period. Based on the information in this table,
we use a single breathing-rate-per-unit-body-weight for male and female adults
of 14 L/kg-h for the daily average air intake.. However, to estimate time-
varying exposures we assume that the infant, child, and adult, respectively,
breathe 29, 24, and 17 L/kg-h during waking hours and 11, 9, and 6.1 L/kg-h
while resting.
Table 6-3 lists reference values for intake of fluids by infants,
children, and male and female adults. We have listed the information by fluid
source. Tap water refers to direct consumption of tap water. Other sources
of fluids refer to all intakes of fluids, exclusive of milk and direct tap-
water consumption. It should be noted that intake of fluid in beverages such
as coffee, tea, and soft drinks may consist of indirect tap-water consumption.
Also listed in Table 6-3 is the fluid intake per unit body weight. For infants
and children, we find this ratio to be relatively high, 0.11 and 10.044 L/kg-d,
respectively. For adults, the ratio of intake varies from a typical value of
about 0.026 L/kg-d to values as high as 0.05 L/kg-d at high environmental
temperatures or for moderately active adults (ICRP, 1975). Appendix B provides
reference values for water consumption in an arid environment. The values of
water consumption discussed in this section are intended for use in assessing
chronic exposures. Values discussed in Appendix B apply to maximum acute
exposure.
Table 6-4 is a summary of the range of values that can be used to estimate
age-dependent intake of various food types. For each category in which a range
of values is listed, the lower value corresponds to a median or "typical" value
associated with the reference adult or child. The upper value corresponds to
a high annual average rate of intake. The high annual average is the value
recommended by the U.S. Nuclear Regulatory Commission (1977) to determine
exposure of the maximally exposed individual from routine radioactive releases.
81
-------
ICRP? ?975).ReferenCe breath1n9 rates f°r infants, children, and adults (from
Activity
Infant Child Adult Adult
, /i1 #,, (1° y> female male
L/h (h/d) L/h (h/d) L/h (h/d) L/h (h/d)
Working, light activity,
or recreation
Resting
Daily average
Daily average breathing
rate per unit body weight
250 (10) 780 (16) 1100 (16) 1200 (1
160
19
L/h
620 850
L/kq - h
19
14
6)
93 (14) 288 (8) 360 (8) 450
(8)
950
13
Table 6-3. Fluid intakes for infants, children, and adults (from ICRP, 1975).
Milk
Tap water
Othera
Total fluids
Fluid intake per unit body weight
a Includes tea, coffee, soft drinks, beer, and other beverages.
Infant
(1 v)
Child
(10 v)
Adult
female
Adult
male
L/d
0.9
0.9
t .0.11
0.45
0.2
0.75
1.4
0.044
0.2
0. 1
1.1
1.4
L/kq-d
0.023
.0.3
0:15
1.5
2.0
0.027 .
82
-------
Table 6-4. Food intake for infants, children, and adults.
Food type
Fruits, vegetables, and grains
Milk and dairy products
Milk fat
Meat and poultry
Freshwater fish
Saltwater fish and other seafood
Infant
(1 V)
-.-
0.90
0.036
—
-r
-'-
Child
(10 V)
kg/d
0.60 to 1.8
0.51 to 1.1
0.021 to 0.044
0.13 to 0.18
0.0077 to 0.044
0.0031 to 0.054
Adult
(male and female)
—
0.52 to 1.6
0.30 to 0.85
0.012 to 0.034
0.21 to 0.30
0.014 to 0.057
0.0074 to 0.071
a The range of values expressed here.reflects the range of values
oublished in ICRP-23 (ICRP, 1975) and in Reg Guide 1.109 (U.S. NRC, 1977).
The lower value expresses the "typical" value; the upper value represents that
suggested to calculate maximum annual average exposures.
WATER-BASED PATHWAYS AND DOSE FACTORS
In this section, we develop the unit-pathway-dose factors for contaminated
tap-water supplies. We consider three pathways—water ingestion, inhalation,
and dermal absorption. For a given pathway,' i (i.e., 1 = ingestion,
2 - inhalation, and 3 - skin absorption), the unit pathway dose factor, F.,
translates the water-supply concentration, Cw, in mg/L into an equivalent
lifetime absorbed dose rate, D., in mg/kg-d,
Ji
Ficw
(6-6)
Because we are interested in the equivalent lifetime dose within a population
composed of three age categories (infant, child, and adult), we calculate the
overall dose factor, F., as the weighted sum of the pathway-dose factors,
fi (age group), for each of the three age categories.
54
F » y -f,(Infant) + f^child) + f,(adult)
(6-7)
83
-------
In this expression, the. factors 2/70, 14/70, and 54/70 reflect the fraction of
time the population cohort spent in each of the age categories. He also
assume that the population is*stationary.
Water Ingestion
For the water-ingestion pathway (i = 1), the unit-pathway-dose factor, f.,
for each age category is obtained by dividing daily water intake by body weight
and multiplying by an absorption factor, a1. The ratio of fluid intake to
.body weight for each age group comes from Table 6-3. We also use data compiled
by the ICRP (1975) on fluid intake by adults at high environmental temperatures
and during moderate activity. The ICRP reports that at high environmental
temperatures (to 32°C) adults consume 2.8 to 3.4 L/d of fluids and that
moderately active adults can consume 3.7 L/d. Using an average adult weight
of 66.5 kg, from Table 6-1, we calculate that this corresponds to a fluid
intake as high as 0.056 L/kg-d. We make the conservative assumption that all
fluids consumed by the members of a household with contaminated water are at
the same concentration level. The absorption of PCE across the gut wall is
assumed to be 100% (see Section 3). We further assume that a reasonably
conservative estimate of the pathway dose factor can be bracketed using the
lifetime average fluid intake per unit body weight with 2-L/d adult fluid
intake as the reference value and the fluid intake by moderately active adults
as the upper bound. Thus; F, is bracketed by
y§ x 0.11 + 75 x 0.044 + |$ x 0.025
0.031 mg/kg-d per mg/L, and
ff
7§ x 0.11 + j^ X 0.044 + |4 x 0.056
*>
0.055 mg/kg-d per mg/L.
(6-8)
(6-9)
The reference value, 0.031 (mg/kg-d)/(mg/L) is similar to the value
obtained under the assumption that intake of drinking water over a lifetime-
approximates 2 L/70 kg or 0.028 (mg/kg-d)/(mg/L). Cothern et al_. (1984) report
that a weighted average derived from water-consumption curves gives a lifetime
84
-------
dose factor on the order of 0.034 (mg/kg-d)/(mg/L). Our upper-bound limit
corresponds to a lifetime average daily fluid intake of 3.8 L/70 kg. To date,
population studies on the variability of water consumption have not been
conducted in the United States (Cothern et a].., 1984). However, the Canadian
Environmental. Health Directorate (1981) has studied the variability of water
consumption in Canada and found 13-16% of adults may consume more than 2 L/d
of fluid. It is not clear how these results apply to drinking-water
consumption in California, but the numbers do suggest it is usefuj to consider
the upper-bound dose factor in ex'posure estimates.
Inhalation Exposure
Several researchers have addressed the relative contribution of the
respiratory pathway to overall human exposures from VOC's in tap water. All
have found that volatile compounds in water supplies can result in inhalation
exposures that are comparable to the ingestion pathway. Prichard and Gessell
(1981) found that public water supplies provide a major pathway for indoor
exposures to radon and measured the amount of radon transferred to air from
water during various household activities. Cothern et aj.. (1984) found tlrat
the respiratory uptake of VOC's from household air attributable to tap wafer
is approximately equal to oral uptake from fluids. • An.de 1 man (1985) developed
a model shower that he used to study human exposures while showering or
bathing. He found that for a volatile pollutant, overall indoor inhalation
exposure may be as much as six times higher than direct ingestion exposures.
Foster and Chrostowski (1986) have developed an integrated household exposure
model for assessing human uptake of VOC's from tap water. Using this model,
they have estimated that the ratio of hjjman uptake by inhalation to uptake by
ingestion is 1.5 for TCE.
McKone (1987) has developed a model that describes the daily
concentration profile of VOC's within the various components of the indoor air
volume of a dwelling. For the mode.l, the indoor air volume is divided into
three compartments—the shower/bath stall, the bathroom, and the household
volume. We use this model to calculate two bounding pathway dose factors that
correspond to the use of contaminated water in the indoor environment. One
corresponds to the average daily lifetime dose absorbed by an individual
85
-------
living in a "typical" California household. The other represents an upper
bound estimate of exposure and dose in which the model parameters are set at
values that provide the upper limit on exposure and dose estimates. For the
typical home, we assume that the household has four occupants and uses 900 L/d
of water contaminated with 1 mg/L of PCE and that PCE has water-to-air transfer
properties similar to radon-222. The water-to-air transfer .efficiency for PCE
is calculated as the product of the efficiency measured for radon and the ratio
of the mass-transfer coefficient for radon to the mass-transfer coefficient for
PCE. The derivation of this estimation technique is found in McKone (1987)".
The time-dependent concentration profile of PCE in shower stall, bathroom;
and household air and the resulting effective lifetime doses were estimated
using two sets of assumptions. These assumptions were intended to define
typical values and a likely upper limit on dose. The two sets of assumptions,
are listed below.
Assumptions used for typical doses:
Occupants spend 100% of their time in the house from 11:00 pm to
7:00 am.
• Bathroom is used for showers/baths from 7:00 am to 8:00 am.
Each adult and child spends 20 min in the bathroom during the period
from 7:00 am to 9:00 am. ,
Each adult and child spends an• additional'20 min in the bathroom
during any 22-h period (excluding the hours 7:00 am to 9:00 am).
• Each adult spends 10 min in the shower or bath.
Adults spend 25% of the time from 7:00 am to 11:00 pm in the house.
Children spend an average of 20 min/wk in showers or baths.
Children spend 60% of the time between 7:00 am and 11:00 pm in the
house.
Infants spend 100% of their time in the house and 2% of that time in
a bathroom.
Fifty percent of the PCE inhaled is available for pulmonary uptake.
Assumptions used for upper-bound doses:
• Each adult and child spends 40 min in the bathroom between 7:00 am
and 9:00 am.
The bathroom is used for showers/baths from 7:00 am to 8:30 am.
86
-------
• Each adult spends 20 min in the shower or bath. . ,
• All age groups spend 100% of their time in the house.
• Children spend an average of 40 min/wk in showers or baths.
• Absorption of PCE in the lung for all age categories is 100% of that
inhaled.
Using this model, we estimate that the pathway dose factor for inhalation
is in the range
F2 « 0.041 mg/kg-d per mg/L for typical households, and
F0 - 0.16 mg/kg-d per mg/L for an upper-bound estimate
(6-10)
(6-11)
(but note the modification of Eq. 6-10 required for the application appearing
at the end of Section 7). These numbers are based on the assumption that an
adult showers every day and that children bathe every second day. Table 6-4
summarizes the relative contribution to the pathway dose factor, F2> from
each age category and household compartment. For adults or children who take
baths instead of showers these numbers are likely to be reduced somewhat. He
have not examined the extent of reduction that taking baths in place of
showers would give. Table 6-4 reveals that exposures to adults in the shower
and bathroom are the major contributors to indoor inhalation exposures
attributable to contaminated water. A local sensitivity analysis of the
different parameters involved in the calculation of the pathway dose factor
for the inhalation route of exposure has been prepared in McKone (1987).
According to that analysis, the pathway dose factor was most sensitive to 1%
increases in the following parameters: the absorption fraction in the lung,
the transfer efficiency of a VOC in shower water to air, the water use of
individual's in showers, and the ratio of breathing rate to body weight (adult).
Although the assumption that adults spend 25% of the time from 7:00 am to
11:00 pm in the house may seem low, we believe this is a plausible value for a
typical adult who spends 10 h per day in work and travel, leaving 6 h of
leisure time of which we assume roughly 2/3 is actually spent in the home.
Furthermore, we assume that roughly 3/4 of all adults work outside the home
87
-------
and that those who do not can be accounted for by the upper bound estimate.
Finally, we have found that the pathway .dose factor is very insensitive to
this occupancy factor.
When calculating "typical doses," we assume that only 50% of the PCE
inhaled is available for pulmonary uptake in order to account for the
observation that.the alveolar volume of the lung — which can be considered
the actual interface between the human system and inhaled air — is
approximately one-half o*f the resting total lung capacity (ICRP, 1975). This
value of 50% is actually somewhat conservative in that it does not take into
account the effect of dead space and lung expansion during respiration. For
example, assuming an average tidal volume of approximately 1.0 L, dead space
of 0.16 L, alveolar volume of 3.0 L, and resting total lung capacity of 5.6 L
(ICRP, 1975), the actual percent of inhaled air ventilated through alveolar
space (assuming instantaneous.equilibrium between alveolar and non-alveolar
air, excluding dead -space) would be approximately 35% to 40% (or from 4.9 to
5.6 L/min assuming a 20 m /d respiration rate).
• Table 6-5 compares the ratio of inhalation intake to ingestion intake
(based on 2 L/d per 70 kg) projected using the McKone model and compares this
to the value of this ratio that has been estimated by other researchers.
Dermal Absorption *
We reviewed the literature on absorption rates of volatile solvents having
direct contact with the skin to estimate the likely value of dermal absorption
from normal daily use of contaminated water. Over the last 20 y, several
investigators have examined the transport of dissolved chemicals across the
skin (Stewart and Dodd, 1964; Riihimaki and Pfaffli, 1978; Brown et a].., 1984;
Cothern et al... 1984; Scheupleih and Blank, 1971; Bronaugh, 1985; and Foster
and Chrostowski, 1986). Although a complex process, dermal uptake of
compounds occurs mainly through passive diffusion through the stratum corneum.
We assume that dermal absorption occurs during bathing and showering. To
determine the pathway dose factor for dermal absorption, we had to make a
number of simplifying assumptions. These are
Resistance to diffusive flux through layers other than the stratum
corneum is negligible.
-------
Table 6-4. Percent contribution to the lifetime average dose factor
(inhalation) from specific age and household exposures.
Exposure
% Contribution
Typical household pathway dose factor
Adult, in
fn
in
Child, in
in
in
Infant, in
in
Upper-bound
Adult, in
in
in
Child, in
in
in
Infant, in
in
shower
bathroom
remainder of house
shower
bathroom
remainder of house
bathroom
remainder of house
pathway dose 'factor
shower
bathroom
remainder of house
shower
bathroom
remainder of house
bathroom
remainder of house
51
20
7.9
5.2
7.2 .
• 7.0
0.17
1.2
100
' 42
25
14
4.2
9.0
4.9
0.16
0.56
100
Table 6-5. Estimates of the ratip of indoor inhalation dose to ingestion dose
for a unit concentration of 1 mg/L.
Inhalation Uptake (mg/kg-d)
Ingestion Uptake* (mg/kg-d)
1.4 to 5.0
1
6
1.5
Compound
PCE
VOC
"Volatile pollutant"
TCE
Reference
LLNL Model (McKone, 1987)
Cothern et aK (1984)
Andelman (1985)
Foster and Chrostowski
(1986)
Assuming 2 L/d for a 70-kg individual.
89
-------
• Steady-state diffusive flux is proportional to the concentration
difference between the skin surface and internal body water.
• An adult spends from 10 to 20 min in a bath or shower each day.
During,bathing, roughly 80% of the skin is in contact with water,
and during showers, roughly 407/of the skin is in contact with water.
Children and infants spend approximately 1 h/wk in bathing or
swimming (U.S. NRC, 1977).
The absorbed dose from dermal absorption is given by the expression
fs SA
(6-12)
where
T
fs
SA
= steady-state flux across the stratum corneum, mg/cm?-h;
= duration in the shower or bath, h;
- fraction of the s'kin surface in contact with water, unitless; and
• surface area of the skin, cm2.
Chemical transport across the skin is assumed to follow Pick's law, so that
the flux J$ across skin tissue is given by
(6-13)
where
AC.
permeability constant across the stratum corneum, L/cm2-h; and
concentration difference of the solute across the tissue, mg/L.
Brown et a].. (1984) have determined, from an analysis of chemical transfer
through the skin layer, that Kp is on the order of 0.001 L/cm2-h for VOC's.
They used this value to characterize Kp for six measured. skin-absorption rates
on four different chemicals. In these experiments, K ranged from 0.0006 to
0.001 L/cm -h. For dilute solutions, AC$ is approximate.^ equal to the
chemical concentration at the skin surface. However, the concentration at the
skin surface is not necessarily the same as the concentration in the water
supply. For showers, we assume that C
Cw, the PCE concentration in tap
90
-------
water. But for bathing, in which water stands for a period of time, we use
C » C where C is the average water concentration over the period of the
bath. Assuming an exponential loss of one-half of the dissolved PCE over a
period of 10 min, we find Cw = 0.72 enduring a 10-min bath and
"w
0.54 C during a 20-min bath (Foster and Chrostowski, 1986).
W
We substitute Eqs. 6-12 and 6-13 into Eq. 6-7 to obtain a lifetime
equivalent dose factor for dermal absorption, .
70 (^ x'fs x m x Vinfant
(
-------
Higher estimate (assuming 20-min bath for adults):
F-. = 0.001 —5— fi <°-17 x °-80 x 490 '- * 0.72 CJ
J cm -h 70 d Kg w
(0.17 x 0.80 x 340 - x 0.72 GW> +
.g (0.33 x 0.80 x 260 - x 0.54
per
= 0.037
(6-16)
Summary of Pathway Dose Factors
Table 6-6 provides a summary of the three pathway dose factors. We list
our best estimate and the "high-average" or upper bound values. The pathway
dose factors are used to convert the water-supply concentration of PCE, Cw,
(mg/L) Into an equivalent lifetime average dose rate in mg/kg-d. Thus, the
pathway dose factor serves to account for the exposure conditions within a
population and the absorbed dose corresponding to an exposure for each pathway.
Table 6-6'also lists the lifetime average fluid intake of contaminated
water that would give the equivalent dose to a 70-kg adult. These values are
listed for comparison. Finally, Table 6-6 lists the percent of the lifetime
equivalent dose attributable to contaminated water that is contributed by each
pathway. '
HUMAN EXPOSURES TO PCE-CONTAMINATED GROUND WATER
In this .subsection, we estimate the integrated population dose
attributable to PCE contamination in California groundwater supplies. To
determine the population doses, we focus on the distribution of contamination
in household water supplies. The pathway dose factors derived above are used
to translate household water-supply concentrations into equivalent lifetime
doses in mg/kg-d. We use AB1803 survey data (CDHS, 1986) to estimate the
distribution of household water-supply concentrations of PCE. Following this,
we use the distribution of household-supply concentrations and dose factors to
calculate the distribution of doses within the California population.
92 *
-------
Table 6-6. Summary of the pathway dose factors for PCE.
Pathway
Fluid Indoor Dermal
ingestion inhalation absorption
Total
Variable
FT
Best estimates:
(mg/kg-d)/(mg/L)
0.031
Equivalent lifetime daily fluid
intake by 70-kg adult, L 2.2
Percent of total 31
Upper bounds;
(mg/kg-d)/(mg/L) 0.055
Equivalent lifetime daily fluid
0.040
2.8
41
0.15
0.028
2.0
28
0.037
0.099
6.9
100
0.24
intake by 70-kg adult, L
Percent of total
3.8
23
10.5
62
2.6 17 .
15 ' 100
At present, the only data available on the distribution of PCE in
household supplies are the AB1803 data (CDHS, 1986). It should be recognized
that this data base has a number of limitations with regard to estimating PCE
concentrations in water supplies. Among the major limitations are • '
• Incomplete sampling data (even though all large systems were
sampled, only a limited number of wells in each system were sampled)
• Limited data on how well-water concentrations change with time.
• No information on the percent of system water supplied-by an
individual well.
• No information on the relation between concentration in individual
wells and the concentration in water-supply lines that enter
individual households.
In spite of these limitations', we estimate the distribution of human doses in
California using the AB1803 data and two sets of assumptions. The first set
of assumptions is
93
-------
Surface water and negative samples are assumed to contain no PCE.
r * A11 We11s within a given water system are assumed to flow at the
same rate.
The average concentration of PCE in unsampled wells is the same as
the average concentration of PCE in sampled wells.
Under this set of assumptions the daily dose rate to people in a given water
system is given by the expression
i =Ci fwi
where
(6-17)
'wi
» F2' F3
daily average lifetime dose rate to individuals obtaining
water from water system i, mg/kg-d;
average concentration of PCE in all wells of system i that
were tested for PCE and are still in use, mg/L;
fraction of the water in system i that is taken from
groundwater supplies (obtained from Appendix C), unitless;
and
pathway dose factors derived above.
The second set of assumptions is the same as the first, with the exception that
Unsampled wells are assumed to contain no PCE.
Under the second set of assumptions the daily dose rate to people in a given
water system i takes the form
* -
+F2.+F3)
(6-18)
where the parameters are the same as above and the additional parameters N.
and N. are the number of wells tested and the total number of wells in
system i, respectively.
9.4
-------
-------
B.4
EXPOSURE ASSESSMENT
Attribute O Central estimates and upper and lower bounds on exposures
or, ifpossibk, the full population distribution of exposures are
described; any preferred estimates are noted, together with
supporting documentation.
SOURCE Case Study K. Tetrachioroethylene (Pages 95-96).
Note Support for the upper bound and "best" estimates can be inferred from
the step-by-step derivation shown in Exposure Assessment Attribute 5.
-------
UCRL-15831
Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
K. T. Bogen, L. C. Hall, T. E. McKone,
D. W. Lay ton, and S. E. Fatten
Environmental Sciences Division
Lawrence Livermore National Laboratory
University of California
P.O. Box 5507
Livermore, CA 94550
April 10,1987
*•
Prepared for
California Public Health Foundation
P.O. Box 520
Berkeley, CA 94701
-------
Table 6-7 shows the distribution of PCE exposures among the California
population as calculated from Eqs. 6-17 and 6-18. The table was constructed
using AB1803 data to obtain the average concentration in tested wells and the
data compiled in Appendix C on the number of people associated with each water
system.
We also used Eqs. 6-17 and 6-18 to calculate integrated population dose
rates attributable to sampled water systems. The integrated population dose
rate (IPO) in person-mg/kg-d is calculated by summing the product of the daily
dose rate and the population over all systems having positive PCE
concentrations
IPD
(6-19)
where
IPD = integrated population dose rate to PCE from large water systems
in California, person-mg/kg-d;
Pi =• population associated with water system i;
D. = lifetime daily dose rate to PCE in system i, calculated using
Eq. 6-17 or 6-18, mg/kg-d; and
n - number of large systems with positive PCE measurements.
Table 6-8 provides a summary of the integrated population dose as it is
calculated, based on the two sets of dose factors and two sets of assumptions
regarding the interpretation of sample data. It should be noted that these
population doses do not include contributions from surface water and private
wells.
95
-------
Table 6-7. Distribution of population doses to PCE from ground*water in
California.
System average dose, mg/kg-d
Population receiving the corresponding dosea
Using best-estimate Using upper-bound
dose factors dose factors
Assuming only positive test wells
are contaminated
10"7 < D < 3 x 10~6
3 x 10~6 < D < 10~5
10~5 i D < 3 x 10~5
3 x 10~5 1 D < 10"4
•1(T4 < D < 3 x 10~4
3 x l(f4 < D < 10~3
10~3 < D < 10"1
10"7 £ D < 3 X 10~6
3 x 1(T6 < D < 10~5
10~5 < D < 10~4
10~4 < D < 3 x 10"4
3 x TO"4 < D < 10~3
10" 3 < D < 10"1
1,000,000
4,400,000
870,000
470,000
60,000
800
, —
Extrapolating average
tested wells
800,000
760,000
5,039,000
. 140,000
77,000
800
710,000
510,000
4, 300 ,-000
950,000
240,000
59,000
800
concentrations in
to' all wells
•
960,000
1,920,000
3,800,000
71 ,000
29,800
a The sum of each column is the number of people who obtain water from a
system having one or more contaminated wells.
Table 6-8 Summary of integrated population dose rates in person-mg/kg-d
calculated from Eq. 6-19.
Integrated population dose (IPD) in person-mg/kg-d
Calculated
using Eq. 6-17
for 0
Calculated
using Eq. 6-18
for 0
Based on best-estimate dose factors
Based on upper-bound dose factors
250
610
92
220
96
-------
B.4
EXPOSURE ASSESSMENT
Attribute 7 Uncertainties in the estimates are described, and the relative
importance of key assumptions and data is highlighted.
SOURCE Case Study D. Formaldehyde (Pages 6-1 to 6-47).
Note None of the case studies presented a cumulative highlighting of key
assumptions or a particularly quantitative treatment of uncertainty.
Excerpts from the formaldehyde report illustrate a qualitative descrip-
tion of factors that would result in variations from the upper bound
estimates.
-------
Assessment of Health Risks
to Garment Workers and Certain Home Residents
from Exposure to Formaldehyde
April 1987
Office of Pesticides and Toxic Substances
U.S. Environmental Protection Agency
-------
6- EXPOSURE ASSESSMENT
6.1. Introduction
The sources of HCHO can. be grouped into two major
categories: commercial production and indirect production. 'The
chemical is not 'imported in. any appreciable quantities.
Commercially, HCHO is produced from the catalytic oxidation
of methanol, using either silver oxide or a nixed-metal oxide as
the catalyst. Processes accounting for the indirect production
of HCHO include the photochemical oxidation of airborne
hydrocarbons released from incomolete combustion processes, the
oroduction of HCHO during incomplete combustion of hydrocarbons
in fossil fuels and refuse, and certain natural processes.
.The 1984 commercial production of HCHO amounted to about 6
billion pounds. The major derivatives are urea-HCHO resins,
phenol-HCHO resins, acetal resins, and butanediol. The urea- a-.d
phenol-HCHO resins account for about 53 percent of HCHO
production. Adhesives and plastics.are the major end uses.
The "consumption" of HCHO can be broken down into three
major categories: nonconsumptive uses, pseudo-consumptive uses,
and consumptive uses. In nonconsumptive uses, the chemical
identity of the HCHO does not change. , In pseudo-consumptive
s&
uses, the chemical identity of HCHO does change, .but it is not
irreversibly altered. Under aporopriate conditions, some or all
of the original HCHO may be regenerated. Consumptive uses, on
the other hand, are those uses in which HCHO serves as a -
feedstock for the preoaration of other chemicals. The
6-1
-------
derivatives are irreversibly formed and usually contain only
residual levels of unreacted HCHO. Under extreme conditions,
such as very high temperatures or highly acidic conditions, some
of the derivatives may degrade and release HCHO.
HCHO's major nonconsumptive uses are (1) disinfectant, (2)
preservative, (3) deodorant, and (4) textile and pacer uses.
'The major pseudo-consumptive uses are (1) urea-HCHO resins
which are used in flberboard, particleboard, plywood, laminates,
urea-HCHO foams, molding compounds, and paper, textiles, and
protective coatings; (2) urea-HCHO concentrates which are used to
produce time-release fertilizers, and (3) hexamethylenetetramine
which is used as a snecial anhydrous form of HCHO to cure resins
and to treat textiles/and rubber.
The major consumptive uses are (1) melamine-HCHO resins
which are used for molding, compounds, fiberboard, particle'ooard,
plywood, laminates, paper'.and textiles, (2) phenol-HCHO resins
which are used in fiberboard, particleboard, plywood moldina
compounds, and insulation; (3) nentae.rythri tol which is used to
produce alky.d resins, (4) 1,4-butanediol which is used to produce
tetrahydrofuran, (5) acetal resins which are used in the
manufacture of engineering plastics, and (6) trimethyloloropane
which is used in the production of urethanes.
6.2. Estimates of Current Human Exposure
To obtain estimates of human exposure to HCHO, the Agencv
commissioned a contractor study (Versar, 1982). This studv
integrated the existing monitoring data, engineering or
6-2
-------
estimates, use data, population estimates, and assessment of 'the
likelihood of exposure from HCHO-related activities into an
exposure assessment detailing those .activities having a high HCHO
exposure potential.„. EPA updated some portions of this assessment
to reflect new data received in response to the FEDERAL'REGISTER
notice of November 18, 1933 and other data gathered by EPA. The
combined data were used as. the basis for the May 1985 draft risk
asses sment. ,
Subsequent to the draft risk assessment, the Agency
commissioned additional contractor studies to assess garment
f
worker (PEI, 1985-) and residential (Versar, 1986a,b,c) exposure
to HCHO in more depth. The exposure estimates from these reports
were used as the primary basis for this risk assessment. The
conclusions of these contractor reports are summarized in this.
document; more detailed information regarding exposure can be
obtained by referring to the contractor reports.
6^3. Populations at Risk
The two populations at risk.-examined here are certain home
residents and garment workers.
6.3.1. Home Residents
Based on a projection of- manufactured housing starts by
Schweer" (1987), it is estimated that•7,800,000 persons may occupy
new manufactured homes during the next ten years. This figure
assumes 295,000 starts per year and 2.64 persons per home.
Similarly, ah estimated 214,000 new conventional homes
containing significant- quantities of pressed wood products as
construction materials will be started each year for the next ten
6-3
-------
years with an occupancy rate of 2.95 persons for a total of
6,310/000 persons.
6.3.2. Garment Workers
The number of potentially exposed garment workers is
estimated to be 777,000 (Versar, 1982) out .of 1,100,000 workers
employed in the U.S. apparel industry (Ward, 1984)'. -This figure
may drop in the future due to increased foreign competition and
the introduction of labor saving equipment.
6.3.3. Summary
*
Table 6-1 presents population, estimates for the two housing
segments. Assuming'that the number of potentially exposed
garment workers remains steady at 777,000, then a total of almost
15,000,000 persons over the next ten years may-have the potential
to be exposed to elevated levels of HCHO.
Category
Manufactured homes
Conventional homes
* Schweer (1987)
Table 6-1.
POPULATIONS 'AT RISK
Population
Estimates
per yr
779,000
631,000
10 yrs
7,790-, 000
6,310,000
-------
6-4. Souroaa of HCHO in Population Categories of Concern
The principal sources of HCHO in the two population
categories of concern are HCHO-based resins, principally urea-
HCHO (UF) resins. In homes, these resins are used -to bond .the
* ...
wood plys used, to make plywood and to bind the wood particle and
fibers used to make particleboard and medium density
fiberbpard. For garments, HCHO-based resins are used to impart
permanent press finishes to the. garments.
6.4.1. Homes Containing Pressed-Wood Products
6.4.1.1. Pressed-woo.d product descriptions
Pressed-wood products are used in flooring, interior walls
and doors,, cabinetry, and furniture. ' The three principal types
of products containing uT-resin are particleboard, medium-density
fiberboard (MDF), and hardwood plywood.
Particleboard is composition board comprised of 6 to 10
percent resin (by weight), and small wood particles. UF resin is
used in the majority of particleboard (about 90 percent of. total
production capacity). The 1983 production of particleboard was
over 3 billion square feet, of which 70 percent was used in
furniture, fixtures, cabinets, and similar products. The
remaining 30 percent was used for construction, including deckinc
in manufactured home manufacture and flooring underlayment in
conventional housing.
Recent data indicate that particleboard is used in home
construction at a rate of 0.16 square feet (ft2) (~ 0.5 ^n2)
cubic foot (ft3) o.f indoor air volume in mobile homes. The
ner
6-5
-------
loading rate (ft2/ft3) in conventional homes is lower on average;
approximately 0.05 ft2/ft3 ("ST 0.17 m2/m3) (see Table 6-2) ,
However, loading rates in conventional homes may vary
considerably from homes that contain-only particlebpard as a
cabinet material to homes whose floors are constructed with
particleboard underlayment.
MDF is also a composition board. It is comprised of wood
fibers and 8 to 14 percent UF resin solids by weiqht.
Approximately 95 percent of MDF production (over 600 million
square feet in 1983) was used to manufacture furniture, doors,
fixtures/ and cabinetry. No data are available on the precise
extent of MDF's use in either mobile or conventional homes.
Unlike the two composition boards discussed above, hardwood
plywood is a laminated product; the resin is used as a qlue to
hold thin layers of wood and veneers together. Of the nearly 4.3
billion sguare feet consumed in 1983, 55 percent was used for
indoor paneling, 30 percent for furniture and cabinets, and 15
percent for doors and laminated flooring.
6-6
-------
Table 6-2 . U» of Pressed-wood Products in How Construction
Cate<»rv
Mew taws
Percent units containing.
Hardwood plywaod.paneling
Particleboard undarlaynwnt
Average loading rates,6 (m^/m3)
Hardwood plywood paneling
Particleboard underlayment
Particleboard shelving*
Particleboard kitchen cabinets
Total particleboard
Honas (Canada) d
Ptrctnt units containing
Particltooard
Av«raot loading rat»s (m2/™3)
Total p«rticj«ooard
Eiistinq Hemrs (U.S.)*
Ptrcont units containing
Hardbood plywood paneling
Parti cleooard
Avcraot loading rat« (m2/™3)
Hardtaood plywood paneling
Particlaboard
7.6
30.5
0.05*
0.118
0.010
0.039
0.167
100
35.5
90.3
0.098
O.OSS
Type o'f taw
TH
9.3
9.2
0.039
0.092
.016
.052
0.
0.
0.160
100
0.145 0.100
f
ff
8.5
1.7
0.049
0.033
0.020
0.059
0.112
100
most.
most
1.0
0.5
100
0.079 0.479
most
most
1.0
0.5
Data rtf1»ct only inttrior us«s of UF pr«s»d wood products.
Loading rates ar* for tnou taws containing tfws* products.
*Sourc«: HPA (1984) and HPHA (1984) for conventional hows - Based on
interpretation of tr» results of a survey of 900 taw builders (103
responses) regarding the extent of use of particleooard and hardwood
plywood paneling in new hones containing these products (NAH8 1984).
bSource:
hows.
«eyer and Hermanns (19eAa), NAH8 (1984). wi (1984) >or mobile
(Footnotes continued on nest page)
6-7
-------
Table 6-2. Footnotes (continued)
of produce surface anea/m3 of indoor air volume.
<%ource: InterArt (1983V - based on in-hcme surveys at 9 SFO, 1 TH, 1 rf
and 1 W. Total loading includas underlaynent, shalving and cabinets.
SFO loading* ranged fron 0.029 to 0.491 n^/m3.
'Sourca: Schutte (1981) - Sased on in-ncme surveys at 31 SFO. Average
loadings based on hows containing those products.
f SFO » Single fanly c^elling
TH • TcMnhouse
HP » ftjltifamily dwelling
m * Mobile hone
6-8
-------
6.4.1.2. HCHO release from oressed-wood products ' -._ -
Each of the pressed-wood products described above contain L'F
resins which release HCHO over'time. The release is attributable
to two basic sources (Podall, 1984):
1. - Free (unreacted) HCHO present as a result of incomplete
crosslinking during resin cure/
2. Decomposition of unstable UF resin or resin-wood
chemical species as a result of their intrinsic '
instability and/or due to hydrolysis.
Free HCHO, which is present in cured resin at low levels (<{
percent).is the most significant source of HCHO release from
pressed-wood products in the initial period after they are
manufactured (Podall, 1984). The specific time period in which
free HCHO dominates releases is not known.
The second source, decomposition and hydrolysis, pertains to
the large proportion of HCHO-bearing species like methylene
ureas, urea methylene ethers, and cellulose-crosslinked species
that may release HCHO for a much longer period of time (Podall,
1984). These species differ in their susceptibility to
hydrolytic attack and decomposition., and their relative rates and
durations of release can only be hypothesized at this time. «...
Release of HCHO from UF-resin containing pressed-wood
products is complex, with numerous interrelated aspects. Th.e
pressed-wood product manufacturing process, and other factors,
affect the amount of each HCHO-releasing species present in the
finished product. The resin formulation has a direct effect on
6-9
-------
release; resins with a low HCHO:urea ratio have, when cured, a ,
lower level of free HCHO but may be less stable and more
susceptible to hydrolysis (Myers, 1984). Other additives to the
resin, such as acid catalysts, chanqe the resin-chemistry and
influence the -release profiles. The conditions under which the
resin is cured affect bond strength, determining to some extent.
the stability of the resin components. The character of the wood
itself also affects HCHO release; the more acidic the wood, the
greater the tendency for acid -hydrolysis and HCHO release
(Podall, 1984).
Under normal use conditions, the release of HCHO decreases
with time, as discussed previously. Emission reductions linked
to product aging relate to a decrease over time in both the HCHO
present in the board as a residual from nanufacturinq and the
latent HCHO present in the board in hydrolyticall.y labile resin
and wood components. The emission rate decay curve for a board
is apparently exponential with time; the resid.ual HCHO is emitted
at relatively high rates followed by a slow, release of latent
HCHO. Althouah the short-term emission rate behavior of boards
*
has been reported in numerous studies, little quantitative
information is available on' the long-term emission rates,
particularly for*newer products made with low HCHO-urea ratio
resins or treated with .scavengers.
6.4.1.3. Other Sources of HCHO
Indoor HCHO concentrations may be attributable to sources
other than pressed-wood products containing UF resin. The other
sources can be characterized as follows:
6-10
-------
o Urea-HCHO foam insulation (UFFI) (existing homes only)
o. Products with phenol HCHO resins (PF)
-- softwood plywood
hardboard
-- • wafer-board •
oriented strand board
fibrous glass insulation
fibrous glass ceiling tiles
o Consumer products that may contain HCHO resins
.— . .upholstry fabric
drapery fabric '
-- other textiles
o Combustion products '
unvented kerosene .a"hd gas appliances
smoke from tobacco products
• combustion of wood or coal in 'fireplace's
o Outdoor air ' . • . .
ventilation system air exchange
Compared to pressed-wood products, with^the exception of
UFFI, the other sources are usually minor contributors to HCHO
concentrations in conventional and manufactured homes.
The Consumer Product Safety. Commission (G-PSC) in-1982
prohibited the installation of UFFI in residential buildings and
schools. Although It was later overturned by a Federal court,
the CPSC ban on UFFI caused the virtual elimination of the UFFI
industry (Formaldehyde Institute, 1984). There is considerable
debate among the regulatory agencies'and the UFFI industry as to
the extent of long-term HCHO emissions from UFFI presently in
place (Hawthorne et al., 1983). UFFI is not discussed in detail
in this section? refer to Versar (1986c) for further information
and references.
Though no residential sources of HCHO have been as' well-
studied as urea-HCHO foam insulation and pressed-wood products
v
made from UF resins, there are fairly complete data on the '.
6-11
-------
importance of pressed-wood products with PF resins, on fabrics
treated with UP resins for permanent press, on fueled appliances,
and on cigarette smoke as sources of residential levels.
Common applications of PF resin pressed-wood products
include roof and wall sheathing, subflooring,'and siding. Small
amounts are used for' shelving, cabinets, indoor paneling, and
«
fixtures (APA, 1934). Phenol-HCHO resins are inherently nore
stable than are UF resins, and pressed-wood products nad-e of ??
resin emit HCHO at much lower rites than do products made with O'F .
»
resins. The small amount of HCHO that is emitted from the panel
products is the result of residual HCHO that remains in the resin
(APA, 1984).
w
There are several published studies on HCHO emissions from
PF pressed-wood panel products. Myers and Nagaoka (1981) found
that HCHO levels in chamber tests rarely exceeded 0.1 ppm in the
presence of PF particleboard at 25°C. Matthews et al. (1983,
reports X-XV) tested PF hardboard and softwood plywood and
obtained similar results. Myers (1983) measured higher levels
(0.3 ppm) initially in tests of waferboard and particleboard made
with PF resins, but levels declined rapidly. The American
Plywood Association (APA,. 1984) has submitted data (reviewed by
Versar, 1986c) indicating that PF-resin pressed wood products
emit little HCHO.
Other generic product lines- containing PF that are used in
construction applications are fibrous glass, insulation and
6-12
-------
ceiling tiles. In 1983, as a result of a study on HCHO release
from consumer products (Pickrell et al., 1982), CPSC decided to
further evaluate HCHO emissions from fibrous glass insulation and
ceilinq tiles. These products, when comoared with other products
tested, were among the hiqhest group of emitters tested by *
Pickrell'ej: a'l. (1982). Concern about these test results arose
because of 'the high loading rates of these'products in homes. *
Under normal use conditions (in attics), insulation would be
subjected to temperatures much higher than normal room
temperatures, thereby increasing ootential HCHO emissions.
Further evaluation by Matthews et al. (1983) and Matthews
and Westley (1983) (under contract to CPSC) indicated that a
predicted increase of no more than 0.022 ppm in indoor HCHO level
would result from.use of new ceiling tiles'or new insulation. As
the- products aqe, the HCHO emission rates and resulting indoor
concentrations would be expected to decline sianificantly.
Available data on treated fabrics (Pickrell et al., 1982,
1984) indicate that, with emission rates only as high as 115 '
ug/m2/hr, these can be relatively important sources in homes only
with large surface areas of furnishings like draperies (at least
when new). The data on combustion appliances show that HCHO
release is a function of whether the applia.nce is tuned and
functioning properly. Gas stoves may emit less . than 2 t.o'nearlv
30 mg HCH'O per hour, of use; ;gas heaters can emit less than 5 to
over 60 mg/hr, depending on"the efficiency of burning; and new '
kerosene heaters emit up to of 3.9 mg/hr of HCHO. (Traynor et al.,
6-13
-------
1982; Girman et al., 1983; Fortmann et al., 1984; Traynor et al. ,
1983; Caceres et al, 1983).
The emissions data on sidestream cigarette stnoke range from
20 ug per cigarette .(Bardana, 1984) to nearly 1.5 rug/cigarette
(reported by Matthews et al., 1984). Several studies, however,
concur on an emission rate of 1.0 to 1.2 mg/c'igarette. The
importance of this source is obviously related to' use patterns.
Studies where numerous persons chain-smoked in a poorly
ventilated room (Timm and Smith, 1979) did indeed show that HGHO
levels were elevated after a short period of time, but other
studies (Traynor and Nitschke, 1984) in the homes of smokers
indicated that, at a smoking rate of 10 cigarettes per day,. HCHO
levels were not elevated over controls with similar loading rates
of other sources.
6.4.2. Garment Manufacture
The principal source of HCHO in the garment manufacturing
workplace, is the release of HCHO from fabric treated with resins
that impart durable or permanent press properties. The textiles
normally treated are blends of'cotton, acetate, and rayon. These
fabrics account for 60-80 percent of the textile produced
annually.
The re«in of choice is dimethyldihydroxyethylene urea
(DMDHEU) and its alkylated derivatives. It is estimated that
approximately 90 percent of the durable press resin market is
accounted for by DMDHEU. Other resins used are urea-HCHO,
melamine HCHO, and carbamate resins, plus a HCHO/sulfur dioxide
vapor phase process.
.. 6-14 ' • ' ' • •
-------
HCHO is released from treated fabric in three phases. In
Phase I, any HCHO loosely held by Van der Waal forces is released
as the fabric is dried. Release of HCHO by this- mechanism is
usually complete by the time garment workers receive the
fabric. Surface desorption occurs during Phase II. This
S?
represents the release of HCHO which is not covalently bound to
the fabric, and can last up to 240 hours. The material is
normally stored during this phase,'and increased ventilation can
increase the rate at which HCHO is desorbed. Phase III, in -which
hemiacetal hydrolysis is the mechanism of release, -is thought to
be the phase of HCHO release which results in worker exposure at
the manufacturing site. Release of HCHO by the hydrolysis
mechanism is. independent of air changes, but dependent on
humidity and temperature (Ward, 1984)
6.5. HCHO Levels in Homes and Garment Manufacturing Sites
6.5.1. HCHO Levels in Homes ' *
Table 6-3 briefly summarizes the residential HCHO monitoring
studies reviewed by Versar (19S6a, c). However, because of the
changing nature -of pressed-wood products with UF resins and the
constant- evolution and improvement in monitoring techniques, this
residential monitoring data base is not the most appropriate for
describing current HCHO exposure in homes. Many data sets are
based on investigation of homes from which complaints of HCHO
symptoms have been filed; these data sets may not be
representative of average exposure because of bias toward high
concentrations. Homes studied before 1980 were built with
6-15
-------
1
1
t
• «•• q
|
•«•
a
1
1
u»
"5
MM*
«J
1
O
y.
I
»5
ri
1
o
O
2
6-
j*
C
3
1
Ol
i
% i
- i
l"
S o
It
ut
at
1
*s
u
J
1
O
1
Study/sapling date(s) Wu
/reference)
^i
e
i%
|
\fi
d
i
d
d "
&
i
i
*
OOUVENIIOUAL HONES
Fleming 1 Associates
^
^
"S
i
j
h
e
•f5
8 M
U -
Primarily
measureme
o
A
«i
d
V-
i
«n
^-
s-
New York Study
(Iraynor t Nitschke 1964)
Univ. Washington (1962-1963)
(Breysse 1964)
• —
-* S
|j
UJ ^
«> j*
e &
•^ i
^» Q
? 2
>. —
2-2
2 7
01 S
Includes
weather \i
the U.S.
Z 2
CM O
d d
2 2
u*i un
l»
!
i
i
is
SS
LBL (1919-present)
(Ginun et al. 1963)
•^5 — /i
^ VI
-"&i.-
•« S j?
-* u S
S c
i -^
s. • •<
5^2
_« -a 'S
- " w
e vi s
1 5 Si
O 2£ vi "O
W VI >»
- ? 1
_aj 5 'a
3 • VI ^
— U 01 1
u * e ^ 9
. «« 4| .S3 » <
S^j -fl * T
is a- !•
a -c s 3 s >,
S *l s & • "§
S* J 22 3
2 «< 3 — O)
-------
I
•~
'I
r*l
1
d
j"»4
_t$
'e-
wt
e
1
£
J
s i
- 8
It!
C Ql
^u ^
S
"a.
5
M
tte
O
|
ut
1
O
Study/sampling date(s)
(reference)
5
j
3
1
' - i
^
2.S
'Z 8
. a. w
i
>•> .••>
** ® 1
O 'fl K
1
1
t
1
*
I
if
if'
1*.
* s
' IT "3 *
^ M
2 *•"
.If | • J "•
§^± -- ™
^ o 2
2. §• i w
«• V i '* •
- s -5 ^
< ^ • U
* ^ si
Is S 5
ill i i!
«> •< — > "S. a
a a -S .£ | v
i I «» f we
si 1 i a
<^< f 9 5 5 -
>,! v v .|
i^I I II
I x S§
d d d d
22 22
0 0 2 0
d d d d
S -. | i
«
rf» . T
|?|
in
2
2
i
S
d
o
2
§
Indiana Board of Health
study (19)9-1983)
(Konopin&ki 1983)
-------
1
^,
i
|
<•*>
1
O
O
"fl
i-1
Connenls
S '
£
£•
ee
M
ll
1 S,
i!
Ul
01
1
*s
1
*
u
i
-j
01
3
1 s| |
if'l- f ' 1
5 <=• S 0 2 . -a •$
f -S 5 o ' 3 T! J ' ' ?
i!1: * . li « •
l|?i 2; |2 "ri la-
1 1 1 11 ! H
Ml I'll* !•!
"8 I 3 S . - "3 22 Is
fill * il-'i P
s^ias 5 3s £ * s
^ 2 ° S
O ° pj
2 2 o 2
••*
S 5 . - 5
, ? ' ° °
ut
s
o
2
8 -5 1 a 5
0 0 | 0 0 —
1 I ! i ! '.I.
f m 4 —
1 * S 3 '*"'—. . i "*
e— s i * s •* 2 * ^
— -J S •— * .£ * S* 8 e
!l § i I] li ll ll
xl ull jg ^m# g WC V* « ^ jl
t-a | i li II IS il
333 3 a- a- .•*•" a •
Assuming 601 of total aldehydes is
formaldehyde, (ton-complaint hones.
2
2
•*»
o
0
„
d
S
N
2
2
,j'
^
«
u» S3
i s|
_ i flj gf
5 §2
i 1-
31 of 822 Measurements >I.O ppn. Co^>!
hones.
3-oonth old home built specifically f"'
o ^
T 0
o e
^tm «•
o ^
v . o,
*
! • ®
«M £
• •
8
1
i
i
^
s <-»
.11
3 S« $
. J c
Men complaint, occupied and nouotcupiei
Concentration by home age evaluated.
*9
2 'S
O *-*
f
2|
!
•
CSI
~
S S
^ 2 .""
i 1 7;
1 '! t
^^•H
-------
I
"Z
%
w
rl
I
a .
s-
1
wt
J
s
gj
s
i i
I1
S o
l!
2
1.
•3
i
wt
1
1
Study/sailing date(s)
(reference)
i s
If
|? 1
S * sf
ii £•
u £ v
"So a
11 *
8 § |
a'.
4
d d
i i
i i
- a
lennes&ee (1982-1983)
(Hodges 1984)
f ?.
1 H
a. |l
Ol M. Al
— . U C
4 * 41
2 - 1
U s| s
«* 2 c
- . 1 5 i
.£ | |§2
II ill
£ 3
— d
2 2
o 3
d d
£3 ^B ;
d d
it
i i
i s
«^
a ?->
i ^
1 ^ . O* *••
^
1
s
i"
i
* *
f
oa
.2
vt
ut
5
d
2
^f
d
^r
d
i
i
«n
I
SAI California survey (
(SAI 1984)
ta» •
S !
?
1
i
o
• il •*'
s .£
| 1
v< e
^ .
d
o
S
o
o
9
I
d
i
3.
ii
California State survey
(Sexton et al. I985a.
i
2
^rf
^ .5
w ••*
41 ^
1 8
1 *
- Insufficient data in
NO = Not Detectable, or
-------
with products made of high HCHO:urea ratio resins that are no
longer on the market; they cannot be considered as baseline
exposures for that reason. The most appropriate data for
describing current exposures in mobile and conventional homes
are, therefore/ those generated by random sampling of
noncomplaint homes after 1980, preferably after 1982 (when
manufacturers began using resins with mole ratios of 1.5 (F:U)'or
less).. These restrictions on t-he "appropriate" data base still
leave a considerable volume of monitoring data on levels in
homes. Table 6-4 summarizes the noncomplaint (random) data on'
HCHO levels in conventional and mobile homes.
6.5.2. Manufactured Homes
HUD has recently promulgated changes in i.ts Manufactured
Home Construction and Safety Standards (24 CFR 3280). The
changes, published in the FEDERAL REGISTER of August 9, 1984 (47
FR 31996), set product emission standards for particleboard (0.3
ppm) and plywood (0.2 ppm). HUD believes that if the. produce
standards are met and no other major emitters of HCHO are present
(e.g., medium density fiberboard), ambient levels will not exceed
0.4 ppm (EPA estimate of 0.15 ppm as a 10 year average) under
certain t«p«rature, humidity, and ventilation rate conditions.
Th© HUD regulations, however, were designed to reduce acute
reactions to HCHO and are not based oh HCHO's potential
carcinogenicity in humans.
6-20
-------
Table 6-4. Sunnary of Residential Monitoring Data from Randomly-Sampled Haws
Nutter
of homes
Conventional
30
40
17
29
3.1
6
120
29
103
78
51
Mobile
2
259
137
121
3
663
Mean (pom)
-- ' •
0.040
' —
0.05
0.060
0.063
0.084
0.09
0.05
0.027
0.07
0.038
0.21.
0.62
0.38
0.18
0.114
0.091
Range (pom)
0.007 - 0.151
0.25
0.013-0.34
0.046 - 0.153
—
0.03 - 0.07
—
<0.008 - 0.29
0.013 - 0.085
0.07 - 0.46
0.02 - 2.9
0.02 - 2.26
0.04 - 0.35
0.068 - 0.144
<0.01 - 0.48
Investigator, date
(date of monitor tinig)
Traynor 1984
Sirmtn 1983 (1979-83)
noschandreas 1978 (1978)
Hawthorne 1984 (1982)
Schutte 1982 (1980)
SAI 1984 (1984)
Koncpinski 1984 (1979-1983)
Godish 1983
Conn 1981
Stock and ntndez 1985 (1980)
(includes apartments and
condominiums)
Sexton et al . 198Sb (1984)
aoschandreas 1978 (1978)
Singh 1962 (1980-1981)
Anderson 1983 (I960)
University of Texas 1983 (1982)
SAI 1984 (1984)
Sexton et al. 19656 (1984)
6-21
-------
EPA estimates a ten-year average ambient HCHO level of. 0.10
Dpm for new manufactured homes. EPA has used this estimate and
the estimated 10-year average for new homes that just meets the
HUD target level of 0.4 com (0.15 ppm) in the quantitative cancer
risk assessment. Another study has reported average levels of
0.54 pom for manufactured homes less than three years old and
0.19 ppm for homes older than three years (State of Wisconsin,
1983). The Exposure Panel of the Workshop (1984) reported
studies that showed average ambient levels of 0.38,ppm for
manufactured homes not subject to complaints about HCHO odor -by
residents, and averages of 0.38 ppm to 0.90 ppm for complaint
homes. Thus, an unrealistic worst case exposure estimate was not
used to estimate human risk. Also, only 10 years of exposure
were assumed for manufactured homes. Specific exposure data
follow.
The average HCHO level in mobile homes appears to have
declined in recent years due to the use of lower-emittina stressed
wood products in mobile home construction and to the natural
decay of HCHO emissions from products in existing mobile hones.
Average levels in the existing stock of mobile homes are now
around 0.2 to 0.5 ppm/ with mean levels in individual- homes
(including-complaint" homes) ranging from less than 0.; to over
1.0 ppm. .
This apparent decline is shown graphically in Fiqure 6-1.
The Conyers (1984) study of complaint mobile homes, initiated in
1980, showed mean HCHO levels of 0.85 ppm in new homes. An
6-22
-------
a ' '
a
2
a . ,
« 2
1 • '» I i i
ft 5 § 5
i - : * i i
a * * -JS s- . >
S * ~ • S • a
< * 3} 2 • j 2
H Si S v 3 ^ 2
« < s < - 8 S
I i \ 1 : 1 I
• • a Z a a _
1 3 i 5 I 5 2
a • '
i !•«•*+$
* .
•
*
;
' •
• 1 1 1 i 1
9 0 6 00 <
•3> ^
T-- •
" *
^ • ^>
^ ^
* • *
4 •
* 4 *
• ^b
^F
• **
• +«
• «^
• *«
• •'
1 i 1 r ^
• rt w ^
i oo «
-I
-m
^ U
^5
kw
- | ^
u.
2
-| I
u.
O
^ 9
" >
O
9 ^*
"™ 2 2
II
mm 3 y S
~ ^ V3
• • i
2 r
K < -,
-S 2 3
is
« 22
•• S > —
•• u
a
-a I
•z
* —
_. 2 .u
2 >
-u
^
-3 ' -
•* i
-------
exponential function describing the relationship between HCHO
level and home age (r2=0.35) for the combined Singh (1982) and
Anderson (1993) data (i.e. the Clayton/Wisconsin data set) (1200
data points) predicts an average level of 0.5 pom in new 1970 to.
1980 vintage mobile homes (noncomplaint). Results of studies
begun in more recent years (University of Texas, 1984; MHI, 1984;
Sexton et al., 1985; Groah et al. , 1985) indicate that initial
HCHO levels in new homes on average fall within the range of 0.2
to 0.3 ppm.
Using the exponential, function describing the Clayton/
Wisconsin data to estimate decay of HCHO emissions over time, 10
year average concentrations can be estimated. For initial
concentrations in new homes of 0.5 ppm (i.e., Clayton/Wisconsin
data set), 0.4 ppm (i.e./ the HUD target level), and 0.25 ppm
(i.e., midpoint of range of recent study of new home levels),, the
10-year average concentration estimates are 0.19 ppm, 0.15 ppm,
and 0.10 ppm, respectively.
The fraction of homes with elevated levels of HCHO also
appears to have declined in recent years. Figure 6-2 shows that
the majority of homes less than 215 days old in the Clayton/
Wisconsin data set had HCHO concentrations above 0.4 ppm. More
recent studies indicate that this fraction is decreasing. The
California survey of 663 mobile homes (Sexton et al., 1985)
reported levels exceeding 0.4 ppm only in two and three-year old
homes. The Texas study (University of Texas, 1984) reported that
-------
i i i
Q 9 f*
- • ' • '
r A -v
'•»!••»•
z z z
5W W W
•:•.•. . -
8 <* • •
§
w
«
'
•
• ' .
•
• •
I
5
•
"* • . <
1
4
«
•
• • . - 4
.
. !
• "
•
• ' • +
'
9«
'
• 4
• 4
• 4
4
. s
s
*•*
-2 .
•«' <
••5
S «•
"I |
• 1
IS
" 2
-1
"™ «
"!
*N
8
> J"
s I
1^1
s «
SC v
6-25
-------
the highest mean in any-=group of homes was 0.35 ppm (ten hones in
one county less than one year old); it is likely that one or more
of these had levels above 0.4 pom, but not approaching 1.0 ppm.
Levels measured at any one temperature and humidity can,
however, be misleading. Table 6-5 which illustrates the effect
of temperature and humidity changes on a 0.4 ppm reading at 25'C
and 50 percent relative humidity (the HUD target) shows' that
under more extreme conditions (30°C/70 percent RH) , the .predicted
level could rise to 0.92 ppm. Because changes in temperature and
humidity occur over the course of a day and with seasonal weather
f luctuationsr homes without constant climate control would
therefore be affected.
These data illustrate clearly tha.t HCHO levels in homes are
the functions of multiple variables; neither age nor temperature
and humidity, nor any other variables can account for all
variations in residential levels (Versar, 1986b).
As the foregoing illustrates, HCHO levels in new
manufactured homes were tending toward 0.4 ppm and in some 'cases
above, until about 1979. After that date, mean HCHO levels in
new manufactured homes began to fall or level off slightly below
0.4 ppm. Even so, peak levels above 0.4 .ppm can be expected at
times du« to adverse temperature and humidity conditions.
The
frequency for such peaks is not known with confidence, but based
on the data available (see. Tables 6-6 and 6-7, and Figure. 6-1)
they could be expected to occur in a substantial fraction of new
manufactured homes.
6-26
-------
Table 6-5. Potential Effects of Temperature and Relative
CMn9»*.an Formaldehyde Air Concentration* (pan)*
Relative hutidity
Temperature 301
S9*f (IS^C) . 0.08
68*f <20«C) 0.15
77»f (25»C) 0.24
9S*F (30*C) 0 . 40
401
0.11
0.19
0.32
0.53
SOI
0.14 .
0.24
0.40
O.M
601
0.17
0.29
0.48
0.79
701
0.19
0.33
0.54
0.92
Calcul*t*d (rting Kjuationsin Myers, l984uAicft MT*
-------
Table 6-6.
FREQUENCY OF OBSERVATIONS FOUND IN CONCENTRATION
INTERVALS BY CLAYTON'ENVIRONMENTAL CONSULTANTS
Concentration
Interval (ppm)
0.0 -
. 11 -
.21 -
.31 -
.41 -
.51 -
.61 -
.71 -
.81 -
.91 -
1.1 -
2.1 -
Number
.10
.20
.30
..40
.50
.60
. 70
.80
.90
1.00
2.00
3.00
of homes
1 Percent of Sampled Homesa
<0.5 yrs > 0.5-1 yr All Hones
3.6
7.9
6.5
7. 2
5.8
6.5
5.8
5.8
6.5
12.2
24.5
7.9
139
8.0
4.0
36.0
16.0
0.0
12.0
16.0
4.0
0.0
4.0
0.0
0.0
25
8". 1
19.7/
14. 3
9. 3
5.0
4.6
4.6
3.9
3.9
7.7
14.7
4.2
259
a 259 "noncomplaint" mobile homes up to eight years old were
sampled in 1980-1981. Three measurements were typically- taken
in each single-wide home and-four measurements were taken in
each double-wide home. The data in the Table reflect the
average concentration measured in each home.
Source: Versar (1986a) statistical analysis of data supplied by
Singh et al. (1982).
-------
Table 6-7.
FREQUENCY OF OBSERVATIONS FOUND IN CONCENTRATION
INTERVALS BY WISCONSIN DIVISION OF HEALTH
Concentration
Interval (ppm)
0.0 -
.11 -
.21 -
.31 -
.41 -
.51' -
.61 -
.71 -
.81 -
.91 .-
1.1 -
2.1 -
Number
.10
.20
.30
• 40
.50
.60
• 70
.80
.90
1.00
2. .00
3.00
of observations
• Per;
_<_0. 5 yrs
*
2. .63'
29.0
0.0
10. 5
10.5
13.2
10.5
7-9
2.6
2.6
10.5
0.0
38
:ent of Observations'* ' '
>0.5-1 yr All -iorr.es
3.8
13.6
21.1
14.6
11.3
12.2
8.9
5.6
3.3
0.0
5.2
0.5
213
14.1
20.4
18.4
14.0
9.2
8.0
5.2
3.6
2.2
0.7
3.8
0-. 3
976
h°mes Up t0 nine vears old were
Each home was sampled at least six
rel13' ^ data in the table reflect
results of 976 measurements.
Source: Versar (1986a) statistical analysis of data supplied by
Wisconsin Division of Health (1984).
6-29
-------
6.5.3. Conventional Homes
The average HCHO levels reported in several monitoring
studies of conventional homes range from less than 0.03 to 0.09
ppm (see Table 6-4). Newer homes and energy efficient homes with
low air exchange rates tend to have hi-gher HCHO Levels (often
exceeding 0.1. ppm) than older homes (Versar, 1986c) . Results of
*
recent studies indicate that initial HCHO levels in new
conventional homes generally fall within the range of 0.05 tp 0.2
ppm; few .neasurements exceeded 0.3 pom (Stock and Mendez, 1985;
Hawthorne et al., 1984; SAI, 1984; Wagner, 1982). Computer
modeling to estimate initial HCHO levels in conventional homes
built 'using significant amounts of pressed wood (i.e., either
underlayment, paneling or both) yields values ranging from 0.1 to
0.2 ppm (Versar, 1986 ). Using the exponential decay function
described in Section 6.5.2, the 10 year average concentration for
a'home with an initial concentration of 0.15 ppm (i.e.,
approximate midpoint of range of new home levels) is estimated to
be 0.07 ppm. Summaries of some of the major HCHO monitoring
studies are presented below. .
The Lawrence Berkeley Laboratory (LBL) has summarized HCHO
concentrations in 40 residential indoor environments since 1979
(Girman et al., 1983). They have found that HCHO concentrations
in homes designed to be energy-efficient are somewhat higher than
concentrations in conventional homes. The maximum reported value
is 0.214 ppm in an energy-efficient home in Mission Viejo,
California. Data are not sufficient to allow calculation of mean
levels. • ' '
-------
As part of the development of an indoor air pollution model
'based on outdoor pollution and air 'exchange rates, Moschandreas
et al. (1978) studied the patterns'of indoor aldehyde'levels
monitored in 17-houses in the U.S. These data can be useful i£
'we assume HCHO constitutes 60 percent of total aldehydes, based
on L8L data (Girman et al., 1983). The 17 houses had an average
aldehyde concentration of 0.09 ppm. Applying the 60 percent
factor, the average HCHO concentration for the houses would be
0.05 ppm. The highest mean for any one home was 0.26 ppm; the
range for that home was 0.2 to 0.45 ppm. Another home with a
mean of 0.20 ppm reported a range of 0.07 to 0.5 pom. For -no
.other conventional home.did levels exceed 0.4 ppm,
A University of Towa Study . (Schutte et al.., 1981), oerforted
for the Formaldehyde Institute, monitored 31 conventional,
81
detached homes not containing urea-HCHO* foam insulation (UFFI)
for HCHO concentrations in the indoor air. Samples were
evaluated, in relation to outdoor HCHO concentrations, age of the
home, and other environmental factors monitored at each of the
sampled homes. The average indoor concentration found in the
homes was 0.063 ppm (standard deviation = 0.064) with a ranrje of
0.013 to 0.34 ppm. In only 5 o^f the 31 homes were average
concentrations higher than or equal to 0.1 ppn.
The 1981 Canadian study (UFFI/ICC, 1981) also studied nor.- '
UFFI homes. Table 6-8 summarizes these data, showing that levels
in none of the 378 homes exceeded 0.2 ppm.
6-31
-------
Table 6-8. Comparison of Non-UFTI Canadian Homes
by Average HCHO Concentration
Avtrag*
foraaloirtyo*
concentration (ppi)
<.01
.01-025
.025-040
.040-055
.055- 070
.070-. 085
.085-. 10
.1-15
.15-. 20
Totals
NuA*r of
taws
48 .
Ill
97
67
30
15
—
9
* 1
378
*~,u.
12.7
29.4
25.7
17.7 .
7.9
4.0
. — .
2.4
0.3
100.1
Cunulativ*
p«re*flUgt .
12.7
42.1
67.8
85. 5
93.4
97.4
—
99.8
100.1
Sour«: UFFI/IOC (1981).
6-32
-------
A report,by Virgil J. Konopinski (1983) of the Indiana.State
Board of Health summarizes the 'results -of a series'of
investigations conducted from, 1979 through 1983 to determine HCHO
levels in conventional homes in Indiana. The mean' HCHO level in
the 120 homes without UFFI was 0.09 ppm .(0.05 for homes-with
UFFI). That mean could be skewed by the maximum concentration of-
1.35 ppm reported in -one home. Neither the age of the homes nor
the age of the UFFI installations was reported.
From April to mid-December 1982, Oak Ridge National
Laboratory (ORNL) with the U.S. Consumer Product Safety
Commission (CPSC) studied indoor air quality in 40 east Tenessee
homes. The objective of the study was to increase the data base
of HCHO monitoring in a variety of American homes and further.
examine the effect of housing types,, inhabitant lifestyles, and
environmental factors on indoor pollutant levels.
Homes to be' sampled were selected based on a. stratification
, . &
to ensure representative home age, insulation types, and heatina
sources. All were voluntarily enrolled. Twice a month, four
samplers at each location monitored HCHO levels in three rooms
and outside the house. Samplers were exposed to the air for
24-hour periods. 'No modifications to the residents' life styles
were requested during these measurements.
Table 6-9 summarizes these data by home age and season
(indicative of temperature-and humidity). HCHO measurements in
the 40-home east Tennessee study led to the following major
conclusions:
6-33
-------
Table 6-9.
ORNl/CPSC «ean Formaldehyde Concentrations (pom)
as a Function of Age and Season (Outdoor Ae«ns Ar«
Less Than 25 ppo Detection Limit)
Age of nous*
«n
0-S years
5-15 years
Older
0-S y»*rs
=
5-15 ywrs
olo»r
all
S«a«en
all
all
all
all
spring
suMwr
fall
spring
sunwr
fall
spring
SUMT
fall
spring
SUMW-
fall
X*
0.012
O.OS4
0.042
0.032
0.087
0.111
0.047
0.043
0.049
0.034
6.036
0.029
0.026
0.062
0.083
0.040
s*
0.077
0.091
0.042
0.042
0.093
0.102
,0.055
0.040
0.048
0.03S
0.051
0.037
0.023
0.076
0.091
0.047
• n
5903 40
3210 18
1211 11
1482 11
1210
1069
931
626
326
259
757
341
384
2593
1736
1574
Hot*: x « BMII concentrations.
s » standard deviation.
• • nuttar of
Include* hows with and without UFfl.
Hawthorne et al. (1984).
6-34
-------
(1) Th« average HCHO levels exceeded 100 ppb .(0.1 ppm) in 25
percent of the homes. .
*
,(.2) HCHO levels were found to be positively related to
temperature in homes.- Houses with UFFI were freauently
found to exhibit a temperature-dependent' relationship
with measured HCHO levels.
(3) HCHO levels generally decreased with increasing age of
the .house. This is consistent .with decreased emission
from materials due to aging.
(4) H.GHO levels were found to fluctuate significantly both
during the day and seasonally.
Studies by Breysse (1984) evaluated conventional/ non^UFFI
homes. The University of Washington studied 59 such homes;
private laboratories, in the state studied an additional 25. The
freauency distribution for measured levels are presented in Table
"6-10. A total of 6 of. the 189 samples.(3.1 percent) were over
0.5 ppm and 56 samples (26.5 percent) were over 0.1 ppm.
Traynor and Nitschke (1984) monitored indoor air pollutants
in« 30 homes with and without suspected combustion (and other)
sources, the average HCHO level observed in all the test homes
was 40'ppb; a high value of 151 ppb was found in one of the
tested residences categorized as containing new furnishings and
new paneling as a suspected pollution source.
6-35
-------
Table 6-10. Frtqmncy Distribution of FonMldthy 1-0
> 0.5 -0.99
> 0.1 • 0.49
< 0.1
TOTAL OBSERVATIONS
2
2
41
«S
113
0
2
9
65 ,
76
1.0
2.1
2*.S
70.4
Sourca: Irtyss* (1964)
6-36
-------
The results can be summarized as follows:
o The 4 homes with no identified source had a range of
means of 0.007. to 0.034 ppm.
o The 3 homes with new furnishings had a range 'of means of
0.015 to 0.061 ppm.
o The 4 homes with cigarette smokers had a range of means
of 0.032 to 0.060 ppm.
o The 18 homes with gas, coal, and wood fueled
appliances/heaters had a range of means of 0.012 to
0.056 ppm.
o. The 12 homes.with a combination of sources reported a
range of means from 0.013 to 0.064.
- *
Variations in home levels could not be attributed to combustion
sources.
Stock and Mendez (1985) measured HCHO concentrations inside
78 homes in the Houston, Texas area during the summer of 1980.
No mobile homes, UFFI homes, or complaint homes were samoled.
Indoor concentrations ranged from less than 0.008 ppm to 0.29 ppm
with an average value of 0.07 ppm for detectable concentrations
(Number of samples, N»75). Three energy efficient condominiums
had, as a housing category, the highest mean level (0.18 ppm).
Condominiums (Mali), apartments (N=*19), and energy-efficient
houses (N*7) represented the mid-range with mean levels of 0.09,
0.08, and 0.07 ppm/ respectively; the mean of 38 conventional
houses was 0.04 ppm.
Wagner (1982) measured HCHO levels in 12 California homes
that fall into a prescribed "worst-case" category~of buildinq and
occupancy characteristics' (i.e.-, low infiltration and ventilation
rates, new construction, presence of gas stoves). Weekly averaqe
6-37
-------
concentrations ranged from 0.078 to 0.163 ppra with a mean of
0.106 ppm.
Sexton et al. (1985) measured HCHO- levels in 51 home
dwellings. Weekly average concentrations ranged from 0.013 to
0.085 ppm with a geometric mean of 0.035 pom and an arithmetic •
mean of 0.038 ppm. Seventy-six percent of the homes were more
than 10 years ,old and only two were less than six -years old.
A downward tend in HCHO levels in conventional homes is seen
in Figure 6-3. The relative proportion of low HCHO lev-els in
homes that have been monitored has increased over the past six
years, and the proportion of high levels have decreased. These
data are limited and caution in interpretation is recommended
(Versar, 1986a).
6.5.4. Garment Worker Exposure
HCHO levels in apparel manufacturing facilities were
generally below 3 ppm prior to 1980 (see Table 6-11). OSHA had
established a 3 ppm TWA (time-weighed average) in 1967. However,
OSHA is presently considering establishing a new level (see 50 FR
50412; December 10, 1985). The ACGIH (American Conference of
Government Industrial Hygienists) recommended level, is 1 ppm
TWA. In r«c«nt years, HCHO levels observed were generally below
1 ppm (see Table 6-12). The data in Tables 6-11 and 6-12 must be
viewed with caution because in 1983, the National Institute for
Occupational Safety and Health (NIOSH) discovered that the
commercially prepared inpregnated charcoal tubes, which had been
used in previous personal monitoring studies were unstable.
5-38
-------
Z
§
i
*M
a TI
1 •!
<
a
< §
: 1
I
s
<
2
5-39
-------
Table 6-11. PRE-1980 MONITORING DATA FOR GARMENT MANUFACTURING AND
CLOSELY RELATED INDUSTRIES
•i
*
et»1««
0.1 - 1.4 (1171)*
1-11
0.3 • 2.7 (TWA-4f*«. IHt)
•0.1 - 1.4 (T\a, 117*)'
o.is - O.M (TV*, mo*
O.I • 2.7 (TWU*rt». 1H4)
o.ooc-4.»*
(i.t. - irfti »»•*». eiatn-
Iftf •*•*!•*» Utr«l. «te
t?
Mi
•s
•s
. Wl)
2.600 •
1179)
2.2M
1*79)
0.030
O.I • 3.3 (Ct1Hfli.«r««. 19t<)
0.13 • O.«i (l»i»)4
0.04 . 0.73 (TWA, U71J*
4)
•S
32
10
1
«c
1
Hi
•S
22
>S
1
1
I
2
i.:
Ml
t
Mi
CM 125.
> t CM 125.
* I CM 31».
eu
J54.
«r
i*
15
••t
5-40
-------
o —
<«
2as
3
LiJ
UJ
ae
_
li
• T
: i
M
«
M
l\
* • *
1 •
*
|
1
s V
•I *
s
«*
»
1
X
3
|
•
o
a
"s
1
1*
w X
P
2 ;
S
i
*? %
MH IJi
s s • • — v
r 3 s s rss
s s s s ?••
« • «• 4 • •
a, - • • «--
»w •' •« <«• • ••
*a
ft ^
«•
*
V • » *
* , * *
J If
* *i
1 J3*
s 5i s s -S3 5
I * .
6 "3
: 212
I £* |8.lt- *3
f 51 I|:| ss-
41 ^E ^ ^ ^ M C * w
^S • 7 ^ • * ^S • J»
Jill iili li
s s
i
1 11
« ' » *
« • *
nil
f_ T-n-i» *
** s;5 «
^J^v ^ ^«i
rjjda
• — o' • o
_,
^
J
*
j
1?
^ a
si
J9
^rf
if
•tf •
wJ
9
2» * * «M«to
S3 3 -
! i
III if
••
• • te •
55- ?5
- ii1. 4i
^ •**
ss: «?
• M • ^^ MA
J82S ?2
— o o' o° * o'
J{
•i
s •
t 1
! r
b b
^ » ^9
V . W
s'i si
« _
fcrf ^rf
I": i«.
a= ii
•il 1!
W X W «
s •
s - L I-
r 1 |s --is
! if |i iiis
1 Ij 41 ill5.
£ S1™* "t?'85SS
• • • ®M * * *
-------
cs
1-1
VO
2 ••( •« »« ••
i S s 1 i
i
*
i
•
t
r
1
•
1
i
>
5
I
i'
S -
11
a c
J r
i
t £ s
I 5 1
7 i -
5 r s
•:'?
. i- i
t -s r-
| as ts X2-
; I! ^! *i
iUr
" *
il^-w,
:53
5-42
-------
Thus, the monitoring data above may be suspect since the Loss of
«
HCHO from the tubes was not consistent. . Consequently, the HCHO
levels recorded most likely represent lower levels than actual
conditions. The NIOSH method at that time was also used by OS HA.
?
NIOSH subsequently developed a stable medium for collecting
the HCHO and did two in-depth industrial hygiene studies. _The
surveys were done at two large manufacturing sites producing
men's dress shirts. HCHO exposure levels were determined for 54
of 72 job titles in two different plants. The number of
individuals within each job title whose exposure levels were
sampled was based on the. total number of employees in that
category and reflect a 95jpercent confidence level that the
highest and lowest exposed individuals were iacluded in the
sampling. A summary of the data are presented in Tables 6-1.3 and
6-14. These tables show that all levels of exposure were less
than 0.51 ppm TWA. Also, as Table 6-13 illustrates, the combined
range of data was very narrow (0.01-0.39 ppm) for 5 of the 6
departments in the two plants. The range of mean concentrations
of all departments (0.13-0.20 ppm) is very narrow and compares
well within the overall combined mean exposure level of 0.17 or.n,
which was used for the quantitative ca.ncer risk assessment. In
addition, the average exposure levels used in EPA's section 4(f)
determination (SPA, 1984), 0.23 ppm (area) and 0.64 ppm
(personal) (Versar, 1982), were also used for this -cancer risk
assessment. , ' . -
6-43
-------
cc
(9
(X
I
ra
*j
m
Q
60
•V4
U w
e c
51
I!
Z .a
vo
-
UJ
• •
'•o £
-------
o
fi£
o
CM
O O
O O
Q.
Q_
CO
en
tn in
—o
vo
I
•w
us
3S
r*r
0
o
1
pH
• o
9
0
4
o
1 .
1-H
o
0
V
^
0
1
O4
o
o
f^
o
1
1— t
o
O
V
K\
p
I
f— t
O
o
-------
All of the determinations made in the NIOSH studies were at
one point in time and may not reflect the variation of exposure
over a longer period. -Factors that could affect variation in
HCHO levels in these plants include variation in ambient
temperature, humidity/ type of fabric or resin system/ and volume
of stored materials or completed work.
The exposure range across departments/ within plants/ as
well as between plants/ appears to be narrow. Both these plants
were large manufacturing sites, producing similar products. Both
plants had central ventilation/cooling systems. This type of
plant may potentially represent only 10 percent of the total
number of manufacturing sites (though up to 25-30 percent of the
workforce may work in such plants) (Ward/ 1984).
6.6. Summary
The data presented above indicate that HCHO levels in new
manufactured homes are generally below 0.5 ppm/ with 10-year
averages for new HUD Standard homes of 0.15 ppm or less.
3*
However/ some fraction of new homes experience ^peak levels that
could exceed 1.0 ppm for periods of time. It would be expected
that as temperature/humidity exceed 75°F/50% RH, HCHO levels
would rise as Table 6-3 illustrates. Thus, depending on heating
and cooling preferences/ HCHO levels in new homes may
substantially exceed the reported mean for new homes.
The situation is similar for conventional homes/ although
reported mean levels are lower/ 0.03 to 0.09 ppm. However/
because conventional housing is much more heterogeneous, peak
6-46
-------
levels in some new homes may substantially exceed reported
means. Although temperature and humidity conditions play a large
role, construction techniques which tend to limit air exchanges,
such as in energy efficient homes, and building product mixes are
also of major importance. The ten-y.ear average HCHO
concentration for a new home built with significant amounts of
pressed wood is estimated to be 0.07 ppm. - ' •
Reported HCHO levels during garment manufacture are below
1.0 ppm and in some plants below 0.5 ppm, and the NIOSH data
indicate rather tight ranges (none exceeding 0.51 ppm). However,
much of the reported monitoring data must be approached with
caution due to the technical fault discussed earlier. Building
design, ventilation, and temperature/humidity changes may be
responsible for daily or seasonal variations.
6-47 .
-------
-------
B.4
EXPOSURE ASSESSMENT
Attribute 8
Research or data necessary to improve the exposure assessment
are described. •
SOURCE Case Study B. TCDD (Page xxii). -
Note Several of the case studies noted lack of data on aspects of exposure but
contained little discussion of approaches. Excerpts from the case study
on TCDD illustrate a limited treatment of the prospect of then on-
going human studies.
-------
ncasl
special repoi
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT. INC, 260 MADISON AVENUE. NEW YORK. MY.
EXECUTIVE SUMMARY
DIOXIN: A CRITICAL REVIEW OF ITS DISTRIBUTION,
MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,
AND THE SETTING OF ACCEPTABLE EXPOSURE LIMITS
SPECIAL REPORT NO, 87-07
MAY 1987
-------
- xxii -
generally accepted as "safe," although it is likely that higher
levels are also safe) That is -gratifying because it supports
the idea that levels of 1 ppb or more are not"widespread in the
environment. -Instead-, they are limited to where .they are
expected—production and disposal sites. The EPA survey also
revealed that some fish taken downstream from some bleached
pulp mills contained measurable levels of dioxin,, Those
results identified a possible yet unconfirmed source.
J. Better Information About Exposure
Dioxin* is stored for long periods of time in the human
body. It is stored in adipose (fat) tissue, and the time
necessary to eliminate 1/2 the body burden of dioxin may be 5
years or longer. Therefore, if a person were heavily exposed
10 or even 20 years ago, he or she might still have elevated
levels of dioxin in his or her fat. Scientists have standard
formulae to calculate what the exposure levels must have been
years ago to produce a given level in fat today if the
half-life for elimination is known.
As was mentioned in connection with the proposed Agent
Orange study, the ability to measure dioxin in blood promises
that much more information about exposures will be ;
forthcoming. Until that method was developed, the necessity of
surgical removal of adipose tissue for dioxin measurements
limited the number of ..samples. Based on measurements from a
handful of studies involving a few score people, the "average"
concentration of dioxin in adipose tissue in North Americans is
estimated at between 2 (perhaps less) and about 6 parts per
trillion (ppt). Those estimates may be too high. It may be
that the persons who have been sampled are not representative
of the United States population. If they are among the
more-exposed people, then the average body burden will be
overestimated. Although perhaps on the high side, those
averages can be taken as a measure of "background" exposures to
dioxin.
As more measurements of dioxin in humans are accumulated,
the estimates of,background exposures may become more precise.
However, it is also necessary to have better information about
the half-life of dioxin in humans to refine the calculations.
The estimate of 5 years, based on three measurements in a
single man, may well change.
-------
-------
B.5
RISK CHARACTERIZATION
Attribute 1
The major components of risk (hazard identification, dose-
response, and exposure assessment) are presented in summary
statements, along with quantitative estimates of risk, to give a
combined and integrated view of the evidence >.
SOURCE Case Study D. Formaldehyde (Pages 1-23 to 1-31)!
Note See Dose-Response Attribute 1 in this Appendix for summary state-
ments on epidemiological data and animal data. Also, see Dose-
Response Attribute 2 for quantitative estimates of risk with a treat-
ment of uncertainty associated with dose-response evaluation.
Another section of the report deals with exposure. Uncertainty in
exposure estimates is not integrated into the risk characterization.
SOURCE Case Study B. TCDD (Pages xxiv to xxvii).
Note These excerpts are from the executive summary. Both Attributes 1
and 2 for risk characterization are illustrated by this text.
SOURCE Case Study H. Methylene Chloride (Pages 108-112)
Note This report did not include an exposure assessment. Thus, the integra-
tion primarily focuses on hazard and unit risk without consideration
of overall risks from specific exposure scenarios. The excerpts also
.illustrate Risk Characterization Attribute 2.
-------
ncasl
special report
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT, INC, 260 MAOtSON AVENUE. NEW YORK, N.Y. 1001
EXECUTIVE SUMMARY
DIOXIN: A CRITICAL REVIEW OF ITS DISTRIBUTION,
MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,
AND THE. SETTING OF ACCEPTABLE EXPOSURE LIMITS
SPECIAL REPORT NO. 87-07
MAY 1987
-------
-xxiv -
decide which model is a better predictor. Although there is no
convincing evidence that dioxin has caused human cancer, the
available data do not permit choosing a model. The failure to
find human cancer could mean that humans are less sensitive to
the carcinogenic effects of dioxin than animals. Or it could
be that humans have not been exposed to doses comparable to
those that cause cancer in laboratory animals. Or it could be'
that dioxin-caused diseases are sufficiently uncommon in number
and common in kind that they have gone undetected. For
example, there is no evidence that dioxin causes lung cancer,
but if it were to, a small increase could go undetected against
the large incidence caused by smoking.
In the absence of human data to validate a model, ENVIRON
Corporation chooses a threshold model because of the scientific
evidence that dioxin does not act through a genotoxic mechanism
and because what is known about its mechanism of action
strongly suggests a threshold dose must be exceeded and
sustained 'before the events leading to cancer are set in
motion. It also recognizes that some scientists and the
regulatory agencies of the United States while accepting that
dioxin is a promoter, prefer to use a no threshold model to
estimate cancer risks. Finally, we. emphasize that we do not
suggest that a threshold model be applied to all carcinogens
that do not have genotoxic activity. Each chemical must be
evaluated individually. However, we suggest use of the
threshold model in the case of dioxin because of the
substantial additional evidence that exists concerning its
mechanism of action as a promoter.
L. Examples of Risk Calculations
The discovery of dioxin-contamination of fish downstream
of some bleached pulp mills focused attention on wastes from
those facilities. Waste sludge from pulp production has been
disposed by land-spreading in Maine. The material is used for
soil enrichment and can be either applied to the land surface
or plowed in.
1. Risks from Consumption of Dioxin Containing Milk
Some sludges ^ave detectable dioxin levels, and the
amounts of dioxin that would be applied to the land in any
sample can be accurately estimated. Using sludge
containing, fifty ppt of dioxin at an application rate of
10 tons/acre, sludge can be spread on land for 45 years.
-------
- xxv -
At the end of that time, metals in the sludge that will be
transferred to the soil will be near acceptable limits in
many circumstances, and application must end.
Computer programs were designed to calculate the
concentration of dioxin in the soil for each of the 45
years of disposal plus each of the next 25 years. The
average concentration over that 70 year span is projected
to be 6.4 ppt.
People can be exposed to the land-disposed dioxin
through ingestion of dust or contaminated food, through
inhalation of contaminated dust, and through absorption
across the skin. According to Envirologic Data (a Maine
consulting firm), the most exposed people would be farmers
and their families who obtain all their food from land
spread with pulp mill sludge. The exposure estimates are
deliberately biased toward the high side; it is unlikely
that a farm family would eat food only from'its own land
and that sludge would be spread across all their land.
Consumers would be exposed to significantly less dioxin.
They obtain their food from a variety-of sources, they are
more distant from the sites of land-spreading, and their
risks would be less.
Dioxin, if taken up at all by plants, is taken up
very poorly, and little enters the food chain that way.
The most important food chain source is from cattle
ingesting dirt, which they do when grazing. Dioxin
absorbed from the ingested dirt will be stored in the
animals' fat tissues and excreted in milk.
The highest exposures were estimated for farmers who
drank milk and ate beef only from cattle grazed on
sludge-augmented lands, who ate corn from the same land,
and who were exposed to the same soils. The largest
single source is from beef consumption.
»
Use of EPA's estimate of dioxin's carcinogenic
potency produces the largest estimated risks. It predicts
about 2 extra cancers per million farmers from drinking
TCDD-contaminated milk at the maximum contamination level
every day for a full 70-year lifetime. Use of FDA's or
CDC's estimate of potency produces calculated risks of
less than one in a million.
-------
- xxvi -
A lifetime risk of (10~6), is a de facto acceptable
dose; concern about carcinogenicity drops off remarkably
when the lifetime risk is less. Only use of EPA's
estimate of carcinogenic potency with exposure estimates
from land spreading generates lifetime risks greater than
10~6. Although EPA's estimate is most generally used in
this country, its scientific and theoretical underpinning
are no better than FDA's or CDC's, and probably poorer
than for the ADI approach. This example illustrates that
risk characterization very much depends on the models used
for estimating risks.
^
2. " Risk from Fish .
Fish are a special problem in considerations of
dioxin. Although the chemical is, for practical purposes,
insoluble in water, it binds strongly to organic materials
in soils and sediments. Dioxin-contaminated sediments in
bodies of water can be ingested by fish, and the chemical
is stored in fish fatty tissue. FDA,-making assumptions
that people eat a variety of fish from different fisheries
and using its estimate of carcinogenic potency, advised
that fish from fisheries where concentrations were less
than 25 ppt in edible portions could be consumed without
limit. EPA, making assumptions that people eat only
contaminated fish and using its estimate of carcinogenic
potency, estimates that fish containing 0.069 ppt (about
1/350 FDA's acceptable limit) pose a 10~6 lifetime risk
of cancer. The best analytical methods can barely detect
0.069 ppt, and routine methods cannot detect that low
level in fish. One conventional response to a
non-detectable level is to assume that the chemical is
present at 1/2 the detection level. Therefore any method
that could not detect 0.14 ppt (essentially 2 x 0.069) can
lead to the assumption that dioxin is present at a level
that poses a 10~6 risk. These very different
conclusions illustrate the importance that assumptions
play in risk assessments even when similar models (in
these cases, no-threshold extrapolation) are used.
>
3. Risk Calculations and Models
No-threshold models are designed to minimize any
probability of understating risk. Therefore they
incorporate assumptions that cause them to err on the
conservative side (that is to predict higher risk than are
actually expected). For instance, EPA's estimates of
carcinogenic potency, including the estimate for dioxin,
are upper bounds on risk. Use of the upper bound makes it
unlikely that the actual risk is higher than estimated,
and EPA states that "true value" for potency "may be much
-------
-xxvii -
lower, with a lower bound approaching zero." In addition
to the assumptions common to all no-threshold models,
EPA's model incorporates other features that cause it to
predict a higher risk than does FDA's or CDC's model. The
importance attached to EPA's model because it is widely
used can be balanced against the fact that the other
models are equally valid.
The threshold model predicts much lower risk. It is
well supported by biological research and tests, and it
has been adopted by foreign countries. Given current
understanding of the mechanism of dioxin toxicity, a
threshold model appears to be sufficiently protective of
human health.
-------
United States
Environmental Protection
Agency
Office of Health and
Environmental Assessment
Washington DC 20460
EPA/600/8-87/030A
July 1987
Extarnal Review Draft
Research and Development
EPA
Update to the Health Assessment Document
and Addendum for Dichloromethane (Msthylene
Chloride): Pharmacokinetics, Mechanism of
Action, and Epidemiology
Review
Draft
(Do Not
Cite or Quote)
(Permission to
reproduce granted
by EPA)
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
defined, and such risk estimates do hot, as yet, take into
account changes in the pharmacokinetic model required by the
adjustment of a particular kinetic parameter.
A comparison of risk estimates made directly from the human
data provided by the Kodak epidemiology study to risk estimates
derived from .the results of the pharmacokinetic model used by
Andersen and Reitz does not show the animal-based risk estimates
to be overestimates, using upper-bound risk estimates from
respiratory cancer deaths or using either maximum likelihood
estimates or upper-rbound estimates from the pancreatic cancer
deaths in the Kodak study. •
In summary, EPA concludes that the animal evidence of
carcinogenicity conforms to the definition for "sufficient" in
the EPA Guidelines for Carcinogen Risk Assessment. The
epidemiology studies, while showing no evidence of either liver
or lung cancer attributable to DCM, are not sufficient to rule
out a risk to humans; the data on deaths from pancreatic cancer
give some weight to the possibility that DCM may cause cancer in
humans at sites other than those found in animal species.
Overall, the epidemiologic data conforms to the definition in the
Guidelines for "inadequate" insofar as the pancreatic cancer
deaths cannot be used to establish a connection between exposure
to DCM and human carcinogenicity, yet neither can the possibility
Thus, DCM meets .
human carcinogen.
The available body of evidence on the carcinogenic mechanism
of a such a connection be entirely discounted.
the Guidelines criteria for Group B2, probable
108
-------
of action of DCM and on'species differences in utilization of the
carcinogenic metabolic pathway are not sufficient to support an
estimate of zero cancer risk to humans. An evaluation of the
weight of evidence does lead to the conclusion, however, that
risks should be estimated on the basis of internal dose of the
GST metabolite (s).. A comparison of the results of the available
studies indicates that the GST pathway is the most likely source
of the excess tumorigenesis observed in the NTP mouse bioassay.
Additional research on pharmacokinetic model parameters and
on the carcinogenic mechanism of action underway -by CEFIC are
expected to lead to refinement of the risk estimates presented in
Chapter 8. it should be noted that data from the experiments may
lead to human risk estimates below those estimated from the
currently available data. To the degree that the estimates of
relative metabolism by the GST pathway change as a result of-
these data, the extrapolated risks will change. (Method 2, which
does not use metabolism to extrapolate across species, would be
expected to give virtually the same risk estimates as before,
irrespective of interspecies differences in metabolism that may
be discovered.)
Using the pharraacokinetic model with its original kinetic
parameters to estimate the internal dose of the GST metabolite*,
then following Method 1, and correcting internal dose for
interspecies differences in sensitivity by using the surface area
correction factor, leads to a unit risk estimate for continuous
inhalation exposure to 1 ug/m3 of 4.7 x 10~7.
109
-------
It would be unwise to read too much importance or
significance into changes in the unit risk of a few fold when
pharmacokinetic data are employed by either Method .1 or Method 2.
The previous chapters have outlined uncertainties in the
structure and parameter values of the model formulated by
Andersen et al. (1986, 1987). Although it is difficult to define
these uncertainties in quantitative terms (such _as confidence
limits), it is clear that model projections of internal doses
could vary, perhaps by up to several fold, without contradicting
currently available model validation data. Moreover, there are
large uncertainties as to the biological effects of those
internal doses that overshadow any error in their estimation.
Species differences in responsiveness—and within-species
differences in susceptibility of various tissues—are-unclear.
Perhaps the largest uncertainty lies in the question of the
relative carcinogenicity of high and low doses, owing to the lack
of knowledge about the mechanism of DCM's carcinogenic action.
It is somewhat ironic that the area of risk extrapolation that
has the least uncertainty as far as pharmacokinetics is
concerned—relative internal doses at high and low exposures—
also has the greatest uncertainty in terms of the degree of
carcinogenic response that those internal doses can be expected
to engender. (Such uncertainty continues to be accommodated by
the use of an upper-bound, linearized multistage model for low-
dose extrapolation, which recognizes that the true dose-response
curve may fall off more rapidly at low doses.)
110
-------
In view of the uncertainties involved, the changes in DCM's
carcinogenic potency that result from different uses of the
available pharmacokinetic information are not, in practical '
terms, very distinct. Discussion of the issues has been
worthwhile because of their theoretical importance rather than
their practical significance in the present case. For other
compounds (or for DCM itself, upon the introduction of new data),
the distinction among extrapolation methods may have much greater
practical consequences. -
Rather than focusing on exactly how much the risk
extrapolation has been changed by the use of pharmacokinetic
information, it is instructive to examine how"little it has been
changed. Perhaps the most important result of the foregoing
analysis is that, in the case of DCM, pharmacokinetic
»
considerations have not revealed a great error inherent in using
applied dose as a surrogate for internal or delivered dose.
According to current understanding as expressed in the
pharmacokinetic model used by Andersen and Reitz, there is little
difference between mice and humans in the proportion of a given
applied dose that is metabolized. There are differences in this
proportion from high to low doses, but they are .riot- especially
large. Having uncovered these pharmacokinetic factors, it is
well to incorporate our best understanding of them into the risk
extrapolation process, despite remaining questions as to their
exact magnitudes. The uncertainty in the resulting potency
estimates is reduced (compared to the extrapolation based on
111
-------
r
applied dose) because the potential for the influence of
' pharmacbkinetic factors has been markedly circumscribed.
112.
-------
B.5
RISK CHARACTERIZATION
Attribute 2 The report ckarly identifies key assumptions, their rationale,
and the extent of scientific consensus; the uncertainties thus
accepted; and the effect ofreasonabk alternative assumptions
on conclusions and estimates.
SOURCE Case Study J. Red Dye No. 3 (Pages 87-97).
Note Each critical assumption is discussed.
SOURCE Case Study B. TCDD (Pages xxiv to xxvii).
Note See Risk Characterization Attribute 1 in this Appendix.
SOURCE Case Study H. Methylene Chloride (Pages 108-112).
Note See Risk Characterization Attribute 1 in this Appendix.
-------
A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO. 3
AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
Prepared by:
Dr. Ronald W. Hart, NCTR/FDA (Chairman)
Dr. Thomas Burka, NIEHS'/NIH
Dr. Stan C. Freni, CEH/CDC
Dr. Robert Furrow, CVM/FDA
Dr. David W. Gaylor, NCTR/FDA .
Dr. Theodore Meinhardt, NIOSH/CDC
Dr. Bernard Sass, NCI/NIH
Dr. Elizabeth K. Weisburger, NCI/NIH
Executive Secretaries -
Dr. Paul Lepore, ORA/FDA
Dr. Angelo Turturro, NCTR/FDA
July, 1987
-------
July, 1987
CHAPTER 8 - RISK CHARACTERIZATION
A. Introduction
Uncertainties are the major source of problems in conducting a risk
assessment. There are uncertainties about the quality of measurements,
such as the degree of accuracy or sampling error. This can be dealt with
by applying a variety of statistical tools. Another kind of uncertainty
presents itself when interpreting study results, when it is unknown what
mechanisms have led to the study results. For instance, how should the
effect of gender and species on the. tumorigenic effect of R-3 be inter-
preted, given that it is unknown what caused this dependency? In some
situations, available additional information may clear up part or all of
the uncertainties.. In other cases, deductions or circumstantial evidence
from analogous studies may be of help in interpreting study results. To
what extent this may occur is largely determined by the willingness and
ability to bridge gaps in knowledge with assumptions, and to weigh the
arguments of .their validity.
1
Many assumptions have, knowingly or unwittingly^ been used in the
preceding chapters. There are a number of situations, however, in which
the Panel felt that there was insufficient scientific information for for-
mulating assumptions with the confidence that the balance of arguments
favoring an assumption would outweigh those of rejecting it. From a
scientific viewpoint, forcing a decision in such situation would render it
more or less arbitrary. It is the Panel's opinion that deciding in a con-
troversial issue with lack of supporting evidence for either of the sides
may be based on health policy or regulatory considerations, which is
clearly within the domain of risk management. ,
To allow the reader to understand the extent and the impact of as-
sumptions used by the Panel in the preceding chapters, and in harmony with
recommendations made in a number of documents (10,15), assumptions, con-
sidered to be of crucial importance in the decision process, are listed
below. Where appropriate, options have been presented to .assist risk
managers in making decisions in issues open to alternative interpretations.
Some of these options may lead to an alternative dose level, and thus to an
alternative risk level. Chapters 6 and 7 provide the- tools to assess the
risk associated with different dose levels. In other cases, the options
87
-------
July, 1987
are limited to validity or invalidity of an assumption, and thus to either
acceptance or ^non-acceptance that there is some risk to humans.
B. Characterization of Assumptions
1. It is assumed that a tumorigenic response in any animal species is an
indication of possible carcinogenicity in humans.
Comment; Most human carcinogens have shown a positive response in
animals. Conversely, most animal tumorigens have not been shown to
have an effect in humans. The latter can be explained in a number of
ways: there is a lack of, or a deficiency in, human studies; the
tumorigenic potency of the chemical is too low to be observed in a
human study of manageable size; or the chemical may not be tumorigenic
to humans. The validity or invalidity of the assumption cannot be
judged on the basis of current scientific evidence.
2. It can be concluded that the tumorigenic response in R-3 treated rats
has been mediated through or induced by TSH. Following the former
assumption, it %is then assumed that R-3 has a tumorigenic response in
humans regardless of whether or not TSH has a pivotal role in the
development of tumors.
Comment; Implicit in the above statement is the assumption that the
target organ in humans is necessarily the same as that in rats. There
is evidence that R-3 acts on the hypothalamus-pituitary-thyroid axis
of the rat, in which TSH and iodine have a central role. Given the
f * L
lack of evidence for an alternative pathway, there would be no evi-
dence that, in humans, organs other than the thyroid may be affected..
.. The evidence that TSH is crucial to the response in rats is not con-
clusive, and neither is the evidence that excess TSH causes thyroid
tumors in humans. It is thus possible that thyroid tumors may develop
in humans even if TSH were not essential in the tumorigenesis. Al-
though the Panel feels that there is some evidence that R-3 is not
tumorigenic to humans, the weight of evidence is insufficient to rule
out tumorigenicity. A decision has therefore- to be made using other
considerations.
3. It is assumed that the tumorigenic response observed at high dose
88
-------
July, 1987
levels in rats can be extrapolated, by means of a mathematical model,
to the dose levels of human dietary intake which are several orders of
/• • ,
magnitude lower.
Comment; Many mathematical models can be fitted equally well to the
various observed data points. A number of studies have been, and are
in the process of being, done to evaluate the use of different models
(157-161). An alternative to biologically-based mathematical models
is the use of a NOEL with safety factors (see assumption 6), with the
understanding that the NOEL in this case does not imply that there is
a threshold for effect. It is presently not clear which approach is
more accurate in defining actual risk in carcinogenesis, and the de-
cision to use either approach, or some other, is a matter of science
policy.
4. It is uncertain whether dose levels associated with decreased body
weight and ultramicroscopical lesions exceeded the MTD. Whether
exceeding the MTD in this way would invalidate the study results is
also controversial.
Comment: .A decrease in body weight (in week 82, in male rats 8.3% and
23.2% in female rats) was significant for the 4% dose level only. In
another semichronic rat study (114), smaller decreases in body weight
were observed for 1% and 2% as well, although they were not always
statistically significant. One method to address this is to disregard
the response at the 4% dose level. One consequence of disregarding
this response is to remove any evidence that R-3 induced carcinomas.
The excess of ultramicroscopical lesions was observed at all dose
levels, but the observations were not quantitative. Although these
lesions appear R-3 related, there is no information suggesting that
they are indicative of an abnormal metabolic mechanism that would not
have occurred at lower dose levels. The Panel feels that, the issue of
weight loss indicating that the MTD is exceeded requires a risk man-
. agement decision. Risk estimates have been provided both with and
without the 4% group.
5. In choosing the NOEL it is assumed that the observation of no effects
in a study of limited size can be applied to the entire US population.
89
-------
July, 1987
Comment: The invalidity of this assumption is clear. One method for
addressing this would be to apply a safety factor to account for the
difference in population size. There are several options open for
selecting a NOEL, each associated with different additional assump-
tions or problems. Although frequently used to indicate that there is
no risk below a certain threshold dose, actually the NOEL does not
imply such threshold. The NOEL (actually the NOAEL or "no observed
adverse effect level") simply means that under the study conditions
"(size, duration of exposuref and observation, exposure level, and route
of exposure) no adverse effect, in this case tumors, have been observ-
ed. It does not rule out that a larger study size or a longer
observation period could have resulted in an observable excess tumor
incidence. Defining the target effect for a NOEL is a difficult and
controversial issue, because of the ambiquity in the term "adverse
effect". The following options are offered:
a. One option is to assume that TSH is the mediator of the tumori-
genic effect of R-3 in rats and humans, and thus an increase in
the TSH-level in humans can be used as a parameter of, effect.
Comment: The NOEL may be set at 20 or 60 mg R-3/day. However, in
part because of the problems associated with interpretating Study
Wl, the Panel cannot conclude with certainty on what is the high-
est observed NOEL in humans. It has been discussed in chapter 6
and in assumption #2 in this chapter that there is evidence that
TSH may not be the mediator of thyroid carcinoma, and, therefore,
tumorigenesis in humans. This evidence is not sufficiently con-
clusive to reject using the assumption of TSH mediation. Thus, as
stated before, this decision requires science policy. If it is
decided that TSH is NOT a mediator in tumorigenesis in humans, a
NOEL for TSH would have no meaning.
b. If it is decided that TSH is not a suitable parameter indicating
the tumorigenic potency of R-3 in humans, one may opt to choose a
NOEL for excess tumor incidence.
Comment: The Panel has developed two-possibilities for illustra-
tion of the technique to use if this option is chosen, reflecting
the science policy nature of this issue.
Possibility 1. Use Study 410-002 in rats (72), which shows an
90
-------
July, 1987
increase in tumor incidence at the lowest dose tested in males,
0.1% (or 44 mg/kg-d). If, traditionally, one divides this value
by safety factors of 10 for each of the following: the difference
in species; for the intraspecies variation; and for the use of an
effect level; then the ADI becomes 44 mg/kg-d / 1000 or 44
ug/kg-d. '
Possiblity 2. Since there has been no long-term human study of
R-3 induced tumors to derive a NOEL, use either the other multi-
generational study in rats, (Study Id in Chapter 4) which did not
show an excess of tumors at the highest dietary dose of 4% R-3, or
the animal study with the largest number of animals per dose group
(Study 2a in Chapter 4) involving mice with no excess tumors at <
the highest dietary dose of 3% R-3 in a one-generation study.
Using a NOEL from a rodent study requires that it is established
in a study of sufficiently long duration and size (current
standard is about 50 animals per dose group).
Study Id above was too short (14 months) and comprised only 25
animals per dose group. On the other hand, this study involved
uhree generations compared to the two-generation IRDC studies. If
.one applies a total safety factor of 100 to account for: a) the
too small study size; b) the too short duration of the study; c)
intraspecies variation among humans; and d) the possibility of
humans being more sensitive to the tumorigenic effect of R-3 (if
the action is not involving TSH); then the ADI derived from this
rat study would be 0.04% of the diet, equal- to approximately 1.6
mg/kg-d (using food intake and body weight observed in the IRDC
study), or the equivalent, for a human, of 100 mg/d.
For study 2a, in mice, one can use a total safety factor of 100
to compensate for: a) the study was not multigenerational; b)
intraspecies variations among humans; and c) possible greater sen-
sitivity of humans. The ADI would then be 1/100 of 3% or (at 5
g/d food intake and 30 g body weight) 50 mg/kg-d, equivalent in
the human to 3 g/d. As stated before, the magnitude of safety
factors and their application is considered to belong in the
domain of risk management. It needs mentioning that the value of
10 for a safety factor is based on tradition. , There would be no
91
-------
July, 1987.
reason why larger or smaller values may not better approximate
reality. There is some evidence that in each of the above cases,
a value of 10 for a safety factor over or under protects against
the effect of the uncertainty of interest.
The use of possibility 1 assumes that one will not accept bas-
ing a NOEL on wholly inadequate studies. Using possibility 2 can
be done if one feels that Study 410-002 is as inaccurate as the
negative studies.
c. If one would opt for R-3 not being a human tumorigen, a NOEL could
then be chosen for any adverse, effect. In that case, the rabbit
study mentioned in Chapter 4 under C2 is the study with the lowest
exposure levels. It appears that 12.5 mg/kg-d of R-3 (equivalent
to 760 mg/d in humans) was the NOEL for fetal toxicity. Again,
there is the question of the appropriate safety factors to use.
When reporting on R-3 studies, it is assumed that the "compound ef-
fects" are, in fact, attributable to the halogenated fluorescein, R-3.
Comment; There is sufficient evidence that excess iodide can cause
thyroid tumors in the rat, while its effects on TSH, T^, and 1^ both
in rats and humans have long been known. In ' this light, most of the
findings of the R-3 toxicity studies can better be explained by al-
ready known properties of. iodide (derived from deiodination or from
free iodide present as an impurity) rather than by assuming that the
halogenated fluoroscein is the responsible factor. The suppression of
T^-T3 conversion appears to be R-3 specific, but this feature has not
been adequately studied for iodide. Although there is evidence that
iodide rather than R-3 may be the oncogen, there is insufficient
evidence to preclude R-3 as the toxic factor. A decision in this
issue hence requires risk management considerations. It is worthy to
mention that, in view of option c for assumption 5, a NOEL for iodide
pertaining to non-cancer effects was established at 0.27 mg iodide/
kg-d from a study involving humans exposed to iodide in drinking water
(64). This is confirmed by a study showing that humans exposed to 1
ppm iodide in drinking water (about 0.3 mg/kg-d) for over three years
did not appear to have an increase of T4 (64).
92
-------
7.
8.
July, 1987
It is assumed that the dose scale for rats and humans is the same, if
the dose is expressed in mg/kg body weight.
Comment; Hoel et^ al. (157) have suggested use of mg/m2 body surface
area as an alternative scale on the basis of the acute toxicity of
anticancer drugs in humans. The use of this metameter would increase
the extrapolated (from rats) cancer risk estimates in Chapter 7 by a
factor of 7. However, that this may not be appropriate for general
use is shown by Crump , _ejt al. (158), who concluded that the NTP
bioassays of over 200 chemicals gave the closest correlation among
animal species when the mg/kg body weight metameter was used. This
was also supported by Crouch and Wilson (153), although there are
qualifications in using this study. The Panel, considering the
available information, has opted for using this metameter.
Any risk assessment has to rely on the accuracy of the data. It is
known that the interpretation of thyroid tumor histology is difficult,
leading to different outcomes. The thyroid lesions of the IRDC study
were not an exception. The Panel felt unable, within the constraints
of its charge, to attempt another round of diagnostic readings by
pathologists other than the four listed in Tables 13-15. Although
there were no sufficient arguments to favor one reading over the
others, risk estimates in Chapter 7 have been based on the CCMA con-
sultant's data for females and .the" FDA data for males because they
were the most complete of the, readings. An estimate of the slope of
the various sets of readings would yield a range of estimate of which
the extremes have a ratio of approximately 2.
The use of the linearized multistage model to extrapolate risk from
high to low dose does not imply that the Panel assumes that this par-
ticular model is more appropriate than others. As stated earlier in
assumption #3, various models may equally well fit the observed data
points. The choice of the multistage model, and its impact on the
risk estimate, has been explained extensively in Chapter 7. '
11. There are several options for selecting the pathology data for risk
estimation:
9.
93
-------
July, 1987
a. Use the sum of follicular adenomas and carcinomas in most sensi-
tive strain/species (rats) with all doses for each sex.
b. Use the sum of follicular adenomas and carcinomas in most, sensi-
tive strain/species (rat) for each sex, excluding the 4% group
because it may have'exceeded the MTD.
c. Use the combined' incidence of follicular adenomas in male and
female rats. The rationale would be that the risk is then
relevant to the combined population rather than to each sex
' i
individually.
d. Use the incidence of all tumors combined, regardless of the histo-
logic type or gender, i.e., add C-cell and follicular tumor
together.
Comment; Current practice does not support option d. The scientific
support for the first three options seems to be of equal weight.
Which of the options a-c should prevail can, therefore, not be decided
on the basis of scientific arguments alone. Chapter 7 provides risk
estimates for the options a and b. Option c was not included because
the doses on a mg/kg basis were different for the two sexes.
C. Analysis
Compared to the R-3 component of a risk assessment of the external us.e
of six dyes (4), this Report seems much less conclusive in its assessment
of the risk to humans from R-3. However, it should be appreciated that in
the former document, the assumptions that were quantified were, essent-
ially, ones involved in exposure assessment of externally applied R-3.
Estimation of exposure is Very much less dependent on assumptions, as
available information usually allow making some rough estimates in the
absence of more accurate data. However, careful reading of the previous
document will show that a number of qualitative assumptions remained. In
addition, the previous document did not address the carcinogenicity of R-3.
It was assumed to be carcinogenic based on a science policy decision that
the Panel felt was made by FDA, that a positive effect in any adequate
chronic study indicated that the compound was a carcinogen. This assump-
tion was not made in- this document. Finally, much less information was
then available to the Review Panel than is now.
94
-------
July, 1987
The assumptions and the options outlined and discussed in this chapter
may seem .to be confusingly complex to the reader not very familiar with the
subject. However, it should 'be recognized that, different from the case of
R-3 for cosmetic use (4), the Panel has now ventured into the uncertain
path of separating science from "science policy" and policy. Its charter
offered the Panel ^sufficient room to conduct a risk assessment based only
on scientific considerations. The Panel has made considerable and consis-
tent efforts to, explicitly indicate where it feels that decisions in
certain issues cannot be( made on the basis of scientific arguments alone.
In doing so, the Panel realizes that this report lacks the simplicity of
the conclusions of the first report on R-4 (4), but at the same time it can
be claimed that this report attempts a comprehensive risk assessment of a
chemical not complicated by policy considerations. Responding to its char-
ter, the Panel has also outlined in the Addendum a number of studies that
may be considered to provide information to fill the gaps in knowledge,
thus reducing the necessity of assumptions.
The above assumptions and options can easily be distinguished in three
categories pertaining to issues dealing with: 1) the question whether or
not R-3 is toxic to humans, 2) the- magnitude of the risk or the exposure,
and 3) how to express risk. With respect to the first category, the op-
tions offered to the risk manager, and their .impact on risk, is clear and
simple, since the choice is between yes or no. If it is decided that a,
given assumption is invalid, e-.g., that animal tumors can be extrapolated
to humans (assumption 1), then the outcome of the risk assessment is simply
that there is no tumorigenic risk to humans. The second category provides
exposure options which would alter the risk estimate in either direction,
but not to the extent that there is no risk. If a change in the exposure
estimate is made due to a decision in one of the controversial issues, the
new exposure estimate can be translated in a new risk estimate by direct
proportionality. The last category of options may be the most difficult
one offered: whether to manage the risk through the NOEL approach or
through mathematical model risk estimates.
The Panel does not express its preference for either, but it would
like to repeat that*a distinction should be maintained between the toxicity
•of iodine and that of R-3 proper. Integrating all available information,
it appears that the preponderance of information is compatible with the
9.5
-------
July, 1987
statement that the toxicity of commercially available R-3 is largely or
entirely due to the iodine component, present as an impurity and from
deiodination of R-3. The important feature not yet explained, by iodide
alone is effect on the TS - TA conversion, which seems an effect of R-3
proper. With the currently available information, ascribing all observed
toxic effects to R-3 proper thus seems largely conjectural. In contrast,
/ f4
reduced body weight despite creased food intake, effect on TSH-T3-T4,
thyroid tumors and hyperplasia, and fetotoxic effects are all well-known
toxic properties of iodine. Although iodine is a recognized animal carcin-
ogen, a Recommended Dietary Allowance of 0.15 mg/d has been issued (161),
equal to 50% of the NOEL, and present in 2 grams of iodized table salt
(which is about the daily per capita used of salt in the USA), while the
range of iodine intake is 0.240 - 0.740 mg/d (64). Another example of
iodine as a food additive is the use o'f iodine containing dough conditioner
in bread. A dose of 1.4 mg R-3 equals an intake of approximately 0.01 mg
iodine, as compared to 0.076 mg in 1 gram of table salt.
D. Conclusions
1. R-3 is a rat oncogen with equivocal evidence of carcinogenicity and
with some evidence for causing benign thyroid tumors.
2. The tumorigenic effect of R-3 on rats is more likely to be the result
of an indirect (secondary) mechanism.
3. It is likely, not certain, that the tumorigenic effect of R-3 is
attributable to its iodine component present in part as an impurity.
4, There is insufficient evidence to support the assumption that R-3 is
tumorigenic to humans, but this possibility cannot be ruled out.
5. The average per capita exposure to R-3 through food and internal drugs
is estimated at 1.41 mg/d. , »
6. If it is assumed that R-3 poses a tumorigenic risk to humans, the risk
from ingesting R-3 containing food and drugs is small, that is, the
96
-------
7.
July, 1987
number of people with R-3 .induced tumors would be too small to be
observed by epidemiologic or other human studies. •
Several uncertainties in the risk assessment of R-3 cannot be solved
on the basis of scientific arguments alone. These situations have been
outlined, and options have been presented for risk management deci-
sions, m addition, Chapter 9 lists a large number of studies that may
be conducted to -provide additional information to clear a part of the
uncertainties.
8- Two options have: been offered as a basis for setting any human exposure
level: the use of biologically-based mathematical models or the NOEL
approach.
97
-------
-------
RISK CHARACTERIZATION
Attribute 3 The report outlines specific ongoing or potential research
projects that would probably clarify significantly the extent of
uncertainty in the risk estimation.
SOURCE Case Study J. Red Dye No. 3 (Pages 98-101).
Note See Hazard Identification Attribute 4 in this Append!
-------
-------
B.5
RISK CHARACTERIZATION
Attribute «l jftg report provides a sense of perspective about the risk
through the use of appropriate analogy.
Note None of the case studies provided an example illustrating the use of
appropriate analogy.
*U.S. GOVERNMENT PRINTING OFFICE: 1990-718-159/20166
-------
-------
-------
""** 03
""* W
75'
= • CD
I $
CD
tn
CD
CO
O
o
<
f
VD
o
u>
r fa?
S *C 3" i-i
3-00
III
2 o
f *
If
(S cn
,-i to
35
5 =
5 m
(D O
— m
?P
5.° 3
§s!
C 03
3 •<
S3-
~ Q)
3" 3
38"
CD Sli
CD
O
5"
D'
O.3
0°
=' 3
3 0)
Q) rt-
— o
O5
00
------- |