UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
\, A^" WASHINGTON, D.C. 204SO
% PHO^
SEP 24 1984
OFFICE QF
AMD TOXIC SUBSTANCES
MEMORANDUM
SOBJECT: Proposed Design for a Retrospective Study
of PMN Hazard Predictions
., -f
';
FROM; John A, Moore
Assistant Administrator
for Pesticides
and Toxic Substances
TQs Terry Yosie, Ph.D.
Director
Science Advisory Board
Attached for review by the Science Advisory Board's
Environmental Health Committee is the Office of Pesticides and
Toxic Substances' proposed design for a retrospective study of
PMN hazard predictions. The purpose of the proposed study is to
obtain some measure of the validity of the Office of Toxic
Substances' use of structure activity relationships in its
assessment of the potential hazards posed by Premanufacture
Notification (PMN) chemicals submitted under section 5 of the
Toxic Substances Control Act.
The document presents an introduction to and background
information on the task facing OPTS in its efforts to assess the
risks posed by "new" chemicals. This is followed by a
description of the experimental design proposed for the study and
includes discussion of the tests selected by OPTS for use in the
study, the strategy for selecting a set of 100 chemicals from the
PMN universe of over 4,000 chemicals, and the analytic approach
which OPTS contemplates using in measuring the correlations
between SAR-based hazard predictions and the results of the
proposed laboratory testing.
-------
In reviewing the proposed design the Environmental Health
Committee is requested to focus on the appropriateness of: (1)
the testing battery selected by OPTS within an approximate
funding limit of $50,000 per chemical; (2) the selection criteria
and strategy for sampling 100 chemicals from the PMN universe?
and (3) the approach proposed for analyzing the concordance of
the SAR-based hazard predictions with the results of laboratory
testing.
The OPTS in preparing the proposed design for the study has
relied heavily on the helpful comments and suggestions offered by
the Environmental Health Committee at the May 10, 1984, public
meeting, in the July 19 letter reporting the Committee's formal
positions, recommendations, and questions, and at the July 24
briefing of the committee by OPTS. I believe that the Science
Advisory Board can continue to provide OPTS with critical review
and comment on the technical merits of the design proposed for
the retrospective study of PMN hasard predictions.
The results of this study are expected to provide
information vital to an evaluation of the strengths and weak-
nesses of the current use of SAR in the PMN program. Moreover,
the .study...might p-ro-v-ide some indication of the costs and benefits
to be derived from a limited set of testing in terms of its
potential for (1) improving the present PMN hazard assessment
process versus (2) creating economic barriers to the introduction
of innovative PMN chemicals. Finally, as noted by the
Environmental Health Committee in its formal comments, the test
data generated through this proposal are expected to provide
useful toxicologic information which could be used to improve the
quality of the overall TSCA section 5 chemical assessment
process.
In closing, I want to bring two OPTS activities to the
attention of the Environmental Health Committee. The first
concerns the possible development in OPTS of an expert system for
hazard analysis, an enterprise that was suggested to OPTS by the
Committee. This past summer, the Office was fortunate to have
the services of Dr. Eden Fisher of Carnegie-Mellon University, an
American Association for the Advancement of Science/EPA
Environmental Science and Engineering Fellow, who prepared an
evaluation of the potential role of expert systems in the new
chemical hazard assessment process. Dr. Fisher is in the process
of preparing a written report of her conclusions, a copy of which
can be provided to the Committee if so desired. The second
activity concerns a workshop that OPTS is sponsoring jointly with
-------
the EPA Office of Research and Development's Health Effects
Research Laboratory to be held In Research Triangle Park on
October 25-26, 1984, The purpose of the workshop is to develop
an Agency research strategy for 'the improved application of
quantitative structure activity relationships (QSARs) and other
computational techniques as tools for predicting health
effects. Invited extramural participants1 at the workshop
include spokespersons for many of the state-of-the-art techniques
presently being employed in computational studies to predict the
biological chemistry of xenobiotic substances.
Ors. Corwin Hansch, Arnold Hagler, Peter Jurs, Gilles Klopman,
Milan Randic, Gilda Loew, Paul Craig, Carrol Johnsonr Harel
Weinstein, Todd Wipke, and Peter Politzer,
-------
Proposed Design for a Retrospective
Study of PMN Health Hazard Predictions
This paper proposes a design for a retrospective study of
the approach used by the Environmental Protection Agency (EPA) in
evaluating the hazards of new chemicals submitted to EPA under
section 5 of the Toxic Substances Control Act (TSCA). More
specifically, the purpose of this study is to obtain some measure
of the validity of EPA's use of structure activity relationships
in its assessment of the potential health hazards posed by "new*
or Premanufacture Notification (PMN) chemicals submitted under
TSCA section 5.
Under the proposed study, EPA plans to conduct a selected
set of laboratory toxicity tests on a representative sample of
the PMN chemicals. Test data obtained from the study will be
compared with previously generated SPA PMN health hazard
predictions in order to determine (1) the concordance of those
predictions with results obtained by testing and (2) the extent
to which EPA's health hazard assessments would have changed if
similar test data had been available at the time that the PMN was
reviewed.* It is hoped that the proposed project can contribute
to the resolution of some of the major uncertainties surrounding
*
the use of structure activity relationships as the basis for the
hazard assessment of new chemicals. In addition to supporting
the objectives of this project, the proposed testing is expected
to generate useful toxicologic information which will expand the
data base available to the overall TSCA section 5 chemical
assessment process.
The testing approach proposed for this study may also be
useful in indicating the types of test data that should be
considered for inclusion in future PMN submissions. Moreover,
the results of such a study might provide some basis for
evaluating the relative costs and benefits of requiring limited
testing on new chemicals.2
-------
INTRODUCTION
The Toxic Substances Control Act was passed by the Congress
and.signed by President Ford In 1976. The stated purpose of the
Act is to "protect human health and the environment by requiring
testing and necessary use restrictions on certain chemical
substances" (P.L, 94-469, 1976). Under TSCA, the Congress
decided to distinguish between so-called "new" and "existing"
chemicals. New chemicals, that is, those not appearing on an
inventory of existing chemicals, are subject to premanufacture
reporting requirements under section S of TSCA. Since publica-
tion of the inventory of existing chemicals in July 1979,
Premanufacture Notifications have been received on some 4,000 new
chemicals (see Figure 1).
TSCA section 5 requires that manufacturers and importers of
new chemicals submit a Premanufacture Notification (PMN) to the
EPA 90 days before they intend to commence manufacture or
import. TSCA thus provides EPA with a 90-day review period,
extendable with cause to 180 days, in wnich to complete its risk
determination on each PMN chemical. (The task before EPA in
regulating chemicals under TSCA is to distinguish between
"reasonable" and "unreasonable" risks; as used in TSCA, "risk" is
some function of hazard, i.e. , toxicity, and exposure, and
includes consideration of economics.) Under TSCA section 5r
certain information must be provided in the notification, as
follows;
-------
0 a description of the new chemical substance, its
chemical identity, molecular structure,, and common or
trade name;
0 the estimated total amount to be manufactured or
processed;
* the proposed categories of use and the estimated
amount to be used for each such category;
0 a description of the by-products resulting from
manufacture, processing, use, or disposal;
* the number of individuals expected to be exposed in
their place of employment and estimates of the
expected duration of exposure?
0 the manner and methods of disposali and
0 any test data in the possession or control of the
notifier related to the health or environmental
effects of the substance (15 O.S.C. 2604(d)).
As can be seen in the above listingt TSCA does not require that
submitters conduct toxicity testing prior to submission of the
PMN; rather they need only supply any health or environmental
test data which are available to them at the time of
submission. Presently, EPA receives test data on fewer than 50%
of all PMNs submitted; when provided, the data most commonly
consi-st of acute lethality and local irritation studies (see
subsequent discussion on this point).
-------
Given the general paucity of submitted test data on PMN
chemicals, EPA has come to rely on structure activity relation-
ships (SAR) in its evaluation of the potential hazards posed by
these chemicals* The analysis performed by EPA in its use of SAR
involves the following components:
0 review of submitted test data,, if any, on the PMN
chemical;
0 review of test data available on structurally
analogous substances (these can be identified by
either EPA or the submitter)?
0 use of quantitative SAR methods where available and
applicable?
0 the professional judgments of scientific assessors in
interpreting and integrating the above, plus
consideration of factors derived directly from the
structure of the PMN chemical. These factors can
include, for example, molecular shape and size,
fundamental physical/chemical properties, log P,
presence and positioning of reactive chemical
functional groups, metabolic pathways, and so on (see
Areas, 1983; Arcos and Auer, in press).
The task before EPA is to determine, despite a paucity
(oftentimes an absence) of test data, whether PMN chemicals,
under their projected conditions of use, manufacturing, or
processing, "may" or "will" present an unreasonable risk of
injury to health or the environment. Under the former (TSC&
section 5e or "may present") finding, the Agency can prohibit or
limit manufacture processing, use, or disposal of the new chem-
ical pending development of test data sufficient to permit a
reasoned evaluation of the risks posed by the chemical (15 U.S.C.
2604{e)). In the event that EPA can support the finding that a
-------
new chemical "will present" an unreasonable risk, the Agency can
take action under TSCA section 5f to prohibit manufacture
altogether or to limit its use or release without a requirement
for development of additional test data (15 u.S.C. 2604(£)). To
date, SPA has issued two TSCA section 5f orders (involving 3 PMN
chemicals)t and 7 unilateral TSCA section 5e orders (involving
15 PMN chemicals); 23 voluntary or consent Se orders (involving
150 PMN chemicals) have also been instituted. In addition, 77
PMN chemicals have been voluntarily withdrawn by the submitter in
the face of a likely TSCA section 5e or 5f order.(EPA, 1984a).
In the event that EPA chooses not to take any action to
control or otherwise limit a PMN chemical, the submitter is free
to manufacture or import the chemical following expiration of the
90-day review period. Upon commencement of manufacture or
import, the submitter is required to providewritten notification
of this fact to the EPA, after which the chemical will be placed
on the inventory of existing chemicals. To date, a notice of
commencement of manufacture or import has been received on
approximately 50% of all PMNs submitted to the Agency (see Figure
2).
EPA's use of SAR in reaching PMN.hazard assessment
conclusions has been the subject of questioning and some
criticism from the Congress, environmental groups, and others
(Office of Technology Assessment, 1983? Government Accounting
Office, 1984; American Chemical Society, 1984) who point out the
many uncertainties associated with the approach, in response to
these concerns as well as EPA'S desire to have some measure of
how well it is doing in predicting new chemical hazards, a design
is hereby proposed for conducting a retrospective study of PMN
hazard predictions. The objective of this effort is to obtain
some indication of the validity of fiPA's use of SAR in assessing
the potential hazards posed by new chemicals.
-------
BACKGROUND
The PMN assessment ^Process
Figure 3 presents a schematic of the overall PMN risk
assessment process employed by EPA. The first step in the
process is a determination that ail necessary information has
been included in the notification. This is followed by a series
of three meetings which bring senior level expertise to bear on
the questions of chemistry, ha2ardf and exposure within the first
15 days of the 90-day period available to EPA for the assessment
of each PMN chemical. Although generally characterized as
"professional judgment" meetings, the initial discussions are
supported by a variety of information gathering activities
including: (1) identification of previously received PMN
chemicals which structurally (or otherwise) resemble the newly-
submitted chemical; (2) substructure and nomenclature-based
searches to identify potential analogues? (3) searches in
handbooks and bibliographic data bases to identify pertinent
literature on the PMN chemical and/or its potential analogues;
(4) critical review of submitted test data with special emphasis
on toxicologic testing? (5) calculation of a variety of physical
chemical properties including water solubility (Banerjee et al./
1980) Koc, vapor pressure (Lyman et__al_., 1982), log P (Chou and
Jurs, 1979), bioconcentration factor (Veith _et_ _aJL. , 1979), etc.
The first of these meetings, known as Chemistry Review and
Search Strategy (CRSS)f (1) confirms the chemical identity of the
substance in terms matching the name with the structural depic-
tion provided, (2) considers (or develops as needed) the PMN
chemical's industrial synthetic process to identify by-products
or impurities which might be present (if by-products and/or
impurities are identified by the submitter, the task is one of
explaining and expanding on the information provided), (3)
collects and/or estimates values of various physical chemical
properties, and (4) attempts to understand how the substance
-------
functions chemically in its projected use applications. The
second professional judgment meeting is known as the Structure
Activity Team (SAT) and is responsible for the initial assessment
of the potential hazards posed by the chemical. The SAT is
composed of senior scientists who specialise in organic
chemistry, environmental chemistry, metabolism of xenobiotic
chemicals, chemical carcinogenesis and mutagenesis, systemic
toxicity, teratogenesis, and environmental toxicology, SAT
consideration of a PMN chemical commences with a presentation by
the CRSS chairman of the findings of that group. This is
followed by discussions of (1) environmental fate, (2) health
effects, and (3) environmental effects. The health effects
discussion considers uptake, metabolism, distribution, excretion,
and any specific toxicities suggested by the PMN chemical's
structure or identified through preliminary information searches;
test data provided with the notice are also discussed and
evaluated at this time. The environmental effects discussion
considers acute and chronic aquatic toxicity, bioconcentration
potential, and terrestrial toxicity. The final professional
judgment meeting is termed the Exposure Analysis Meeting (EXAM)
and is responsible for consideration of the nature and magnitude
of (I) occupational exposures associated with manufacture,
processing, and use of the new chemical, (2) environmental
releases resulting from these activities as well as disposal of
chemical wastes, (3) consumer and environmental exposures
(including drinking water), and {45 environmental fate (transport
and transformations).
The three professional judgment meetings are followed by the
initial risk assessment meeting, termed the Focus meeting. The
Focus meeting is chaired by a regulatory decision-maker and
brings together the chairpersons from the chemistry, hazard, and
exposure meetings for synoptic presentations of the conclusions
from each previous meeting followed by open discussion of the
risk potential presented by the PMN chemical. Using the
information and/or judgments developed to this point, a decision
-------
is reached at the Focus meeting to either drop the chemical from
further consideration (approximately 60% (EPA, 1984a) of all
cases) or to pursue further review of the chemical, a process
known as a "Detailed Review."® Thus at the crucial initial
stages of the PMN review period, EPA has devised a "screening"
procedure which relies on senior expertise to identify those PMN
chemicals which are potentially hazardous and/or might have
significant exposure and release potential, from other chemicals
considered unlikely to present a significant risk of injury to
health or the environment.
During the Detailed Review, those aspects of the case
identified for further consideration at the Focus meeting are
examined in depth by discipline experts. All available data on
the PMN substance and its analogues are gathered and reviewed by
the experts who prepare a written evaluation of the data and make
appropriate recommendations. These recommendations are then
reviewed by senior scientific and managerial staff who determine
the nature and magnitude of potential hazards and then examine
whether there is sufficient exposure such that the chemical might
pose a significant risk. If the decision is reached that the
chemical might pose such a risk, the case is presented to senior
regulatory decision-makers at a "Disposition" meeting who decide
whether the risk is "unreasonable," The possible outcomes of
this meeting include; (1) control the chemical pending completion
of needed testing under TSCA section 5e, (2) directly control the
chemical under TSCA section 5f, or (3) drop the case from further
consideration. All of the above decisions turn on the strength
of the case made for an "unreasonable risk" determination with,
as noted before, TSCA section 5e requiring a "may present"
finding, TSCA section 5f a "will present" finding, and a drop
resulting from the inability to meet either of these tests.
-------
Test Data Received inPMWs
As noted earlier, TSCA section 5 does not require the
provision of test data with incoming PMN submissions. Rather,
submitters must provide any test data that are available to them
at the time of submission. Table 1 presents summary statistics
for test data submitted with*PMNs, As can be seen, 52% of all
PMNs contain no test data of any type? the situation is somewhat
better for nonpolymeric PMN chemicals, of which only 391 do not
contain any test data. In general, health test data are more
commonly provided than ecotoxicological or environmental fate
data. Among the various types of health test data provided, the
most commonly received studies are various acute tests
(specifically, oral and dermal acute toxicity and skin and eye
irritation studies), with mutagenicity, dermal sensitization, and
"other" health, studies, appearing less, frequently.. "Other" health
test data include repeated dose toxicity studies, teratogenicity
assays, phototoxicity studies, and a variety of other toxicity
studies, (The reader is also referred to an earlier evaluation
of the test data provided in PMNs prepared by the Office of
Technology Assessment, 1983.)
Table 2 provides a summary of the combinations of different
health test data that are most commonly encountered among PMNs
containing health data. Thus, this table gives an indication of
the extent of the testing that is undertaken for those cases that
contain health test data. The most frequently encountered combi-
nations of tests submitted with PMNs consist of acute toxicity
(any route) and local irritation studies (34% of the 1560 PMN
cases- containing health test data); these for the most part take
the form of the battery required for labeling purposes under the
Federal Hazardous Substances Act (16 CFR 1500). An equal number
(522 or 33%) of the health data-containing cases include acute
toxicity (any route) and irritation studies (both skin and eye)
in combination with any one or more of sensitization,
mutagenicity, or "other" health test data. Sixty (or 4%) of the
-------
data-containing cases contain onlysensitization, mutagenicity,
and/or "other" health test data (thus they do not contain acute
or local toxicity testing). Table 3 reports the number of PMNs
which contain one category of health test data, two categories of
test data, and so on through the eight health test data
categories included in TDIS, an EPA in-house PMN information
system.
The amount and types of health test data provided with PMMs
can be contrasted with the data estimated (national Research
Council, 1984) to be available on chemical substances contained
in the TSCA inventory of existing chemicals. Table 4 presents
estimated percentages of TSC& existing chemicals which have
undergone various types of health effects testing with the
results of the testing being publicly available ti.e., additional
test data may exist in restricted access files). The table also
presents corresponding figures for the set of PMN chemicals. As
Table 4 illustrates, compared with a published estimate of avail-
able data on TSCA existing chemicals (NRC, 1984), relatively more
data are available on the set of PMN chemicals than is the case
for TSCA existing chemicals. However, in absolute terms,
available data on PMN chemicals for endpoints other than acute
toxicity and local irritation are limited. This is a serious
limitation since studies such as sensitization, mutagenicity,
reproductive/developmental toxicity, repeated dose toxicity, and
so on, are generally viewed as being more critical to an overall
assessment of health hazards than are acute studies.
The Use of BAR in PMN•Health Hazard Assessments
Given the.limitations of the test data which are provided
with PMNs, EPA has evolved a reliance on structure activity
relationships (SAR) in its evaluation of the potential hazards
posed by PMN chemicals. EPA's operational definition of
"structure activity relationships" was described above as being
comprised of four components. The first of these components.
10
-------
submitted test data, has already been discussed. The second of
the components, data on analogous substances, deserves some
comment. In order for an analogue to be useful to EPA, it must
resemble the PMN chemical in one or more critical aspects (e.g.,
structurally* substructurally, physicochemically, etc.) and at
the same time must have pertinent toxicologic data available on
it in the scientific literature. It has been EPA's experience,
and the point was recently confirmed by the NEC (1984), that
available test data are very limited for most chemicals, but
especially so for TSCA chemicals. This factor becomes a major
limitation on the usefulness of many potential analogues.
Analogous substances may be identified by either EPA or the
submitter. The latter instance most often takes the form of test
data on structurally related substances which are existing
chemicals produced by the submitter., .The, data provided tend to
resemble in scope the types described earlier as typically
accompanying PMN submissions (i.e., acute toxicity and local
irritation studies/ occasionally inutagenicity or repeated dose
toxicity studies), EPA relies on two sources in its internal
efforts to identify chemical analogues. These consist of
analogue recommendations offered by members of the Structure
Activity Team and other technical staff/ and structural analogues
retrieved from several publicly available automated chemical
o
substructure and nomenclature search systems. In the former
instance, the proposed analogues often provide a rich source of
pertinent information which can be applied to the assessment
effort* The Structure Activity Team and other technical staff
also provide guidance in constructing the strategy for the
automated analogue searches. This guidance consists of the
identification, based on experience and professional judgment, of
the putative toxicophore(s) within the PMN chemical's
structure. Potential analogues resembling the PMN chemical in
ft
the structure or function* of the putative toxicophore(s) are
then identified via an automated searching capability. When
attempting to select analogues for literature searching and
11
-------
subsequent assessment purposes, similarity in the structure or
function of the toxicophore is an essential element. Physical
chemical properties, especially those that are known or suspected
of contributing to a chemical's biological activity, as well as
other major aspects such as projected metabolic pathways for the
potential analogues as compared with those of the PMN chemical
are also considered in selecting potential analogues.
One of the major limitations of the available substructure
and nomenclature search systems is that they restrict the
searchable parameters to chemical names, name fragments,
substructural components, molecular formulae, and so on.
Sizeable data bases of physical chemical properties are not
available in a readily searchable environment, although several
are under development (Saktn and Johnson, 1981? Milne and Heller,
1980} Magnuson _ejt jj^., 1981; Howard ^et^., 1982? Page, 1983;
Page and Kissman, 1984). Thus automated screening of the
analogue search outputs using quantitative measures of physical
chemical properties to select potential analogues is not
presently available. Consequently EPA relies on manual screening
using a variety of quantitatively and/or qualitatively applied
factors which can be used for comparison between the potential
analogues and the PMN chemical. These factors can include
relative differences in, for example, molecular weight, molecular
topology, log P (Hansch and Leo, 1979), presence and positioning
of reactive or potentially reactive groups, possible steric
effects, presence of aromatic systems, and presence of ionizable
(Perrin, 1965? Perrin, 1972| Serjeant and Dempsey, 1979) or
zwitterionic groups* Once potential analogues have been
selected, they are subjected to automated literature searching
.using a variety of readily available bibliographic systems and
data bases in the hopes of uncovering pertinent toxicity
information. Only those analogues yielding test data are carried
forward in the assessment process.
12
-------
The use of the third component, quantitative SAR methods, is
at present limited to the estimation of certain physical chemical
properties (such as water solubility (Banerjee et al«, 19805,
Koc, vapor pressure (Lyman et al., 1982), log P (Chou and Jurs,
1979) ) and ecotoxicity (prediction of acute LCSOs in aquatic
organisms (Veith et al., 1983? Lipnick and Dunn, 1983; Hermans,
1983; Konemann, 1981; Hutzinger _ej: jl_. t 1978) and estimation of
bioconcentration factors (Veith et^^l^. / 1979)).
The last of the four components, the knowledge and profes**
sional judgments of scientific assessors in the interpretation
and integration of available information, is most critical in
terms of the overall success of the evaluation effort. Given
that the three preceding components, even when combined, will
generally produce a limited set of useful information, the
importance of _ the...knowledge and professional judgments of the
scientific assessors becomes apparent* This is especially so in
the case of information developed on analogous substances which
must be critically evaluated and interpreted in terms of the
weight that should be applied to each analogue as a function of
of its "closeness" to the PMN chemical. Thus the assessors1 task
is to evaluate the toxicologic potentialities of the PMN chemical
using submitted data and extrapolations of data available on
suitable analogues. In performing this task the assessors must
consider a variety of parameters as they apply to the PMN
chemical and, in comparison, to the analogue chemicals. These
parameters include those applied previously in selecting
potential analogues as well ass
0 potential for skin, pulmonary, and gastrointestinal
absorption;
0 biotransformation pathways;
0 distribution and excretion;
13
-------
0 consideration of the possible mechanisms of toxicity,
and other parameters (Areas, 1933? Arcos and Auer, in
press?).
Recently, an evaluation of IPA's use of SAR in the PMN
process was prepared by Adrien Albert (1983) at EPA*s request*
Professor Albert, author of Selective Toxicity: the Physico"
Chemical Basis for Therapy, is one of the foremost international
authorities on structure activity relationships. The report
identified four major approaches or tools that are available for
predicting biological activity from chemical structure and
physical chemical properties;
0 professional expertise of scientific assessors;
0 attempting, to relate the whole molecule to a class of
chemicals for which adequate biological data exist
(analogues are chosen to be as close as possible in
size, overall molecular structure, and component
parts)? consideration of biotransformation enters
here?
9 searching for, structural analogues that have a
"domain" (or substructure) similar to the domain
*
thought to be responsible for biological activity in
the submitted chemical (in other words, the putative
toxicophore); biotransformation is a factor in this
approach as well?
.° quantitative structure activity relationships (QSARs)
as exemplified by Hansch analysis.
Albert concluded that EPA relied to the greatest degree on
the first 3 of the available approaches in assessing the hazards
of PMN chemicals. While he recognized and accepted the reasons
for EPA's limited use of QSARs, Albert recommended that EPA
14
-------
attempt to expand its utilization of QSAR approaches to toxicity
assessment. The major factor identified by Albert as limiting
EPA's use of QSARs was the need, as yet largely unmet, to develop
a large data bank containing toxicologic test data, physical
chemical property data, and QSAR descriptors. Albert, in
addition, called attention to the limitations of and potential
pitfalls in an over reliance on the "domain" approach.
In making projections about the toxicity of the PMN
chemical, EPA is hampered by the limited data that are generally
available on the PMN chemical and its analogues. Because of
this, EPA's hazard predictions tend to take the form of
conservative, worst case analyses which reflect the uncertainties
inherent in a process which uses limited test data. The TSCA
section 5e "may present" language recognizes the uncertainties
that are likely to confront EPA in assessing PMN chemicals and
thus allows a less robust regulatory finding to suffice in
requiring the development of the test data needed to adequately
assess the risks posed by new chemicals.
PROPOSED DESIGN10
The stated purpose of this retrospective study of PMN health
hazard predictions is to obtain some measure of the validity of
EPA's use of structure activity relationships in its assessment
of the potential hazards posed by PMN chemicals. The study is
not intended to validate PMN hazard predictions in an absolute
sense. Rather the study is more limited in scope and is intended
to provide EPA with an indication of the validity of its hazard
predictions relative to a limited set of toxicity testing. In
this way, EPA may begin to determine the strengths and weaknesses
of the present SAR-based approach under TSCA section 5. The
study will also provide an indication of the types of toxicity
testing, if anyr that can best complement, or, if indicated, are
needed to supplement, EPA's assessment capabilities. Finally,
15
-------
given that EPA's charge under TSCA is to protect public health
and the environment without unduly impeding or creating economic
barriers to technological innovation, the study attempts to
provide an indication of the relative costs and benefits of
limited testing if such a requirement were to be contemplated for
future PMN chemicals.
The discussions that follow lay out the major factors that
were considered in formulating the study and proposes a study
design which in EPA's judgment can best meet the objectives of
the project within the limitations outlined,
General Design Considerations
The major factors serving to limit the scope of the retro-
spective study are the resource limitations of available funding,
time, and personnel that can be committed to the project. At
this time EPA envisions a project that would involve
approximately one hundred (100) PMN chemicals* each subjected to
around $50/000 worth of testing (EPA, 1984c), for an initial cost
of $5.0 million. Subsequent analysis of the data, report
preparation, and other costs associated with the study are
expected to approximately double the cost to S10.0 million for
the completed project,
in an initial exposition presented to the EPA Science
Advisory Board's Environmental Health Committee (SAB-EHC) on the
design of a retrospective study of PMN health hazard assessments,
EPA (1984b) proposed focusing the effort on determination of a
"false negative" rate among EPA's PMN hazard predictions. A
false negative prediction is one that, either by omission or
commission, incorrectly characterizes a "toxic" chemical as
nontoxic in one or more effect areas. The rationale offered for
this proposal was that false negative hazard predictions are of
public health concern due to their potential for contributing to
underestimation of the risk potential posed by such chemicals.
16
-------
False positive predictions, on the other hand, were characterized
as contributing to, if anything, overestimation of risk potential
(and possibly to unwarranted regulation). Several reviewers
(SAB-EHC, 1984} of the initial discussion of the project (EPA,
I984b) noted that, while determination of a false negative rate
is certainly desirable, to focus tha effort on this aspect to the
exclusion of determining a false positive rate would weaken the
overall study, EPA has decided to broaden the study to include
testing of both positive and negative hazard predictions.
Approach to Data Analysis
in general, the analysis of the data from the study will
focus on the following basic -objectives!
" 1, estimate the degree of agreement in the"" scoring of
chemicals by the two scoring methods (i.e., EPA's SAR-
based hazard predictions versus laboratory testing);
2. test whether the estimated degree of agreement is
statistically significant (i.e., could the degree of
agreement be attributed solely to chance);
3. if the laboratory tests can be viewed as a more
reliable scoring method, then estimate the false
positive and false negative rates of the SAR-based
method relative to the laboratory tests. Note that
these rates would have a limited meaning since the
reliability of the selected laboratory tests {see
subsequent discussion of the tests chosen for this
study) with respect to the actual response of the
chemicals in confirmatory assays would not be known.
17
-------
Let the following table represent the cross classification
of the SAR-based hazard predictions and the corresponding labora-
tory test results for a selected sample of PMN chemicals where a,
b, ct and d represent the number of PMN chemicals in each
quadrant and N represents the total number of PMN chemicals in
the sample.
Test Results
SAR-Based
Hazard
Predictions
-1*
-
-1*
a
c
" a *• c
*"*
b
d
b 4- d
a + b
c + d
N
Also let
be the proportion of PMN chemicals in the sample which are scored
the same by both scoring methods.
= (a + c)(a * b) + (b + d){c + d)
is the proportion of PMN chemicals that would be expected to be
scored the same by chance alone. The statistic K, where
- P
18
-------
represents the degree of agreement after allowing for chance. If
K < 0, this means that the sample had less than•the number of
agreements than would have been expected by chance alone. If K >
Q, this indicates the observation of more than chance agreements.
K = O is then indicative of only chance agreement,
The statistical significance of K can be tested by
calculating
"X? = K2
Var(K)
where Var(K) =
P- + P^2 - V**3 I (a + c) (a + b) (2a + b + c) + (b + d) (c + d) (2c + b + a) ]
\* C
N {1 -pc'j2 '
2
If %- exceeds 3.84, the value needed for significance at the 0.05
level, the conclusion would be that K is significantly different
from zero.
If one views the lab tests as a more reliable method, then one
can estimate the false positive and false negative rates of the S&R-
based predictions viz a viz the laboratory tests when
false positive = b
rate a + b
false negative = c ; ...._
rate c * d
and their standard errors of estimate are, respectively,
f
H
ab and I cd '
(a + b)3 ^ (c + d)3
19
-------
The preceding is the simplest kind of analysis that could be
done and it is presented to illustrate the general approach which is
contemplated for data analysis. The actual analysis of data would
probably include more complex versions of this basic analysis.
Examples includes
1. applying the analysis for each of several different
toxicolofical endpoints and/or types of laboratory tests
or combinations of laboratory tests,
2. classifying the outcomes of the SAR-based predictions
and/or the laboratory tests into more than two categories
(for example, the 2X2 cross classification matrix
described above could be expanded into 2X3, 3X3, etc.
matrices? a simple example involves keeping the present
. . , classification of, ,SAR-b.a,s..e<3., predictions and classifying
the test results into + , -, and ± categories yielding
a 2 X 3 matrix).
3. expanding the dimensions of the analysis! an example is
to include consideration of chemical class.
Selection of Chemicals
One of the major design factor in this study concerns selection
of the PMN chemicals that will be tested as part of the effort. In
the initial discussion of the project, EPA (1984b) proposed a series
of exclusion criteria which could be used to eliminate certain PMN
chemicals from consideration on the basis of low hazard potential or
low risk potential. The S&B-EHC (1984) in written comments
expressed concern regarding the proposed exclusions and requested
that EPA reconsider the issue. EPA has done so and the following
discussion presents EPA's planned approach to chemical selection.
20
-------
The PMN Universe
Over 4,000 chemicals have been submitted to EPA under section 5
of TSCA since mid-1979. The ratio of polymeric to npnpolymeric PMN
chemicals is approximately 40i60. For the set of nonpolymeric PMN
chemicals, the mean molecular weight is approximately 450 daltons
while the mode is between 200 and 300 daltons? the range of
molecular weights extends from 104 to over 4000 daltons (see Figure
4). Information as to health test data availability on the set of
PMN chemicals was previously discussed and presented in Tables 1
through 3.
Considerations in Developing the Samp ..n Strategy
Among the PMN chemical categories, proposed for exclusion in the
initial discussion of this project (EPA, 1984b) was a subset of the
polymeric PMN chemicals. This subset of polymers can be charac-
terized as high molecular weight (less than 5 weight percent below
500 daltons) f nonreactive, and essentially insoluble in water. For
the set of polymeric PMN chemicals {n = 1739), approximately 600 (or
35%) meet these criteria. Essentially all polymeric PMN chemicals
meeting these requirements have been identified by EPA as presenting
a low degree of toxicity based primarily on the view that they will
not be bioavailable. EPA has decided to exclude this subset of
polymers from the sampling universe due to the expectation that they
are likely to possess a low degree of toxicity.
A major consideration in developing the sampling strategy for
the project is the question of the likelihood of obtai-ning specimens
of PMN chemicals for testing. An important limitation in this
regard is the fact that in the absence of Agency receipt of a notice
of commencement of manufacture or import, only "small quantities" of
the PMN chemicals will be available. EPA anticipates that it
would encounter substantial difficulties in obtaining specimens of
chemicals which have not entered commercial production. • EPA has/
21
-------
therefore, decided to include in the sampling universe only those
PMN chemicals which have entered into commercial production as
evidenced by receipt of a notice of commencement of manufacture or
import (known hereafter as "NOG chemicals"). At present, the set of
NOC chemicals contains 1504 or 50% of the PMN chemicals (both
polymeric and nonpolymeric) received by the Agency. Since the
restriction to the set of NOC chemicals could introduce biases into
the overall result, an attempt will be made to study if such a bias
does exist and, if so, to estimate its magnitude. Various analyses
are presently underway to determine the presence of any significant
bias(es) between the set of NOC chemicals and the balance of the
PMNs. In these analyses, the set of NOC chemicals will be con-
trasted with the set of "non-NQC* PMN chemicals in terms of such
parameters as the presence of health test data, the presence of
health concerns (Structure Activity Team "level of concern" rankings
will be used to identify cases having health concerns! see discus-
sion of this point below), and the ratio of polymers to
nonpolymers. If there is any indication of significant bias(es}
between the sets of NOC and non-NOC chemicals, an appropriate number
of additional chemicals will be sampled from the set of non-NOC PMN
chemicals (once again, excluding certain polymers) and a substantial
effort will be made to obtain specimens of these chemicals. The
subset of non-NOC chemicals would then be subjected to the testing
scheme described previously and appropriate analysis of the testing
results undertaken.
Employing the exclusions as described above is estimated to
yield a sampling universe containing somewhere between 1,200 to
1,400 PMN chemicals. Note that these figures do not include PMN
chemicals received in fiscal year 1984; this will eventually add
approximately 500 chemicals to the sampling universe.
22
-------
Sampling of PMN Chemicals
A valid probability sampling will be made of the PMN sampling
universe to select chemicals for testing. A simple random sampling
procedure would achieve data analysis objectives (1) and (2)
(described above in the section entitled "Approach to Data
Analysis")! however, to best achieve data analysis objective {3) as
well as objectives (1) and (2) a stratified random sampling
procedure will be used. First, two strata will be defined; one
stratum will consist of chemicals in the sampling universe for which
EPA identified one or more adverse health concerns (i.e., "positive"
SAR-based hazard predictions) and the other stratum will consist of
chemicals in the sampling universe for which EPA had low or no
health concerns (i.e., "negative" SAR-based hazard predictions).
The total sample will then be allocated to the two strata in
proportion to the relative size of these strata in the PMN universe
as long as both groups are large enough so that this approach will
not result in too small a sample to calculate one or the other of
the error rates, A systematic sampling with a random start (NRC,
1984, p. 45) will then be made from a chronologically ordered
listing of the PMN chemicals within each stratum? this latter step
will be, in effect, a further stratification of the sample through
time. Additional strata will also be developed if indicated; for
example, it may be useful to stratify on the basis of polymeric
versus nonpolymeric PMN chemicals.
PMN chemicals having "positive" SAR-based hazard predictions
will be separated from those having "negative" SAR-based hazard
predictions on the basis of the health effects "level of concern"
estimate provided on each PMN chemical as part of the deliberations
of the structure Activity Team. The SAT level of concern estimates
were originally developed for use by management in assigning
assessment resources to individual PMN cases. The operational
assumption was that if the senior scientists on the SAT identified
certain PMN chemicals as being of "low" health concern relative to
other PMN chemicals, then one might want to direct technical
23
-------
resources toward other PMN chemicals which received a "moderate"
or "high" health concern ranking from the SAT. Although the SAT
level of concern rankings have limitations in separating "toxic"
from "nontoxie" chemicals/ for the purpose of stratifying PMN
chemicals as described above, the health level of concern
rankings provided by the SAT are thought to be an adequate
discriminator for the purposes of this project.
Selection of Tests
in the initial discussion of the "Selection of Tests" issue,
EPA (1984b) suggested that one could structure the retrospective
study of PMN health hazard predictions "(1) to test for a
predetermined set of toxicologic endpoints.,« or (25 to test for
the set of toxicologic endpoints which are identified by SAR as
being-0f potential concern.». „" The decision reached by EPA is
to structure the study to incorporate both of these approaches,
with emphasis placed on the former. Accordingly, the study will
employ a "core set" of laboratory toxicity tests that will be run
on every chemical selected, but will allow some additional
testing which is to be undertaken as described below.
"Core Set* of Testing
The task before EPA in devising a core set of toxicity
testing is to develop the best set of tests that can be assembled
within a given dollar amount. EPA proposed (1984c) and has
adopted $50,000 as the testing budget for the-core set o£ tests
on a per chemical basis. Final recommendations for a $50,000
core set of tests were developed at a working meeting held at the
National Institute of Environmental Health Sciences (NIEHS) in
Research Triangle Park, NC in August 1984. Participants at the
meeting included personnel from EPA, PIEHS, the National
Toxicology Program, and the Chemical Industry Institute of
Toxicology (EPA, 1984d). At the meeting, general agreement was
reached to include in the core set of testing a series of
24
-------
mutagenicity assays which would serve to identify potential
carcinogens, an acute toxicity screen, a 28-day repeated dose
toxicity study, and a dermal sensitization assay.
The decision was reached at the meeting to table discussion
on selection of the short-terra test(s) for developmental
toxicity. The meeting concensus was that none of the available
short-term assays for developmental toxicity have been widely
accepted by the scientific community as being adequately and
appropriately -validated for use as a screening tool. The
participants recommended, however, that the issue of selecting an
adequate screening assay for developmental toxicity be revisited
within the next year* Another aspect of the decision to table
consideration of a developmental toxicity assay is the fact that
the other tests recommended at the meeting already totaled more
than the targeted $50,000 per chemical*
The specific assays recommended at the August meeting are
listed below along with the basis for their selection. The
recommended testing was as follows:
o A series of three short-term mutagenicity screening tests
for identifying potential carcinogens.
Ames SaImone11a/mamma1ian microsome test using 5
strains of Salmonella, both with and without metabolic
activation (EPA, 1983a}« Estimated cost $2,000,
~ in vitrg_ sister chromatid exchange (SCE) assay using
the CHO line of Chinese hamster cells, both with and
without metabolic activation (EPA, 1983b), Estimated
cost; $9,000.
"* in vitro gene mutation test using the L5178Y line of
mouse lymphoma cells, both with and without activation
(EPA, I983c). Estimated cost: $12,000.
25
-------
o • Two general toxicity assays for acute and repeated dose
toxicity in the rat.
acute oral toxicity test (EPA, 1984e). Estimated
cost: $2,100.
- 28-day repeated dose oral toxieity study using a
modified OECD (1981a) protocol. Estimated cost:
$25,000.
o A dermal sensitization assay in the guinea pig*
the Buehler (1965) or "closed patch" test. Estimated
cost: 53,500.
The total cost for the above set of tests is estimated to be
$53,600 on a per chemical basis.
The tests chosen for identification of potential carcinogens
include two tests for gene mutations, one in prokaryotes (Mies
Salmonella/mammalian microsome test) and the second in mammalian
cells in culture (L5178Y mouse lymphoma test), and an in vitro
SCE test. Although the ultimate mechanism of SCE formation is
unknown, the endpoint is visualized as effects on the
chromosome. These three assays were chosen because they are
routinely available, relatively inexpensive, have a large data
base, of tested chemicals, and are performed under standard
protocols allowing for interlaboratory replication of data
(Brusiek and Auletta, 1984). Most importantly these assays
correlate well with in vivo carcinogenicity. Phase II of the
Gene-Tox Program (EPAr 1983d) has found that the Ames assay has
an overall correlation with carcinogenicity of 81% (121/151
carcinogens tested were correctly identified), the L5178Y mouse
lymphoma test has an overall correlation of 90% (18/20
carcinogens correctly identified), and the in vitro SCE assay has
26
-------
an overall correlation of 97% (40/41 carcinogens correctly
identified). The combination of the Mies assay with an assay in
mammalian cells in culture is known to yield increased sensiti-
vity over that of the Ames assay alone. For example, the
International Collaborative Study (de Serres and Ashby, 1981)
found that the use of a mammalian cell assay in addition to an
Ames assay enhanced detection of known carcinogens, so that the
majority of carcinogens not detected in the ftmes assay were
detected in eukaryotic in vitro systems. The mammalian cell
assays examined by de Serres and Ashby included assays for both
gene mutation and effects such as cytogenetics and SCE. Finally,
in conducting the in, vitro SCS assay, the technique used to
prepare cells for SCE analysis is compatible with the way cells
are prepared for chromosomal analysis, thus both SCE and
chromosomal aberrations can often be seen and scored on the same
slide. The combination ..of..the -Mies .Salmonella/mammalian
ntierosome assay, the in jfitro SCE assay, and the in vitro gene
mutation assay in L5178Y mouse lymphoma cells covers a spectrum
of genetic events and enhances the possibility of detecting
potential carcinogenic agents over that which exists with any one
of the tests. It should be remembered that one is dealing here
with sensitivity (the ability to detect known carcinogens) and
that an assessment of specificity (the ability to accurately
assess noncarcinogens) is limited by the paucity of valid
^
negative in vivo, data. The Gene-Tox data base (EPA, 1984f), for
example, lists fewer than 10 chemicals as having been tested with
enough rigor to be classified as noncarcinogens. However, the
assays selected will detect potential _irtr vivo carcinogens and
within these limits they are valuable aids in identifying
potentially hazardous agents.
The utility of determining acute effects of industrial
chemicals to assess potential effects of both routine and
accidental exposure, especially in the workplace, serves as the
basis for inclusion of the rat acute toxicity test (3 dose
groups, 5 animals per sex per dose, 14-day observation period).
27
-------
This test will, in addition, provide information needed to set
the doses for the 28-day repeated dose study. The 28-day
repeated dose oral (gavage) toxicity study in the rat {3 dose
groups plus control, 5 animals per sex per dose) is a screening
test for potential chronic effects other than cancer, mutagenic-
ity, and teratogenicity. The 28-day study is seen as a less
costly but a less detailed alternative to the 90-day subchronic
toxicity test, and, despite its limitations, is expected to
identify most of the organs or systems that will be affected
following repeated exposures over a limited time period. One of
the limitations of the 28-day study is that, unlike the 90-day
subchronic assay, it is not viewed as providing reliable informa-
tion regarding a chronic No Observed Effect Level (NOEL) and thus
will not give an indication of acceptable lifetime exposure
levels for humans* Based on their studies, Weil.and McCollister
(1963) and Weil et_£l_. (1969) have suggested that one may
transfer the results of shorter term animal tests with measured
confidence into a prediction of the "no ill effect" levels for
the corresponding longer term animal studies.
The Buehler sensitization study was selected because it is:
widely used; does not involve intradermal injection (and thus is
more difficult to vitiate by improper technique); does not
require highly trained animal handlers/technicians? requires
fewer animals and is the least expensive of the seven methods
considered acceptable by both EFA (1982) and the OECD (1981b);
and provides data on both the incidence and severity of sensi-
tization reactions.
"Tailored" Testing
In a Status Report presented to the S&B-EHC, EPA (1984c)
noted that it was considering the use of tailored testing in the
retrospective study. The "tailored" tests would be run in addi-
tion to the core set of tests and would be selected on the basis
of the specific chemical or chemical class and the predicted
28
-------
effect(s) of concern. At that time EPA indicated that tailored
testing needs would be identified on a case by case basis. At
the August meeting at NIEHS (SPA, 1984d}f it was recommended that
tailored testing decisions operate at a chemical class or
physical cheroical property level rather than on a case by case
basis. Thus, for example, organophosphates might be identified
as a class for determination of blood cholinesterase levels {Weir
and Hazleton, 1982) as part of the clinical biochemistry in the
28-day repeated dose toxicity study. Glycol ethers, on the other
hand, might be identified as a class for which teratogenicity and
reproductive toxicity testing may be indicated (EPA, 1984g). The
specific applications of tailored testing in the retrospective
study will be developed at a subsequent point in the project.
Operational' Procedures.
Questions remain as to the actual procedures that will be
used to obtain the testing outlined in this report. EP,A recently
approached the National Toxicology Program with a request for
assistance in this project and the preliminary response was
favorable (EPA, 1984i). Laboratory testing will most likely be
performed by laboratories under contract to the government; the
specifics have not been worked out. EPA has an item in its
preliminary budget for fiscal year 1986 to provide support for
the project? some fraction of the funding will likely go to
support laboratory testing, while the balance will be used to
support management of the project and data evaluation.
Evaluation of laboratory studies and comparison of testing
results with EPA's hazard predictions is anticipated to proceed
in the following general fashions
1, Laboratories performing the testing will be responsible
for evaluation of the individual test results (using
29
-------
established criteria provided by EPA; these are yet to
be developed) and for preparation of laboratory
reports. The laboratories will be blind to the chemical
structures and to EPA's hazard assessment documents.
2. A second group (likely a contractor Bunder EPA
supervision) will next evaluate the set of test results
using established criteria provided by EPA and will
prepare a written evaluation of the toxicity of each
tested chemical. This group will be aware of chemical
structures and will interpret the results of the testing
using the knowledge of any correlations known to exist
between the chemical class tested and its response in
the short-term tests used in this study. They will not,
however, have access to EPA assessment documents.
3. A third group (once again, likely an EPA contractor)
will review EPA's PMN hazard assessment documents on
each tested chemical (while being blind to the results
of the testing) and extract conclusions as to the
presence or absence of concerns about health effects for
each tested PMN chemical.
4. Corellations will be derived (as described previously)
between the results of groups (1) and (3) and groups (2)
and (3), above. The results of these analyses will be
used to determine the validity of EPA's SAR-based hazard
predictions as compared with the results of a limited
set of testing,
5. Other-data analyses may also be undertaken.
30
-------
Table 1. Test Data Submitted with Premanufaeture Notices*
Type of Data
Percent of PMNs
Health data (some)
Acute Toxicity
Oral
Dermal
Inhalation
Local Toxicity
Eye irritation
Dermal irritation
Sensitization
Mutagenicity
Other0
Scotoxicological data (some)
Acute lethal vertebrate
Acute lethal invertebrate
Fate data (some)
Biodegradation
Log P
No test data of any type
All
46%
39%
21%
8%
34%
37%
9%
13%
8%
10%
7%
3%
10%
6%
3%
52%
Non- Polymer
58%
51%
26%
10%
45%
481
12%
19%
121
14%
11%
4%
121
9%
6%
39%
Polymer
30%
24%
14%
7%
22%
23%
5%
5%
4%
. ...£%...
4%
2%
5%
3%
1%
68%
%ased on the full set of 3,578 PMNs received to date (6/84).
The "other" health category includes acute toxicity studies by
other routes (ip, iv, etc.), repeated dose toxicity studies
(generally 28 days or less in duration), teratogenicity
studies, phototoxicity, neurotoxicity, and a variety of other
endpoint studies.
(EPA, 1984h)
31
-------
Table 2. Most Frequently Encountered Combinations
of Health Test Data Submitted with PMNsa*°
Specific Testing Combinations
Number of
PMNs
256
177
121d
100
72
70
63
50
34
33
31
28
28
28
26
25
24
21
20
19
19
18
17
Acute Ttoxicity
DHL DRM IHL
X
X X
XXX
XXX
X
X X
X
X
X X
XX X
X
X
X
X X
X
X
X X
X X
X X
X X
Irritation
DIM EYE SENS MUTA
X
X
X
X
X
X
X
X
X
x
X
X
X
X
X
X
X
X
X
X
X
X
XXX
X
X X
X X
X X
X
X
~x x
X , X X
X X
X X
X X
X X
XXX
X
X
XXX
OTHR
. x
X
X
X
X
X
a Based on 1646 IMS cases containing some type of health test data (8/84).
b Abbreviations as follows: QKD-oral? DRM-derroaly IHL-inhalation; EYE-eye?
SENS-sensitization? fCJTA-rautagenicityf CflHR-other health studies.
c Approximately 400 cases contain 94 other combinations of health test data*
d includes 106 synfuels received as a group,
(EPA, 1984h)
32
-------
fable 3. The Number of PMNs Containing 1 Through n
Different Categories of Test Dataa
Number of Test
Data Categories
Submitted"
Number of '•
PMNs 151 109 383! 370 333 126 53 121*
a Based on 1646 PUN cases containing some type of health test data (8/84).
" Test data categories includes acute lethal oral, acute lethal dermal, acute lethal
inhalation, eye irritation, local skin irritation, sensitization, mutagenicity,
and other. This table does not reflect the possible submission of multiple tests
in a single category (for example, a notice containing a single rautagenicity study
is not distinguished from other notices containing more than one such study).
c includes 106 synfuels received as a group.
{EPA, 1984h)
33
-------
fable 4. Ccnpaclson of Available Wealth Taat Data tor TSCft
Enisling Ctenleula versus BM Chemicals
Sample Size"
TEST TWE
Acute*
Subctwonie*
Chronic f
Repro/Develop."
Hutagenicity
Minimal "toxic Ity
Information'
Ho Hoxlclty
Inforwatlcti
i 106 Ib/vr
259
20(15-25)
10(7-14)
6(3-9)
9(6-13)
22(18-26)
78173-841
Existing Chanigalaa
flepocted production Waltme
< 10* U>/yr
PaCEOT MMH
22O5-29J
a (s-131
K2-6J
10(5-15)
24118-30)
76(69-83)
m» Chemicals'*
Nbt
Available
PRSSCHIBED "HSf/-;
L5tV-2H
113-111
3CO-6!
7(3-12)
18(12-24)
82(76-89 J
Ml
Cases,
3989
mfcasvtemP
40
11
< I
< 1
42
-S*
tton-
2J51
51
16
< 1
< 1
19
54
42
Polymers
1634
24
S
< 1
< 1
S
25
70
* rtguees taken fnon MBC (1984, p. 84).
** FlqutiBS taken from EPA (1984h).
c par enlstlng ctienlcals, the oanpla elze reports the nunber of cbanicalfl from the full TSCA inventory
of 4&,523 chemicals which underwent * standardised Bcreetilng procedure to identify available bonclelty
Information {see HFC, 19B4, p. 45-50), Fbr R*f chemicals, the nurtjer roprasenta the count as-of July 1984.
^ exist In? chemloal eatimbeB ate fxpmtsaeS at imana with upper antf lower 901 oonfldenoe limits. Por »W ctanicala
the values reported Include test data submitted with the notice and thus do not reflect arty testing that Is
subsequently undertaken,
* JK»t* toxleitf italtea ans defined as alngle administration via any twite within 21 hours, subdiwnlc
toxicity studies Include 20- and M-day repeated dose studies (oral, c5errral) and guinea pig senaitiKation
tests (frora «ffC, 1984, p. 47}.
' SfipioductiwB/aevelopiBntal tauicSty. ,
" we (1984, p. 47) defined "minimal toxlcity Infarmation" foe T9C& existing chewlcaia as the availability
of any one or note of the S atu% types included In this table,
h For iSCft existing chemicals, MRC U984I noted that *Mltional information might e*l»t In teslaricted eocene
fllen.
34
-------
Figure 1. Annual Receipt of PMNs from FY79 Through F¥84a
1400
1200
1000
800
600
400
200
(11)
FY79
a Fiscal Year
(EPA, 1984h)
(580)
(281)
1
(839*
(1301)
1
PY80
FY81
FY82
FY83
(1200)
est.
1
FY84
35
-------
Figure 2. Statistics on Beceipt of tfotices of Gmmencement of
Manufacture or Import (HCC)a for PMNs Received
Between FY79 and EY83
100
80
% Of SWJs
having an
NOC
60"
40"
20"
#i#
##t
###
a oo
OO«
oiiO
///
///
* * *
* * #
* » p
* » »
* » *
* * #
59%
HI
tit
i«
aoo
9OQ
///
///
///
# » *
* 4 *
# * »
* * *
* * *
« » *
* * *
61%
44*.
it*
ooo
f>oa
///
///
///
* * *
* * *
• * *
* * t
* # *
• » *
* 4 •
56%
Iff
ooo
///
* * *
* * *
* » *
* * «
V * *
* » *
39%
f*t
ooo
eoo
049
///
//L
* * *
* • *
* * *
* # *
* * P
* * *
• « *
FY79-8Q FY81 F¥82 fY83 Ht>tal
(292) (580) (839) (1301) (3013
50%
Year of EMN SuhnissJbn
("total Number of EMN Cases)
Leg^id: Days after day 90 that the Notice of Oanmencement is
###
7/7
days > 365
days 181-365
days 91-180
days 1-90
a see explanation of an IOC in the text,
Ti - Fiscal Year
(EPA, 1984h)
36
-------
piguse 3, : TON Assessment" Prooeea
Day 10
Paperwork
to ConCirra
Conpletcnegs
1
Inerarpli
Return to
Submitter
X
a be
Jf
CRSS
{Chemistry:
I
Day 11
SW
(Hazard)
1
4,
Day 14
-mm
(Exposure)
j
^U Day 15 Days 16-90 Day 72-80
FDCUB Detailed Review Disposition Uireasonable Control action
* IRisJc) ^ (Exposure > fi&etirvf Risk ^ Dnfler
* and ' (RiskJ fan SSe or 5f
,, Kazani)
/* LW 1
/ risk 1 tow
/ 1 risk
\ ' *•*
Drop Case Drop Case
f cutt • d^ffi
Review Review
Legendj CRSS - Chemistry Review and Search Strategy
SW - Structure activity "team
EKRM - Exposure Analysis Meeting
37
-**:
-------
Figure 4, Molecular Weight Distribution (in daltons) for
Nonpolymeric Chemicals (EPA, 1984h)
400 - -
350
300
250 -
Number of
PMN Cases
(1979 -
1984) 200 -
ISO
100
50
•i
t
n = 1639
range - 104-4660
mear
1 = appro x
, 450
•
•d
_<100 101- 201- 301- 401- 501- 601- 701- 801- >9Q0
200 300 400 500 600 700 800 900
38
-------
FOOTNOTES
extent to which testing obtained under this project indicates
that certain health hazard determinations were not appropriate does
not in isolation indicate that the overall risk decision made for
each such case was wrong. Risk decisions must consider both hazard
(i.e., toxicity) potential and exposure potential. An analysis to
determine the effect of additional test data on historical risk
decisions is beyond the scope of this project.
2&s noted, one of the major objectives of this study is to determine
the extent to which EPA's predictions of toxicity would have changed
if the results of certain tests were included in each PMN.
Information obtained via this project might be useful in analyses to
study whether the submission of test data should be required under
section 5 of TSCA. In the course of such analyses the benefits of
the information gained by such testing must be weighed against their
costs and the impact of these costs on the introduction of
innovative PMN chemicals,. For example, while long term bioassays
directed at multiple endpoints would provide much greater certainty
concerning the toxicity of new chemicals, the cost to industry of
requiring such testing on all new chemicals militates against such a
requirement. For this reason; ...... the- study design presented in this
paper focuses on test methods which are relatively inexpensive in
comparison to long term bioassay tests.
^pesticides, drugs, foods, food additives, cosmetics, and certain
other chemicals which are controlled by other statutes are not
within the purview of TSCA (see 15 U.S.C. 2602(2)}.
4The TSCA "preroanufacture" reporting requirements can be contrasted
with the European Communities (EC) "premarket ing" notification
requirements {EC Directive 79/831/EIC, 6th Amendment, 1979-) . As the
terms indicate, premanufacture notification under TSCA is required
at an earlier point in the development of a chemical than is the
case for the EC's premarket notification procedure. Many of the
information reporting requirements under the EC's directive, are
similar to those outlined earlier in TSCA with the major difference
being that the EC Directive (1979) requires as a mandatory part of
the premarketing notification a specified "base set* of health,
environmental, and physical chemical test data. A minimum set of
test data is thus available on premarket notification (EC)
chemicals, whereas the hazard assessment of TSCA premanufacture
chemicals often starts out with fewer or no data,
39
-------
^A "consent" 5(e) order is one in which in which EPA negotiates the
terras of the order with the company that submitted the PMN. The
company agrees to be bound by the order and waives its rights to
file objections to the order. This waiver does not affect any other
rights that the company may have under TSCA, The company at a later
date can request a modification of the consent order. This is
contrasted with a unilateral order under which EPA takes action to
restrict or prohibit the manufacture or use in commerce of the PMN
substance.
"Another option which is available at Focus (as well as subsequent
points in the assessment process) is to enter the chemical into the
"Followup" program which would consider the need for promulgation of
a "significant new use rule" under TSCA section 5(a)(2) or an
information reporting rule under section 8(a). Chemicals controlled
under section 5(a)(2) are subject to the same 90-day notification
requirement as for new chemicals, except that the need to report is
triggered by the development of one or more significant new uses not
outlined in the original premanufacture notification submitted to
EPA. These new uses are then examined to determine if they "may" or
"will" present an unreasonable risk under section 5. Section 8(a)
reporting rules, on the other hand, only require the submission of
.cejr.ta.in. information and do not trigger...automatic review under
section 5*
"*
'It should be noted that the approaches employed in EPA's
applications of SAR differ to a substantial degree from the
approaches which are typically associated with the use of
quantitative structure activity relationships (QSARs) and other
computational techniques to predict the biological chemistry of .
xenobiotic substances. Examples of these approaches are Hansch
analysis, molecular and quantum mechanicst pattern recognition, and
so on (see Golberg (1983) for a general treatment of these
topics). The chief constraint on the use of these approaches by the
EPA at present is that the problem confronting SPA differs from the
earlier formal applications of these techniques (for example, in tlte
pharmaceutical industry) in two important ways. First, the desired
goal (prediction of adverse health effects) is not a single endpoint
but depends on multiple interactive paths and the mechanisms
involved are often not well understood. Second, there is little
experimental data available or easily obtainable not only relative
to health effects of the PMN chemicals but also relative to all but
the simplest physical chemical properties. Further, the chemicals
confronting EPA are often not the simple extension of a family of
similar chemicals where a great deal of data exist for other family
members (as is often the case in pharmaceutical applications of
these techniques).
Examples of available substructure and nomenclature search systems
include SANSS (Structure and Nomenclature Search System) in the
NIH/EPA Chemical Information System, CAS-QNLINE available from
Chemical Abstracts Service, and DARC, a French system available from
Questel.
40
-------
n
A "functionally similar" chemical is one which, although it differs
substructurally , can be considered a functional equivalent of the
PMN chemical. Examples of functional equivalents, depending on the
specifics of the case, might include acceptance of an aromatic amine
substituent in lieu of an aromatic nitro {based on the expectation
o£ bio trans format ion) or a chloro (but generally not a fluoro) in
lieu of a bromo substituent (based on the concepts of isosterismj
see Burger, 1970 and Mathison _etijl_. , 1976),
•*• Definitions of terms used in this section are as follows;
PMN universe - the full set of PMN chemicals submitted to EPA under
TSCA section 5j sampling universe - the subset of the PMN universe
remaining after certain practical exclusions have been made;
sample - the one hundred (100) PMN chemicals selected from the
sampling universe (specimens of these chemicals will undergo the
laboratory testing described in the paper); specimen - physical
sample of a PMN chemical that will be obtained from the manufacturer
for use in testing,
(1984) in Appexdix E of its report describes what it terms "a
plausible approach — based on economic theory — to the assignment
of costs to errors in classification" of chemical toxicity (e.g. ,
identifying toxic chemicals as being nontoxic) . The reader is
referred to the cited document for details of the analysis
undertaken. In brief, the NRC concluded from its analysis that "the
social cost of underregulating a chemical is much greater than that
of overregulation. "
i215 U.S.C. 2604(h)(3) states that for chemicals not listed on the
TSCA inventory small quantities can be manufactured or processed
solely for purposes of scientific experimentation or chemical
research. '
41
-------
BIBLIOGRAPHY
Albert A, 1983. tfAR Program Evaluation Report, Final Report from
Work Assignment No. 17, USEPA Contract No, 68-01-6554.
/
Albert A. 1979. Selectlve_Toxicity; The Physieg—Chemica1 Basis_o£
Therajay;, 6th edition. Chapman and Hall, London.
American Chemical Society. 1984, Conference on Structure Activity
Relationships and Toxicity Assessment, June 6-8, 1984, Gaithersburg,
MD. Panel discussion on Thursday afternoon, June 7, 1984.
Arcos JC. 1983. Comparative requirements for premarkettng/
premanufacture notifications in the EC countries and the USA, with
special reference to risk assessment in the framework of the U.S.
Toxic Substances Control Act (TSCA). J. Am. Coll, Toxicol. 2(1},
131-145,
Arcos JC and Auer CM. In Press. Scientific basis and procedural
implementation of the assessment of premanufacture notifications
within the U.S. Toxic Substances Control Act. Commission of
European Communities Workshop (1982) on "Decision-making process in
the evaluation of new chemicals in the EC and the USA." Published
for the CEC by the Istituto Superiore della Sanita, Rome.
Banerjee S, Yalkowsky SH, Valvani SC. 1980, Water solubility and
octanol/water partition coefficients of organics. Limitations of
the solubility-partition coefficient correlations. Env. Sci« Tech.
14*1227-1229.
Brusick D and Aaletta A, 1984. Developmental status of assays for
genotoxicity, A report of Phase II of U.S. EPA's Gene-Tox
Program. Mutat, Res, In Preparation,
42
-------
Buehler EV. 1965* Delayed contact hypersensitivity in the guinea
pig. Arch. Dermatol* 91:171.
Burger A. 1970. Relation of chemical structure and biological
activity, in Medicinal Chemis.trrfc, A. Burger, ed., 3rd edition.
Wiley-Interscience, New York.
Chou JT and Jurs PC. 1979. Computer assisted computation of
partition coefficients from molecular structures using fragment
constants. J. Chem. Inf. Comput. Sci. 19:172-178.
de Serres F and Ashby J. 1981. Evaluation of short-term tests for
carcinogens. Report of the International Collaborative Program.
Elsevier/North Holland. New York, Amsterdam, Oxford.
Eakin DR and Johnson DQ.. 1981,. SPHERES Scientific Parameters for
Health and the Environment, Retrieval and Estimation. Requirements
Analysis and Examination of Alternatives. Final Report from Work
Assignment No. 41, USEPA Contract No. 68-01-4795.
EPA. 1982. New and revised health effects guidelines. HG-
Qrgan/Tissue-Derraal Sensit. Office of Pesticides and Toxic
Substances, USEPA. EPA 560/6-83-001. PB 83-257691.
Ibid. 1983a. HG-Huta-5. typhimuriam.
Ibid. 1983b. HG-Chromo-Sister-Chrom-In vitro.
Ibid. 1983c. HG-Gene Muta-Somatic cells,
EPA. 1983d. Gene-Tox computer printout, retrieved May 1983.
Office of Pesticides and Toxic Substances, OSEPA.
43
-------
EPA* 1983e. New and revised health effects guidelines. HG-
Neurotoxieity-Functional Observational Battery, pp* 4-5. Office of
Pesticides and Toxic Substances, USEPA, EPA 560/6-83-001. PB 83-
257691.
Ibid, 1983f, HG-Neurotoxicity-Neuropathology, pp. 3-6.
EPA. 1983g. Experimental test design for new chemical review
program. Final report from Work Assignment 3-10, USEPA Contract No.
68-01-6287.
EPA. 1984a. Monthly PMN Summary Report, August 1984. Office of
Pesticides and Toxic Substances, USEPA.
EPA. 1984b, Design Options for a Retrospective Validation Study of
PMN—Health-Hazard Assessments.- Submitted to the Environmental
Health Committee, Science Advisory Board, EPA, in May 1984. Office
of Pesticides and Toxic Substances, USEPA. (The reader is also
referred to a previous report (EPA, 1983g) which discusses possible
designs for a retrospective validation study.)
EPA. 1984c. Status Report on "Design Options for a Retrospective
Validation Study of PMN Health Hazard Predictions." Presented to
the Environmental Health Committee, Science Advisory Board, EPA on
July 24, 1984. Office of Pesticides and Toxic Substances, OSEPA.
SPA. 1984<3» Meeting to discuss EPA project to validate PMN
toxicity predictions with emphasis on "selection of tests" issue,
held at the National institute of Environmental Health Sciences,
Research Triangle Park, NC, on August 13, 1984. Participants
included personnel from EPA/Office of Pesticides and Toxic
Substances (J. Moore, W. Parland, W. Waugh, D. Seal, A. Auletta, C.
Auer, J. Carra), NIEHS (B. Schwetz, R. Tennant)» National Toxicology
Program (E« McConnell, J. Huff), EPA/Office of Research and
Development (R. Cortesi, M. Waters, R. Kavlock, J. Rabinowitz), and
Chemical Industry Institute of Toxicology (J. Gibson). Office of
Pesticides and Toxic Substances, USEPA.
44
-------
EPA. 1984e. New and revised health effects guidelines. HG-Acute-
oral. Office of Pesticides and Toxic Substances, USEPA, EPA 560/
6-83-001. PB 83-257691.
EPA, 1984f. GENE-Tox computer printout, retrieved September
1984. Office of Pesticides and Toxic Substances, OSEPA.
EPA. 1984g. Risk assessment o£ 2-methQxyethanol, 2-ethoxyethanol,
and their acetates. Office of Pesticides and Toxic Substances,
USEPA.
EPA. 1984h. Data printouts retrieved from the Technical Data
Indexing System (TDIS), the EPA in-house PMN information system.
Office of Pesticides and Toxic Substances, USEPA.
EPA. 19841, ..Personal., comraunciation from Dr.., JQhn,_AJ, ..Moore,..
Assistant Administrator, Office of Pesticides and Toxic Substances,
OSEPA.
European Communities Directive 79/831/EEC, as amended by a Council
Directive for the sixth time. 1979. Annex VII, Information
required for the Technical Dossier ("Base Set") referred to in
Article 6(1) of the Directive, specifies the following set of
testing requirements* Physical Chemical Properties! melting point,
boiling point, relative density, vapor pressure, surface tension,
water solubility, fat solubility, partition coefficient, flash
point, flammability, explosive properties, auto-flarnmability, and
oxidizing properties? lexicological Studies: oral LD50, inhalation
LD50, dermal LD50 (route(s) for administration will be selected
based on physical chemical properties and intended use), skin
irritation, eye irritation, skin sensitization, 28-day sub-acute
toxicity study (dosed 5 to 7 days per week for at least 4 weeks with
route selected on basis of intended use, acute toxicity, and
physical/chemical properties), and mutagenicity (2 tests, one
bacteriological, with and without metabolic activation, and one non-
bacteriological)? Ecotoxicological Studies: acute LD50 (fish and
daphnia) and degradation (biotic and abiotic).
45
-------
Federal Hazardous Substances Act (FHSA), Title 16, Section 1500 of
the Code of Federal Regulations,
Golberg L. , ed. 1983. Stxjjct.ar e^ _Act ivi _ty_ _Cg r re lat ion_ _as_ a_
F u ng am_e_n t a 1 s_> Mg t h od, _ _ a n d_
Hemisphere Publishing Corporation. Washington B.C.
Government Accounting Office (GAG). 1984. Assessment of New
Chemical Regulation Under the Toxic Substances Control Act.
Prepared in response to a request from Senate Environment and Public
Works* Subcommittee on Toxic Substances and Environmental Oversight
and the House Energy and Commerce's Subcommittee on Commerce,
Transportation, and Tourism* GAO/RCED-84-84.
Hansch C and Leo A. 1979. Su^^'itjieji,^C^^^^^^^r^^rw^^OLn
_An a l^si s_jLnr_ghein j g t_ry a nd_ Jj io^lgj^, John Wiley and Sons, New York,.
Hermans J» 1983, The use of QSAR in toxicity studies with aquatic
organisms. Correlation of Toxicity of Different Classes of Organic
Chemicals with Poet, pKa and Chemical Reactivity. Proceedings
Fourth European Symposium on Chemical Structure-Biological Activity
Quantitative Approaches, Bath, England, September 6-9, 1982, In
JC Dearden, ed. Elsevier,
New York, pp. 263-264.
Hutzinger 0 j^t_a^, 1978, Chemicals with pollution potential, in
A^uatic^PQll_utants_« Pergamon Press, Hew York, pp. 13-31,
Howard PH, Sage GW, LaMacchia A, Colb A. 1982, The development of
an environmental fate data base, J. Chem, inf. Comput. Sci. 22:38-
44.
Konemann H. 1981. Quantitative structure-activity relationships in
fish toxicity studies. Toxicology 19:209-221.
46
-------
Lipnick RL and Dunn WJ III. 1983. An MLAB study of aquatic
structure toxicity relationships. Proceedings Fourth European
Symposium on Chemical Structure Biological Activity Quantitative
Approaches, Bath, England, September 6-9, 1982. In Quantitative
Approaches to Drug Design, JC Dearden, ed* Elsevier, New York, pp.
265-266.
Lyman WJ, Reehle WF, Rosenblatt DH. 1982. Handbook .of Chemical
Property Estimation Methods. McGraw Hill, New York.
Magnuson VR, Harriss DK, Fulton MS, Anderson EG, 1981. ISHOW,
Information System for Hazardous Organics in Water. 182nd Natl.
Meet., Aug. 23-28, Amer. Chen. Soc. , Paper Mo. CINE 22.
Mathison IW, Solomons WE, Morgan PH, Tidwell RR. 1976. structural
.features and, pharnaacologic activity, in Principles of Medicinal
.7 W*Q» Foye, ed. Lea and Febicjer, Philadelphia, PA,
Milne GWA, Heller SR. 1980, NIH/EPA Chemical Information System.
J. Ghent. Inf. Comput. Sci. 20:204-211.
national Research Council (NRC). 1984. Toxicity Testing;
Strategies to Determine Needs and Priorities. National Academy
Press, Washington, DC. 382 pp.
v>
Office of Technology Assessment (OTA). 1983. The Information
Content of Premanufacture notices — Background Paper. Prepared in
response to a request from the Subcommittee on Commerce,
Transportation, and Tourism of the House Committee on Energy and
Tourism. OTA-BP-H-17.
Organization for Economic Cooperation and Development (OECD).
1981a* OECD Guideline for Testing of Chemicals. "Repeated Dose
Oral foxicity - Rodent: 28-day or 14-day study," adopted May 12,
1981, modified as follows: (1) Section 2.B. Experimental
animals. The rat will be used. (2) Section 2.B. Test
conditions. The "Limit Test" will not be used? the observation
47
-------
period will be for 28 days. (3) Section 2.B. Procedure. Animals
will be dosed 7 days per week. (4) Section 2.8. Clinical
examinations, (a) Hematology will include only hematocrit and
hemoglobin concentration. (b) Clinical biochemistry will be done
based on target organs defined by the acute study or by EPA's
assessment of the PMN chemical's toxicity. (c) Urinalysis- will not
be done. Also follow recommendations made in HG-Neurotoxicity-
Functional Observational Batteryr (EPA, 1983e). (5) Section 2.B.
Pathology. Gross necropsy to include weighing of ovaries in
females? the organs taken for histopathologic examination will be
extended to include the testes/ovaries and lungs. As part of the
histopathology, follow recommendations made in HG-Neurotoxicity-
Neuropathology (EPA, 1983f), but limited to examination of 6 animals
in the highest dose group with observations followed to lower doses
Ibid. 1981b. Skin-Sensitization. -
Page N. 1983. NLM's toxicity information pcogram helps in
nationwide cleanup of hazardous wastes. NIH Record. November 22,
1983, p. 11.
page N and Kissraan H, 1984. Qn-Line Sources for Toxicity and
Safety Data. Proceedings of the Annual Conference on Hazardous
Wastes and Environemntal Emergencies. Houston, TX, March 12-14,
1984, pp.409-412.
perrin DD» 1965. Dissociation Constants of OrganicBasesin
AqueousSolution, IUPAC Chemical Data Series. Buttersworth,
London.
perrin DD. 1912. Dissociation Constants of Organic Bases in
Aqueous Solution. IUPAC Chemical Data Series: Supplement 1972.
Buttersworth, London.
Public Law 94-469, 90 Stat. 2003, October llr 1976.
48
-------
Science Advisory Board, Environmental Health Committee (SAB-EHC),
EPA. 1984, Letter plus attachments to Dr. John A. Moore, dated
July 19, 1984.
Serjeant EP and Dempsey B, 1979. Ionizat ion _Constant s of' Organ1 c
Acids in Ague_Qus_ Solution. IUPAC Chemical Data Series. Pergamon
Press, New York.
Veith GD, De Foe DL, Bergstedt BV. 1979. Measuring and estimating
the bioconcentration factor of chemicals in fish. J. Fish Res. Board
(Canada). 36:1040-1049.
Veith GD, Call DJ, Brooke LT. 1983. Structure-toxicity
relationships for the fathead minnow, Mmephales gromelas; Narcotic
Industrial Chemicals, Can. J. Aquat. Sci. 40*743-748.
Weil CS and McCollister DD. 1963. Relationship between short- and
long-term feeding studies, in designing an effective toxicity test.
J. Agr. Food Chera. 11:486-491.
Weil CS, Woodside MD, Bernard JR and Carpenter CP. 1969,
Relationship between single-peroral, one-week, and ninety-day rat
feeding studies. Toxicol. Appl. Pharmacol. 14:426-431.
Weir RJ and Hazleton LW. 1982. Organic Phosphates, in Patty_'s
Indagtrial^Hygiene jandviiToxicology (Vol. 2C), GD Clayton and FE
Clayton, eds., third revised edition. Wiley-Interscienee, New York.
IS U.S. Code, Section 2602(2), 1976.
15 U.S. Code, Section 2604(d), 1976.
15 U.S. Code, Section 2604(e), 1976.
15 U.S. Code, Section 2604(f), 1976.
15 U.S. Code, Section 2604(h)(3), 1976.
... 49
------- |