DRAFT
Background Document on the Development and
Use of Reference Doses
Part I: Data Needs and Apportionment
Prepared for:
Office Of Solid Waste
U.S. Environmental Protection Agency
Washington, D.C.
Prepared by:
ENVIRON Corporation
1000 Potomac Street, N.W.
Washington, D.C. 20007
December 20, 1985
-------
DRAFT
Table of Contents
Page
1. Introduction - Purpose and Organization 1
2. Introduction to the Concept of Acceptable Daily Intake (ADI) 4
2.1 Origins 4
2.2 Use by EPA 7
2. 3 Media-Specific Limits 9
2.4 Utility and Limitations of the ADI 9
3. Minimum Data Needs for Establishing ADIs 11
3.1 Introduction 11
3.2 Utility and Limitations of Various Types of Toxicity Tests 12
3.2.1 Basic Concepts 12
3.2.2 Acute Toxicity Studies 14
3.2.3 Subchronic Toxicity Studies 16
3.2.4 Chronic Toxicity Studies 18
3.2.5 Reproductive Toxicity Studies 19
3.2.6 Teratology Studies 21
3.3 Summary and Conclusions... 21
4. Special Issues in the Use of Toxicity Data to Derive ADIs 26
4.1 ADIs for Essential Nutrients 26
4.2 Mixtures and Toxicological Interactions 27
4.2.1 General Types and Mechanisms of Interaction 27
4.2.2 Interactions in Contaminated Air or Water 29
5. Apportionment of RfDs and RSDs 31
5.1 Introduction 31
5. 2 Apportionment Among Media and Sources ' 31
5.3 Relationships Between Air or Water Concentration and Human Dose 37
5.4 Apportionment Between Air and Water 42
6. Conclusions and Recommendations 52 -- (0O
-------
DRAFT
1. Introduction Purpose and Organization
The Office of Solid Waste (OSW) of EPA is proposing certain restrictions
on the land disposal of hazardous wastes. The principal concern of these
restrictions is the problem of long-term, low-level release of hazardous
chemicals from land disposal sites that may arise because of the deterioration
of containment systems. To ensure protection of human health, OSW proposes to
place limits on the extent of air or water contamination that may result from
any such releases.
Limits are to be proposed for individual chemicals to protect humans from
the possible adverse effects of repeated, low level exposure (chronic
exposure). (The Agency has already promulgated regulations dealing with
single or infrequent, high-level exposures that may arise because of
accidents.) The two principal determinants of these limits are:
(1) for substances not known to display carcinogenic properties, the
acceptable daily intake (ADI), hereafter to be referred to by EPA as
the Reference Dose (RfD).
(2) for substances known to display carcinogenic properties, the
lifetime average daily dose corresponding to a specific level of
excess lifetime cancer risk, hereafter, the Risk-Specific Dose (RSD).
These two determinants are well-established and widely-accepted health
protection criteria. They satisfy the goal of protecting humans from chronic
exposures to chemicals that may be released from various sources, to the
extent current scientific knowledge can allow (NRC, 1980; 1983). EPA is
abandoning the use of the term "Acceptable Daily Intake", because it may be
read to imply that doses in excess of it are necessarily "unacceptable." As
will be seen, this is an incorrect interpretation, and the Agency believes use
0774S/122085 - 1 -
-------
DRAFT
of the more neutral term "Reference Dose" avoids this difficulty. The
principles upon which RfDs are based, and the data used to derive them, are
identical to those traditionally used to derive ADIs.
The term "Risk Specific Dose" has not previously been used; but it is
simply a convenient way of identifying the dose of a carcinogen corresponding
to a specified level of lifetime risk.
EPA is relying on a number of expert scientific reviews and agency
documents to support use of these two health protection criteria. But there
are several aspects of the proposal that require review and analysis not found
in any existing documents. EPA thus asked ENVIRON to prepare such a review,
focusing on the following principal issues:
(1) EPA proposes to develop toxicity data on chemicals for which limited
or no data are currently available. It is thus necessary to assess
available test methodologies and to identify those suitable for
developing data from which RDs can be established. (No similar
review is needed for RSDs, which are developed from carcinogenesis
bioassay data.)
(2) Chemicals released from waste sites may enter both air and water,
creating two possible routes of human exposure. In addition,
chemicals found at waste sites may also be present in other media
(e.g., a pesticide that is also present in the diet). It is thus
necessary to decide whether and how to apportion RfDs or RSDs among
the several possible human exposure media.
In addition to these two major issues, a number of ancillary points arise
in the approach proposed by EPA. These include: 1) the scientific basis for
the RfD as a protective device; 2) methodology for deriving RfDs from various
types of toxicity data; 3) the accuracy and precision of RfDs; 4) the
development of RfDs for certain metals that are also essential nutrients
(e.g., copper, selenium, chromium); and 5) the problem of interactions among
0774S/122085
-------
DRAFT
chemicals. Although ENVIRON's report is organized around the two major
issues, these additional points will be included in the discussion.
In the next section, a broad introduction to the concept of the RfD is
provided. This is followed by a discussion of the types of toxicity data from
which RfDs can be established, along with a presentation of the strengths and
limitations of various types of data. The purpose of this section is to
identify the types of data believed necessary to develop a reliable RfD.
The report then moves to a discussion of establishing RfDs from various
types of data and of the several ancillary issues relating to RfDs described
earlier. We then examine the apportionment issue, and describe the options
available to EPA and the strengths and weaknesses of each. All of these
issues are presented as Part I of this report.
Because all of the published scientific literature pertaining to RfDs
refers to ADIs, we retain the latter term in the following discussion, even
when we refer to EPA's own earlier literature. It should be noted that EPA
has altered only the label attached to the term, and has not altered its
underlying basis. It is for this reason that all of the information relating
to ADIs is directly relevant to EPA's proposed development and use of RfDs.
A discussion of the various considerations influencing the design of
protocols for toxicity testing, provided to guide identification of the most
cost-«ffective means to collect toxicity data, is presented in Part II of this
report.
0774S/122085
-------
DRAFT
2. Introduction to the Concept of Acceptable Daily Intake (ADI)
2.1 Origins
According to a Committee of the National Academy of Sciences
"The acceptable daily intake (ADI) of a chemical
is defined as the dose that is anticipated to be
without lifetime risk to humans when taken
daily. It is not assumed that this dose
guarantees absolute safety (NRC, 1980)."
This definition is essentially the same as that given by the World Health
Organization (FAO/WHO, 1958; 1965), the EPA (see, e.g., various Health Effects
Documents), and the FDA (FDA, 1982).
Experimental data on toxicity is typically collected in small groups of
experimental animals at doses sufficiently high to produce directly observable
forms of toxicity. Such experimental studies can reveal the dose-effect
relation for the chemical, as well as the maximum dose* at which toxicity is
not observed (termed the no-observed-effect level, NOEL).
Faced with this type of data for several food additives, Lehman and
Fitzhugh (1954) proposed that ADIs could be established by dividing the
experimental NOEL by a "safety factor." These authors (who were FDA
officials) cited acute toxicity data suggesting that, for some substances.
* Because of practical limitations on the number of dose levels used in an
experiment, it is usually not possible to identify the true maximum NOEL.
The measured NOEL is, in many cases, less than the true maximum NOEL.
0774S/122085 - 4 -
-------
DRAFT
small groups of relatively homogeneous experimental animals were ca. 10-fold
less sensitive to their toxic effects than were members of the general human
population, and then reasoned that the variability in response expected among
members of the human population might make some members ca. 10-fold more
sensitive than the "average". These notions, coupled with the long-standing
idea that chemical toxicity did not become manifest until the dose exceeded a
threshold value, led Lehman and Fitzhugh (and the FDA) to the conclusion that
they could estimate a human population ADI by dividing the chronic,
experimental NOEL by a "safety factor" of 100.
The FDA recognized that the ADI was not a guarantee of absolute safety.
They also recognized that human exposures for many substances might well
exceed the ADI by some (undefinable) amount for extended periods without
resulting in human chronic toxicity. That is, it was recognized by the
original developers of the ADI that the figure was only an estimate based on
incomplete knowledge, and that it should not be considered a sharp dividing
line between "safe" and "unsafe" chronic exposures (Lehman and Fitzhugh, 1954;
FDA, 1982; Rodricks and Taylor, 1983). Instead, the "NOEL-safety factor"
approach is a practical device for deriving acceptable exposure levels, for
various regulatory and public health purposes, in the face of limited
scientific information and knowledge.
FDA has also derived chronic ADIs for substances for which chronic (i.e,
lifetime) toxicity data were not available. When, for example, the only data
available for a substance revealed the effects of subchronic exposure (e.g.,
90-day exposure studies in rodents), FDA incorporated an additional 10-fold
safety factor to derive an ADI. Thus, the NOEL from subchronic studies was
0774S/122085 - 5 -
-------
DRAFT
divided by 1000 to establish the chronic ADI. The agency (and other
investigators as well, see below) justified this practice on the principle
that, with the exception of carcinogenicity, there is very high confidence
that all of the major toxic effects of a chemical can be found in carefully
designed subchronic studies, and that chronic studies would merely extend the
dose-response curve (by ca. 10-fold) for the effects observed after subchronic
exposure. (Additional, detailed discussion of this point is presented in
Section 3, below). The additional 10-fold safety factor was thus used as a
substitute for the dose-response data that would be obtainable at the lower
doses used in chronic experiments.
Scientists associated with other national and international organizations
have also adopted the concept of ADI as a health protection device.
Scientists associated with the World Health Organization (WHO) and the Food
and Agricultural Organization (FAO) have further justified a 100-fold safety
factor for food additives based on differences among species in body size,
food requirements, water balance exchange, and variations in susceptibility to
the toxic effect. This rationale and approach were also accepted by the
FAO/WHO Expert Committee for Pesticide Residues (FAO/WHO, 1965).
A committee of the National Academy of Sciences (NRC, 1977) estimated
ADIs for contaminants in drinking water using an approach similar to that of
FDA, but used "uncertainty" (rather than "safety") factors to account for the
limitations in the data base and in our knowledge of inter- and intra-species
variability in response.
OT74S/122085 - 6 -
-------
DRAFT
2.2 Use at EPA
Several uncertainty factors have been used to estimate ADIs depending on
the type and quality of available human or animal toxicity data. At EPA, the
magnitude of the chosen uncertainty factors depends on the differences between
the human exposure characteristics and the conditions of the experimental
studies used to derive the ADIs (now, RfDs). Further, if the no-observed-
effect level (NOEL) is sufficiently close to the ambient exposure level, and
there is no evidence of adverse effects at these levels, then relatively small
uncertainty factors have been used. Also/ detailed knowledge of a chemical's
mechanism of.toxicity, critical effect, and pharmacokinetic behavior in humans
and experimental animals may permit modification of the standard (generic)
uncertainty factors for some substances. Such information is, however, seldom
available to influence estimation of the ADI.
An uncertainty factor of 10 is used by EPA to estimate ADIs from
appropriate human data; its purpose is to account for intraspecies variability
in response to the adverse effects of a chemical. An uncertainty factor of
100 is used with relevant (with regard to duration and route of exposure)
animal data from properly conducted chronic studies; this factor accounts for
both intra- and inter-species variability. If only marginal data are
available (e.g., data from subchronic studies in animals), an uncertainty
factor of 1000 is used; this figure incorporates the uncertainty in
extrapolating from one duration of exposure to another and also accounts for
intra- and inter-species variability. This approach essentially matches that
of FDA, WHO, and the NRC.
0774S/122085 - 7 -
-------
DRAFT
Additional uncertainty factors have been used to compensate for other
short-comings in experimental information. These additional uncertainty
factors were incorporated when the only data available revealed a lowest-
observed-effect level (LOEL) rather than a NOEL, when subchronic data were
used to project potential chronic effects for humans, or when there were other
deficiencies in the data base upon which decisions had to be made. Recently/
Oourson and Stara (1983) demonstrated that some of these additional factors
(which typically range from two to ten) have some experimental support and are
likely to be highly protective for many chemical substances.
ADI's have been developed by EPA's Environmental Criteria and Assessment
Office (ECAO) (EPA, 1982; 1984) and EPA's Office of Pesticide Programs (OPP).
The following guidelines for deriving ADIs from toxicity data have been
adopted by some groups at EPA (EPA, 1980a).
Doses associated with an increase in frank toxic effects, such as
mortality or convulsions, are not suitable for derivation of an ADI,
A free-standing NOEL is unsuitable for derivation of an ADI. If
multiple NOELs of equal quality are available without additional
data on LOELs, NOAELs, or LOAELs, the highest NOEL shoud be used to
derive an ADI.*
A NOAEL, LOEL or LOAEL can be suitable for an ADI derivation. A
well-defined NOAEL from a chronic or subchronic study can be used
directly, applying the appropriate uncertainty factor, and is
preferred. For a LOEL, a judgment must be made as to whether it
actually corresponds to a NOEL or a LOAEL. In the case of a LOAEL,
an additional uncertainty factor is applied; the magnitude of the
additional uncertainty factor is not to substitute levels for which
severely adverse effects are seen. (For some groups at EPA, no
differentiation is made between NOEL and NOAEL or between LOEL and
LOAEL.)
* The NOAEL and LOAEL include the additional adjective "adverse"; in many
cases an effect may be observed at a given dose, but may not be adverse
to health. If- this is the only effect observed, the dose may be labeled
NOAEL.
0774S/122085 - 8 -
-------
DRAFT
If — for reasonably closely spaced doses - only a NOEL and a LOEL
of equal quality are available, the appropriate uncertainty factor
is applied to the NOEL.
As an additional general rule, EPA does not ordinarily consider it
appropriate to use recommended occupational exposure levels, such as Threshold
Limit Values (TLVs), for directly deriving ADIs; nevertheless, the data bases
which were used to derive TLVs or other occupational exposure levels may be
appropriate for use in deriving an ADI. In some instances the TLV may be
directly useful, if its derivation was based on the same general principles
used to devise ADIs.
2.3 Media-Specific Limits
For many chemicals, human exposure may occur through several media (air,
water/ food/ direct soil contact). In such cases it is important that the total
exposure from all media not exceed the ADI. Human intake or contact with
various media must thus be taken into account when estimating the maximum level
that can be tolerated in each medium without the ADI being exceeded for any
individual. Several methods have been developed to deal with this issue, and
they will be discussed in Section 5, on apportionment.
2.4 Utility and Limitations of the ADI
Since its introduction the ADI has been widely used as a practical,
health-protection device. For this reason, it appears to be the appropriate
criterion for establishing limits for substances migrating into air or water
from hazardous waste sites. While no better means for accomplishing EPA's
health protection objectives is available, the limitations in the ADI should be
recognized. The following list of limitations and other characteristics of the
0774S/122085 - 9 -
-------
DRAFT
ADI has been assembled from all of the various sources cited in the foregoing
discussion.
(i) The ADI does not represent a sharp dividing line between "safe" and
"unsafe" exposures.
(ii) There is no readily definable way to estimate the magnitude of the
uncertainty in any given ADI. Uncertainties arise from many sources,
and are related to the quality and completeness of the toxicity
data, the uncertainty in the dose-response information and the NOEL,
and the lack of chemical-specific data on intra- and inter-species
variability in response. There is no means available to quantify
accurately the accuracy and precision of ADZs.
(iii) The safety (or uncertainty) factors in common use are probably
overprotective for some substances, because they have been selected
to protect against "worst-case" substances. It is possible,
however/ that in a few cases they are underprotective.
(iv) Brief excursions above the ADI can probably be tolerated without
harmful effect by all members of the general population. There is
no precise definition of "brief".
(v) Consistent protection at or near the ADI ensures that individuals
will be protected from the acute effects of all chemicals.
Protection against acute toxic effects usually requires safety
factors no larger (and sometimes smaller) then those used for
establishing the ADI. Because the chronic NOEL will always be a
lower dose that the minimum effective acutely toxic dose, the ADI
will clearly protect against acute toxicity.
(vi) ADIs have not been established or used for carcinogens, on the
ground that threshold doses may not exist for this class of agent
(i.e., carcinogens pose a finite risk at all finite doses, with the
risk increasing with dose).
These characteristics and limitations of ADIs apply, of course, to RfDs,
and should be kept in mind when using them or explaining their basis. Several
will be discussed more fully in the Sections to follow.
0774S/122085 - 10 -
-------
DRAFT
3. Minimum Data Meeds for Establishing APIs
3.1 Introduction
Some toxicity data are available from controlled studies in exposed
humans, but these are typically limited to short-term exposures, producing
rapidly reversible effects. Epidemiological data are available for a
relatively large number of important chemicals, but such data usually lack
quantitative dose-response information and, in many cases, are ambiguous with
respect to the issue of causation. In all but a few cases, toxicity data from
human studies are either not available or inadequate to establish ADIs (NRC,
1980; EPA, 1984). Because of this it has become necessary to rely upon data
from studies in experimental animals. In this section we describe the various
types of experimental tests available for collecting toxicity data and the
types of information provided by each of the various tests. We also describe
the limitations in each of the types (i.e., what they cannot reveal). The
discussion is limited to types of toxicity tests that have been sufficiently
validated for use in regulatory standard-setting.
The purpose of this discussion is to identify the minimum amount and type
of data necessary to establish an ADI for chronic human exposure. It should
be noted that identifying minimum data requirements is not a strictly
scientific undertaking, because it is possible, as a policy matter, to use
safety or uncertainty factors to compensate for almost any kind of data gap
(and some degree of scientific support can probably be found for such
selections, see ENVIRON, 1985). While we adhere to these general concepts, we
point out in the closing evaluation subsection what appears to be the current
0774S/122085 - 11 -
-------
DRAFT
consensus in the scientific and regulatory policy communities regarding
minimally acceptable data for establishing ADIs.
3.2 Utility and Limitations of Various Types of Toxicity Tests
3.2.1 Basic Concepts
A fundamental assumption in the estimation of human health risks posed by
chemicals is the ability to extrapolate animal test results to predict human
response. This is the cornerstone of most regulatory decisions regarding the
safety of substances in the environment, in the food supply, or in drugs.
This assumption, however, is not based on complete certainty regarding the
predictive power of animal models. Rather, it is based on the widely-accepted
view that well designed animal studies provide an indication of potential
human toxicity and that the strength of the indication is a function of the
rigor, completeness, and reproducibility of the test animal studies. This
section contains an examination of the various standard toxicity studies
currently accepted by various regulatory and public health agencies and a
summary of the information each provides (and does not provide) about
potential human toxicity. This review will proceed from least to most complex
test type, and emphasizes the confidence that can be placed in the results of
each type of test. It will be made clear what types of effects might not be
detected at each level of testing, and what uncertainty would remain if
testing were to cease at a given level. A tabular summary of this information
will be provided, with some estimate of the cost of moving from one level of
testing to the next. The cost figures allow some judgment regarding the value
of obtaining new information, or of the cost-effectiveness of each test type.
0774S/122085 - 12 -
-------
DRAFT
Required or widely accepted general toxicity tests are classified as
acute, subchronic, and chronic. These tests are "general" because they are
designed to identify the full range of toxicities associated with a chemical.
Additional tests, however, have been found necessary to identify specific
effects not readily observable in the general toxicity studies. Tests for
reproductive injury, teratogenicity, and genetic effects are among those
widely used for such purposes. Cancer bioassays are generally considered a
subcategory of the chronic test. There are tests available for other specific
endpoints (e.g., behavioral and neurological injury and adverse effects on the
immune system) that have not yet been widely accepted as valid indicators of
human toxic potential - i.e., they are still in the developmental state.
Metabolism and pharmacokinetic studies are becoming increasingly important
components of a toxicity profile.
The specific protocols for each of these types of tests vary somewhat
among agencies, but they nevertheless provide the same basic information.
In the ideal, it would seem that determination of an AOI for a chemical
to which humans would be chronically exposed would reguire chronic animal test
data in several species. These studies are used to identify the range of
chronically toxic doses, and to establish the NOEL or NOAEL. Fiscal,
manpower, and legal constraints often require that the "ideal" level of
testing be adjusted to a more realistic level. Most regulatory agencies have,
therefore, adjusted the level of testing required for a compound according to
the magnitude of expected human exposure and the outcome of previous,
less-than-lifetime toxicity studies. Current toxicity testing strategies are,
therefore, hierarchical sequences of tests designed to develop a profile of a
0774S/122085 - 13 -
-------
DRAFT
substance's toxicity. The hierarchy generally consists of several levels, or
tiers, of tests. The lowest or initial tier consists of relatively rapid,
inexpensive tests intended to identify the acute toxicity of the compound.
Often included in the initial tier are short-term and genetic toxicity tests
that rapidly provide information about potential carcinogenic effects. This
first tier information, although not directly useful in predicting chronic
adverse effects in humans, can be used to guide decisions about the need for
and type of more extensive testing. The second tier of testing may include
subchronic, whole-animal types of tests that require 1-3 months to conduct.
Later tiers are intended to yield direct information on the chronic toxicity
of the substance, and on its effects on reproduction or development (FDA,
1982).
3.2.2 Acute Toxicity Studies
Acute toxicity studies are used to provide an estimate of the adverse
effects that would be associated with a single exposure to a chemical. In
addition, they provide an estimate of the relative susceptibilities of various
species and sexes, identify target organs, suggest mechanisms of action, and
assist in selection of dose levels to be used in longer-term studies. The
most common measure of acute toxicity is the median lethal dose (LDSO or
LCso). It should be emphasized that estimation of an LDso is not
eguivalent to describing the acute toxicity of a compound. A well designed
acute toxicity study will also include consideration of non-lethal parameters
of morbidity or pathogenesis.
In general, a battery of acute exposure studies is usually used to
describe the acute toxicity of a compound by several routes of administration.
0774S/122085 - 14 -
-------
DRAFT
These include tests by the oral, dermal, and inhalation routes; skin and eye
irritation studies are also considered at this phase of testing. These data
are necessary to protect workers and others who may be exposed during
production, transport, use, and disposal of a chemical. For the toxicologist,
acute studies can provide information on the possible mechanism of action of
the compound, its target organs, the reversibility of the effects, and on
structure-activity relationships. Such information also assists, indeed is
necessary, in the design of longer-term studies.
It should be emphasized that acute studies do not provide any information
about the cumulative effects from subchronic or chronic exposure to a compound,
reproductive effects, teratologic effects, or carcinogenic effects. In other
words, they reveal nothing about the nature of toxicity that will arise after
repeated exposures.
McNamara (1976) examined a series of 122 non-carcinogenic compounds and
concluded that the LD5o could be divided by 100 to estimate a subchronic
NOEL or by 1,000 to estimate a chronic NOEL. Furthermore, in a report
prepared by ENVIRON (1985), it was shown that for 85 non-carcinogenic
compounds, the LD50 could be divided by 119 to estimate a subchronic NOEL
and 3,120 to estimate a chronic NOEL. Use of such safety factors to develop a
chronic or subchronic NOEL from LD50 data, based on these type of empirical
analyses, has not been accepted by the general toxicology community and is
thus not yet considered a reliable method for estimating ADIs, at least for
deriving health-protection limits. An ADI derived on the basis of these types
of empirical observations, which concern only quantitative factors and which
are not based on any real knowledge of toxicity, would seem to be useful only
0774S/122085 - 15 -
-------
DRAFT
in cases where decisions clearly have to be made before more appropriate data
could be collected.
3.2.3 Subchronic Toxicity Studies
Subchronic toxicity studies are used to determine the toxic effects that
occur from repeated exposure for various fractions of an animal's life span,
and to identify the NOEL for these effects. Such studies provide information
about target organs, physiologic and metabolic capacity of the animal to
tolerate prolonged exposure, and cumulative toxicity.
An important component of Subchronic studies is the use of a broad screen
of measures which will detect most forms of toxicity. These include daily
behavioral observations, periodic physical examinations, body weight and food
consumption monitoring, analysis of hematologic parameters, and clinical
screening of blood and urine. Of most importance is the conduct of gross and
histopathologic examinations of animals, and collection of organ weight data
at sacrifice.
The period of exposure for a Subchronic study is dependent on the species
of animal used and how the study will be used. In general, rodents are main-
tained on test for 3 months while longer lived animals, such as dogs and
monkeys, for one year or more. If the Subchronic study is being used as a
range finding study for selection of doses that will be administered in
reproductive, chronic, or carcinogenesis studies, then a one-month exposure
period is probably adequate for most compounds.
0774S/122085 - 16 -
-------
DRAFT
Well designed and conducted subchronic studies have been found to be
reliable predictors of most forms of toxicity except for carcinogenic, terato-
genic, or reproductive effects. The FDA (1982), in its toxicological
principles for safety assessment, suggests that if a compound tested in a
subchronic study is found to cause focal hyperplasia, metaplasia,
proliferative lesions, or necrosis, then a carcinogenicity study in two rodent
species is indicated. Finally, if a subchronic study indicates reproductive
organ toxicity, then a two-generation reproduction study with a tetratology
phase may be appropriated. This type of approach (which may be most
appropriate for substances requiring premarket approval) implies that only
under limited circumstances is chronic or reproductive toxicity data necessary.
Because of the enormous cost of conducting chronic studies, several
authors have examined the question of what additional information is gained by
extending the subchronic study. Wei11 aad McCollister (1963) compared the
results of 90-day studies to those obtained in 2-year studies for 33 chemicals
tested in rats. Only body weight gain, relative weight changes of the liver
and kidney, and liver and kidney pathology were monitored. They found that,
for 95% of the studies, the 90-day maximum effect level was only 6 times larger
than the 2 year maximum effect level. Peck (1968) examined eleven drugs tested
for periods of up to 2 years and found that only one study showed additional
new forms of toxicity after 3 months and only four showed additional new
toxicity after 2 months. The author supported the use of 3-6 month studies
for detecting long term effects.
McNamara (1976) examined data on 122 compounds for which subchronic and
chronic studies were available. Of these, only 2.5% produced previously
0774S/122085 - 17 -
-------
DRAFT
unnoted toxic effects after 3 months of exposure. For another 6.5% of the
compounds, effects were found in less than 3 months at the highest dose, but
effects at a lower dose were then seen after 3 months. In almost all cases,
new toxicities, not found before 3 months of exposure, did not appear after
longer periods of exposure.
McNamara also estimated the relationship between the chronic, and sub-
chronic NOELs. He concluded that, for 95% of the chemicals, the subchronic
NOEL will be no more than ten times larger than the chronic NOEL. McNamara
concluded that the 90 day subchronic NOEL could reliably predict the chronic
toxicity and the NOEL. It should be noted that this finding may more simply
reflect the relative design characteristics (specifically, dose-spacing) of
subchronic and chronic studies. Of course, the findings remain useful as long
as the two types of studies continue to be designed as they have been and now
are.
In a review by EPA (1980), the work of several authors (Barnes and Denz,
1954; Boyd, 1968; Davey, 1964; Peck, 1968; WHO Technical Report, 1966) was
reviewed, and was found to support the hypothesis that tests of 3-6 months can
predict chronic toxicity and NOELs. EPA (1980) also reviewed several primary
studies and found that 90-day studies were reliable predictors of chronic
effects.
3.2.4 Chronic Toxicity Studies
The chronic toxicity study is used to determine the effects of a
substance after repeated exposure for the major portion of an animal's
lifetime. There are two forms of the chronic toxicity study. One is
0774S/122085 - 18 -
-------
DRAFT
concerned with establishing NOELs for toxic endpoints which have a long latent
period or are cumulative in nature. Such a protocol will monitor the animal
throughout its lifetime for general toxicity, including neurological,
physiological, biochemical, and morphological measures. The second type of
chronic study is designed to determine whether a compound can induce cancer
after near-lifetime exposure. This cancer bioassay does not involve
monitoring the animals for general toxicity, except as it relates to
shortening the lifespan of the animal. The chronic cancer bioassay is thus
more limited in the information it provides about general toxicity.
It is possible to combine a chronic toxicity study with a carcinogenesis
study but this requires adding more animals and including interim sacrifice
groups and additional measurements. In most cases a carcinogenesis study
cannot be used to replace a chronic toxicity study. However, a positive
finding of carcinogenicity in a chronic toxicity study can be used as evidence
of carcinogenesis.
As was true for subchronic studies, chronic studies do not provide
information on potential reproductive or teratogenic hazards. Any suggestions
of reproductive organ toxicity would suggest the need for a reproductive study.
3.2.5 Reproductive Toxicity Studies
Multigeneration reproduction studies are designed to assess reproductive
function of an animal by evaluating effects on gonadal function, estrous
cycle, mating behavior, conception, parturition, lactation, weaning, and
postnatal growth and development of the offspring. Most guidelines recommend
continuous exposure•in two or three generations of animals, with careful
0774S/122085 - 19 -
-------
DRAFT
monitoring of the reproductive performance of the parents. The design of
these studies provides a qualitative indication that reproduction is being
adversely altered, but usually cannot provide information on the specific
mechanisms causing these effects. Further tests will usually be required to
ascribe the reproductive effects specifically to male or female influences.
The traditional three-generation reproduction study required two litters
per generation. This was used because the first litter of a new generation
was usually considered highly variable in response. Furthermore, it was felt
that three generations were needed to detect transmitted genetic damage and
cummulative effects that occur due to this damage. Today, both the EPA (1978,
1979) and FDA have decreased the number of required generations to two and
decreased the number of litters needed per generation to one. These changes
have considerably decreased the cost of a reproduction study and are now
thought to provide data as reliable as that provided by the former protocols
(Dixon and Hall, 1982).
Although multigeneration reproduction studies can give an indication of
the presence of a potent teratogen, they are not well-suited to measure
teratogenicity. For this reason, separate teratology studies are often
conducted. However, a multigeneration reproduction study can be expanded to
include a full teratology screen at a cost savings to the testor. FDA (1982)
suggests the use of such a protocol. Furthermore, FDA (1982) suggests that,
if reproductive toxicity is found in the two generation study, a teratology
study should be conducted.
0774S/122085 - 20 -
-------
DRAFT
3.2.6 Teratology Studies
The purpose of teratology studies is to determine the effects of exposure
of the embryo and fetus to the substance. Such studies are conducted for a
period of time that includes the stages of organogenesis for the particular
species being used. For rats and mice the period of exposure is usually 6-15
days past conception and for the rabbit 6-18 days. One day prior to birth the
dams are sacrificed and the fetuses are removed for examination of gross,
visceral, and skeletal abnormalities.
Teratology studies are also performed in conjunction with multigeneration
reproduction studies. In these cases exposure to the substance is continuous
and would occur before, during, and after conception and would continue until
one day before the dam was to deliver, at which time she would be sacrificed
and the fetuses removed for full examination.
Teratology studies can provide some information about reproductive
function as it relates to preimplantation loss. This measure is, however,
only one of many causes of infertility and cannot be used as a replacement for
a full reproduction study.
3.3 Summary and Conclusions
Five major types of animal toxicity studies are routinely used by
regulatory agencies to establish acceptable daily intakes of a substance.
Starting from acute studies, each successive level of testing provides more
reliable information with which to determine an ADI. Table 1 presents these
five major toxicity tests and indicates the primary information provided by
the tests, the debatable information provided, the critical information
0774S/122085 - 21 -
-------
DRAFT
lacking for determination of an ADI, and the agencies with standard protocols
published. Through the use of this table a decision can be made as to the
degree of confidence one wishes to achieve and at what cost.
In general, it appears subchronic testing provides the most information
about toxicity and can, in most cases, reliably be used to predict chronic
(non-carcinogenic) toxicity. Moreover, the chronic NOEL can be estimated from
the subchronic NOEL with relatively high reliability. When this is considered
in relation to the relative costs of subchronic and chronic tests (Table 1),
it appears that subchronic tests are a considerably more cost-effective means
of collecting data suitable for chronic ADI estimation than are chronic
tests.
There appears to be substantial empirical support for the proposition
that subchronic toxicity data can be reliably used to establish ADIs. At the
same time, it needs to be recognized that for some substances, certain
findings from such studies, or from other studies reported in the scientific
literature, may suggest the possibility of effects not detectable in
subchronic studies. Whenever such findings are reported, it is probably
prudent to consider an ADI based on subchronic studies to be tentative, and to
seek additional toxicity data.
For example, subchronic studies do not provide information about
reproductive and teratogenic effects, but certain results from them may
suggest that a substance may cause the former. If adverse reproductive
effects are suggested at relatively low doses, it may be appropriate to
consider a two-generation reproduction study. If reproductive damage is seen
0774S/122085 - 22 -
-------
DRAFT
Table l
Toxicity Information Provided by Various
Types of Toxicity Studies
Test Type
Acute Exposure
Subchronic
Exposure
Chronic
Exposure
Primary Information
Provided
1) LOso. LCso
2) Irritation
Potential
3) Target Organ
Toxicity
4) Dose Response
Information for
selection of
doses to be used
in longer term
studies
1) Determines sub-
chronic NOEL for
estimating AOI
2) Target organ
toxicity
3) Cumulative
toxicity
4) Reversibility of
effects
5) Physiologic and
metabolic
tolerance to
dos i ng
1) Determine chronic
NOEL for
estimating ADI
2) Detect toxicities
with long latent
periods
3) Carcinogenic
potential of the
compound
Debatable Information
Provided
1) Estimate subchronic
NOEL by dividing LDSO
by 100
2) Estimate chronic NOEL by
dividing LOso by
1.000-3.000
1) Estimate chronic NOEL
by dividing subchronic
NOEL by 10
2) Predict all chronic
effects except cancers
3) Preneoplastic changes
suggest need for
carcinogenic testing
4) Gonadal changes suggest
need for reproductive
testing
1) Chronic studies do not
detect any new non-
carcinogenic toxicities
than subchronic studies
2) A chronic carcinogenesis
study cannot be used to
Information Cost Per Study Agencies with Published
Lacking Protocols
1) Cumulative toxicity $5.000 for two EPA
2) Subchronic toxicity species FOA
3) Chronic toxicity OECO
4) Reproductive effects
5) Teratogenic effects
6) Carcinogenic effects
1) Carcinogenic effects Mouse, $35-40,000 EPA
2) Reproductive effects Rat. $50-90.000 FDA
3) Teratogenic effects OECO
1) Reproductive effects Mouse. $200-250.000 EPA
2) Teratogenic effects Rat. $300-325.000 FOA
oeco
replace a chronic toxicity
study unless properly
modified.
-------
QRAFT
Table I (continued)
Test Type
Primary Information
Provided
Debatable Information
Provided
Information
Lacking
Cost Per Study
Agencies with Published
Protocols
Reproduction
Studies
1)
2)
3)
1)
2)
Determine NOEL
for reproductive
impairment and
and use for
estimating ADI
Reproductive
impairment measured
Postnatal growth
and development
evaluated
4) Effects on lactation
measured
Estimate teratogenic ))
potential of a compound 2)
Ascribe effects to the 3)
male or female 4)
Teratogenic effects
Subchronic toxiclty
Chronic toxicity
Carcinogenic potential
Rodent. $40-50.000
EPA
FDA
OECD
Teratology
Studies
1) Determine NOEL
for teratogenic
effects and use
for estimating
AOI
2) Measure effects on
organogenesis
3) Observe gross fetal
abnormalities
4) Measure skeletal
abnormalities in
the fetus
5) Observe visceral
abnormalities in
the fetus
2) Provide some estimate
of reproductive
impairment due to pre-
implantation loss
1) Reproductive effects
2) Subchronic toxicity
3) Chronic toxicity
4) Carcinogenic potential
Rodent. $25-30.000
EPA
FDA
OECO
-------
DRAFT
in a two-generation reproduction study, then a teratology study might be
needed to measure definitively the teratogenic potential of a compound.
Similarly, carcinogenicity can be detected and measured only through the
conduct of a chronic bioassay. Before proceeding with or recommending
carcinogenicity studies, however, the results of previous tests should be used
to decide on the advisability and priority with which scarce monetary
resources will be committed to their conduct. For example, the FDA (1982) has
suggested that if a subchronic study demonstrates a substance causes focal
hyperplasia, metaplasia, proliferative lesions/ or necrosis, then priority
should be given to conducting a carcinogenicity study. Likewise, the results
of short-term mutagenicity studies have been suggested as a screen to select
compounds for cancer bioassay (Food Safety Council, 1980; FDA, 1982).
Because acute toxicity data do not provide information about the effects
of repeated exposure, and because predicting subchronic or chronic NOELs from
LDso values is not considered a validated methodology in the scientific
community, it would appear that subchronic toxicity studies constitute the
minimally necessary data for establishing reliable ADIs. Subchronic studies
reveal a great deal about the toxic properties of chemicals and at relatively
modest cost. Thus, in most cases, subchronic studies are not only minimally
necessary, but are also entirely adeguate to establish reliable ADIs. In
addition, for the cost and in the time necessary to develop chronic toxicity
data on a single chemical, ADIs can be developed on the basis of subchronic
data for several (perhaps 4-6) substances.
0774S/122085 - 25 -
-------
DRAFT
4. Special Issues in the Use of Toxicity Data to Derive APIs
4.1 APIs for Essential Nutrients
Several metals that exhibit toxicity at high doses are known to be
essential nutrients for humans. Among these are zinc, copper, selenium, and
chromium. Selenium is not only toxic to the liver, but also induces
hepatocellular tumors in experimental animals. Chromium (VI) is carcinogenic
in humans, at least by inhalation (NRC, 1980).
For nutrients that are not known to be carcinogenic, the application of
the standard uncertainty factors may well lead to ADIs below the recommended
dietary intake level. For carcinogens, application of standard extrapolation
models will reveal a finite risk of cancer at the recommended intake level.
Should this be the case, it should not be inferred that a toxic risk exists at
and below the recommended nutrient intake level. Rather, it suggests that the
standard uncertainty factors are unnecessarily large, and, for the
carcinogens, that a non-threshold model may not be appropriate for these
categories of elements, probably because mammalian systems have developed
homeostatic mechanisms for dealing with the toxic properties of these elements
when exposures are at or near the nutrionally necessary levels (Stults, 1981;
Roberts, 1981). In the case of, at least, chromium, it may also mean that the
route of exposure is critical •-- i.e., that carcinogenicity is not expected
for ingested, rather than inhaled chromium.
ADIs have to be established for essential elements on the basis of
case-by-case analysis. Judgments have to be made by first examining the
toxicity data and NOELs and comparing the NOEL with the recommended daily
0774S/122085 - 26 -
-------
DRAFT
intake. If a very wide margin exists, it may be possible to apply the
standard extrapolation factors and derive an ADI that is greater than the
recommended intake level (one that allows the recommended intake to be
exceeded by a. significant degree). In other cases, where the margin is
relatively small, it will be necessary to decide what intake in addition to
that recommended for nutritional well-being can be tolerated before toxicity
will almost certainly arise (i.e., it will be necessary to allow some
additional intake beyond that which is essential if the ADI is to be a figure
other than zero). The Safe Drinking Water Committee of the National Academy
of Sciences has undertaken these types of analyses for several substances
(copper, chromium, selenium, iodide, fluoride*, phosphorous, etc.) and their
work can be consulted for guidance (NEC, 1977; 1980; 1983).
4.2 Mixtures and Toxicoloqical Interactions
At most hazardous waste sites and in many other situations, there are
many chemicals that may enter air or water simultaneously. It is thus of
interest to examine the questions of possible biological interactions among
these substances to determine whether ADIs should be adjusted to account for
them.
4.2.1 General Types and Mechanisms of Interaction
Some chemicals may interact in ways such that the risk to health from
exposure to a combination of chemicals differs, either qualitatively or
quantitatively, from the estimated risk from exposure to each chemical by
* Fluoride is not essential, but is added to water supplies at a level
sufficient to reduce the incidence of dental cavities.
0774S/122085 - 27 -
-------
DRAFT
itself. Such interactions can be synergistic or antagonistic, and can occur
during absorption, distribution, metabolism, or excretion, or at the site of
biological action (target site).
One Striking and well-studied example of interactions among hazardous
agents is the marked synergism between cigarette smoking and asbestos in the
induction of lung cancer. Epidemiologic evidence indicates that an asbestos
worker who smokes cigarettes has 8 times the risk of smokers of the same age
who do not work with asbestos, and 92 times the risk of men who neither work
with asbestos nor smoke cigarettes (NRC, 1980a).
Although the ways in which chemicals interact are complicated and
incompletely understood, three general mechanism by which chemicals can
interact have been identified:
• Chemical-Chemical Reactions
A chemical may react with another in such a way that: (1) the
potentially injurious chemicals(s) never reach the target site(s) in an
active form; or (2) the chemical products reach the target site(s) and
cause enhanced injury or an altered form of injury (NRC, 1980a).
• Chemical Competition at Macromolecules
This mechanism of chemical interaction involves the competition for
binding or reaction at a limited number of reaction sites or cellular
macromolecules. These sites may control absorption, activation,
detoxification, injurious action, or excretion, with competition for
these sites resulting in either enhanced or reduced toxicity. This
mechanism generally requires that the interacting chemicals or
derivatives be present in the organism at the same time.
• Altered Cellular Responsiveness or Reactivity
A cell or tissue may be altered by one chemical in such a way that the
cell's or tissue's response to a second chemical is altered, even if the
first chemical is no longer present. This mechanism is demonstrated by
the initiation-promotion theory of carcinogenesis. Administration of a
promoting agent weeks or even months after administration of an
0774S/122085 - 28 -
-------
DRAFT
initiating agent generally enhances tumor formation, while administration
of the promoter before the initiator has little or no effect.
4.2.2 Interactions in Contaminated Air or Water
In most cases people exposed to contaminated air or water will be exposed
to a mixture of chemicals rather than to a single substance. Because of the
possibility of interactions of the types described above, a question arises
about the total risk due to all the contaminants present. At least four
possibilities arise:
1. The total risk is equal to the sum of the risks of each of the
chemicals (more specifically — the risks of agents showing similar
effects would be strictly additive, so that one could calculate, by
addition, total carcinogenic risk, or total risk of liver damage,
etc.).
2. The total risk is greater than the risks obtained by addition (this
represents the phenomenon of synergism).
3. The total risk is less than the risks obtained by addition (this is
the phenomenon of antagonism).
4. Within a given combination of chemicals, various combinations of
synergism, antagonism, or strict additivity may occur for different
toxic effects.
It might be possible to subject these four possibilities to empirical
tests in properly designed animal experiments. Any such experiments would,
however, be extraordinarily costly and results from them are likely to be of
limited generality. Thus, experiments may be conducted on certain commonly
occurring combinations of chemicals (the effects of which would have to be
compared with the effects of the individual constituents of the combination),
but it is unlikely that any such results would be clearly applicable to other
combinations of chemicals, or to the same chemicals occurring in different
proportions. This view may be unduly pessimistic, however, and it may be
useful to consider Ulis testing question further.
0774S/122085 - 29 -
-------
DRAFT
Even if major research efforts were initiated on the question of
interactions, no useful data would likely be available for several years. No
generally applicable protocols for such tests appear to be available, and
problems of study design, conduct and interpretation have only been discussed
from a theoretical viewpoint (NEC, 1980a; 1982). If risk assessments or ADI
derivations are to take into account possible interactions, they must for the
present be based on consideration other than empirical evidence (EPA, 1984).
Unless the use of uncertainty factors in the derivation of ADIs is
considered in part to compensate for the possibility of interactions, then
there is no area of risk or safety evaluation that has, as a matter of course,
taken interactions into account. The major reason for this is that there is
very little data available to demonstrate toxic interactions, especially of
chemicals found at waste sites (most data come from studies of drugs,
Calabrese, 1983).
It thus appears that, unless specific data become available that reveal
the mode of interaction among groups of chemicals that will be known always to
co-occur in the same relative proportions, and that will reveal the
quantitative effect of one upon the others, it is probably not possible to
take interactions generically into account in deriving AOIs. Under such
circumstances, it would appear that treating each substance as an
independently-acting toxicant would be the course most in keeping with current
scientific understanding.
0774S/122085 - 30 -
-------
DRAFT
5. Apportionment of RfDs and RSDs
5.1 Introduction
The final issue concerns apportionment of RfDs and RSDs (we shall now use
the preferred EPA terms) among sources and between air and water. To ensure
that potentially exposed individuals do not experience intakes significantly
in excess of the RfD, it is necessary to establish limits on the
concentrations of individual substances in air and water.
In deriving such limits, two major issues must be considered:
1. Apportionment of RfDs among media and sources; and
2. Relationships between concentrations in air and water, and human
doses.
5.2 Apportionment Among Media and Sources
The RfD for a chemical represents the maximum allowable daily dose of the
chemical that is anticipated to have no adverse effect in humans following
chronic exposure. For most chemicals, it is the systemic dose that is
important in defining the RfD; the route of exposure, whether it be by
inhalation, ingestion, or dermal contact, is not of major concern.
Since exposure to a chemical may arise from different sources (that is,
not solely as a result of waste disposal) and via different routes of exposure
(for example, inhalation from the air, ingestion in food or drinking water, or
dermal contact), the RfD must represent the limit on total dose received from
all sources and via all routes. Derivation of limits on concentrations
0774S/122085 - 31 -
-------
DRAFT
of substances in a particular medium (for example, air or water) must allow
for the possibility that other sources and routes of exposure to the substance
exist.
In the present case, we are interested in deriving limits on
concentrations of chemicals from waste sites in air and water. Chemicals
migrating from waste sites may enter either or both of these media to varying
degrees depending on chemical-specific physical characteristics. For example,
a chemical which is highly volatile but poorly soluble in water, such as
dichlorodifluoromethane, will tend to migrate into the air, while a water
soluble chemical of low volatility, such as many inorganic salts, will tend to
migrate into water. However, contamination of water and air from a chemical
waste disposal site is not the only way in which people may be exposed to a
chemical. Some compounds of wastes (for example, certain pesticides and
organic solvents) are widespread environmental contaminants to which human
exposure may occur via air, food, or consumer products. Others, for example,
many inorganic substances such arsenic and cadmium, occur naturally in the
environment, and again may reach people through food and several other media.
In setting maximum allowable concentrations of chemicals in water and
air, allowance must.be made for potential exposure from other sources and by
other routes. Since exposure may occur by several routes, and the RfD
represents, by definition, the total allowable exposure, the RfD must be
apportioned over the various possible routes of exposure.
The concept of apportionment of a chemical by medium and by route of
exposure is not new*. The NEC Safe Drinking Water Committee (NRC, 1980)
0774S/122085 - 32 -
-------
DRAFT
calculated suggested no-adverse-response levels (SNARLs) for chronic exposure
to non-carcinogens in drinking water while incorporating an "arbitrary
assumption" that 20% of the intake of the chemical was from drinking water.
The EPA in setting maximum contaminant levels (MCLs) for chemicals in tap
water, also selects a fraction of the ADI (usually 20% if there are no data to
suggest some other fraction).
Another use of apportionment was a risk evaluation procedure developed
for EPA's Office of Emergency and Remedial Response to evaluate and manage the
risks for specific remedial action sites. This procedure apportioned
concentrations equally in environmental media (e.g., air and water) as an
initial basis for calculating an allowable rate of release to the environment;
at times, unequal apportionment was selected, if there were significant cost
and feasibility differences in controlling exposures via the different
pathways (ENVIRON, 1983).
The Food, Drug and Cosmetic Act [Section 409(c)(5)] specifies that in
deciding whether a proposed use of a food additive is safe the FDA must
consider certain relevant factors. Included as one of these factors is a
consideration of "the cumulative effect of such additive in the diet of man or
animals, taking into account any chemically or pharmacologically related
substances or substances in such diet." This language has been interpreted by
the FDA as requiring that all sources of exposure must be combined in order to
estimate the total exposure to an additive. This total exposure level is then
compared to the ADI to decide if the proposed uses are safe.
0774S/122085 - 33 -
-------
DRAFT
In Section 706(b)(5)(A) of the Food, Drug, and Cosmetic Act it is also
specified that for a color additive the FDA will determine if a color additive
is safe after specific factors are considered. The factors include "the
probable consumption of, or other relevant exposure from, the additive and of
any substance formed in or on food, drugs, devices, or cosmetics because of
use of the additive. In addition the Secretary must consider, "the cumulative
effect, if any, of such additives in the diet of man or animals, taking into
account the same or any chemically or pharmacologically related substance or
substances in such diet." As for food additives, colors must be reviewed for
safety after all uses of the color have been combined and compared to the ADI.
In its review of lead in the human environment, the NRC (1980) stressed
that one step in a comprehensive risk assessment requires that the level of
exposure be estimated quantitatively for each pathway that affects the target
population. Quantitative assessment of exposure for lead required estimation
of population exposure from dust, air, soil, water, paint, food, and
cosmetics. Although uncertainties are associated with exposure estimates for
these various sources, it was agreed that all contributions to the "total
exposure" were needed to establish the safety of lead.
NRC (1980) suggested that although it is difficult to measure specific
lead source contributions directly, various modeling techniques can assess the
relationship between lead in the environment and human exposure levels of
lead. The EPA's Office of Drinking Water had developed a detailed model for
estimating the relative contributions of air, water, and other sources of lead
to the total exposure (Drill et al., 1979 as cited in NRC, 1980) for various
populations.
0774S/122085 - 34 -
-------
DRAFT
It should also be noted that if the total RfD is used for hazardous waste
sources of a chemical, then the only scientifically supportable decisions
regarding other sources of the chemical is zero exposure. While the issue of
apportioning allowable exposure levels among various sources and media has
strong scientific justification, and considerable precedent, the choice of
fractions of total exposure to allot to the various media is less clear. In
the context of the present work, two issues are important in making this
choice. The first relates to how much of the total allowable exposure may
come from sources other than water and air, and the second relates to how
exposure should be apportioned between water and air.
Many of the chemicals of concern are common environmental contaminants.
The most rigorous procedure, scientifically, would be to analyze on a case-by-
case basis each potential exposure situation to determine background levels of
exposure to the substance in question and allot an appropriate fraction of the
RfD on that basis, retaining the remaining fraction for air and water exposure
resulting from escape of the substance from the waste disposal site. This
would entail a level of effort that is out of proportion to its importance in
protecting public health, and, because of data gaps for many chemicals, is not
likely to be productive. As an alternative, alloting 50% of the RfD to
background exposures and 50% to waste site-related air and water exposure
seems a useful first approximation. A more rigorous apportionment is not
called for in light of the facts, noted previously, that the RfD is itself
subject to considerable uncertainty, and that occasional excursions above the
RfD, for a relatively small number of substances, will not likely produce
excess toxic risk (see Section 2). It would be prudent, however, to
reconsider the 50% figure whenever readily available data on background levels
0774S/122085 - 35 -
-------
DRAFT
of specific substances reveal it to be too generous, and to use an alternative
value if sufficient data on environmental distribution of the substance are
available to justify an alternative.
This 50% allotment of the RfD is probably not necessary for carcinogens.
For such substances, the equivalent dose (the RSD) is estimated by a procedure
which introduces unavoidable uncertainties. The procedure used is
deliberately selected to be conservative; so that a twofold difference in dose
is well within the margin of uncertainty of the estimated RSD.
Moreover, for carcinogens, the determinate of risk is the daily dose
averaged over a full lifetime. Small variations around the daily dose have
little effect on the lifetime risk, as long as the average is not affected.
For this reason, a two fold reduction in the RSD is relatively insignificant.
For non-carcinogens, it is possible that not applying the 50% reduction (the
indirect effect of which is to permit an approximate doubling of the ADI) may
cause the threshold to be exceeded on some or even many days of the human
exposure period. Exceeding the threshold of effect may have significant
health consequences for some individuals. Thus, there appears to be
justification for treating non-carcinogens differently from carcinogens with
respect to this apportionment issue.
Before turning to the question of apportionment between air and water, we
first discuss the interrelationship between the concentration of a substance
in air or water and the human dose of the substance resulting from drinking
the water or breathing the air.
0774S/122085 - 36 -
-------
DRAFT
5.3 Relationships Between Air or Water Concentration and Human Dose
The RfD or RSD is defined in terms of a daily dose or daily intake,
generally measured in mg/kg/day. To define maximum allowable concentrations
of a chemical in environmental media (in this case, water and air) it is
necessary to know the relationship between the concentration in each medium
and the daily human dose resulting from normal intake of that medium. For
many substances, the daily intake (in mg/kg/day) may be calculated simply by
multiplying the concentration of the chemical in the medium by the daily human
intake of that medium and dividing by the human body weight:
Intake of chemical = concentration in medium (mg/1) x daily intake of
(mg/kg/day) medium (I/day)/body weight (kg)
Adjustments to this simple equation may be necessary to account for
incomplete absorption of the ingested or inhaled chemical. However, if the
RfD is based on intake by the same route as that by which the human intake is
being calculated, and if it can be assumed that the degree of absorption
occurring in the experimental situation is the same as that in the humans of
concern, such an adjustment will not be necessary (EPA, 1984; NRC, 1983).
It is generally assumed that the greatest contribution to the exposure
from tap water results from direct ingestion and that inhalation of vapors and
aerosols of water contaminants while showering, or dermal absorption of those
substances during bathing, are relatively trivial contributors to exposure
(NRC, 1983). It is advisable, however, that empirical verification or
refutation of this premise be sought/ because it is not clear it would hold
for all chemicals (especially highly volatile ones).
0774S/122085 - 37 -
-------
DRAFT
If the RfD is based on exposure through one medium, say air, and we are
attempting to derive a safe concentration in another medium, say water,
adjustment for absorption will be necessary, unless the degree of absorption
of the chemical when inhaled is the same as the degree of absorption of the
chemical when ingested in water.
Specific absorption information is known in experimental animals for
numerous compounds, but in humans for far fewer substances (Calabrese, 1983).
However, the data from experimental animals represent reasonable
approximations of those parameters in humans (Calabrese, 1983). In general,
absorption of retained foreign compounds is greatest (i.e., both rate and
efficiency) via the lungs, less by water in the gastrointestinal tract, still
less by food in the gastrointestinal tract, and least by the skin (Klaassen,
1980).
Absorption data are not available on all substances of interest by all
routes of exposure. The conversion from RfD to media concentrations,
therefore, necessarily relies on knowledge from similar compounds for which
such information is available. For example, among the chlorinated alkanes, it
is possible to approximate crudely the degree of absorption of members of the
class whose absorption is not known from absorption data on those members of
the class for which absorption is known in experimental species considered
predictive of chemical behavior in humans. Similar estimations are possible
among inorganic metals (Calabrese, 1983; NAS, 1975).
For airborne particles, it is important to distinguish between
alternative sites of deposition in order to apportion systemic doses since the
0774S/122085 - 38 -
-------
DRAFT
site of deposition influences how and to what extent a substance may be
absorbed. Although deposition is determined by a number of variables
including particle charge and shape, aerodynamic diameter can be used to
estimate the most likely site of deposition. For airborne materials, those
that are in particulate form are often not retained, and hence not absorbed,
if smaller than 2-5 microns in diameter. Those between 5 and 20 microns in
diameter are efficiently retained for absorption. Those greater than 20
microns in diameter are generally deposited in the upper respiratory tract
from which they may be cleared by the mucocilliary escalator and swallowed.
For such substances, therefore, it is gastrointestinal absorption rather than
absorption through the lung that is critical in defining absorption.
As noted earlier, the conversion of the RfO to concentrations in
environmental media requires knowledge of the extent of human exposure to the
media themselves in addition to knowledge of the extent of absorption. Where
specific data of this type exist, they are incorporated into the analysis of
daily exposure. However, in the absence of such data, assumptions must be
made. In the present case a variety of assumptions have been made to convert
between daily human dose levels and media concentrations (see Table 4).
The previous discussion dealt with agents that produce injury through
systemic distribution and selective affinity and injury to specific tissues.
Some compounds, such as acids and alkalies, when present in adequate
concentrations will damage the tissues with which they come into direct
contact. Such effects are unusual for substances that have migrated from
waste sites because of dilution, buffering, and other physical and biological
influences, and are' not considered further in this document.
0774S/122085 - 39 -
-------
DRAFT
The data in Table 4 (or other specific data if available) may be used to
convert a daily human dose to a corresponding concentration in air or water
using the general equations below.
Conversion of Daily Human Dose to Equivalent Air Concentration
Air Concentration = Daily dose (mg/kg/day) x body weight (kg) x correction factor
(mg/m3) m air breathed/day
Conversion of Daily Human Dose to Equivalent Water Concentration
Water Concentrations = Daily dose (mq/kq/day) x b/w (kg) x correction factor
(tng/1) liters water consumed/day
The procedures described above for interconverting between media
concentrations and human doses are typical of those commonly used by
regulatory agencies and other bodies for such purposes. For example, EPA
currently used identical procedures to set ambient water quality criteria
(USEPA 1980) and maximum contaminant levels.
The National Research Council (1977) describes calculations directly
analogous to those above for deriving acceptable water concentrations from
acceptable daily doses. Likewise, the American Conference of Governmental
Industrial Hygienists (1980) uses similar procedures to derive maximum
permissible air concentrations (TLVs) for substances for which the only
relevant data are derived from studies in which the substance in question is
administered in a different route (for example, orally).
These procedures are also virtually identical to those originally
published in 1958 by Stokinger and Woodward (1958). The validity of this
procedure was discussed in a recent EPA conference (USEPA 1984), which
0774S/122085 - 40 -
-------
Table 4. Assumptions for Converting Daily Human Doses
to Media Concentrations
DRAFT
Adult male body weight
Adult male body surface area
Volume of air breathed by adult
male per day
Efficiency of pulmonary absorption
Amount of water consumed by adult
male per day
Efficiency of gastrointestinal
absorption
Correction factor
70 kg (ICRP, 1975)
1.8 m2 (ICRP, 1975)
23 m3 (ICRP, 1975)
100% (unless data to the contrary)
(Calabrese, 1984)
2 liters (ICRP, 1975)
100% (unless data^ to the contrary)
(Calabrese, 1984)
1.0*
Adjusts for different extents of absorption if the RfD is based on a
route of exposure other than that for which the RfD is being derived. If
differences in the extent of absorption have not been reported, they are
assumed to be identical, and therefore, the correction factor is 1.0.
0774S/122085
- 41 -
-------
DRAFT
concluded that it provides a reasonable first approximation, though estimates
obtained are likely to be somewhat inaccurate if factors such as absorption
and pharmacokinetics are not taken into consideration.
In the present context, the "daily dose" which is being converted to an
air or water concentration would be the portion of the RfD or RSD that is
alloted to the medium. What proportion of the total RfD or RSD is alloted to
each medium is the subject of the next section.
5.4 Apportionment Between Air and Water
We have already noted how it is appropriate to allot just a portion (50^>)
of the total RfD to air and water contamination resulting from escape of
chemicals from hazardous waste disposal sites to ensure that the total RfD is
not exceeded in the likely event that some exposure to the chemicals of
concern occurs via other media, particularly food. In this section we
describe how the portion of the RfD that is alloted to air and water can be
partitioned between these two media. In the end it must be ensured that the
maximum concentrations permitted in air (in mg/m3) and in water (in mg/1)
yield a total exposure no greater than 0.5 RfD.
Many volatile and semivolatile chemicals may be present in both air and
water, which present dual pathways of exposure. Atmospheric dispersion may
substantially reduce the concentration of chemicals in air to a much greater
extent normally than would be expected in surface water or ground water
systems. However, airborne chemicals may accumulate in poorly ventilated
0774S/122085 - 42 -
-------
DRAFT
areas such as enclosed buildings where significant airborne exposures may
occur.
There are analytical methodologies that could represent the individual,
complex processes that affect the partitioning of chemicals between air and
water, and the transportation of these chemicals from the source area to the
receptor. However, no analytical approach would be suitable to all such
processes. To make such predictions would require site and chemical specific
data and reliable, verified models of nonconservative atmospheric and water
borne transport processes. The application of such models for the purpose of
partitioning the RfD is not justified in most instances, given the lack of
data and verified models at most sites. Moreover, use of such refined models
would seem incompatible with the relatively crude approximations used to
derive the RfD.
There are two physical characteristics of chemicals that describe their
behavior in air and water. The octananol-water partition coefficient (K0w)
is a measure of a chemical's partitioning between water and an organic phase
(approximately represented by octanol). The K0w for a chemical provides an
indication of its solubility in water, and may describe its behavior in an
environment likely to be present at a hazardous waste site. Specifically,
chemicals with large values of K0w are preferentially retained in an organic
phase and only poorly soluble in water. Conversely, substances with small
values of K0w are more readily soluble in water than in an organic phase.
It would thus appear that chemicals with small values of K0w (e.g., high
solubility chemicals such as phenols and halogenated phenols) are more likely
to escape a landfill in an agueous phase than those with large values of Kow.
0774S/122085 - 43 -
-------
DRAFT
Once present in water, a chemical may partition between air and water.
For a dilute solution the ideal gas vapor pressure (an alternative expression
of concentration) of a volatile solute is proportional to its concentration in
the solution. The gas vapor pressure, or airrwater partition, is described by
Henry's law (Tinsley, 1979), which can be expresssed as:
Ca = Hc Cw
where Ca is the chemical concentration in the gas (air) phase (mg/m3),
Cw is the chemical concentration in the liquid (water) phase (mg/1), and,
Hc is Henry's Law constant (mg/m3/mg/^). It is important to note that
Henry's Law applies rigorously only to dilute solutions where solute-solute
interactions are negligible. Hc is customarily expressed as atm-m3/mole
when Ca is expressed in terms of partial pressure (atm). Using Boyle's Law
for ideal gases, PV = NRT, the conversion of partial pressure (P) to
moles/m3 can be achieved by dividing P by RT. The gas constant, R, is 8.205
x 10~s atm mVmoles0 K, and the ambient temperature T is normally assumed
to be 20°C (293°K). Therefore, to convert Hc from units of atm-m3/mole to
mg/m3/mg/£ (i.e., ^/m3), one must multiply by 1/RT (i.e. 4.16 x 104).
The relative air and water concentrations of a chemical, at equilibrium
will be indicated by the value of Henry's Law constant Hc, (units of
mg/m3/mg/£ or atm m3/mol). Chemicals with large values of Hc will
have a tendency to exist predominately in air, whereas those with low values
will partition preferentially to water.
0774S/122085 - 44 -
-------
DRAFT
It is recognized the values of K0w and Hc for specific chemicals are
obtained in the laboratory under idealized conditions, and that their
application in a setting as complex as a hazardous waste site can not be
expected to yield accurate predictions of the behavior of a chemical. It is
nevertheless true that these two physical constants will predict trends in
behavior. That is, chemicals with low values of K0w and low values of Hc
are highly soluble in water and poorly volatile, so that their concentrations
are likely to be high in water and low in air, relative to chemicals with high
values of the two constants, assuming that the source is not limiting. One
approach to deciding how to partition a chemical between air and water would
depend on these trends in behavior, as described by K0w and Hc. There
are, of course, no standard definitions of "high" or "low" with respect to the
two physical parameters Kow and Hc. Values of K0w and Hc for selected
chemicals that have been reported in the literature are presented in Table 5.
The chemicals are those considered as especially important by EPA. A simple
scheme to partition the RfD using K0w and Hc is presented in Table 6. The
partition chosen is meant only to reflect the general direction of expected
migration of the chemical from a source to water and air.
A model more refined than that shown in Table 6, i.e, depending only on
ROW and Hc, and which results in a partitioning of the RfD into more than
three broad groups (partitioning mainly into air; partitioning mainly into
water; and approximately equal partitioning between the two media) would not
appear to be warranted, given the uncertainties inherent in the RfD and the
0774S/122085 - 45 -
-------
DRAFT
Table 5. Henry's Law Constants and
Octanol-Water Coefficients for
Selected Hazardous Chemicals
Chemical
Carbon Disulfide
Chlorobenzene
Cresols
1,2 Dichlorobenzene
Methylene chloride
Trichloromonofloromethane
Isobutyl Alcohol
Methyl Ethyl Ketone
Nitrobenzene
Pyridine
Tetrachloroethylene
2,3,4,6 Tetrachlorophenol
Toluene
Methylchloroform
Trichloroethylene
2,4,5 Trichlorophenol
2,4,6 Trichlorophenol
Pentachlorophenol
1,2,2 Trichloro-1,2,2,
Trifluoroethane
Ethylbenzene
Henry's Law Constant
atm mVmole
Octanol-Water
Coefficient
1.68E-02
3.46E-03
5.05E-06
1.88E-03
3.19E-03
8.02E-01
1.23E-05
2.61E-05
2.40E-05
1.95E-07
2.87E-02
4.53E-06
5.93E-03
2.76E-02
1.17E-02
2.84E-05
1.77E-05
4.62E-06
9.00E+00
6.44E-03
6.99E-t-02
1.44E+02
2.10E-01
7.82E+01
1.33E+02
3.34E+04
5.12E-01
1.09E+00
9.98E-01
8.11E-03
1.19E+03
1.88E-01
2.47E-.-02
1.15E+03
4.87E+02
1.18E-01
7.36E-01
1.92E-02
3.74E+05
2.68E+02
1.45E*02
7.41E+02
1.41E+02
3.80E+03
1.80E+02
3.31E*02
5.50E*00
2.00E+00
7.94E+01
4.79E+00
5.80E*02
2.14E+04
6.61E+02
3.16E+02
2.29E+02
7.24E+03
2.93E+03
1.15E*05
1.26E+03
1.41E+03
0774S/122085
- 46 -
-------
DRAFT
Table 6: RfD Partition
Between Water and Air using Kow and Hc
Low
Hc
Low
High
Air:Water
50:50
Air:Water
80:20
High
Air:Water
20:80
Air:Water
50:50
0774S/122085
- 47 -
-------
DRAFT
low predictive power of Hc and K0» in the context of a hazardous waste
environment.
A further simplification of the partitioning scheme can be envisioned.
The ultimate question to be addressed in partitioning the RfD is the relative
rather than absolute concentrations in air and water that a chemical might
achieve at equilibrium. The relative concentrations can be indicated by the
use only of Hc•
The RfD is comparable to the total dose (or intake) of a specific
chemical by the receptor. The dose is a function not only of the chemical
concentration in air and water, but also of the breathing and ingestion rates,
and absorption through the respiratory and gastrointestinal systems. Lacking
specific data, inhaled or ingested chemicals are assumed to be totally
absorbed. However, breathing and ingestion rates are well established and can
be considered in the partitioning of the RfD.
The total dose of a specific chemical by the combined air-water pathways
is given by:
Total Dose = dose inhaled + dose ingested by consumption of water
= BR * Ca + IR * Cw
where BR is the breathing rate (m3/day), IR is the water ingestion rate
(^/day), and Ca and Cw are as defined previously.
It is assumed that at equilibrium the chemical concentration in air, Ca
(mg/m3) is related to the chemical concentration in water, Cw (mg/^), by
the Henry's law constant, Hc (^/m3). The dose model can be rewritten as:
0774S/122085 - 48 -
-------
DRAFT
Total Dose = BR * Hc * Cw + IR * Cw
= (BR * Hc + IR) * Cw
where the air and water based doses can be represented by (BR * Hc * Cw)
and (IR * C»), respectively.
Using the dose model, the partitioning of the RfD for each chemical
between the air and water pathways can be calculated by the proportion that
each contributes to the total dose.
(Air) (Water)
BR * Hc IR
BR*HC+IR : BR*HC*IR
When calculating the air:water partition, breathing and ingestion rates of
23 mVday and 2 I/day, respectively, can be assumed (see previous
section). The partition model can then be rewritten as:
(Air) (Water)
23 * Hc 2
23 Hc+2 : 23
where Hc must be expressed in units of mg/m3/mg/^,
An even more simplified approach might involve using the dose model
described above with ranges of Hc, in which the air: water partition could be
represented by high, moderate and low ranges. Such an approach is illustrated
in Table 7. Although this approach may be less precise than a chemical
specific calculation, it may more reasonably reflect the inherent
uncertainties in these simplistic models of complex natural processes.
0774S/122085 - 49 -
-------
DRAFT
Table 7
RfD Partition Between Air and Water Using
Henry's Law Constant (Hc)
Hc
Range Air:Water
(l/m3) Partition
>0.35 80:20
0.35 - 0.02 50:50
<0.02 20:80
* The values of Hc shown in this Table were derived by assuming the
air:water partition values (e.g., 80:20, air to water) and calculating
Hc from:
(Air) (Water)
23 * Hc 2
23 Hc+2 : 23
0774S/122085 - 50 -
-------
DRAFT
The representation of the air:water partition by the equilibrium
relationships described herein is admittedly a somewhat simplistic
approximation to the partitioning of the RfD. The approach may accurately
represent the relative contribution of air and waterborne chemicals only in
close proximity of the chemical source. At greater distances the predicted
air:water partition would be less precise. However, it is intended that the
partitioning of the RfD will be established within reasonable limits that are
chemical specific, but not to be reevaluated on a site by site basis. The
proposed approaches accomplish this purpose.
0774S/122085 - 51 -
-------
DRAFT
6. Conclusions and Recommendations
1) Since its introduction in the early 1950s, the ADI has been widely
used by regulatory and public health agencies as a practical, health
protection tool. It appears to be the appropriate criterion for establishing
limits for substances that may migrate into environmental media to which
humans may be chronically exposed. EPA now proposes to adopt the term
"Reference Dose" (RfD) to replace ADI. All references to ADI in the following
also describe the RfD.
2) The ADI should not be considered a sharp dividing line between "safe"
and "unsafe" exposures. It is a practical tool, subject to considerable
scientific uncertainty, and occasional excursions above the ADI should not be
considered cause for concern. It is not possible, however, to quantify the
magnitude of uncertainty associated with any given ADI.
3) Consistent protection at or near the ADI ensures that individuals will
be protected from the acute effects of all chemicals. Protection against
acute toxic effects usually requires safety factors no larger (and sometimes
smaller) then those used for establishing the ADI. Because the chronic NOEL
will always be a lower dose that the minimum effective acutely toxic dose, the
ADI will clearly protect against acute toxicity.
4) In general, it appears subchronic testing can be used to predict
chronic (non-carcinogenic) toxicity. When this is considered in relation to
the relative costs of subchronic and chronic tests (Table 1), it appears that
0774S/122085 - 52 -
-------
DRAFT
subchronic tests are a considerably more cost-effective means of collecting
data suitable for ADI estimation than are chronic tests.
5) Although the data base supporting the conclusion (4) is relatively
extensive, it is not without uncertainty. Evaluations of the need to move
beyond subchronic should be made on a case-by-case basis, giving due
consideration to the results of acute and subchronic studies, chemical
structure, and metabolic information.
6) Subchronic and chronic tests do not provide information about
reproductive injury and teratogenic effects. Conduct of a two-generation
reproduction study can provide most of this information as well as an
indication of frank teratogenic effects. If reproductive damage is seen in a
two-generation reproduction study, then a teratology study would be needed to
measure definitively the teratogenic potential of a compound.
7) Carcinogenicity can be measured only through the conduct of a chronic
bioassay. Before proceeding with cancer bioassays, however, the results of
previous tests should be used to decide on the advisability and priority with
which scarce monetary resources will be committed to their conduct. For
example, the FDA (1982) has suggested that if a subchronic study demonstrates
focal hyperplasia, metaplasia, proliferative lesions, or necrosis, then
priority should be given to conducting a carcinogenicity study. Likewise, the
results of short-term mutagenicity studies have been suggested as a reliable
screen to select compounds for cancer bioassay (Food Safety Council, 1980; FDA
1982).
0774S/122085 - 53 -
-------
DRAFT
8) Because acute toxicity data do not provide information about the
effects of repeated exposure, and because predicting subchronic or chronic
NOELs from LDso values is not considered a validated methodology in the
scientific community, it would appear that subchronic toxicity studies
constitute the minimally necessary data for establishing reliable ADIs.
Subchronic studies reveal a great deal about the toxic properties of chemicals
and at relatively modest cost. In most cases, data from such studies should
be fully adequate to establish an AOI. Decisions about the need for
additional toxicity data, and their value relative to the costs involved,
should be made on a case-by-case basis, under the general criteria described
above (items 5, 6, 7).
9) ADIs should be established for essential elements on the basis of
case-by-case analysis. Judgments will have to be made by first examining the
toxicity data and NOELs and comparing the NOEL with the recommended daily
intake. If a very wide margin exists, it may be possible to apply the
standard extrapolation factors and derive an ADI that is greater than the
recommended intake level. In other cases, where the margin is relatively
small, it will be necessary to decide what intake in addition to that
recommended for nutritional well-being can be tolerated before toxicity will
almost certainly arise (i.e., it will be necessary to allow some additional
intake beyond that which is essential if the ADI is to be a figure other than
lero).
10) Since most people will be exposed to mixtures of chemicals rather
than single substances, the possibility of interactions among chemical raises
questions about the' total risk due to all the contaminants present.
0774S/122085 - 54 -
-------
DRAFT
It appears that, unless there are specific data available that reveal the
mode of interaction among groups of chemicals that will be known always to
co-occur, and that will reveal the quantitative effect of one upon the others,
it is probably not possible to take interactions genetically into account in
deriving ADIs.
11) The final issue concerns apportionment of ADIs between air and
water. To ensure that potentially exposed individuals do not experience
intakes in excess of the ADI, it is necessary to establish limits on the
concentrations of individual substances in air and water. Because indivuduals
may be exposed through other media, only a portion of the ADI can be used for
allocation to air and water contamination. There are several precedents for
making such allocations for non-carcinogenic chemicals.
12) While the issue of apportioning chemicals between air and water is
susceptible to some degree of analytic examination, there appears to be no
readily definable means to select prospectively the portion of the ADI that is
allotted to "all other exposures." In the general case, allocation of 50% of
the ADI for air and water at hazardous waste sites and the remaining 50% to
other exposures would, in the absence of data to the contrary, seem to
represent a reasonable first approximation. It would probably be prudent,
however, to reconsider the 50% figure whenever readily available data on
background levels of specific substances reveal it to be incorrect.
0774S/122085 - 55 -
-------
DRAFT
13) In many (perhaps most) cases, substances will likely enter both air
and water. The 50% of the ADI alloted to waste site exposures thus must- be
partitioned between the two media, and maximum allowable concentrations (in
mg/1 water and mg/m3 air) must be established to ensure total exposure does
not exceed the acceptable limits. The octanol-water partition coefficient
(K0w) together with Henry's Law Constant (Hc), can be used to indicate
trends in partitioning. Alternatively, it is also possible to use a simpler
scheme, involving only Hc, to decide on the approximate apportioning
specific chemicals will assume.
0774S/122085 - 56 -
-------
DRAFT
References
American Conference of Governmental Industrial Hygienists 1980.
Documentation of the TLVs. ACGIH, Cincinnati.
Barnes, J.M. and Denz, F.A. 1985. Experimental Methods used in Determining
Chronic Toxicity. Pharmcol. Rev. 6:191-242.
Boyd, E.M. and Lloyd, G.R. 1965. Minimal Oral Doses of Benzylpenicillin
Producing Signs of Toxicity in Albino Rats. Arch. Int. Pharmacodyn.
156:172-179.
Calabrese, E.J. 1983. Principles of Animal Extrapolation. John Wiley and
Sons, New York. Pp 55-69.
Carcinogen Assessment Group. 1985. Relative Carcinogenic Potencies Among 53
Chemicals Evaluated by the Carcinogenic Assessment Group as Suspect Human
Carcinogens. in EPA's Health Assessment Document for Dichloromethane
(Methylene Chloride). February, 1985.
Davey, D.G. 1964. The Study of the Toxicity of a Potential Drug: Basic
Principles. Eur. Soc. Study Drug Toxic. Suppl. 6:1-13.
Dixon, R.L. and Hall, J.L. 1982. Reproductive toxicology. In Hayes, A.W.,
ed. Principles and Methods of Toxicology. Raven Press, New York.
Doull, J., Klassen, C.D., and Amdur, M.O., eds. 1980. Casarett and Doull's
Toxicology. Second Edition. Macmillan Publishing Co., Inc., New York.
Dourson, M.J., and Stara, J.F. 1983. Regulatory History and Experimental
Support of Uncertainty (Safety) Factors. Regulatory Toxicology and
Pharmacology 3:224-238.
Environ Corporation. 1985. Documentation of the Development of Toxicity and
Volume Scores for the Purpose of Scheduling Hazardous Wastes. Prepared
for EPA, Office of Solid Waste, Contract No. 68-01-6861. March 28, 1985.
Environmental Protection Agency (EPA). 1978. Proposed guidelines for
registering pesticides in the United States; hazatd evaluation: Humans
and domestic animals. Fed. Reg. 43(163):3733-37403.
Environmental Protection Agency (EPA). 1979. Support Document Test Data
Development Standards: Chronic Health Effects. EPA-560/11-001.
Environmental Protection Agency (EPA). 1980. Water Quality Criteria
Documents; Availability. Fed. Reg. 45:79318-79379.
Environmental Protection Agency (EPA). 1980a. Appendix C - Guidelines and
Methodology Used in the Preparation of Health Effect Assessment Chapters
of the Consent Decree Water Criteria Documents. 45 FR 79347-79357.
0774S/122085 - 57 -
-------
DRAFT
Environmental Protection Agency (EPA). 1980a. Proceedings of the Workshop
on Subchronic Toxicity Testing; Denver, Colorado, May 20-24, 1979.
EPA-560/11-780-028.
Environmental Protection Agency (EPA). 1980b. Scientific Rationale for the
Selection of Toxicity Testing Methods: Human Health Assessment.
EPA-560/1-80-001.
Environmental Protection Agency (EPA). 1982. Health Effects Assessment
Summary for 300 Hazardous Organic Constituents in Support of Regulatory
Impact Analysis of the Land Disposal Branch (LDB) and Interim Final
Incinerator Regulations of the Technical Branch (TB) of the Office of
Solid Wastes, OSWER/EPA, Draft, September 20, 1982.
Environmental Protection Agency (EPA). 1984. Approaches to risk
assessment for multiple chemical exposures. Proceedings of a workshop.
EPA-600/9-84-008.
Environmental Protection Agency (EPA). 1984. Summary of Current Acceptable
Daily Intakes (ADIs) for Oral Exposure. Internal Review Draft.
Environmental Criteria and Assessment Office, Cincinnati, Ohio.
February, 1984.
FAO/WHO Expert Committee on Food Additives. 1958. Procedures for the Testing
of Intentional Food Additives to Establish Their Safety for Use. World
Health Organization Technical Report Series. Rep. 229. Geneva, pp 104.
FAO/WHO Expert Committee on Pesticides Residues. 1965. Report of the Second
Joint Meeting of the FAO Committee on Pesticides in Agriculture and the
WHO Expert Committee on Pesticide Residues.
Fears, T.R., and Douglas, J.F. 1977. Suggested Procedures for Reducing
the Pathology Workload in a Carcinogen Bioassay Program, Part I: J.
Environ. Toxicol. 1:125-137.
Fears, T.R., and Douglas, J.F. 1978. Suggested Procedures for Reducing
the Pathology Workload in a Carcinogen Bioassay Program, Part II:
Incorporating Blind Pathology Technigues and Analysis for Animals with
Tumors. J. Environ. Toxicol. 1:211-222.
Food, Drug, and Cosmetic Act, as Amended January 1980. Published by DHHS.
Publication No. (FDA 80-1051).
Food and Drug Administration (FDA). 1982. Toxicological Principles for the
Safety Assessment of Direct Food Additives and Color Additives Used in
Food. FDA, Bureau of Foods.
Food Safety Council. 1980. Proposed System for Food Safety Assessment.
Washington, D.C. pp 160.
International Commission on Radiological Protection (ICRP). 1975. Report of
the Task Group on Reference Man. Pergamon Press, New York.
0774S/122085 - 58 -
-------
DRAFT
Klaassen, C.D. 1980. Absorption, Distribution, and Excretion of Toxicants.
In Doull, J., Klaassen, C.D., and Amdur, M.O., eds. Casarett and Doull's
Toxicology. Second Edition, Macmillan, New York. Pp 28-55.
Kulwich, B.A. Hardisty, J.F., Gilmore, C.E., and Ward, J.M. 1980. Correlation
Between Gross Observations of Tumors and Neoplasms Diagnosed
Microscopically in Carcinogenesis Bioassays in Rats. J. Environ. Pathol.
Toxicol. 3:281-287.
Lehman, A.J. and Fitzhugh, O.J. 1954. 100-Fold Margin of Safety. Assoc.
Food Drug. Off. U.S.Q. Bull. 18:33-35.
Loomis, T. 1978. Essentials of Toxicology, 3rd edition. Lea and Febiger,
Philadelphia, PA. pp 245.
McNamara, B.P. 1976. Concepts in Health Evaluation of Commerical and
Industrial Chemicals. In New Concepts in Safety Evaluation (Men1man,
M.A., Shapiro, R.E. and Blumenthal, H., eds.) Adv. Mod. Toxicol. 1:61-140.
National Academy of Sciences (NAS). 1975. Principles for Evaluating
Chemicals in the Environment. National Academy of Sciences, Washington,
D.C.
National Research Council (NEC). 1977. Drinking Water and Health - Volume
1. National Academy of Sciences, Washington, D.C.
National Research Council (NRC). 1977a. Principles and Procedures for
Evaluating the Toxicity of Household Substances. National Academy of
Sciences, Washington, DC.
National Research Council (NRC). 1980. Drinking Water and Health, Volume 3.
National Academy Press, Washington, D.C.
National Research Council (NRC). 1980a. Principle of Toxicological
Interactions Associated with Multiple Chemical Exposures. Panel on
Evaluations of Hazards Associated with Maritime Personnel Exposed to
Multiple Cargo Vapors, Board on Toxicology and Environmental Health
Hazards, and Committee on Maritime Hazardous Materials, National
Materials Advisory Board, National Research Council. National Academy
Press, Washington, D.C.
National Research Council (NRC). 1980b. Lead in the Human Environment.
National Academy Press, Washington, D.C.
National Research Council (NRC). 1982. Strategies to Determine Needs and
Priorities for Toxicity Testing. Volume 2: Development. National
Academy Press, Washington, DC.
National Research Council (NRC). 1983. Drinking Water and Health. Volume 5.
National Academy Press, Washington, D.C.
National Toxicology Program (NTP). 1984. Report of the NTP Ad Hoc Panel on
Chemical Carcinogenesis Testing and Evaluation, pp 280.
0774S/122085 - 59 -
-------
DRAFT
Organization for Economic Co-operation and Development (OECD). 1981. OECD
Guidelines for Testing of Chemicals. Paris, France.
Peck, H.M. 1968. An Appraisal of Drug Safety Evaluation in Animals and The
Extrapolation of Results to Man. In Importance of Fundamental Principles
in Drug Evaluation (Tedeschi, D.H. and Tedeschi, R.E., eds.) pp. 449-471,
Raven Press, New York.
Roberts, H.R. 1981. Food Safety in Perspective. In Food Safety edited by
Howard Roberts. John Wiley and Sons, Inc., New York.
Rodricks, J. and Taylor, M.R. 1983. Application of Risk Assessment to Food
Safety Decision Making. Regulatory Toxicology and Pharmacology 3:273-307.
Sontag, J.M., Page, N.P., and Saffiotti. 1976. Guidelines for Carcinogen
Bioassay in Small Rodents. U.S. Department of Health, Education, and
Welfare Publ. No. (NIH)76-801. pp 65.
Stokinger, H.E. and Woodward, R.L. 1958. Toxicologic methods for
establishing drinking water standards. J. Am. Water Works Assoc.
50:515-529.
Street, A.E. 1970. Biochemical Tests in Toxicology. In Methods in Toxicology
(Paget, G.E., ed.) pp. 313-337, F.A. Davis Company, Philadelphia PA.
Stults, V.J. 1981. Nutritional Hazards. In Food Safety edited by Howard
Roberts. John Wiley and Sons, Inc., New York, NY.
Tinsley, I.J. 1979. Chemical Concepts in Pollutant Behavior. John Wiley
and Sons, Inc., New York;
Weill, C.S. and McCollister, D.D. 1963. Relationship Between Short and
Long-term Feeding Studies in Designing an Effective Toxicity Test. Agric.
Food Chem. 11:486-491.
World Health Organization (WHO). 1966. Principles for Pre-Clinical Testing
of Drug Safety. WHO Tech. Rep. Ser. 341, Geneva, pp. 3-12.
World Health Organization (WHO). 1978. Principles and Methods for Evaluating
the Toxicity of Chemicals: Part I. Environmental Health Criteria 6.
Geneva. 273 pp.
Zbinden, 6. 1963. Experimental and Clinical Aspects of Drug Toxicity. Adv.
Pharmacol. 2:1-112.
0774S/122085 - 60 -
-------
Background Document on the Development
and Use of Reference Doses
Part II; Considerations Related to the
Development of Protocols for
Toxicity Studies
Prepared for:
Office Of Solid Wastes
U.S. Environmental Protection Agency
Washington, D.C.
Prepared by:
ENVIRON Corporation
1000 Potomac Street, N.W.
Washington, D.C. 20007
December 20, 1985
-------
Table of Contents
Page
1. Introduction 1
2. Species Selection 1
3. Selections on the Basis of Sex 4
4. Number of Dose Levels 5
4.1 Sample Size 7
5. Examination of Test Animals 8
5.1 Hematology 8
5.2 Blood Chemistry , 10
5.3 Urinalysis 12
5.4 Pathology 14
5.4.1 Gross Pathology 14
5.4.2 Histopathology 15
6. Conclusions 19
References 20 "" 2- v
-------
1. Introduction
In Part I of this report, we identified the types of toxicity studies
minimally acceptable for establishing AOIs. In this Part we provide a
detailed examination of the protocols to be used for such tests. Specifically,
we examine the options available for selecting the species and sex, species
number, number of dose levels, sample size, and extent of animal examination,
toward the end of identifying the minimally acceptable protocol for conducting
a required toxicity test. As in the first report we set forth the options
available and identify the strengths and weaknesses of each. Based on this
information, and on the judgments made by various regulatory and public health
bodies in the past, we recommend how decisions about experimental design
should be made to achieve the maximum amount of relevant information at
minimal cost.
2. Species Selection
A basic premise in toxicity testing is that laboratory animals can be
used to predict toxic responses in humans (NRC, 1977). It is also widely
recognized, however, that animal models are not infallible predictors of human
toxic response. Qualitatively different responses are sometimes seen in
different species. More often, however, different species have similar
responses to a given substance and differ only in the doses which elicit the
toxic response (WHO, 1978).
0870S - 1 -
-------
The reason for the differences in toxic response is most often attributed
to differences in metabolism (e.g., detoxification and activation) among
species. Absorption, distribution, and elimination of a given compound can
also vary among species and can produce differences in toxic responses. And
for reasons that are not always understood, a given target organ in different
species will exhibit differences in sensitivity or response to a given
substance.
One possible error in toxicity testing is choosing a test species that is
less sensitive than humans to the test substances. Knowledge about the
pharmacokinetics of the test substance and the availability of previous test
data can reduce the chances of making this error. For example, if it is known
that a particular substance is metabolized much differently in a rat than in a
human, then a different species would probably be chosen. Similarly, if the
rat is known to exhibit a sensitivity substantially different from humans to
the anticipated effects of a substance, then another species would probably be
chosen. Another way of reducing the chances of making this error is to test
the substance in more than one species and to use the result from the most
sensitive species, unless it is known to be substantially more sensitive than
humans. The use of data from the most sensitive species is particularly
popular when an NOEL is being identified. If information on the test
substance itself does not exist, knowledge of the pharmacokinetics or toxic
effects of structurally similar compounds may help in the choice of a test
species and in the decision to test in more than one species.
Toxicity tests are usually conducted using one or two species.
Occasionally, tests'are conducted in three species. As a general rule, most
0870S - 2 -
-------
test protocols recommend the use of at least two species. For some tests
(e.g., acute toxicity, carcinogenicity), this may be two rodent species. In
others (e.g., subchronic or chronic toxicity), this may be a rodent and a
non-rodent (most commonly the dog). As previously discussed, the primary
reason for testing in more than one species is to identify quantitative and
qualitative differences in response. Presumably, results can be extrapolated
to humans with more certainty if responses in different animal species are
similar. For at least some tests, however, testing in an additional species
has little marginal benefit with regard to identifying either NOELs or
potential toxic responses. For example, Weil and McCollister (1963) evaluated
21 chemicals that had been studied in rats, for 2 years, and also in dogs, for
at least 1 year. They found that in none of the 21 cases was the dog more
sensitive than the rat. In a similar study, Aviado (1978) evaluated 110
chronic studies performed on both the rat and dog and concluded that the use
of both species was unnecessary.
Thus, if the purpose of toxicity testing is to estimate NOELs and to
establish an ADI, the primary reason to test in more than one species of test
animal is to reduce the uncertainties associated with extrapolating the
results to humans. However, as mentioned above, the additional information
gained from testing in a second animal species may not be very substantial,
particularly if the first species is the rat and the second is the dog. Other
information concerning the toxic effects or metabolism of the test substance,
or structurally similar substances, can also be effective in reducing the
uncertainty of extrapolating results to humans. Thus, the decision to incur
the additional expense of testing in more than one species needs to be
evaluated in light of what additional information the test is likely to reveal
0870S - 3 -
-------
and what is already known about the substance. For example, if the test
substance produces a toxic effect for which the test species is known to have
a substantially different .sensitivity than humans, then testing in a second
species may be warranted.
3. Selections on the Basis of Sex
There are many examples of chemicals that produce substantially different
toxicities in males and females. This includes both carcinogenic and non-
carcinogenic effects (Barnes and Denz, 1954; EPA, 1980b). It is the opinion
of at least some toxicologists, however, that males and females of the same
strain and species usually exhibit only slight differences in toxic response
(Doull, 1980).
The mechanisms of all sex-related toxicity differences are not known
(EPA, 1980b). Sex hormones are thought to play an important role, either by
being the target or by modifying the toxic response (Chan et al., 1982). Sex-
related differences in the biotransformation of foreign substances appear to
be the most common reason for the development of different toxic responses
(Doull, 1980). Such differences appear at puberty in some mammals (Dauterman,
1980). Differences between males and females in fat-free bodyweight and food
consumption may also influence toxic responses.
Because there are many examples of sex-related differences in toxic
responses, most protocols suggest or require testing in both sexes (EPA, 1973;
EPA, 1979; EPA, 1980a; FDA, 1982; HAS, 1975; NRC, 1977a; OECD, 1981; WHO,
1978). Barnes and Denz (1954) suggested that it may not, however, be
0870S - 4 -
-------
necessary to test equal numbers of each sex. Until a quantitative evaluation
can convincingly demonstrate that the fraction of chemicals with significantly
different toxicities in males and females is very small, testing in only one
sex is likely to evoke criticism of the tests' reliability.
4. Number of Dose Levels
The ability to characterize a dose-response function and to estimate a
NOEL from toxicity test results are strongly affected by the number of dose
levels used in the test. The dose levels tested usually range from a high
dose level which produces toxic effects but minimal mortality, to a low dose
level which produces no signs of toxicity (EPA, 1978; OECD, 1981; WHO, 1978).
The number of intermediate dose levels tested is variable, and the testing of
more intermediate levels provides a better characterization of the
dose-response relationship. The clear demonstration of a dose-response
relationship allows increased confidence that the relationship is not spurious
(WHO, 1978). Consequently, when more intermediate dose levels are used, the
estimate of the threshold dose for the test species can be made with more
precision and confidence; and the human health risks extrapolated from such
results can also be made with more precision (EPA, 1979).
For the reasons discussed above, virtually all protocols for subchronic
and chronic toxicity tests either require (EPA, 1978) or recommend (EPA, 1979;
EPA, 1980b; NAS, 1975; NEC, 1977a; FDA, 1982; FSC, 1980; OECD, 1981) that at
least three dose-levels (i.e., high, low, and one intermediate) be tested.
While tests for effects other than cancer have been conducted using only one
or two dose levels (Barnes and Denz, 1954), there are two serious drawbacks to
0870S - 5 -
-------
using less than three dose levels. The first is that two dose levels are not
sufficient for characterizing a dose-response function. The second
consideration is that the use of only two doses allows little margin of error
if the dose range is incorrectly chosen. For example, if the tested dose
range is too high, the NOEL may be missed at the low dose, or the high dose
nay have such high mortality that too few survivors remain for meaningful
statistical evaluation. Thus, the use of fewer than three dose levels
increases the chances that a study will have to be repeated.
If only two dose levels are to be tested then it would be important to
conduct a prior range-finding study. Pharmacokinetics data allowing a
determination of the test substance's accumulation potential would also be
useful. Even with this information, an element of luck remains (Barnes and
Denz, 1954). Testing at less than three dose levels is discouraged by the
fact that the additional effort and expense of testing a third dose level is
saall in comparison to the effect and expense of having to repeat an entire
study.
Thus, the use of more dose levels can improve the characterization of the
dose-response relationship and can reduce the chances of having to repeat a
study. The use of more dose levels is, however; limited by practical
considerations; it can either require the use of unmanageable numbers of
animals or the use of small groups of animals, which may be unsatisfactory for
purposes of statistical evaluation (Barnes and Denz, 1954). Unless there is
special need for a particularly well characterized dose-response function,
there appears to be little reason to test at other than three dose levels.
0870S - 6 -
-------
4.1 Sample Size
The number of animals used at each dose level is a compromise between the
need to have a sufficient number of animals to allow adequate statistical
analysis of results and practical considerations of needing to limit the cost
and the workload. (EPA, 1980b; EPA, 1979; WHO, 1978; Barnes and Denz, 1954).
Benitz (1970) commented that the use of a large number of animals may diminish
the thoroughness and care needed for a good study. It is his opinion that
more useful information is obtained from a thorough study using relatively few
animals than is obtained from an incomplete experiment using larger numbers of
animals. An EPA-sponsored conference on the subchronic toxicity test came to
a similar conclusion with its statment that the use of animals in excess of
the recommended numbers would substantially increase the study cost and
diminish the efficient use of facilities and personnel (EPA, 1980a).
Several current protocols and recommendations call for the use of at least
10 rodents of each sex per dose level (EPA, 1980a; OECD, 1981; Chan et al.,
1982; Loomis, 1978; WHO, 1978; NTP, 1984). A few recommendations state that
20 rodents of each sex should be used (EPA, 1978; FDA, 1982; FSC, 1980). When
non-rodents are used, recommendations for the number of animals of each sex to
be used at each dose level fall to 3 to 8 (EPA, 1978; FDA, 1982; EPA, 1980a;
FSC, 1980; Chan et al., 1982). Chronic toxicity protocols may recommend the
use of more animals, particularly in carcinogenicity studies (EPA, 1980b; FDA,
1982; OECD, 1981; EPA, 1978; FSC, 1980).
The number of animals used at each dose level is a choice based primarily
oa practical compromise rather than on theoretical principle. In spite of
this, there is a general consensus, within a fairly narrow range, on the
0870S - 7 -
-------
number of animals to be used. Using more animals than current standard
practice would entail a substantial cost increase and only a marginal
improvement in the ability to detect low-incidence effect (Barnes and Denz,
1954). Using fewer animals would produce a study result with less statistical
confidence than has become the accepted norm.
5. Examination of Test Animals
The discovery and understanding of toxic effects depend, in large part,
on the methods used to examine the test animals. Because it is not feasible
to apply all known tests for toxicity or to examine every cell for signs of
pathology, choices must be made concerning the methods and extent of
examination of the test animals. The choices made play an important role in
determining the reliability of a study. The following section discusses some
of the factors to consider when making these choices. The methods discussed
include hematology, blood chemistry, urinalysis, and pathology.
5.1 Hematoloqy
Hematology tests are essential for the detection of toxic effects to the
hematopoietic system. Minimum testing should provide information on cell
damage and hemorrhagic effects (EPA, 1980b; NTP, 1984). This usually includes
looking for signs of anemia, changes in leukocytes, and some indicator of
clotting ability (Bushby, 1970).
Some of the most commonly performed hematology tests are listed in Table
2. As is shown in the table, some variation exists in what is considered to
be a minimum set of'hematology tests. For purposes of pesticide registration,
0870S - 8 -
-------
Table 2. Hematology Measurements Often Performed
in Toxicity Investigations
Hematocrit ' • 2 • * >s • 6
Hemoglobin '.z.3.".5.6
Erythrocyte count '•2'4'5«6
Total and differential leukocyte counts '-Z'3.
Platelet count 1>3>6
Reticulocyte count
Prothrombin time
Packed cell volume
Mean corpuscle hemoglobin
Mean corpuscle volume
Methemoglobin
Thrombocyte count
Sedimentation rate 3
Sulfhemoglobin
Examination of stained film for polychromasias
and abnormal leukocytes and platelets 3
1 recommended by EPA (1980b)
2 recommended by McNamara (1976)
primary tests recommended by Bushby (1970)
primary tests recommended by Zbinden (1963)
5 recommended by NRC (1977a)
' required by EPA (1978)
0870S - 9 -
-------
the EPA (1978) requires determinations of hematocrit, hemoglobin, erythrocyte
count, total and differential leukocyte counts, platelet count, and, if signs
of anemia are present, reticulocyte count. In addition, the EPA requirements
specify the timing of the test. For example, in a 90-day rodent study, they
require hematology determinations to be made before the dosing begins, at
30-day intervals, and upon termination of dosing (EPA, 1978). A microscopic
examination of the bone marrow is also included in some recommendations for
minimum hematology examination because it may reveal hematotoxic effects
(e.g., anemia) which often only slowly appear in the circulating blood cells
(Egan et al., 1980; WHO, 1978).
As indicators of toxicity, the hematology tests are insensitive; they
rarely are the effect seen at the lowest toxic dose (Weil and McCollister,
1963). Performing less than the minimum hematology determinations would leave
the blood cells and clotting mechanism unexamined and would, therefore,
increase the chances of missing toxic effects in them. In some instances more
than the minimum tests may be indicated by effects that appear during the
course of the toxicity test or by previous indications of hematotoxicity.
5.2 Blood Chemistry
Many chemical analyses can be performed on the blood as indicators of
toxicity to organs, especially the liver and kidneys. Opinions vary as to the
sensitivity, specificity and overall value of the individual test (Tyson et
al., 1985; WHO, 1978). Fluctuations in the chemical indicators may result
from transient changes in organ homeostasis rather than toxic lesions (WHO,
1978). Many toxicologists prefer that blood chemistry indicators of toxicity
0870S - 10 -
-------
be confirmed by histopathology or other evidence of toxicity (EPA, 1980b;
Benitz, 1970).
Among the most commonly applied tests are serum glutamate-pyruvate trans-
aminase (SGPT), serum glutamate-oxaloacetate transaminase (SCOT), sorbitol
dehydrogenase, alkaline phosphatase, blood-urea nitrogren (BUN), and creatinine
(EPA, 1980b; WHO, 1978; Environ, 1985a). The EPA Pesticide Registration
protocol for subacute tests states that the following tests shall be used;
calcium, potassium, serum lactate dehydrogenase (SLOH), SGPT, SCOT, glucose,
BUN, direct and total bilirubin, serum alkaline phosphatase, total cholesterol,
albumin, globulin, and total protein (EPA, 1978). Additional tests many also
be chosen based on the institution and experience of the researcher.
While various individuals and institutions have their preferred tests,
there does not appear to be a strong consensus as to what constitutes a
minimum battery of blood chemistry tests. There is little advantage to
eliminating very many of the tests since automated equipment allows for quick,
inexpensive test results (Tyson et al., 1985); and the results can be helpful
in deciding which organs to look at in the histopathology examination. It
should be mentioned, however, that time spent for interpretation of the test
results can be espensive (Tyson et al., 1985). Unfortunately, currently
available data do not allow determining the cost-effectiveness of the
individual tests (Tyson et al., 1985).
In conclusion, the blood chemistry tests can provide some indication of
organ toxicity and can help guide the expensive histopathology exam. Because
they are relatively"inexpensive, there is little incentive to perform fewer
0870S - 11 -
-------
tests than are mentioned in the minimum test recommendations described above.
The choice of individual tests will depend on the judgment of the researchers
responsible for the toxicity testing.
5.3 Urinalysis
The purpose of urinalysis is to detect toxic effects in the kidneys. As
with the previously discussed clinical tests/ there are many specific analyses
which can be performed (Environ, 1985a). Some of the most commonly used and
recommended tests are listed in Table 3. There is very little consensus in
the literature as to which tests are the most useful or as to the value of
urinalysis in general. Individuals and institutions disagree. For example/
Berndt (1976) believes that urinalysis is superior to histology and anatomical
techniques for detecting nephrotoxicity/ whereas Grice (1972) believes that
histopathology is more sensitive than urinalysis. Similarly, the EPA (1979)
states that routine urinalysis is a significant early indicator of renal
»
damage whereas the National Academy of Science (NRC/ 1977a) sees little value
in routine urinalysis, particularly for subchronic tests. The objections of
the National Academy of Science are primarily due to problems associated with
sample collection and interpretation of results (NRC, 1977a).
It appears that urinalysis can be useful in determining the nature or
specific location of nephrotoxic lesions. Its usefulness as a sensitive
screen for detecting nephrotoxicity is, however, questionable. It would
appear to be of most value when nephrotoxicity is suspected and the individual
test would be chosen on the basis of the researchers' judgment regarding the
possible lesions.
0870S - 12 -
-------
Table 3. Urinalysis Measurements
Often Performed in Toxicity Investigations
Color3
PH3'10
Volume3>6>?
Concentration/specific gravity1'2>3'6>7'8'10
Enzyme activities
urinary glutamic oxaloacetic transaminase (UGOT)l
lactate dehydrogenase4'5
alkaline phosphatase"'5
acid phosphatase
glutamate dehydrogenase
leucine aminopeptidase5
Protein2'3'6'8'9'1*
Creatinine
Ketones'°
6,9,10
Glucose
Urobilinogen1°
Bilirubin '10
Addis count/formed elements including casts1'2'3'9'10
1 recommended as most sensitive by Balazs et al. (1963, cited in EPA 1980b)
2 recommended by Hoe and O'Shea (1965, cited in EPA 1980b)
3 recommended by Street (1970)
4 recommended by Wright and Plununer (1974, cited in EPA 19805) for acute
renal damage •
5 recommended by Cottrell et al. (1976, cited in EPA 1980b) for acute renal
damage
' recommended by Berndt (1976)
7 recommended by the National Toxicology Program (1984)
8 described by the Food Safety Council (1980) as beneficial when renal
toxicity is present
9 recommended by World Health Organization (1978)
10 required by EPA Pesticide Regulation Guidelines (1978)
0870S - 13 -
-------
5.4 Pathology
The pathology exam looks for changes in the structure and function of the
organs and tissues of the test animals. This generally includes a visual
examination of the intact organs (i.e., gross examination) as well as the
examination of tissue under a microscope (i.e., histopathologic examination).
The pathology exam is usually the most valuable method for detecting toxicity,
a consideration which favors a more extensive examination. However, it can
also be the most expensive element in a toxicology study/ a consideration which
favors limiting the pathology examination.
5.4.1 Gross Pathology
The gross pathology examination includes a visual examination of each
animal and organ for signs of toxicity, and usually includes the weighing of
individual organs. Visual examination of the organs can identify some toxic
effects, supplement information from clinical tests, and help determine which
tissues to examine microscopically. There is wide consensus on the need for
gross examination of all organs from all animals at all dose levels (EPA, 19SOb;
EPA, 1978; EPA, 1979; NRC, 1977a; WHO, 1978; FSC, 1980; Barnes & Denz, 1954).
Changes in the absolute organ weight and ratio of organ weight to body
weight are commonly used as indicators of possible toxicity. Weil and
McCollister (1963) found changes in liver and kidney weight, for example, to be
sensitive indicators of toxic effects. However, it must also be recognized
that changes in organ weight may be the result of functional hypertrophy,
metabolic overloading or changes in body weight, rather than the result of a
specific toxic effect (EPA, 1980b). The significance of absolute and relative
organ weight changes varies form organ to organ and is argued in the literature
0870S - 14 -
-------
(EPA, 1980b; Benitz, 1970). At least, organ weight change is useful as a
guide when choosing organs for histopathology examination.
A common recommendation is for the weighing of all major organs, with the
definition of "major" left to the judgment of the researcher. Other organs
should also be include if toxic effects are suspected (EPA, 19805). In its
subchronic and chronic test protocols for pesticide registration, the EPA
specifies which organs shall be weighed (EPA, 1978).
The gross examination of organs is a relatively fast, effective screen
for toxic effects; reductions of this step would not result in much cost or
time savings and would substantially reduce the reliability of the study.
5.4.1 Histopathology
Microscopic examination of tissue can identify toxic effects not seen in
other examinations, confirm toxicity indicated by other tests or exams, and
provide an indication of any dose-effect relationship. Because it is a
time-consuming and expensive process, much thought has gone into ways to
conduct an efficient histopathology examination. The two variables that are
most often considered in discussions on efficient histopathology examinations
are: 1) the number of dose levels at which all animals are to be examined,
and; 2) the number of tissues to be examined per animal.
Recommendations from the National Cancer Institute for cancer bioassays
called for histopathologic examination of all test and control animals,
although positive controls may be exempted (Sontag et al., 1976).
0870S - 15 -
-------
Recommendations by the EPA for chronic (EPA, 1978; EPA, 1979) and subchronic
(EPA, 1978) procedures repeat the recommendation that all animals be
examined. Other scientific panels recommend the routine examination of all
animals only in the high-dose group and in the control group (NRC, 1977a; WHO,
1978; FDA, 1982; OECD, 1981; FSC, 1980). Most of these recommendations also
specify that any organs of animals at intermediate dose levels should be
examined under a microscope if gross examination indicates that toxic effects
may be present. Histopathological examination of the tissues of animals that
die prior to the end a of study is also a common recommendation. Zbinden
(1963) suggested that examining only 25-50% of the control animals would be a
reasonable way to reduce the histopathology workload.
The number of tissues recommended for histopathologic examination in
various necropsy protocols varies substantially. Examination of all (up to
41) tissues is called for in some protocols (EPA, 1979; EPA, 1978; Benitz,
1970) while others call for examining only selected tissues (as few as 16-18)
(NTP, 1984; EPA, 1980b). The rationale for examining all tissues is to
maximize the sensitivity of the test (EPA, 1979; Benitz, 1970). The rationale
for examining only selected tissues is that some tissues are sufficiently poor
indicators of toxicity that their examination is not justified by the cost
(EPA, 1980b; Barnes & Oenz, 1954).
The recommendations that only a selected number of organs be examined are
based primarily on the likelihood of a positive finding, (i.e., sensitivity)
(EPA, 1980b; Zbinden, 1963; NTP, 1984). Tissues that have rarely exhibited
signs of toxicity in previous studies in a wide range of compounds are
eliminated from the"list of tissues recommended for routine microscopic
0870S - 16 -
-------
examination. Other considerations are the desire to examine tissues
representing all organ systems and the frequency with which a particular
tissue has been examined in past studies (EPA, 1980b). The exact number of
tissues to be examined will also depend on the researcher's judgment as to any
other target organs the test substance may affect. In addition to the
examination of the pre-selected tissues, any tissues showing indications of
toxicity (e.g. grossly observed lesions, weight chantes, etc.) are also
usually recommended for histopathologic examination (EPA, 19805). Other
methods for reducing the pathology workload, such as the use of random
sampling techniques have also been proposed (Fears and Douglas, 1977 and 1978).
A third approach to selecting tissues is to perform no routine
examination of any particular tissue and to only look at tissues if there are
other signs of toxicity (Barnes and Oenz, 1954). This approach could include
the evaluation of a tissue at all dose levels once an effect is seen at any
dose level. For example, as a result of the gross observation of liver
lesions, in the high-dose group, the livers of all animals at all dose groups
would be subjected to microscopic examination. Prepared slides from all
tissues and all animals could also be prepared and saved in case reason to
evaluate the slides arises at a later time. Rulwich et al., (1980) and Frith
et al., (1979) compared gross and microscopic examination results and found
that reliance on gross examination to identify neoplastic lesions would have
missed many (50% or more in some tissues) of the lesions identified when at
least one histological section was examined from each organ. The nature of
the lesion and the size of the organ affected the correlation between gross
detection and microscopic detection of lesions (Kulwich et al., 1980; Frith et
al., 1979), but the"two studies illustrate the point that some toxic effects
0870S - 17 -
-------
are likely to be missed if gross examination alone is relied on to select
tissues for microscopic examination.
The extensiveness of the histopathology examination represents a balance
between the desire for sensitive detection of toxic effects and the high cost
of histopathology examinations. Evaluation of all tissues in all animals
maximizes the sensitivity of the study but is the most expensive option.
Complete reliance on gross pathology to identify tissues for examination under
a microscope will diminish sensitivity to a point some would consider
unacceptable.
The design of an intermediate approach depends, at least in part, on the
information about the test chemical that exists before testing begins. If
very little is known about the possible toxic effect of the chemical, routine
examination of a wider range of tissues may be warranted. If there is reason
to believe the chemical affects a particular organ, a more focused
histopathology design may be possible. Before the pathology component of the
study can be designed, the investigator needs to know if the microscopic
examination of all tissues in all animals is going to be required for the
establishment of an Acceptable Daily Intake or whether a more selective
approach will be acceptable. If a more selective approach will be acceptable,
then the design will largely depend on the availability of pre-existing
information about the chemical and the investigators' judgment.
0870S - 18 -
-------
6. Conclusions
*
The importance and expense of toxicity testing has stimulated much
thought and discussion over the last forty years on how to make these tests
more cost-effective. As a result of these efforts, each test variable in the
subchronic toxicity test, for example, has a fairly well defined range within
which an individual investigator must exercise judgment in designing and
conducting a study for a given chemical. Even though each variable has a
circumscribed range of adjustment, the cumulative effect of adjustments of all
the variables can substantially influence the sensitivity and overall cost of
the study.
The most important single decision in the study design is probably the
extent of histopathology to be reguired. Pathology can account for as much as
40% of the overall cost of a toxicity study (EPA, 1979). Thus, the decision
to require microscopic examination of all tissues from all animals at all dose
levels or to allow a more selective approach could substantially affect the
study cost.
The other factors which recommended protocols leave to the investigator's
judgement can also cumulatively have a significant effect on the cost and
sensitivity of the study. The previous discussion of test variables described
how the information available to the investigator could affect such
decisions. This information would include previous toxicity and metabolism
test data when available. However, in many cases very little of this
information will be available, and for these chemicals, knowledge about the
toxic effect and metabolism of structurally related compounds may be valuable.
0870S - 19 -
-------
References
Aviado, D.M. 1978. Overwhelming Similarities and Minimal Differences in
Toxic Responses of Rats and Dogs (Part I: 1974-1978 and Part II:
1966-1974 (Unpublished; cited in EPA, 1980b).
Barnes, J.M. and Denz, F.A. 1954. Experimental Methods used in Determining
Chronic Toxicity. Pharmcol. Rev. 6:191-242.
Benitz, K.-F. 1970. Measurement of Chronic Toxicity. In Methods in
Toxicology (Paget, G.E., ed.) pp. 82-131, F.A. Davis Company,
Philadelphia, PA.
Berndt, W.O. 1976. Renal Function Tests: What Do They Mean? A Review of
Renal Anatomy, Biochemistry/ and Physiology. Environ. Hlth. Pers.
15:55-71.
Bushby, S.R.M. 1970. Haematological Studies During Toxicity .Tests. In
Methods in Toxicology (Paget, G.E. ed.) pp. 338-371, F.A. Davis Company,
Philadelphia, PA.
Chan, O.K., O'Hara, G.P., Hayes, A.W. 1982. Principles and Methods for Acute
and Subchronic Toxicity. In Principles and Methods of Toxicology (Hayes,
A.W., ed.) pp. 1-51, Raven Press, New York, fTf.
Dauterman, W.C. 1980. Physiological Factors Affecting Metabolism of
Xenobiotics. In Introduction to Biochemical Toxicology (Hodgson, E. and
Guthrie, F.W., eds.) pp. 133-142, Elsevier Science Publishing Co., Inc.
New York, NY.
Doull, J. 1980. Factors Influencing Toxicology. In Casarett and Doull's
Toxicology. (Doull, J., Klaassen, C.D., and Amdur, M.O., eds.) pp.
70-83. Second Edition. Macmillan Publishing Co., Inc., New York.
Egan, G.F., Lewis, S.C. and Scala, R.A. 1980. Experimental Design for Animal
Toxicity Studies. In Experimental and Clinical Neurotoxicology (Spencer,
P.S. and Schaumburg, H.H., eds.) Williams and Wilkins, Co., Baltimore, MD.
Environmental Protection Agency (EPA). 1978. Proposed guidelines for
registering pesticides in the United States; hazard evaluation: Humans
and domestic animals. Fed. Reg. 43(163):3733-37403.
Environmental Protection Agency (EPA). 1979. Support Document Test Data
Development Standards: Chronic Health Effects. EPA-560/11-001.
Environmental Protection Agency (EPA). 1980a. Proceedings of the Workshop
on Subchronic Toxicity Testing; Denver, Colorado, May 20-24, 1979.
EPA-560/11-80-028.
0870S - 20 -
-------
Environmental Protection Agency (EPA). 1980b. Scientific Rationale for the
Selection of Toxicity Testing Methods: Human Health Assessment.
EPA-560/1-80-001.
Environ Corporation. 1985. Issues Relating to the Development and Use of
Reference Doses. Draft Report. Prepared for EPA, Office of Solid
Waste. November 12, 1985.
Environ Corporation. 1985a. Severity of Toxic Effects Ranking Scheme. Draft
Report. Prepared for EPA, Environmental Criteria and Assessment Office.
August 5, 1985.
' Fears, T.R., and Douglas, J.F. 1977. Suggested Procedures for Reducing
the Pathology Workload in a Carcinogen Bioassay Program, Part I: J.
Environ. Toxicol. 1:125-137.
Fears, T.R., and Douglas, J.F. 1978. Suggested Procedures for Reducing
the Pathology Workload in a Carcinogen Bioassay Program, Part II:
Incorporating Blind Pathology Techniques and Analysis for Animals with
Tumors. J. Environ. Toxicol. 1:211-222.
Food and Drug Administration (FDA). 1982. Toxicological Principles for the
Safety Assessment of Direct Food Additives and Color Additives Used in
Food. FDA, Bureau of Foods.
Food Safety Council. 1980. Proposed System for Food Safety Assessment.
Washington, D.C. pp 160.
Frith, C.H., Booth, A.D., Greenman, D.L., and Farmer, J.H. 1979.
Correlations Between Gross and Microscopic Lesions in Carcinogenic
Studies in Mice. J. Environ. Pathol. Toxicol. 3:139-153.
Grice, H.C. 1972. The Changing Role of Pathology in Modern Safety
Evaluation. CRC Crit. Rev. Toxicol. 1:119-152.
Kulwich, B.A. Hardisty, J.F., Gilmore, C.E., and Ward, J.M. 1980. Correlation
Beteewn Gross Observations of Tumors and Neoplasms Diagnosed
Microscopically in Carcinogenesis Bioassays in Rats. J. Environ. Pathol.
Toxicol. 3:281-287.
Loomis, T. 1978. Essentials of Toxicology, 3rd edition. Lea and Febiger,
Philadelphia, PA. pp 245.
McNamara, B.P. 1976. Concepts in Health Evaluation of Commerical and
Industrial Chemicals. In New Concepts in Safety Evaluation (Mehlman,
M.A., Shapiro, R.E. and Blumenthal, H., eds.) Adv. Mod. Toxicol. 1:61-140.
National Academy of Sciences. 1975. Principles for Evaluating Chemicals in
the Environment. National Academy of Sciences, Washington, D.C.
National Research Council (NRC). 1977. Drinking Water and Health - Volume
1. National Academy of Sciences, Washington, D.C.
0870S - 21 -
-------
National Research Council (NRC). 1977a. Principles and Procedures for
Evaluating the Toxicity of Household Substances. National Academy of
Sciences, Washington, DC.
National Toxicology Program (NTP). 1984. Report of the NTP Ad Hoc Panel on
Chemical Carcinogenesis Testing and Evaluation, pp 280.
Organization for Economic Co-operation and Development (OECD) . 1981. OECD
Guidelines for Testing of Chemicals. Paris, France.
Sontag, J.M., Page, N.P., and Saffiotti. 1976. Guidelines for Carcinogen
Bioassay in Small Rodents. U.S. Department of Health, Education, and
Welfare Publ. No. (NIH)76-801. pp 65.
Street, A.E. 1970. Biochemical Tests in Toxicology. In Methods in Toxicology
(Paget, G.E., ed.) pp. 313-337, F.A. Davis Company, Philadelphia PA.
Tyson, C.A., Story, D.L., Green, C.E., and Meierhenry, E.F. 1985
Traditionally used Indicators in Relation to Pathologic Lesions. In
Organ Function Tests in Toxicity Evaluation (Tyson, C.A.. and Sawhney,
D.S., eds.) pp. 23-89, Noyes Publications, New Jersey.
Weill, C.S. and McCollister, D.D. 1963. Relationship Between Short and
Long-term Feeding Studies in Designing an Effective Toxicity Test. Agric.
Food Chem. 11:486-491.
World Health Organization. 1978. Principles and Methods for Evaluating
the Toxicity of Chemicals: Part I. Environmental Health Criteria 6.
Geneva. 273 pp.
Zbinden, G. 1963. Experimental and Clinical Aspects of Drug Toxicity. Adv.
Pharmacol. 2:1-112.
0870S - 22 -
------- |