EPA/630/R-92/003
September 1992
REPORT OFTHE
NEUROTOXICITY RISK ASSESSMENT GUIDELINES
PEER REVIEW WORKSHOP
June 2-3,1992
Washington, DC
Prepared by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02173
EPA Contractor
RISK ASSESSMENT FORUM
U.S. Environmental Protection Agency
FINAL REPORT
August 24, 1992
Printed on Recycled Paper
-------
-------
-. ;'; '.-'.; -'" .. CONTENTS /.-.- . '-'. ';.."-. -, r ..- ;" ": .'. ' '; ;
NOTICE '..-. ....... . . .. . ........ . : .. . ... ... ................ . . v
PREFACE . . . .V, . ....,;.... ..:... ..... ...... V.. ....... . ... ....... vi
1. INTRODUCTION .......... ..V. ........... 1-1
1.1 Risk Assessment Guidelines Program .... ..... . ... ..... .... ... 1-1
1.2 Neurotoxicity Guidelines Peer Review Workshop . . ... . ... .... .... ... 1-1
1.3 , Organization of this Report ........ ............ .......-., ......... . 1-2
2. ISSUES PAPER .... ......-..................................... ... 2-1
2.1 Panel 1: Neurotoxicity as an Appropriate Endpoint
for Environmental Risk Assessment ........................... .v. 2-1
2.2 Panel 2: Interpretation of Neurotoxicity Data
When Effects Are Transient .. ..^...........,..............2-2
2.3 Panel 3:' Agents Acting Through Indirect, as Well
as Direct, Means Can Be Considered Neurotoxic .......... i . ., . . . . . . 2-2
2.4 Panel 4: Extrapolation of Neurotoxicity Data
from Laboratory Animals to Humans ....................... ....... 2-3
2.5 Panel 5: Interpretation of Behavioral Data .... .. .. . ....... . ....... . 2-4
3. CHAIRPERSON'S SUMMARY OF THE WORKSHOP
William F. Greenlee ............................................... 3-1
3.1 Workshop Overview ... . . ... ... ... . . . . ... .......... .. ..... ... 3-1
3.2 Neurotoxicity as an Endpoint . .'.... ... 3-3
3.3 Other Concerns ...V....... ,....../ '3-4
3.4 Risk Characterization ............. -... . . . . . ... . . ... . . . 3-6
4. SUMMARY OF OPENING PRESENTATIONS . . . . . . . . ... .... . . ...... 4-1
4.1 Introductory Comments .................... . . ................. 4-1
4.2 Welcome and Objectives ........ ..... ......................... 4-1
4.3 Neurotoxicity as an Endpoint (Panel 1) .. . . ..... . .. . ...... . . . . . .... 4-4
4.4 Transient and Persistent Effects (Panel 2) .......... ............ 4-7
4.5 Direct and Indirect Effects (Panel 3) . . . . .. . . . ...-.._. ....... ..... 4-13
4.6 Animal-Human Extrapolation (Panel 4) ............. . . . . 4-17
4.7 Behavior (Panel 5) ..... . . ....... ....... ......... .... .4-22
-in-
-------
5. WORKGROUP REPORTS 5-1
5.1 Transient and Persistent Effects Panel John O'Donoghue 5-1
5.2 Direct and Indirect Effects Panel Barry Wilson 5-8
5.3 Animal-Human Extrapolation Panel Shayne C. Gad 5-14
5.4 Behavior Panel John L. Orr 5-18
APPENDIX A AGENDA A-l
APPENDIX B LISTS OF PARTICIPANTS AND OBSERVERS B-l
APPENDIX C PREMEET1NG COMMENTS C-l
APPENDIX D POSTMEET1NG COMMENTS D-l
-IV-
-------
NOTICE
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. ' -
This workshop was organized by Eastern Research Group, Inc., Lexington, ;
Massachusetts, for the EPA Risk Assessment Forum. ERG also assembled and produced this
workshop report. Sections from individual contributors were edited somewhat for clarity, but
contributors were not asked to follow a single format. Relevant portions were reviewed by each
workshop chairperson and speaker. Their time and contributions are gratefully acknowledged.
The views presented are those of each contributor, not the U.S. Environmental Protection
Agency.; ' ;..' . " -y" ; '-- ."'-. ,- ;'"-- ,:-..:..'.;,',.' "v .,= '." :
-v-
-------
PREFACE
On June 2 and 3, 1992, the U.S. Environmental Protection Agency's (EPA's) Risk
Assessment Forum sponsored a workshop for peer review of draft EPA guidelines for
neurotoxicity risk assessment. The meeting was held in Washington, DC, and was chaired by
Dr. William Greenlee of Purdue University (57 Federal Register 21086, 18 May 1992).
Participants from academia, industry, and state and federal government brought expertise from
a wide range of disciplines to the discussion. Members of the public and EPA scientific staff
attended the workshop as observers. The Agency is using the peer review comments to help
complete a proposal for neurotoxicity risk assessment guidelines that will be published for public
comment and reviewed by EPA's Science Advisory Board during the coming year. This
**
workshop report presents information on issues discussed at the workshop, identifies
participants, and summarizes workgroup conclusions.
-VI-
-------
.".." :. .:.",..... -..'-1. INTRODUCTION ' . : .''"'-:
1.1 RISK ASSESSMENT GUIDELINES PROGRAM
In its 1983 book Risk Assessment in the Federal Government: Managing Jhe Process, the
National Research Council recommended that federal regulatory agencies establish "inference
guidelines" to promote consistency and technical quality in risk assessment, and to ensure that
the risk assessment process is maintained as a scientific effort separate from risk management.
A task force within the U.S. Environmental Protection Agency (EPA) accepted that
recommendation and EPA embarked on a long-term program to develop such guidelines. The
first guidelines were published 3 years later (51 Federal Register 33992-34054, 24 September
1986); two of the 1986 guidelines were recently revised (56 Federal Register 63798-63826, 5
December 1991; 51 Federal Register 22888-22938, 29 May 1992). Currently, six other guidelines
are in various stages of development and review.
1.2 NEUROTOXKTTY GUIDELINES PEER REVIEW WORKSHOP
An EPA work group, chaired by William Sette and Suzanne McMaster, prepared draft
neurotoxicity guidelines for peer review. The purpose of the guidelines is to describe the
principles, concepts, and procedures that EPA will follow in evaluating data on potential
neurotoxicity associated with exposure to environmental toxicants. Like EPA's other risk ,
assessment guidelines, the draft neurotoxicity guidelines are organized around the National
Research Council's paradigm for risk assessment.
On June 2 and 3, the EPA sponsored a workshop to peer review the draft guidelines.
The meeting opened with discussion of key features of the draft guidelines, including areas of
expected controversy, followed by workshop review of the scientific foundation for each element
in the guidelines. Workshop participants from academia, industry, and government (state and
federal) brought expertise in a wide range of relevant disciplines to the discussion.
The workshop did not attempt to address all of the principles, concepts, and methods
that are important for neurotoxicity risk assessment. Rather, EPA asked for expert opinion on
the logic, scientific validity, and utility of the principles proposed in the workshop draft as
.1-1-
-------
general guidance for EPA risk assessors. Workshop participants were asked to review the draft
guidance with these objectives in mind. To help focus the review, EPA distributed an issues
paper highlighting issues raised during EPA reviews of earlier drafts.
1.3 ORGANIZATION OF THIS REPORT
This workshop report presents the issues paper (Section 2); the overall workshop
summary prepared by the chairman, Dr. William Greenlee (Section 3); a summary of the
opening presentations (Section 4); and reports of the four workgroups (Section 5), including
their conclusions and recommendations to EPA regarding the draft guidelines for neurotoxicity
risk assessment. The workshop agenda, list of panel members, participants and observers, and
premeeting and postmeeting comments are provided in Appendices.
1-2
-------
2. ISSUES PAPER: PEER REVIEW WORKSHOP ^
DRAFT EPA GUIDELINES FOR NEUROTOXICITY RISK ASSESSMENT
The purpose of this workshop is twofold: (1) to develop expert information and opinions
on the risk assessment guidance presented in the draft guidelines, "Proposed Guidelines for
Neurotoxiciry Risk Assessment," and (2) to ascertain scientific consensus among workshop
participants on principles and-methods proposed as guidance for EPA Neurotoxicity Risk
Assessment. Each Peer Review Panel should examine the conclusions, suppositions, and
limitations stated below for their consistency with available data arid applicable scientific
principles. ':.'
2.1 PANEL 1: NEUROTOXICITY AS AN APPROPRIATE ENDPOINT FOR
ENVIRONMENTAL RISK ASSESSMENT --'-"V' ' \ " ' ' -:" '"-
The workshop draft concludes that neurotoxicity is an appropriate endpoint for
environmental risk assessment.
Conclusions and Suppositions Used in Reaching This Position
Proper functioning of the nervous system is an essential element of health.
A wide variety of agents is known to cause neurotoxicity.
Human exposure to neurotoxic agents may be significant.
Areas of Special Focus ;, ;,
Inadequate toxicological information is available for a vast majority of chemicals.
Standards for evaluating neurotoxic potential have not been firmly established.
2-1
-------
2.2 PANEL 2: INTERPRETATION OF NEUROTOXICITY DATA WHEN EFFECTS ARE
TRANSIENT
The Workshop draft concludes that both reversible and irreversible effects of chemicals
on the nervous system should be considered adverse.
Conclusions and Suppositions Used in Reaching This Position
The nervous system contains billions of cells wired in complex patterns and is
known to be resilient to environmental and toxicological insult by a process
known as compensation or adaptation.
Once damaged, nerve cells have limited capacity for regeneration.
Apparent recovery actually represents activation of reserve capacity, decreasing
remaining potential adaptability.
Areas of Special Focus
Traditionally, effects of toxicants are considered to be persistent or long lasting,
while pharmacological effects are considered to be transient or short-acting.
An effect that appears to be transient in an unchallenged organism may be
revealed as long lasting through an environmental or pharmacological challenge.
It is not known whether transient effects observed following developmental
exposures should be evaluated at specific points in the life span.
2.3 PANEL 3: AGENTS ACTING THROUGH INDIRECT, AS WELL AS DIRECT, MEANS
CAN BE CONSIDERED NEUROTOXIC
The Workshop draft concludes that chemicals may produce neurotoxic effects by direct
and by indirect means.
Conclusions and Suppositions Used in Reaching This Position
Distinctions between direct and indirect action on the nervous system are the
2-2
-------
same as those sometimes referred to as primary and secondary effects,
respectively. v
Agents such as glutamate damage specific neurons through direct stimulation of
receptors, whereas agents such as carbon monoxide act indirectly to kill neurons
by decreasing oxygen availability.
Effects on endpoints of neurotoxicity produced through direct or indirect means
are functionally equivalent.
It is logically inconsistent to say that a compound produces neurotoxicity but is
not a neurotoxicant. , .
Any compound delivered in a high enough dose will be lethal, and, by definition,
neurotoxic. :
Areas of Special Focus
The kind of information available to risk assessors rarely permits a firm
determination of primary versus secondary sites of action.
Compounds damaging the liver, or producing diabetes, may produce nervous
system damage as a secondary consequence of the primary damage.
2.4 PANEL 4: EXTRAPOLATION OF NEUROTOXICITY DATA FROM LABORATORY
ANIMALS TO HUMANS
. The workshop draft notes that EPA must frequently make risk assessment judgments
regarding the potential neurotoxicity of a substance for which the human data base is absent or
inadequate. The draft concludes that with an adequate animal data base, as defined in the draft
guidelines, such judgments may be scientifically valid.
Conclusions and Suppositions Used in Reaching This Position
Substances producing neurotoxicity in humans also result in neurotoxicity in
other species. -< r ~
Compared with human studies, animal studies are more often available and
provide more precise dose information and better control for environmental
factors.
2-3
-------
Many diagnostic procedures employed to evaluate neurotoxicity in humans have
corresponding animal models.
The range of uncertainty factors used to extrapolate from animal data to human
risk for other endpoints of toxicity are applicable for neurotoxicity risk
assessment.
Areas of Special Focus
The full range of human behaviors, for example language, is not present in other
species.
Factors such as differences in metabolism can result in differences among species
in sensitivity to a compound.
The most sensitive species may not be the species affected most like the human.
2.5 PANEL 5: INTERPRETATION OF BEHAVIORAL DATA
The workshop draft concludes that behavioral changes can provide evidence of
neurotoxicity in the absence of additional data.
Conclusions and Suppositions Used in Reaching This Position
Behavior is often regarded as one of the most sensitive indicators of toxicity.
Evaluations of behavior have played an important role in neuroscience research
efforts to understand brain function.
Most behavioral evaluations conducted in humans have well established animal
counterparts.
Behavioral effects often appear prior to measurable effects on physiological or
morphological endpoints.
The primary effects of developmental exposure to some chemicals may be
behavioral.
2-4
-------
Areas of Special Focus
Some behavioral changes may represent non-specific effects such as sickness or
malaise. .
It is difficult to identify a maximum tolerated dose for behavioral studies, since
most chemicals produce behavioral effects at high doses.
Toxicant-induced changes in behavior can result from a variety of physiological
changes in addition to effects on the nervous system.
2-5
-------
-------
3. CHAIRPERSON'S SUMMARY OF THE WORKSHOP
William F. Greenlee, Ph.D., Workshop Chair
3.1 WORKSHOP OVERVIEW
The Environmental Protection Agency (EPA) invited 13 scientists from academia,
industry, and government to participate in a workshop convened to review the draft document
"Proposed Neurotoxicity Risk Assessment Guidelines." This draft, prepared by an EPA
Workgroup headed by Drs. William Sette and Suzanne McMaster, was structured in accordance
with the risk assessment paradigm set forth by the National Research Council (NRC) in its 1983
book, Risk Assessment in the Federal Government: Managing the Process. The draft guidelines
were distributed to participants prior to the workshop, along with a series of issues papers
designed to focus participants' attention on topics of particular concern to EPA. Preliminary
comments on the guidelines were submitted by each participant and were also distributed prior
to the workshop.
- . " - . ". " - ;" .- " ' ; - - "..--. "'--" ^
The workshop was held at the Omni Georgetown Hotel in; Washington, D.C. on June 2
and 3, 1992. To promote discussion of issues appropriate for the peer review process, the
Agency designated five working panels to meet during the course of the workshop:
» Panel 1: Neurotoxitity as an Endpoint (Dr. William Greenlee, Chair)
Panel 2: Transient and Persistent Effects (Dr. John O'Donoghue, Chair)
Panel 3: Direct and Indirect Effects (Dr. Barry Wilson, Chair)
Panel 4: Animal-Human Extrapolation (Dr. Shayne Gad, Chair)
Panel 5: Behavior (Dr. John Orr, Chair) .
The infrastructure of the workshop was thus a matrix in which panels formed to consider
particular issues were assigned the task of reviewing a document organized around the various
stages of the risk assessment process (see Figure 1). Individual panels were asked to review the
draft guidelines from the perspective of issues related to endpoint selection (Panels 1 and 5),
issues concerning the nature of neural responses to environmental toxicants (Panels 2 and 3),
and issues involving the extrapolation of experimental results from one species to another
3-1,
-------
Hazard
Identification
Exposure
Assessment
Dose-
Response
Assessment
Risk
Character-
ization
Endpoints
Neurotoxicity (Panel 1)
Behavior (Panel 5)
Response
Transient/Persistent
Effects (Panel 2)
Direct/Indirect Effects
(Panel 3)
Extrapolation
Animal-Human
Extrapolation (Panel 4)
Figure 1
3-2
-------
(Panel 4). Each of these issues, in turn, was to be addressed across all four components of risk
assessment defined by the NRC: hazard identification, exposure assessment, dose-response
assessment, and risk characterization. At the conclusion of the workshop, panel chairs were
asked to prepare summary reports of the discussions that took place in their respective
workgroups (see Section 4 of this report).
3.2. NEUROTOXICITY AS AN ENDPOINT
Workshop participants were in general agreement that neurotoxicity is an appropriate
endpoint for risk assessment, and that regulatory action may be warranted for substances found
to be neurotoxicants under certain conditions of exposure. There was considerable discussion
and debate, however, regarding the definition of adversity as it relates to effects of a toxicant on
the nervous system. Panels 2 and 3, for example, proposed a three-tiered system for classifying
nervous system effects according to their degree of adversity. One goal of this classification
scheme was to provide a basis for distinguishing between substances that are neuroactive or
behaviorally active and those that are neurotoxic. Panel 5 concluded that behavioral changes
could provide evidence of neurotoxicity even in the absence of other data, but the panel did not
provide guidance as to how behavioral endpoints might be incorporated into routine chronic
toxicity bioassays.
~ * " ' " . ;
Participants also agreed that is important for the best available data and scientific
judgment to be brought to bear on decisions about a substance's potential neurotoxicity. It is
important, for example, to be able to distinguish actions of a chemical on the nervous system
that are causally linked to a clinically observable effect versus those that are not. Toward this
end, participants thought that more explicit guidance may be needed if non-neuroscientists will
be expected to use the guidelines to assess the adequacy of available data in establishing
whether or not a given substance should be regarded as a neurotoxicant. For example,
participants suggested that a short list of quantifiable endpoints be developed in lieu of the
"laundry list" of potential endpoints contained in Table 1 of the draft guidelines. This "short
list" would be comprised of those effects that are most likely to arise in association with chronic,
low-level exposure to environmental agents. One panel recommended that the guidelines also
encourage risk assessors who lack expertise in the relevant neurosciences to seek appropriate
3-3
-------
outside help if they encounter difficulties in dealing with data or situations that are not
clear-cut.
During the final session of the workshop, after all of the panels had met and their
conclusions had been discussed by the group as a whole, the Chair asked workshop participants
whether there are features of the nervous system that present unique challenges in the
development of risk assessment guidelines. A number of features were discussed, including:
The fact that the normal range of functioning is narrower for the nervous system
than for other organ systems.
The need to consider toxicant levels in extracellular spaces (e.g., at the
myoneural junction) as well as within the target tissue.
» The confounding role that conditioning (and/or other aspects of an organism's
history of interaction with the environment) can play in the interpretation of
behavioral data.
The greater subjectivity involved in assessing the functional status of the nervous
system than in assessing the function of other organ systems.
The fact that biomarkers of nervous system function are fewer in number and are
less well understood than biomarkers of function for other organ systems (e.g.,
the liver).
*.
Participants agreed that additional research is needed in all of these areas.
33 OTHER CONCERNS
In the opening session of the workshop, the Chair raised the question of whether
additional sources of information about the effects of chemicals on the nervous system might be
useful in assessing the neurotoxic potential of environmental agents. Although the importance
of well-controlled epidemiologic studies is widely recognized, for example, it is also possible that
data from pharmacologic studies could provide important insights into the neurotoxic potential
of certain chemicals or classes of chemicals. Similarly, the veterinary literature could serve as an
important adjunct to the types of information currently obtained from animal toxicity studies.
3-4
-------
Throughout the workshop, there was a fair amount of discussion regarding the relevance
of animal models for predicting neurotoxicity in humans. Panel 4 concluded that appropriate
animal models do exist for many if not most human behaviors. Differences in toxicant
metabolism, distribution, and other pharmacokinetic properties, however, are important
determinants of interspeeies differences in sensitivity to potential neurotoxicants. Participants
agreed that there are currently major gaps in our understanding of the pharmacokinetic
determinants of neurotoxicity, and suggested that additional research is needed, particularly in
the area of low-dose exposures to known or suspected neurotoxicants. Advances in the field of
pharmacokinetics have significantly improved our ability to study the uptake, distribution, ;and
metabolism of potential neurotoxicants. Based on this new understanding, physiologically-based
models of neurotoxicant handling should be developed, incorporating what we now know about
the metabolic properties of specific cell populations within the nervous system, factors that
control circulating concentrations of a toxicant, and factors related to the uptake and
localization of toxicant molecules within the nervous system. These pharmacokinetic data, ;
which are usually generated early in the process of characterizing a compound, should be
integrated with available dose-response data throughout the risk assessment process.
While noting that both animal and in vitro models could be useful in this regard,
participants were in general agreement that in vitro systems should probably be viewed mainly
in terms of their potential to provide supporting information about a toxicant's mechanism of
action, rather than as screening tools in a more general sense. At the same time, contemporary
cell and molecular biological techniques could be used to gain new types of knowledge about
nervous system function that might be relevant to the risk assessment process. As factors that
influence the neurotoxic potential of chemicals continue to:be identified, comparative
pharmacokinetic studies should be conducted in vivo and in vitro to determine the extent to:
which these factors are or are not species-specific. Hepatic metabolism and elimination, blood
flow to target tissues, interaction with the blood-brain barrier and/or brain lipids, and
high-affinity binding to target proteins are all examples of pharmacokinetic processes that need
to be studied in this way. It is important that differences in toxicant handling by humans and
rodents be incorporated into pharmacokinetic models used to predict the effect of a given
neurotoxicant on a nervous system target. At the same time, interspeeies differences in
metabolism and elimination need to be considered in determining appropriate doses for toxicity
3-5
-------
testing, particularly in studies that attempt to address the effects of low levels of exposure to a
suspected neurotoxicant.
Participants also suggested that it is important to recognize the speed with which
advances continue to occur in the field of neuroscience. Because of this, there was general
agreement that the guidelines should be structured to facilitate the incorporation of new
knowledge into the decision-making process.
3.4 RISK CHARACTERIZATION
As noted previously, discussion during the workshop focused largely on issues related to
the definition and classification of adverse effects. These discussions reflected both the inherent
complexity of the nervous system and the difficulty that inevitably arises when one attempts to
form conclusions about the functioning of this complex system on the basis of isolated
observations. The workgroup did not resolve all of the outstanding issues related to hazard
identification, but did recognize the need to avoid an inappropriately rigorous classification
scheme for known or suspected neurotoxicants. Criteria need to be established to provide
guidance in determining the adequacy of results obtained from chronic bioassays for assessing
the neurotoxic potential of a given agent.
The group also agreed that the goal of risk characterization in the area of neurotoxicity
should be to develop a mechanisms-based approach that incorporates our current understanding
of the biology of the nervous system to the greatest extent possible. Given the complexity of
this system, it is essential to focus on endpoints that are quantifiable and that are linked to a
clinical outcome under conditions of exposure that are likely to occur in the environment.
Toward this end, physiologically-based pharmacokinetic models should be used to develop
quantitative descriptions of the behavior of neurotoxicants at low doses. At the same time,
quantitative descriptions of the molecular and biochemical interactions that occur between a
neurotoxicant and its nervous system target need to be fully elaborated and, together with
pharmacokinetic models, incorporated into the risk characterization.
3-6
-------
4. SUMMARY OF OPENING PRESENTATIONS
4.1 INTRODUCTORY COMMENTS
Dr. William Greenlee, Purdue University
Dr. William Greenlee, Chairman of the Peer Review Workshop, opened the workshop
by welcoming participants and observers. Dr. Greenlee noted that the purpose of the workshop
was to provide a scientific peer review of the draft Neurotoxicity Risk Assessment Guidelines.
While noting that discussion among workshop .participants was expected to comprise the bulk of
the meeting, Dr. Greenlee also indicated that there would be an opportunity for comments from
observers at the conclusion of the Opening Plenary session. Observers wishing to comment at
that time were requested to add their names: to a sign-up sheet. Following-this introductory
statement, Dr. Greenlee asked Dr, William Wood to give an overview of the Risk Assessment
Forum's goals for the peer review meeting.
4.2 WELCOME AND OBJECTIVES
Dr. William Wood, U.S. EPA Risk Assessment Forum
Dr. Wood began by thanking workshop participants for agreeing to take part in the peer
review process. This review represents a first step in the Agency's efforts to develop risk
assessment guidelines in the area of neurotoxicity. The purpose of his presentation was to give
the group some generaf background regarding the guidelines development process and to
answer any questions participants might have about the objectives or intended audience for the
draft guidelines the group had been asked to review.
The Risk Assessment Forum was established in response to a 1983 study by the National
Academy of Sciences (NAS). Among other recommendations, the NAS study suggested that
each agency develop what were called "inference guidelines" to foster consistency within and
among federal agencies in the area of risk assessment. Originally a group of nine scientists
drawn froni various EPA programs and laboratories, the Risk Assessment Forum has since
grown to involve more than 30 members, including specialists in the areas of exposure
assessment and ecological risk assessment. The purpose of the Forum, however, has remained
4-1
-------
essentially the same: to develop Agency-wide guidance to promote consistency in the
application of risk assessment methodologies.
Dr. Wood noted that the 1983 NAS study identified four general elements of the risk
assessment process:
Hazard identification
Dose-response assessment .
Exposure assessment
Risk characterization
Although some of the,newer guidelines involve slight modifications to this general scheme, the
NAS paradigm has largely been preserved in all of the risk assessment guidelines issued to date,
including the draft neurotoxicity guidelines.
Dr. Wood also pointed out that the NAS study made a clear distinction between risk
assessment and risk management. Risk characterization represents the area of overlap in which
the "hand-off between risk assessment and risk management occurs. As such, much of the work
that has taken place in guidelines development over the past several years has been directed
toward improving our ability to describe or characterize risk.
The original set of risk assessment guidelines, issued by EPA in 1986, consisted of five
guidelines. Three of these (carcinogenicity, mutagenicity, and developmental toxicity) were
related to specific endpoints, while two others (exposure assessment and chemical mixtures) cut
across multiple endpoints. Within days of their publication, recommendations for revisions to
these guidelines began to be considered by working parties within the Risk Assessment Forum.
Revised versions of some guidelines have subsequently been published (e.g., developmental
toxicity and exposure assessment), while others are at various stages of review and revision (e.g.,
carcinogenicity and chemical mixtures). The point is that the guidelines development process is
a dynamic one, in which revisions and updates to existing guidelines are constantly being
considered. Similarly, efforts to develop new guidelines in other areas are ongoing. In addition
to the draft neurotoxicity guidelines currently before the group, for example, Dr. Wood
4-2
-------
indicated that efforts are also underway to develop risk assessment guidelines in the areas of
reproductive effects, quantification of non-cancer effects, immunotoxicity, and ecological risk.
Dr. Wood noted that, in reviewing the proposed neurotoxicity guidelines, it would be
important for workshop participants to keep in mind what the guidelines are not intended to be.
The guidelines are not intended to serve as a step-by-step "cookbook" in risk assessment
methodology, as a rigid and inflexible "rule book," or as a comprehensive "textbook" that reviews
all of the relevant literature. Rather, the guidelines are intended to provide a framework that
establishes the boundaries of acceptable scientific methodology, while allowing as much
flexibility; to risk assessors as the state of the science merits. In this way, the guidelines are
intended to provide assistance in moving from a body of unanalyzed data toward a meaningful
characterization of risk. As such, Dr. Wood requested that reviewers critique the draft
guidelines mainly in the following terms: ; ;>
Does the document provide guidance on how to think about the types of
information that are likely to be available to a risk assessor? _- :
Does the document provide guidance regarding judgments about the adequacy of
available methodologies and the significance of the data various methods
produce? . , --..
Does the document provide guidance regarding the appropriate use of available
data as the risk assessment, process moves from hazard identification toward a
more complete characterization of risk? .
What kind of guidance does the document provide for integrating available
information into a meaningful characterization of risk?
What kind of guidance, if any, does the document provide regarding risk
communication between the risk assessor and a risk manager?
Dr. Wood pointed out that the primary audience for the final guidelines will be Agency
scientists who conduct risk assessments or who review risk assessments submitted to EPA.
Although ideally one would prefer to have neuroscientists performing this task, he noted that :it
is likely that the task will often fall to a team of generalists in toxicology. For this reason, it is
important for the guidelines to lay out areas in which there is general scientific agreement about
acceptable and/or preferred approaches, areas in which several different approaches may be
4-3
-------
equally appropriate, and areas in which the science is simply not advanced enough for a
particular type of information to be of predictive value at this time.
Dr. Wood noted that the scientific peer review represents a relatively early stage in the
guidelines development process. Based on the peer review group's report, the draft guidelines
will be revised before being submitted to internal reviews by the Risk Assessment Forum and
the Risk Assessment Council, a senior management group that considers the implications of
proposed risk assessment guidelines for program offices within the Agency. Following this
internal review, there will be a 90-day period for public comment on the proposed guidelines, as
well as a review by the Science Advisory Board. Another round of revisions will then take
place, culminating approximately a year later with publication of final guidelines in the Federal
Register. Altogether, Dr. Wood estimated that it will take another 2 to 3 years for final
neurotoxicity guidelines to be published.
Following Dr. Wood's presentation, Dr. Greenlee reviewed the proposed agenda for the
two-day peer review meeting. He noted that the workshop was designed to follow an iterative
process, in which small-group discussion sessions would alternate with plenary sessions involving
all participants. He said that the goal of this process is to assure adequate discussion of specific
issues as well as to integrate these issues into more general guidance for the Agency in its
continuing development of the proposed guidelines. To begin this process, panel chairs had
been requested to provide an opening presentation summarizing the issues and concerns likely
to be addressed during each panel's individual deliberations.
43 NEUROTOXICITY AS AN ENDPOINT (PANEL 1)
Dr. William Greenlee
Dr. Greenlee began his presentation by noting that the whole process of guidelines
development is based on the general agreement that neurotoxicity is, in fact, an appropriate
endpoint for environmental risk assessment. The challenge to the peer review workshop,
therefore, is to identify ways of bringing the best available science to bear on the risk assessment
process. More specifically, the group's task is to conduct a scientific peer review of the draft
guidelines in order to provide guidance to the Agency in its continuing effort to develop a
4-4
-------
scientifically-based risk assessment process for neurotoxicity. This guidance should include
recommendations about how best to use the existing neurobiological knowledge base, as well as
identifying areas of research that are likely to fill existing gaps in our ability to assess neurotoxic
risks. - : ' ^ : - ' -' '-
Dr. Greenlee noted that the overall purpose of risk assessment guidelines is to reduce
the uncertainty in assessing the potential of a physical, chemical, or biological agent to produce
an adverse response. He said that this is a particularly challenging task in the area of
neurotoxicity, because the extreme complexity of the nervous system poses challenges beyond
those typically confronting the toxicologist. Dr. Greenlee suggested that toxicology falls short of
its goals when it does not keep pace with the current state of biological understanding. This
failure may occur either because lexicologists attempt to extend their conclusions beyond the
current state of biological understanding or because their conclusions do not fully incorporate
all available data. One task of the workshop, therefore, would be to address both sides of this
question by determining areas in which the proposed neurotoxicity guidelines might fall short of
our current understanding of neurobiology as well as areas in which the proposed guidelines
might extend beyond current neurobiological knowledge. ,
Dr. Greenlee noted that, in his view, a scientificallyrbased neurotoxicity risk assessment
process is necessarily a biological mechanisms-based approach, couched within the .
exposure-dose-response paradigm. In this context, he speculated that three types of efforts
might be especially likely to improve our understanding of risk assessment in this area:
Physiologically-based pharmacokinetic studies, particularly those that would
improve our ability to predict tissue doses based on exposure to an ' "
environmental agent
Development of pharmacodynamic models, particularly those that would improve
our ability to quantify biological processes at both the cellular and the
whole-animal level ' .
Mechanistic studies, particularly those that would increase our understanding of
the biological determinants of tissue-or species-specific actions of an
environmental agent
4-5
-------
Dr. Greenlee suggested that another important topic for the group to consider is the
extent to which existing resources are or could be exploited in identifying the neurotoxic
potential of an environmental agent. Although we recognize the importance of controlled
epidemiologic studies, for example, he wondered whether we are taking full advantage of other
types of human studies, such as pharmacologic studies that might offer important insights into
the molecular determinants of toxicity. Similarly, while we recognize the importance of
well-controlled animal toxicity studies, we may not be taking full advantage of the wealth of
information that is available in the veterinary literature. He cautioned that the results of in vitro
studies may have to be considered somewhat more rigorously, since in vitro systems are clearly
more limited in their ability to reflect the complexities of the intact nervous system. In vitro
studies can play an important role in exploring mechanisms of toxicity, but only when the
relationship of the in vitro system under study to responses in the whole animal is reasonably
well understood. The results of in vitro testing should not be over-interpreted.
Given the variety of approaches that can be taken to study nervous system function,
Dr. Greenlee suggested that one goal of risk assessment is to incorporate data obtained from
human, animal, and in vitro studies into an integrated understanding of the biological
mechanisms of neurotoxicity. Effective integration of all available data, in turn, is most likely to
occur when biologists, mathematicians, and risk assessors are able to work together throughout
the risk assessment process. In this way, it might be possible to develop quantitative
descriptions of the complex biological processes that are characteristic of the nervous system,
including the potential for interactions of this system with environmental agents. Although it is
important for the workshop to focus on specific issues of concern, Dr. Greenlee expressed the
opinion that it would also be important for the group to keep in mind how its review of the
proposed neurotoxicity guidelines could foster this larger, overarching goal. Part of his role as
Chairman, he concluded, would be to assure that the group not lose sight of this aspect of its
charge.
4-6
-------
4.4 TRANSIENT AND PERSISTENT EFFECTS (PANEL 2)
Dr. John O'Donoghue, Eastman Kodak Company
Dr. O'Donoghue's presentation focused on participants' pre-meeting comments on the
interpretation of transient and persistent effects of an environmental agent on the nervous
system. Dr. O'Donoghue said that identifying areas of agreement and disagreement proved to
be somewhat difficult, since there was a fair amount of imprecision and ambiguity in the way
similar terms were used by different reviewers. Transient effects, for example, are not
equivalent to reversible effects, although these terms were used interchangeably in some
reviewers' comments, including his own. Likewise, use of the terms "persistent" and ;
"irreversible" was sometimes confusing, as was use of the terms "effect" and "toxicity."
In an effort to clarify the distinctions among these terms, Dr. O'Donoghue proposed a
series of definitions for consideration by the group: ;
Transient effects are temporary, fleeting in time, or short-lived. These effects can
usually be measured in minutes, hours, or at most a few days. Mechanistically,
transient effects involve the direct exposure of the target system to a chemical; as
a result, they generally persist only as long as the chemical is present in the body.
Persistent effects, on the other hand, continue beyond the pharmacologicallife
span of the causal agent, even if .only for a short period of time. As an example,
Dr. O'Donoghue described a temporary impairment of vision by a flash of light
as a persistent effect, since it lasts beyond the period of the flashy .
Reversible effects are those that can be corrected or rectified by repair processes
that enable the body to return to its original state. Although reversibility is
usually thought of in both structural and functional terms, it is often difficult to
determine that a return to the original state has occurred in both of these areas.
Irreversible effects are effects that cannot be corrected or repaired. As a result,
irreversible effects result in some permanent or long-lasting change in the
structure or function of the nervous system.
Latent effects are those effects appearing long after the last contact with the
causal agent, usually after the agent is no longer present in the body. Residual
effects, on the other hand, are a subtype of irreversible effects in which damage is
incompletely repaired. These effects may be readily apparent or "they may be
silent, reappearing only at some later time when the organism is challenged.
Structural damage; that persists after clinical recovery has occurred is an example
, of a residual effect, since this type of damage could be expected to reduce the
residual capacity of the organism to accommodate subsequent insults.
4-7
-------
Dr. O'Donoghue suggested that recognizing the differences among these terras is critical if a
consensus is to be reached about how each type of effect should be interpreted and dealt with in
the risk assessment process.
There was a general consensus among participants regarding the importance of
distinguishing among different levels or degrees of adversity in characterizing the effects of a
chemical on the nervous system. In the area of pathology, for example, Dr. O'Donoghue noted
that there are many different types .of changes that could be observed, not all of which would be
considered equally strong evidence of toxicity by most pathologjsts. Some pathologic changes,
for instance, mimic artifactual responses so closely that it may be difficult to tell whether a
pathologic change has actually occurred. Because of their ambiguity, these types of effects
should be considered relatively weak evidence of toxicity. Quantitative changes in morphology
or in the rate or extent of functional processes that normally occur in the organism, on the
other hand, might be considered stronger evidence of toxicity. More significant still would be
clearly pathologic events involving structural or functional changes that are never seen in a
healthy animal. Similarly, in the area of behavior, important distinctions can be made among
effects of a chemical that interfere with the animal's ability to survive, effects that perturb the
animal's ability to interact with the environment, and behavioral effects that are not specific to
the action of the chemical. Dr. O'Donoghue held that qualitative differences in the degree of
adversity of various effects should be taken into account when neurotoxicity data are evaluated.
Another general area of concern raised by many participants was the difficulty inherent
in attempting to determine whether a chemical is or is not a neurotoxicant, given that toxicity is
so intimately related to the conditions of exposure to a chemical. One way of dealing with this
problem is to incorporate assumptions about exposure into the classification system. For
example, a substance could be classified as an occupational neurotoxicant if it produces adverse
effects on the nervous system at exposure levels expected to occur in the workplace. By this
definition, acrylamide and n-hexane would both be classified as occupational neurotoxicants.
Environmental neurotoxicants, on the other hand, would be those substances that produce
adverse effects on the nervous system at exposure levels occurring in the ambient environment.
Lead and mercury would fit the definition of environmental neurotoxicants. Dr. O'Donoghue
said that focusing on these types of classifications is important, since the guidelines are
4-8
-------
ultimately intended to protect people from effects associated with occupational or
environmental exposure to the substance under study.
In considering primary versus secondary effects of a chemical on the nervous system,
Dr. O'Donoghue suggested that the group adopt the traditional toxicologic distinction between
primary and secondary toxicants. Thus, a primary neurotoxicant would be a material such as
2,5-hexanedione that interacts directly with target sites in the nervous system. A secondary
neurotoxicant, in contrast, would be a material such as n-hexane, which must be metabolized
before it can interact with its nervous system target.
Dr. O'Donoghue then provided a summary of the extent to which pre-meeting comments
addressed each of the issues before the Transient and Persistent Effects panel.
Issue #1: The workshop draft concludes that both reversible and irreversible effects of
chemicals on the nervous system should be considered adverse. .
There was little agreement regarding the extent to which all transient, persistent, and
reversible effects of a chemical on the nervous system should be considered adverse. Most
commenters agreed that some effects would be of greater concern than others, but opinions .
diverged about whether all types of effects should be considered adverse. Similarly, there was
general agreement that all chemicals could be expected to impact the nervous system at some
dose levels, but participants generally did not think that this observation was particularly useful.
More important, most agreed, is the likelihood that a substance will produce transient or
persistent effects at exposure levels expected to occur in an occupational or environmental
setting. If an agent produces adverse effects on the nervous system only at exposure levels
significantly higher than those expected to occur in these settings, participants did not agree that
the agent should be classified as a neurotoxicant.
Issue #2: The nervous system contains billions of cells wired in complex patterns and is
known to be resilient to environmental and toxicological insult by a process known as compensation or
adaptation.
4-9
-------
Dr. O'Donoghue indicated that this issue seemed to generate a fair amount of confusion,
mostly because participants seemed to have some difficulty understanding the significance of the
statement. At least some participants held that metabolic or behavioral adaptation should not
necessarily be considered adverse, even if the adaptation is in some sense pathological. To
resolve this issue, Dr. O'Donoghue suggested that the panel would probably want to try to
address where the boundary between adverse and non-adverse forms of adaptation might lie.
Issue #3: Once damaged, nerve cells have limited capacity for regeneration.
There was agreement that once neurons are killed or central neural processes damaged,
regeneration is very limited, if indeed it occurs at all. The capacity for repair is, however, much
greater following transient effects or effects that do not involve structural damage to central
neural processes. On the functional level, in fact, the capacity for repair often is substantial.
SP
Issue #4: Apparent recovery actually represents activation of reserve capacity, decreasing
potential adaptability.
There was general agreement that the reversibility of effects involving cell death or
destruction may represent an activation of reserve capacity which in turn could decrease future
adaptability. There was not, however, a consensus that reserve capacity is diminished by
neurotoxicants acting by other mechanisms, such as those producing strictly neurochemical or
functional changes.
Issue #5: Traditionally, effects of toxicants are considered to be persistent or long-lasting,
while pharmacological effects are considered to be transient or short-acting.
Participants agreed that transient effects occurring at high exposure levels are not
necessarily indicative of environmental neurotoxicity. The group also agreed that transient,
short-acting, or pharmacologic effects should not be put in the same category as permanent or
irreversible effects. In general, commenters recognized a need for more than two categories of
classification i.e., neurotoxicant or non-neurotoxicant. To adequately address all available
4-10
-------
information, most participants felt that there would need to be a broader range of possible
classifications. :
Issue #6: An effect that appears to be transient in an unchallenged organism may be revealed
as long-lasting through an environmental or pharmacological challenge.
Issue #7: It is not known whether transient effects observed following 'developmental exposure
should be evaluated at specific points in the life span. ,
Most workshop participants did not address either of these issues in their pre-meeting
comments, so it was not possible to get a sense of the group's thinking about these topics.
Following Dr. O'Ponoghue's presentation, there was discussion among workshop
participants of several of the issues he raised. One participant suggested that another term that
could be added to Dr. O'Donoghue's list of definitions is progressive effects, which might be
defined as effects that continue to evolve even in the absence of continued exposure to a
toxicant. An, example of a toxicant producing this type of effect is carbon monoxide, which may
continue to produce progressive deterioration in the nervous system, despite an apparent
recovery of function after an acute exposure. Progressive deterioration can also occur late in
life, when the results of prior exposure to a chemical combine with age-related heuronal
attrition to produce an accelerated or progressive loss of function. ,
The same participant also noted that, in addition to addressing the need to specify
whether an effect is primary or secondary, some people had also brought up the need for
distinctions between direct and indirect effects. The examples that Dr. O'Donoghue had
provided for primary and secondary neurotoxicants were both direct-acting agents, as opposed
to agents that exert their effects on the nervous system indirectly for example, via effects on
energy metabolism. The commenter wondered whether it might be useful to consider primary
and secondary effects as subcategories of both direct and indirect effects of a chemical ori the
nervous system.
4-11
-------
Third, this participant suggested that it might be useful for the group to think about
classifying some substances as neuroactive agents, as distinct from neurotoxicants.
Dr. O'Donoghue indicated that he thought all of these issues would be discussed during
the workgroup's deliberations. Noting that a separate panel had been established to consider
the question of direct and indirect effects, he predicted that there would be a fair amount of
overlap among the issues to be addressed by the various workgroups. He suggested that it
would be difficult to divorce transient and persistent effects from direct and indirect effects, for
example.
Also important in the consideration of direct and indirect effects is the issue of
identifying neurological signs and symptoms that may be incidental to processes occurring in the
rest of the body. Citing a passage from an 1897 text on poultry pathology, Dr. O'Donoghue
pointed out that it has long been recognized that there are a number of characteristic behaviors
that animals exhibit when they are sick, regardless of the nature of the illness, and that these
types of effects need to be distinguished from either direct or indirect effects of a chemical on
the nervous system. One participant agreed, noting that the definition of neurotoxicity in the
draft guidelines might be overly broad. This participant suggested that more categories or levels
of concern would probably be necessary to reflect important differences between a chemical that
has a short-term, transient effect on nervous system function and one that causes widespread
structural damage within the brain. Dr. O'Donoghue wondered whether part of the problem
might be the draft guidelines' overwhelming emphasis on hazard identification as opposed to
other elements of the risk assessment process, such as dose-response relationships, exposure
assessment, and risk characterization.
Another participant suggested that the guidelines might need to be broadened to provide
guidance to individuals designing and performing neurotoxicity studies as well as those
interpreting the results of these studies. It might be possible to describe a more complete
spectrum of possible approaches and possible results in neurotoxicity testing, if in the process it
is specified how different types of findings might contribute to an overall assessment of a
substance's neurotoxic potential. Dr. O'Donoghue agreed that there might t>e some merit to
such an approach, but wondered whether this type of discussion would be appropriate, given
4-12
-------
that the group's assigned task was to review the scientific underpinnings of the existing draft of
the risk assessment guidelines. ' ; ' . "
4.5 DIRECT AND INDIRECT EFFECTS (PANEL 3)
Dr. Barry Wilson, University of California (Davis)
Dr. Wilson began his presentation by summarizing the proposals and statements laid out
for consideration by the Direct and Indirect Effects panel. ;These statements included a
proposal that agents acting through indirect as well as direct means could be considered
neurotoxic, that direct and indirect, effects of a chemical are equivalent to primary and
secondary effects, and that effects on the endpoint of neurotoxicity are "functionally equivalent,"
whether they arise from direct or indirect effects of a chemical on the nervous system. In
addition, Dr. Wilson noted that the panel had been asked to concentrate on two areas of special
focus: a caveat that information available to risk assessors might not be adequate to distinguish
primary from secondary actions of a chemical, and a statement that hepatic toxicity, which could
secondarily damage the nervous system, might or might not represent a neurotoxic effect.
Finally, Dr. Wilson indicated that the panel had been asked to consider a syllogism in which the
logical inconsistency of claiming that a compound can produce neurotoxicity but not be a
_ neurotoxicant is linked to the conclusion that any compound, at a high enough dose, must be a
considered a neurotoxicant because of its potential to produce lethal effects.
Participants' pre-meeting comments were in general agreement regarding the need to
classify as neurotoxicants those chemicals that directly affect the nervous system. There were,
however, a number of conditions that various participants thought should be added into,the
equation. Most people agreed, for example, that in order to qualify as evidence of
neurotoxicity, an effect should occur at low doses of a chemical and should exhibit a
dose-response relationship. Similarly, most participants thought that the observed effect should
fit a restrictive definition of adversity, although there was a wide range of opinions regarding
how restrictive this definition should be. Dr. Wilson pointed out that the definition of adversity
proposed in the draft guidelines is any alteration in the structure or function of the central
nervous system and/or peripheral nervous system that diminishes the ability to survive,
reproduce, or adapt to the environment. Noting that some participants would not consider this
4-13
-------
definition restrictive enough, Dr. Wilson suggested that coming up with an acceptable definition
of adversity might be one of the most important tasks before the workshop as a whole, since this
definition is in a sense where the process of risk assessment begins.
Participants were not in agreement regarding the extent to which chemicals with indirect
or secondary effects on the nervous system should be regarded as neurotoxicants. Similarly, no
consensus was apparent on the question of whether any chemical-induced change in the
structure or function of the nervous system should be considered an adverse effect, regardless of
the nature of the change.
In order to sort out these differences, Dr. Wilson speculated that the Direct and Indirect
Effects panel would need to address a number of important issues. For one thing, it would
probably be important to try to come up with more precise definitions for many of the
non-trivial terms used in the guidelines, including "adverse effect" as well as terms such as
"direct" and "indirect" or "primary" and "secondary." In this respect, the panel's efforts would
complement some aspects of the work that Dr. O'Donoghue had proposed for the Transient
and Persistent Effects panel.
Second, Dr. Wilson said that the panel would probably want to consider the dose level at
which an assessment of risk should begin. Given the important role of the nervous system in
modulating and integrating homeostatic responses, he suggested that it is not immediately
obvious whether risk assessment should begin at the dose level where an effect is first detected
or some higher dose level where the observed effect exceeds some physiologic limit or limits. In
this sense, he suggested that there might be a fair amount of room for a chemical to produce
effects on the nervous system that do not represent damage.
Finally, based on a general sense of participants' concerns about the definition in the
draft guidelines, Dr. Wilson proposed an operational definition of neurotoxicity in which a
chemical would be classified as a neurotoxicant if it had its major action on the nervous system,
whether via primary or secondary mechanisms, such that at low dose levels it produces effects
that injure the short- and long-term health of the organism. While acknowledging that this
definition might not address all of the participants' reservations about the existing definition of
4-14
-------
adversity, Dr. Wilson suggested that it might provide a reasonable starting point for further
discussion.". . - ' ''- .... ;'"- " ' ' :".':-. ::*--"-' '-"..: :/-'- _ -.:',-'-' '.-. -^
Following Dr. Wilson's presentation, other workshop participants offered their views on
issues to be considered by the Direct and Indirect Effects panel. One participant suggested that
the difficulty people were having with definitions used in the draft guidelines might have to do
with the fact that the focus was on trying to label compounds rather than describing observed
effects. It is more important that exposure to a toxic compound be limited than whether the
compound is classified as a neurotoxicant or a hepatotoxicant. In this sense, he said, it is the
effect rather than the underlying mechanism that should form the basis for regulation of a "
chemical. Dr. Wilson agreed that it is important to keep the focus of the discussion on the,
ultimate goals of the process, but noted that .there is a great potential for misunderstanding if
one group of people uses the term "indirect" to refer to substances that must be metabolically ..
activated while another group uses the same term to refer to substances that act mainly on
non-neural systems. ":' : '- -. '.''. , ''" : /
Another participant asked representatives from EPA to clarify the extent to which the
guidelines were or were not intended for the protection of species other than humans, and the ;
extent to which the neurotoxic potential of biological agents should be taken into; account. An
EPA representative indicated that the guidelines are intended to address risks to humans.
Although attention in this area has traditionally focused on chemical and physical agents, it is
increasingly the case that the Agency is being asked to evaluate the risks associated with
biological agents.
-': The same participant also asked for clarification regarding the breadth of the Agency's
definition of neurotoxiciry. He wondered, for example, whether the neuroendocrine system
should be considered part of the nervous system. Similarly, he wondered whether agents that
act principally on muscle should be considered potential neurotoxicants, since many agents that
act in this way are known to produce behavioral and neurologic effects in humans. Dr. Wilson
echoed this concern, noting that damage to target organs of the nervous system often results in
damage to the nerve or nerves innervating that organ. ,; , -_
4-15
-------
An EPA representative responded that the Agency had not directly considered effects on
muscle or on the neuroendocrine system, and would leave it to the expert reviewers to
recommend whether these types of effects should be incorporated into the definition of
neurotoxicity. The participant who had raised the question indicated that he thought that
effects on the neuroendocrine system and on muscle should be included within the definition of
neurotoxicity. Dr. Wilson wondered whether there might not be a property, such as excitability,
that would link all of the various systems that had been being discussed. Another participant
argued that it did not seem productive to attempt to look at the nervous system in a vacuum.
This person suggested that the point should be to identify the lowest-dose effect of a chemical
and regulate it on that basis. It doesn't matter whether the basis of the regulation is one effect
or the other, so much as it matters that exposure to the toxicant be limited. It is important to
keep in mind that neurotoxicity will not be the only basis upon which exposure to a chemical
can be or should be regulated.
Dr. Greenlee said that this issue will need to be addressed by one or more of the
individual workgroups, probably in the context of how specific endpoints do or do not reflect
complex interrelationships between the nervous system and other organ systems. Another
participant observed that this problem is not unique to the nervous system. Noting that in a
biological system everything is related to everything else, this commenter pointed out that it is
equally difficult to determine, for example, whether hypertension should be considered a disease
of the cardiovascular system or the central nervous system. In this person's view, the problem
may be that the draft guidelines are attempting to fit too many different types of effects into the
definition of neurotoxicity.
Another participant suggested that the whole process of risk assessment and regulation
could be thought of as a diagnostic exercise, in the sense that it is sometimes possible to treat a
disease specifically and sometimes possible only to treat it symptomatically. The parallel to
specific treatment is that it will sometimes be possible to regulate a chemical because its role in
producing signs or symptoms of neurotoxicity is well understood. In other cases, however, it
may be necessary to make a judgment about which "symptoms" are important enough to trigger
regulation of the chemical, even though the mechanism by which the chemical causes those
symptoms may not be well understood. Dr. Wilson expressed some reservations about this
4-16
-------
analogy, noting that it probably would not be desirable to have one way of handling risk
assessment for compounds with known mechanisms of action and another way for compounds
that are not as well studied.
4.6 ANIMAL-HUMAN EXTRAPOLATION (PANEL 4)
Dr. Shayne Gad, Becton Dickinson Research Center
Dr. Gad began his presentation by noting that the Animal-Human Extrapolation panel
had been asked to consider the conclusion that, if an animal data base exists and is adequate
according to the definition provided in the guidelines, it is possible to make judgments about
the potential toxicity of an agent for which no or inadequate human neurotoxicity data exists.
To set up the discussion of this and other issues expected to arise during the workgroup's
deliberations, Dr. Gad provided a summary of pre-meeting comments related to each of the
conclusions and suppositions that had been used by the authors of the guidelines to reach this
general conclusion.
Statement #1: Substances producing neurotoxicity in humans also result in neurotoxicity in
other species, '
The importance of this statement is that in most cases the overwhelming majority of
toxicity data available in assessing a chemical's potential toxicity will have been gathered in
animals. Because hazards are almost always identified first in animals, the relevance of animal
data in assessing a chemical's potential toxicity to humans is a question of pivotal importance.
Dr. Gad noted that participants were divided in their opinions about the validity of this
statement. Although no one disagreed that animal data could be used to identify neurotoxic
agents, there was a considerable disagreement regarding appropriate ways of addressing
differences between animals and humans that might limit the direct applicability of animal data
to an assessment of a chemical's potential risk to humans. Concerns about differences in routes
of exposure, metabolism, and the time scale of effects were prominent in this regard.
Statement #2: Compared with human studies, animal studies are more often available and
provide more precise dose information and better control for environmental factors.
4-17
-------
Dr. Gad reported that there was broad, substantial agreement with this statement among
workshop participants. Virtually all participants recognized animal models as a primary tool for
hazard identification in neurotoxicology as in other areas of toxicology. The problem, however,
is that limitations of individual animal models may sometimes make it difficult to distinguish
neurologic from other target organ effects. This may be especially difficult in the sense that risk
assessors are typically asked to assess toxicity on the basis of data that were generated by
someone else. Although it is true that precise control of dosing is usually possible in animal
studies, the trade-off is that our understanding of pharmacokinetic or metabolic differences
between the test animal and humans may be rudimentary, at best. These gaps in our
understanding, in turn, sometimes make it very difficult to determine whether an observed effect
is a nervous system effect or an effect on some other system.
Statement #3: Many diagnostic procedures employed to evaluate neurotoxicity in humans
have corresponding animal models.
There was some divergence of opinion among participants regarding the validity of using
the term "many" or "most" in this statement. Participants generally agreed, for example, that
diagnostic procedures requiring cognitive interactions with a patient do not have corresponding
animal models. Another issue that arose in connection with this statement was the caution that
laboratory animals are not and should not be considered "little humans." The fact that many
human diagnostic tools have corresponding animal models does not mean, for example, that it is
possible to use most of these tools in any one animal model. It might be easiest to assess a
compound's effects on cognitive function by testing it in pigeons, for example; it would be
difficult to assess the compound's more general toxicity in this system, however, since clinical
chemistries and many of the other parameters one might wish to consider are not as well
characterized in pigeons as they are in rats and mice.
Statement #4: The range of uncertainty factors used to extrapolate animal data to human
risks for other endpoints of toxicity are applicable for neurotoxicity risk assessment.
While noting that participants were in substantial agreement with this statement, Dr.
Gad indicated that there also was a strong sentiment that efforts should continue to improve
4-18
-------
precision in this area. In general, participants agreed that the more we know about the
biological basis of an observed effect, the more likely we are to understand the relevance of the
effect across species. With this increased understanding, in turn, our assessments of risk are
likely to become-significantly more precise and, as a result, more effective in achieving the goal _
of appropriate regulation.
Dr. Gad then turned to the three areas of special focus assigned to the Animal-Human
Extrapolation panel. This task was more difficult, he said, since he was not sure that the areas
of focus identified in the Issues Paper were in fact the areas that he would consider key to the
panel's deliberations. Nevertheless, he had attempted to summarize participants' views on the
areas of speciar focus identified by the authors of the draft guidelines. ^
Focus Area #1: The jutt range of human behaviors, for example language, is not present in
other species, , '
This statement raised again the issue of cognitive endpoints,-which most participants
agreed did not have corresponding animal models at least not in any one non-human species.
On the other hand, participants agreed that it is possible to model most human behaviors by
looking broadly across a variety of animal models. For this reason, and because animal data are
generally what we have to work with, it was not clear to some other participant^ what the
significance of this statement might be.
Focus Area #2: Factors such as differences in metabolism can result in differences among
species in sensitivity to a compound.
Participants were generally in agreement with this point, although Dr. Gad cautioned
that the statement should be read very carefully. It should not be interpreted to mean that
metabolism is the only or even the most important difference between species. He emphasized
that it is also important to recognize the many other factors that can lead to differences in
sensitivity to a chemical among different species as well as among individuals of the same
species, including differences in age, nutritional status, gender, race, health status, and other
characteristics.
4-19
-------
Focus Area #3: The most sensitive species may not be the species most like the human.
It is generally accepted that, in the absence of other information, the species that is
phylogenetically closest to humans should be expected to predict human responses best. This
assumption may be in conflict with the approach to risk assessment in which the lowest observed
effect level is modified by application of a safety factor to determine acceptable levels of human
exposure to a chemical. Humans are not necessarily the most sensitive species, and the species
that best predicts human responses may not be the most sensitive species. This is another area
in which our ability to assess the relevance of animal data is directly related to our
understanding of the biology underlying an observed effect. Dr. Gad emphasized that only by
understanding why an effect occurs in an animal can we truly understand whether and how the
effect might be relevant to humans. We should also be clear about which segment of the
population we are talking about in assessing the relevance of animal data, given the broad range
of sensitivities to a chemical that can be expected to occur in the human population.
In addition to the issues and areas of focus raised by the authors of the guidelines, Dr.
Gad also brought up other issues that the Animal-Human Extrapolation panel might address.
First, Dr. Gad suggested that it might be important for the guidelines to offer some guidance
regarding the selection of animal models. Identifying species that are most likely to provide
useful data might be one way of doing this. Although model selection is an important concern
in any area of toxicology, Dr. Gad suggested that it is particularly important in neurotoxicology,
which attempts to examine very complicated systems and behaviors. In his view, even the
statement that a broad range of animal models may be needed to assess certain types of nervous
system effects might be a useful addition to the draft guidelines.
Second, Dr. Gad raised a concern about the treatment of in vitro models in the draft
guidelines, a topic that a number of other participants had also addressed in their pre-meeting
comments. As currently written, some people thought that the draft guidelines were very
confusing in their treatment of in vitro systems, suggesting both that the results of in vitro tests
are of little predictive value and that these results could be suggestive of neurotoxicity.
Workshop participants seemed to be divided in their opinions about the relevance and utility of
4-20
-------
in vitro models per se, but most seemed to agree that this type of discussion added little of
scientific value to the draft guidelines.
Following Dr. Gad's presentation, workshop participants engaged in a discussion of
issues related to topics before the Animal-Human Extrapolation panel. The_ Chair asked
whether there might be prototype animal models that could be recommended for use in '
studying the neurotoxic potential of specific classes of chemicals. Dr. Gad responded that there
are good models for several classes of known toxicants; however, in the larger sphere of hazard
identification, one is usually dealing with classes of chemicals or biological agents about which
much less is known. In commenting on Dr. Gad's observation that humans are not always the
most sensitive species, the Chair noted that much potentially useful information can be lost if
the focus is exclusively on finding the most sensitive animal model or, in fact, on finding a
response to a chemical. In many cases, asking why a given species does not exhibit a particular
response can also be an important way of investigating mechanisms of chemical toxicity.
Another participant recommended that the Animal-Human Extrapolation panel spend
some time defining more clearly those areas in which animal models of human behavior are and
are not thought to exist. From a behavioral perspective, for example, language is a form of
social behavior, and animal models for social behavior do exist. This participant suggested that
considering language in terms of the class of behavior in which it falls would probably be more
useful than focusing on aspects of language that are related mainly to the characteristics of
human vocal cords. Another participant suggested that the symbolic representation of events
and objects in verbal behavior may be peculiar to humans, but the first participant thought that
even this type of behavior had been shown to occur in non-human primates.
Another workshop participant expressed the view that it would also be important for the
Animal-Human Extrapolation panel to. consider in more detail the advantages and limitations of
in vitro tests, both as a screening tool and as away of investigating mechanisms of toxicity. This
person predicted that more and more in vitro data will be coming in to regulators over the next
few years, and risk assessors will probably need guidance on how to use this information. Dr.
Gad agreed, noting that in vitro models are widely used in the pharmaceutical industry and .:«,
elsewhere to try to understand what worked and what didn't. He said that the problem is
4-21
-------
slightly more complicated with respect to in vitro tests of neurotoxicity, however, since most of
the systems proposed as in vitro models of neurotoxicity are actually models of more general
cytotoxicity.
Another participant pointed out that there is at least one tissue culture system that
reproduces the anatomy, ultrastructure, physiology, and pharmacology of the nervous system
well enough to reproduce in vitro the precise pattern of changes that one sees in humans and
animals exposed to a neurotoxicant. This system has been used for more than 30 years, both as
a screening device and as a tool for investigating mechanisms of neurotoxicity. A third
participant noted that he, too, has been working with tissue culture systems throughout his
career. In this person's view, in vitro systems are particularly useful in studying direct actions of
a toxicant on a living system, independent of metabolic transformations and other detoxification
mechanisms that might come into play in the whole animal. Then, by working backward, it is
often possible to gain some insight into why a particular effect that was observed in culture did
or did not take place in vivo. Dr. Gad pointed out that the regulator will be asked to make
judgments on the basis of the available data and will usually not have the luxury of going back
to ask additional questions of the in vitro system.
4.7 BEHAVIOR (PANEL 5)
Dr. John Orr, Southwest Research Institute
To begin his presentation, Dr. Orr observed that the Behavior panel was the only panel
charged to look at an endpoint that is unique to neurotoxicity. In this sense, he said, the
Behavior panel will be considering an endpoint that is in some sense the fodder from which
other panels will start their deliberation.
Noting that the definition of adverse effects had been near the top of the other panel
chairs' lists of issues, Dr. Orr indicated that this definition is one that the Behavior panel would
also need to consider carefully. A discussion of how various endpoints relate to the notion of
adversity could easily fill the whole of the panel's allotted time. In this regard, he predicted that
the sensitivity and specificity of behavioral endpoints might be a particularly important area for
discussion during the Behavior panel's deliberations.
4-22
-------
While noting that most participants' pre-meeting comments focused on topics other than
behavior, Dr. Orr indicated that a few issues of central importance to the Behavior panel had
been raised. One such issue was the question of how to evaluate the whole body of behavioral
data that might be available for a chemical under study. Particularly important in this regard is
the relative weighting of different types of information that comprise the "input" side of the risk
assessment process. How this hypothetical weighting should be accomplished for example,
whether data should be evaluated in terms of clusters or functional domains of behavior
would probably be an important topic for the Behavior panel to consider. .-",-.
Echoing earlier concerns about the difficulty of determining whether a substance is or is
not a neurotoxicant, Dr. Orr concluded his presentation by suggesting a model that he thought
could be used to address this dichotomy. In this model, there would be one threshold above
which a chemical would be classified as a neurotoxicant and an opposing threshold below which
one could be relatively confident that a chemical would not exhibit significant neurotoxic effects.
Beginning at a point somewhere between these two extremes, each piece of data could be
evaluated in terms of the size of the step it would warrant toward one or the other threshold. A
change in a single endpoint out of a functional observation battery, for example, might move the
overall assessment only slightly closer to the threshold for neurotoxicity. ,A finding of frank
neuropathology, on the other hand, would merit a much larger step, in most cases one that
would push the assessment over the toxicity threshold. Using this model, some lines of evidence
might point toward a finding of neurotoxicity, whereas others might lead to a conclusion that no
significant risk of neurotoxicity exists .for a particular chemical under specified exposure
conditions. While acknowledging that he had not worked out the details of this model, Dr. .Orr
suggested that it might provide a useful basis from which the panel's discussion of behavioral
endpoints could proceed.
Following Dr. Orr's presentation, participants discussed issues related to topics that the
Behavior panel would address. One participant asked whether and to what extent epidemiologic
approaches to risk assessment have begun to supplant the types of animal behavior studies that
the group had thus far been discussing. Another participant observed that the guidelines as
written place a great deal of emphasis on epidemiologic evidence of neurotoxicity. In this
person's view, in fact, the guidelines place too much emphasis on this type of evidence, given
4-23
-------
that epidemiology as a science cannot prove that a causal relationship exists between a chemical
and its effects. He suggested that much more valuable information could be obtained from
reports of side effects in the therapeutic drug literature, for example, since these studies usually
relate adverse effects to exposure to a specified dose of an agent for a known period of time.
Uncontrolled human studies can also be very useful; evidence of the neurotoxic potential of
methyl phenyl tetrahydropyridine (MPTP), for example, initially came from clinical neurology
studies performed at two centers. For hazard identification purposes, this person argued that it
might be useful to provide a weighting system for the various types of studies that might have
been performed to assess the toxicity of a chemical. If such a system were adopted, he
recommended that the greatest weight be given to controlled human studies and controlled
animal studies, and the lowest weight be given to tissue culture studies and uncontrolled
epidemiologic observations.
Another participant asked for guidance from EPA representatives regarding the
intended outcome of the hazard identification process in the area of neurotoxicity. Looking at
Table 7A, for example, it seemed that the guidelines were moving toward a multi-category
approach similar to that-used to classify carcinogens, whereas earlier in the document it had
seemed that calculation of a reference dose would be the ultimate goal of the risk assessment
process. The Chair agreed that this is an important question, noting that hazard classification
schemes had been proposed by a number of workshop participants as a way of getting around
the difficulty of stating unequivocally that a particular agent is or is not a neurotoxicant. An
EPA representative indicated that it had been the intention of the authors to avoid a
classification system like that used for carcinogens, because of the many problems that system
has historically engendered. Rather, like proposed guidelines for other non-cancer endpoints,
the neurotoxicity guidelines were premised on an assumption that available data would usually
be adequate to support calculation of a meaningful reference dose. Although it is appropriate
for the group to consider whether this assumption is warranted, it would probably be more
useful for the Agency for the workshop to focus on ways of evaluating the value of specific types
of data rather than attempting to rank the relative usefulness of different types of data. This
distinction Is important, since the task of the risk assessor will be to evaluate available data,
rather than to determine whether other types of data might be more useful.
4-24
-------
Another workshop participant suggested that part of the problem might be that there are
many different types of data that might go into the risk assessment process, and the process
might produce many different outcomes. The real goal is to find a way of expressing confidence
both in the quality of the data going in and in the quality of the judgment coming out of the
risk assessment process. The point should not be whether the data are sufficient or insufficient,
adequate or inadequate in supporting a particular conclusion i.e., that a substance is or is not
a neurotoxicant but rather to clearly identify the level of confidence we have in whatever :
conclusions we can draw from the data. A substance that is not problematic under one set of
conditions may become problematic if the conditions change. This participant maintained that
attempts to justify a black-and-white conclusion that a substance is or is not neurotoxic are likely
to obscure this very important truth.
Another participant suggested that an important part of this debate is who the guidelines
are intended to assist. If, as Dr. Wood had suggested, many non-neuroscientists will be using
the guidelines, this person thought it very important to provide more specific guidance about
how to interpret different types of data. When an effect becomes adverse; when it becomes
neurotoxic, and how either adversity or neurotoxicity relate to exposure scenarios are all
questions that need to be addressed in more detail if non-neuroscientists are the main audience
for these guidelines. ,
At the conclusion of the panel's discussion, the Chair recognized an observer who had
requested the opportunity to comment on the draft guidelines. This individual recommended
that the labeling of any substance as a neurotoxicant be approached very judiciously, since this
label is likely to be taken very seriously by the general public. In this regard, the observer
recommended that the panel confine its discussion to effects of a chemical on nervous system
endpoints, which she distinguished from indirect effects of a chemical on behavior or other
more functional endpoints. She maintained that subchronic and chronic general toxicity studies
should be adequate t6 protect endpoints involving secondary or indirect effects of a chemical on
the nervous system. She predicted that an overly broad definition of neurotoxiciry will produce
a climate in which everything is seen as potentially neurotoxic. If this occurs, industry will be
less likely to explore the interaction of a chemical with the nervous system, since even a small
effect would cause the compound to be classified as a neurotoxicant. ,
4-25
-------
After thanking the observer for her comments, the Chair reminded participants that the
main purpose of the workshop is to provide a scientific peer review of the draft neurotoxicity
guidelines. Toward this end, he suggested that a number of important issues had been raised
during the morning's discussion that could be addressed in more detail in each panel's individual
deliberations. At the same time, it would be important for the group to maintain a sense of its
overall mission, particularly during the plenary sessions that would serve to summarize the
conclusions and recommendations emerging from each panel's efforts. Following several
procedural questions from workshop participants, he announced the time and place for each of
the afternoon panel meetings, and the Opening Plenary Session was adjourned.
4-26
-------
5. WORKGROUP REPORTS
5.1 TRANSIENT AND PERSISTENT EFFECTS PANEL
John L. O'Donoghue, V.M.D., PhJD., Workgroup Chair
The Workgroup met on June 2 to discuss the Draft Guidelines, focusing mainly on issues
related to transient and persistent effects of substances that act on the nervous system. On the
following day, the Direct and Indirect Workgroup met and continued the discussion of topics
relating to both areas. --,-.
Overall, it was apparent that the Agency had done a considerable amount of work in
putting the Draft together and had carefully considered the scientific issues relevant to
establishing risk assessment guidelines for neurotoxicity. The Workgroup was in agreement that
neurotoxicity is an important endpoint for health effects evaluation and that development of
risk assessment guidelines is appropriate.
The Workgroup was of the opinion that some of the terminology used in the Draft could
be clarified by the addition of a lexicon of terms. This lexicon might reduce or eliminate the
ambiguity sometimes encountered in discussions about neurotoxicity risk assessment. The
Workgroup has provided some definitions for the Agency?s consideration in Table 1.
The Workgroup was concerned that the purpose of the guidelines and how they will be
used were not explicitly stated in the Draft. Discussions about neurotoxicity frequently involve
issues requiring expert judgment. The Workgroup recommends that the Draft indicate that
experts in the field of neurotoxicology should be involved in making decisions about the
potential neurotoxicity of agents. If the guidelines are to be used by individuals who are not
expert in the relevant neurosciences, the Draft should recommend that the risk assessor consult
with an expert when dealing with situations that become ambiguous.
The Workgroup recognized that while the Draft followed the basic NRC approach to
risk assessment, it focused almost exclusively on the Hazard Identification Step (Section 3, 38
pages) and provided less guidance on Dose Response Assessment (Section 4, 4 pages), Exposure
5-1
-------
Table 1. Terms Used to Describe Neurotoxicants or Their Effects
Transient effects are temporary, fleeting in time, or short-lived. Their existence is measured
in minutes, hours, or perhaps a few days. Their duration is frequently related to the
pharmacokinetics of the causal agent and its presence in the body.
Persistent effects continue for a period of time which exceeds the pharmacological life span
of the causal agent.
Reversible effects are those that can be corrected, allowing the organism to return to its
original state.
Irreversible effects are those that cannot be corrected, resulting in a permanent change in the
organism.
Latent effects are those that occur at a time distant from the last contact with the causal
agent.
Progressive effects are those that continue to worsen even after the causal agent has been
removed.
Residual effects are those that persist beyond a recovery period. These effects may range
from obvious functional or structural deficits to subtle changes that may become evident only
at a later stage of life or when the individual is further challenged.
Occupational neurotoxicants are those agents that produce adverse effects on the nervous
system under conditions of exposure which occur in the workplace.
Environmental neurotoxicants are those agents that produce adverse effects on the nervous
system under conditions of exposure which occur in the ambient environment.
Primary neurotoxicants are those agents that do not require metabolism prior to interacting
with their target sites in the nervous system.
Secondary neurotoxicants are those agents that require metabolism prior to interacting with
their target sites in the nervous system.
Direct neurotoxicants are those agents or their metabolites that produce their effects
primarily by interacting directly with target sites in the nervous system.
Indirect neurotoxicants are those agents or their metabolites that produce their effects
primarily by interacting with target sites outside of the nervous system. This interaction then
secondarily results in damage to the nervous system. Indirect effects should be differentiated
from remote effects on the nervous system, which occur when the effects of a chemical on
tissues outside of the nervous system have their principal expression through the nervous
system.
5-2
-------
Assessment (Section 6, 2 pages), and Risk Characterization (Section 7, 2.pages). Expansion of
the Draft to include more guidance in these other areas should be seriously considered.
The Workgroup was concerned about Section 5 of the guidelines, which considers the
adequacy of available information for hazard assessment and dose-response assessment. The
categorization scheme presented in the Draft was seen as inappropriately two-dimensional,
allowing only a yes or no answer (neurotoxic or not neurotoxic). The particular concern was
that, while there are some agents that are clearly neurotoxic, there are also many agents that
produce sonie perturbation in the function or structure of the nervous system that should not be
considered neurotoxicants. The range of classifications needs to be enlarged to reflect the fact
that there may be adverse functional consequences of exposure to an agent that should be dealt
with by the Agency outside of these guidelines, and that there may be some functional
consequences of exposure to an agent that should not be considered adverse at all. Also, the
category of agents for which there is sufficient evidence to conclude that they are not neurotoxic
should be more explicitly identified.
The group discussed the importance of exposure level in determining whether an effect
is neurotoxic. For example, one participant pointed out that he would be concerned about a
work environment that contained 100 ppm of n-hexane for a year, but he would be less
concerned about a work environment that contained 1000 ppm for a brief period (minutes).
The Draft should discuss the importance of distinguishing between effects seen at high and low
exposure levels. There should also be some discussion of the need for caution in trying to
extrapolate effects seen at high exposure levels to real world exposure scenarios. The opinion
was voiced that, rather than considering chemicals neurotoxic, we should consider chemicals to
have the potential for neurotoxicity at certain dose levels and under certain exposure conditions.
The Workgroup discussed at length the time course of effects associated with exposure
to environmental agents. The group concluded that, in determining whether or not the
consequences of an exposure constitute neurotoxicity, the first consideration is whether or not
the effect is adverse in a dose-dependent manner. The group agreed that consideration of an
observed effect should include a multifactorial analysis of features that contribute to adversity
(Figure 1). The group decided that the severity of an effect, the length of the exposure^
5-3
-------
)
on
CO
5-4
-------
necessary to produce an effect, and, importantly, the exposure concentration all need to be
considered in determining whether an effect should be considered adverse.
Based on its discussion of degrees of adversity, the Workgroup developed levels of ' -
concern for classifying effects observed in health effects studies. The highest level of concern
was based on two criteria: the effect must be irreversible; and the effect must involve a clear or
demonstrable change in either the structure or function of the nervous system at some time
during the life span of the species under consideration. The second level of concern included
effects that are slowly or incompletely reversible and that clearly or demonstrably impair the
structure or function of the nervous system at some time during the life span. The third level of
concern included effects that are rapidly reversible and that clearly or demonstrably impair the
structure or function of the nervous system at some time during the life span. Structural effects
were defined as morphological changes occurring at any level of nervous system organization.
Functional changes were defined to include both electrophysiological and behavioral effects.
While the group agreed that only an irreversible "change" had to be demonstrated at the
highest level of concern, the majority of the group thought that some "impairment" of structure;
or function should be demonstrated at the lower levels of concern. There was a minority
opinion that considered any "change" in structure or function, particularly if non-volitional,
sufficient for an effect to be considered adverse at the lower levels of concern.
The group recommended that dose-response relationships and exposure conditions be
evaluated before deciding whether the evidence is sufficient to consider an agent a
neurotoxicant. -
Based on these considerations, the group decided that reversible effects could be
considered adverse if they clearly or demonstrably impair the structure or function of the
nervous system in a dose-dependent manner and if they occur at relevant exposure levels.
Issue: The Workshop Draft concludes that both reversible and irreversible effects of
chemicals on the nervous system should be considered adverse. , ;
5-5
-------
The Workgroup agreed that irreversible effects should be considered adverse, if they
involve a clear or demonstrable change in either-the structure or function of the nervous system.
The group also agreed that reversible effects could be considered adverse if they clearly or
demonstrably impair the structure or function of the nervous system. There was a minority
opinion that any "change" in the structure or function of the nervous system could also be
considered "adverse." For an adverse effect to be considered evidence of neurotoxicity, further
considerations such as dose-response relationships and exposure conditions need to be taken
into account.
Issue: The nervous system contains billions of cells wired in complex patterns and is
known to be resilient to environmental and toxicological insult by a process known as compensation or
adaptation.
There was agreement that healing in the nervous system can occur to varying degrees
depending on the seriousness of the impairment. When structural damage is incompletely
reversed, functional changes that allow the individual to adapt to its environment or compensate
for the residual damage may or may not occur.
Issue: Once damaged, nerve cells have limited capacity for regeneration.
There was agreement that once neurons are killed or central neural processes are
severely damaged, regeneration of the cell bodies is ruled out and therefore repair is likely to be
incomplete. There was also agreement that the capacity for repair is not as limited (and can, in
fact, be complete) if effects do not involve the destruction of neurons or central neural
processes.
Issue: Apparent recovery actually represents activation of reserve capacity, decreasing
remaining potential adaptability.
There was agreement that the reversibility of effects resulting from cell death or from
the destruction of cell processes may represent an activation of repair capacity, decreasing
5-6
-------
future potential adaptability, but that this is not necessarily true for agents that operate by other
mechanisms of action.
Issue: Traditionally, effects of toxicants are considered to be persistent or long-lasting,
while pharmacological effects are considered to be transient or short-acting.
The Workgroup recognized that there'should be different levels of concern for different
types of adverse effects. There was agreement that it is important to distinguish between agents
that clearly damage the nervous system and those that do not. The group also agreed that
adverse effects that might not be considered indicative of neurotoxicity should not be ignored,
but rather that these effects should be considered under a different heading and not regulated
as neurotoxicants. In some cases these effects have been referred to as "pharmacological
effects" to distinguish them from neurotoxic effects. There was no agreement on what to call
these other effects, but a clear concern was expressed about lumping large numbers of chemicals
under the term "neurotoxicant," especially when the evidence of neurotoxicity is obtained in
experimental situations that involve exposure .to high concentrations of the toxicant.
Issue: An effect that appears to be transient in an unchallenged organism may be revealed
as long-lasting through an environmental or pharmacological challenge. . ..
This issue was only briefly discussed. There was a consensus that residual effects of a
chemical may appear to have resolved during a recovery period yet re-emerge as significant
when the individual ages or is later challenged. Transient effects that are recoverable and that
do not result in residual lesions would not be expected to be revealed as long-lasting through
either an environmental or pharmacological challenge.
Issue: It is not known whether transient effects observed following developmental exposure
should be evaluated at specific points in the life span.
This issue was discussed only briefly because the Workgroup was not aware of any data
that could be used to identify specific time points in the life span that should be examined. This
issue probably needs additional research before it can be adequately addressed.
5r7
- V '
-------
5.2 DIRECT AND INDIRECT EFFECTS PANEL
Barry Wilson, Ph.D., Workgroup Chair
The panel reaffirmed that chemicals that directly affect the nervous system should be
subject to the proposed guidelines. Generally, it was accepted that the data submitted to the
Agency would usually consist of multiple tests at more than one level of the nervous system,
often including behavioral, biochemical and histopathological studies of one or more animal
species. Human studies were not discussed in detail. Even so, several scientists did not agree
with the idea that epidemiological studies by themselves are sufficient for risk assessment
purposes.
The recommendations of the panel were based upon an appreciation of the difficulties
facing a risk assessor who may not have had extensive training in neurotoxicology or
neurobiology. The nervous system is extraordinarily complex; on the one hand it exhibits a
remarkable plasticity in its ability to adapt to many stimuli while, on the other hand, it may be
extremely sensitive to damage by a number of chemicals. Integration of the nervous system with
the rest of the body makes it difficult to sort out specific from non-specific effects, and the role
of the nervous system in maintaining homeostasis makes it hard to decide when responses have
exceeded normal limits. The panel cautions the risk assessor that simple litmus tests usually
cannot be used either to classify a chemical as "neurotoxic" or to dismiss a compound as
harmless.
5.2.1 Indirect Effects
Although the premeeting comments revealed little agreement as to whether chemicals
that do not act directly on the nervous system could be considered neurotoxic, the panel agreed
that some could. A number of examples were provided by panel members. These included
spasms in blood vessels (such as those caused by cadmium) and other effects on the
neurovascular system; disruption of the oxygen-carrying capacity of the blood (such as that
caused by carbon monoxide); perturbation of metabolic pathways (such as those caused by
5-8
-------
dichloroacetate); rhabdomyolysis (such as that induced by naphthalene), peripheral neuropathy
(such as that induced by cyanide), and encephalopathy (such as that occurring secondary to
alcoholic damage to the liver). Other agents may exert indirect effects on the nervous system by
affecting membranes, causing post-synaptic or sarcolemmal blockade, producing functional
denervation of a target organ, or causing damage to the vertebral column.
In general^ the panel agreed that agents affecting target organs of the nervous system
(e.g. muscle), should be of serious concern, as should systemic toxicants to which the nervous
system is peculiarly sensitive (e.g., metabolic poisons). Since their effects are often expressed
through the nervous system, these substances should be subject to the proposed guidelines.
However, the panel warned that the search for indirect effects can lead to a long list of possible
hazards, exposing the assessor to the risk of "analysis paralysis." The panel noted that one
criterion for accepting an indirect effect,as "neurotoxic" is whether the chemical produces a
special effect on the nervous system, regardless of its other effects on the rest of the body. In
many cases, however, the chain of events leading to toxicity may not be fully understood, as in
the Spanish rapeseed oil incident. r ,
5.2.2 Adverse Effects
The discussion of what constitutes an adverse effect vis-a-vis the nervous [System, which
began in Panel 2, continued in Panel 3. The finalproposal categorized adverse effects on the
nervous system into a three-tiered hierarchy encompassing both direct and indirect effects. The
effects may be either irreversible or reversible (i.e., recoverable) at the cell and organ levels.
For the purposes of risk assessment, an adverse effect was operationally defined as "a
demonstrably recognizable and dose-dependent impairment in the structure or function of the
nervous system arising at any stage in the life history of the organism." Effects in Category I
deviated slightly from this definition, however, since irreversible changes in structure or function
were considered adverse whether or not they could be established as injurious to the animal in
other respects.
5-9
-------
Category I: irreversible effects on the nervous system, whether or not identified as
impairments (e.g., as caused by methyl mercury).
Category II: slowly or incompletely reversible effects on the nervous system (e.g., as seen
following exposure to certain organophosphate esters).
Category III: rapidly reversible effects on the nervous system (e.g. as seen following
exposure to neuroactive solvents).
The term "reversible" refers to situations in which full recovery of form and function
occurs. Effects in all three categories must be dose-related to be considered evidence of
neurotoxicity.
The panel emphasized that chemicals per se are not neurotoxic; it is the action of a
chemical at a particular concentration in the nervous system or in other parts of the body that
ultimately results in neurotoxicity. Some panelists did not accept restricting "adverse effects" to
impairments; they argued that any significant change in the nervous system is an "adverse
effect." The panel advises risk assessors to avoid being too exclusive by narrowly defining
"impairment" and to avoid being too inclusive by failing to recognize that some "changes" fall
within the normal physiological range.
were:
The panel assigned relative levels of concern to the three levels of adverse effects. They
Category I:
Category II: ""to*
Category III:
5-10
-------
These weightings were not equivalent to uncertainty factors; instead, they were intended
to express the relative importance that the panel attached to irreversible versus recoverable
changes in the nervous system.
' ' The panel thought of the hierarchy of adverse effects as a first step in assessing potential
neurotoxicity. The goal was to help the risk assessor separate long-term damage to the nervous
system from short-term, readily recoverable effects. Although the initial level of concern is
indicated by the position of an effect in the tier, the relative severity of any given effect was to
be assessed later in the process, when dose-response curves, pharmacokinetic data and
structure/activity relationships are specifically considered.
The panel also discussed issues related to hazard characterization, in which "real world"
considerations play a role. For example, there was some discussion of a compound that causes
Organophosphate-Induced Delayed Neuropathy (OPIDN), but only at doses above the lethal
level. To see these delayed effects at all, the animal must be protected from the acute toxicity
of the compound by pre-treatment with atropine. On the basis of these data, the chemical
would be considered to fall within Category I, since the effects were irreversible. However, the
assessor might not consider it to be a serious risk after subjecting the data to hazard
characterization. Knowledge of the potential for neurotoxicity at very high doses might,
however, prompt the risk assessor to request additional study of this compound in order to
determine whether chronic exposures at sublethal levels might also produce damage to the
nervous system. .
The statement in the document that any compound will be neurotoxic at a high enough
dose because it is lethal to the organism was not considered germane, and was set aside during
the considerations. Some panelists did not agree with this statement regardless of the context in
which it would appear. The panel was more interested in directing a risk assessor to events that
occur at relatively low doses than in focusing attention on those effects that occur at lethal
levels.
5-11
-------
The panel chair briefly discoursed on Neuropathy Target Esterase (NTE), both to set
the record straight in the premeeting comments and to suggest OPIDN as one of the possible
models for risk assessment that the Agency could use to illustrate the panel's proposals.
Irreversible inhibition of NTE is accepted as a biomarker of exposure to organophosphate
agents that may cause OPIDN. Recent findings suggest that inhibitions of NTE as low as 40%
may be associated with lesions indicative of OPIDN in the spinal cords of hens. Indeed, lesions
have been detected even after diisopropyl fluorophosphate (DFP) treatment at doses that do
not produce ataxia.
5.2.3 Unresolved Issues
During the panel session and the plenary meeting that followed, several additional issues
were raised. Although time constraints precluded resolution of these issues, they are important
for the Agency to consider. One was concern about the role envisioned for behavioral tests in
the risk assessment process. Behaviorists are well aware that not all behavioral changes are
evidence of neurotoxicity. For example, they are very careful not to use unhealthy animals in
neurobehavioral studies, recognizing that the responses of these subjects will be abnormal. At
the least, the risk assessor should be aware of the limitations of behavioral tests and be chary
about labeling a compound as neurotoxic based on behavioral evidence alone, especially if the
evidence is obtained from only a few tests. Some panelists felt that a second category such as
"behavioral toxicant" might be needed to handle compounds with behavioral effects that are
thought to stem from systemic or, at any rate, non-neural events. Others felt strongly that this
category would not be useful.
Another category that was discussed but not adopted was one that would indicate
suspected but unproven neurotoxicants; this category was discussed as being analogous to the
Scottish verdict of "Not Proven." A majority of the panel felt this category to be an unnecessary
addition, arguing that such chemicals would be addressed somewhere else in the risk assessment
process.
5-12
-------
The importance of factors such as tolerance, repeated and chronic exposures and
differential sensitivity during different stages in the life history of an animal were raised but
were not discussed in any detail. An animal may respond to a chemical in different ways
depending on the stage of its life history, from embryonic life, through neonatal growth,
adolescence, maturity, and senescence. Risk assessors need to be aware of the special
sensitivities that may be present at different stages of life. Similarly, the issue of the effects of
mixtures was raised, but not discussed in detail. "Real world" scenarios frequently involve more
than one agent, and it is important to consider synergisms, especially if one of the agents is an
inducer of liver enzymes. Exposure to a chlorinated hydrocarbon that increases liver P450 and
monoamine oxidase (MAO) activities could affect the toxicity of other chemicals that require
bioactivation. For example, the neurological problems found among workers in a plant
synthesizing the neurotpxic agent leptothos were never satisfactorily attributed to exposure to
leptothos, to the solvents used during the synthetic process, or to a combination of leptothos
and the solvents. . , ;."".../: ; -".'.. ' '.
5.2.4 Closing ,
Agency personnel familiar with the carcinogenicity/mutagenicity risk assessment process
may be struck by the apparent looseness of the guidelines recommended here. This is more due
to the nature of the nervous system, rather than to the inability of neuroscientisits to agree.
What constitutes the normal limits of response to a chemical is different for different parts of
the nervous system; for example, the resting potential of a nerve is regulated closely, while the
avoidance response of an animal to a stimulus may be more variable. Labeling a compound as
a potential "neurotoxicant" is likely to have a dramatic effect on potential consumers, and the
risk assessor needs to obtain the best data and background information possible before drawing
a conclusion.
5-13
-------
5.3 ANIMAL-HUMAN EXTRAPOLATION PANEL
Shayne C. Gad, Ph.D., D.A.B.T, Workgroup Chair
The panel charged with reviewing and commenting on the animal-human extrapolation
aspects of the EPA proposed guidelines for neurotoxicity risk assessment used as its working
matrix the statements presented in the issues paper provided before the review session. The
panel's comments and suggestions on the points contained in the issues paper were as follows:
5.3.1 Conclusion: With an adequate animal database, as defined in the draft guidelines, risk
assessment judgments, even in the absence of human data, may be scientifically valid.
Conclusions and Suppositions Used in Reaching This Position
Substances producing neurotoxicity in humans also result in neurotoxicity in
other species.
Panel members generally agreed with this statement, but only with some qualifications
related to the importance of differences in routes of exposure and times to effect. There are
appropriate animal models for neurotoxicity, but not all species will be similarly affected all the
time. Currently, animal models and studies constitute the best and most likely form of hazard
identification available.
Compared with human studies, animal studies are more readily available and can
provide more precise dose information and better control of environmental
factors.
There was substantial agreement on this point. However, limitations of individual
models may make it difficult to differentiate neurotoxicity due to direct effects on the nervous
system from neurotoxicity secondary to other target organ effects.
Many diagnostic procedures employed to evaluate neurotoxicity in humans have
corresponding animals models.
5-14
-------
There is a divergence of opinion on the meaning of "many" in this statement. Diagnostic
procedures requiring verbal or written interaction with a "patient" may not universally fit this
description. In addition, it must be noted that rats are not little humans with tails. Also,
selection of the appropriate species must be carefully considered. Finally, it should not
necessarily be expected that there will be a one-to-one concordance between endpoints observed
in animals and diagnostic endpoints that are relevant to humans.
The range of uncertainty factors used to extrapolate risk from animal data to
humans for other endpoints of toxicity are applicable to neurotoxicity risk
i- assessment.
There is substantial agreement with this contention from the point of view of categories
of factors. At the same time, it is clearly hoped that more precision might be achieved. Also,
there is a notable lack of comfort with any set magnitude for these uncertainty factors. The
more completely understood the biological basis of observed effects, the more likely it is that
data may be made more precise and its relevance to human risks may be clarified/This will
lead to refinement of uncertainty factors, which under no circumstances should be arbitrary
numbers that fail to take all available information into account.
Areas of Special Focus
The behavioral domains for animals and humans are similar. However, the
complexity of behavior may vary precluding an exact concordance of effects.
Factors such as differences in metabolism can result in differences among species
in sensitivity to a compound.
There was substantial agreement on this point. The statement, however, should not be
read as focusing solely on metabolism. A lot of other factors not only lead to differences in
sensitivity among species but also among various human "populations" (based on age, nutrition,
race, health status, sex, etc.).
5-15
-------
The most sensitive species may not be the species most relevant to predicting
risks in humans. In the absence of information to the contrary, however, the use
of the most sensitive species is warranted.
There was substantial agreement on this point but its relevance is unclear. The issue of
which human population we are "modeling" or concerned about protecting was also raised.
Other Issues
No general guidance is offered on the determination that findings from animal models
are relevant for the risk assessment process. Such factors as sensitivity, stability of the model,
availability of techniques and baseline data, and overall relevance of the model for the endpoint
of interest should be considered.
Substantial disagreement exists on the relevance and utility of data obtained from in
vitro models in the human risk assessment process. Findings from adequate in vivo studies
should take precedence over in vitro findings. In the face of mixed in vitro and in vivo findings,
the generally accepted relative strengths, as summarized in Table 2, should be considered.
The value of an integrated, full range of data (i.e., neurochemical, pathology, behavioral
and physiological data) needs to be explicitly recognized in the guidelines.
The guidelines should also address where the risk assessment process should begin in
terms of whether uncertainty factors should be applied to the ED10 or NOAEL for the endpoint
of concern.
5-16
-------
Table 2. Potential Advantages and Disadvantages of In Vitro Toxicity Tests
Advantages
1. Avoid complications and potential confounding or "masking" findings of in vivo
studies. '.-:'...
2. Exposure levels and conditions at target sites can be better controlled.
3. Test condition standardization can be better than for in vivo studies.
4. Reduction in animal usage and/or in pain to experimental animals.
5. Ability to directly study some target tissue effects on a real time basis.
6. Reduced requirements for text agents.
Disadvantages
1. Lack of ability to evaluate longer term effects.
2. Limited ability to simulate and evaluate integrated organismic level effects.
3. May not reflect influence of agent absorption, distribution, metabolism, and
excretion effects.
5-17
-------
5.4 BEHAVIOR PANEL
John L. Orr, Ph.D., D.A.B.T., Workgroup Chair
Unlike the other panels, which were charged with issues that are important in many
areas of toxicology (animal-to-human extrapolation, direct and indirect effects, and transient and
persistent effects), the focus of the behavior panel was on the set of functional endpoints that
reflect the status of the nervous system. The conclusions, recommendations, and statements of
the panel touch on issues involving the definition of adversity, the importance of transient
effects, the utility of behavioral data, and interpretation of behavioral data in the presence of
other evidence of toxicity.
5.4.1 Data Submission Scenario
To focus the panel's discussion, data scenario was envisioned in which a set of data
arising from a Toxic Substances Control Act (TSCA) neurotoxicity test rule was submitted. The
data package was defined to contain a report with behavioral data arising from the functional
observation battery (FOB), motor activity tests (MA), and neuropathology. This scenario served
as a useful device to frame the situation and focus discussion, since it was assumed to represent
a minimal data scenario in which the risk assessment guidelines might be used. The use of this
scenario to focus the discussion does not imply that the panel felt the scene-setting scenario to
be optimal for establishing NOAELs or conducting risk assessment.
5.4.2 Issues Paper Conclusions
The panel worked through the items in the Issues Paper and reached consensus with the
conclusion that". . .. behavioral changes can provide evidence of neurotoxicity in the absence of
additional data."
The panel had some recommendations to fine-tune the conclusions and suppositions
listed in the Issues Paper. The panel concluded that:
Behavior can be a sensitive indicator of toxicity
5-18
-------
Behavioral evaluations have played an important role in efforts to understand
brain function
The behavioral domains for animals and humans are similar, but the complexity
of behaviors may preclude an exact concordance of effects
Data from measures of behavior can be available prior to data from physiological
or morphological studies -, , '
The above four conclusions apply to both adult and developing organisms.
5.4.3 Response to Areas of Special Focus
The panel discussed the three areas of special focus in the Issues Paper: non-specific
effects, high-dose effects, and the possibility of indirect effects. The panel endorsed the
following paragraph in the specific context of the Draft Neurotoxicity Risk Assessment
Guidelines: :, : '
Behavior is an indication of the well-being of an organism. Changes can
arise from a direct effect on the nervous system or indirectly from effects on
other physiological systems. It is appropriate to use behavioral changes for
the determination of reference doses, but such changes may not be sufficient
to establish an agent as a neurotoxicant. Our understanding of the
interrelationship between systemic toxicity and behavioral changes is limited,
(e.g., the relationship between changes in body weight and activity).
Interpretation of such relationships should include consideration of factors
including experimental design, dose-effect information, chemical class, and
other relevant toxicological information. The presence of systemic toxicity
complicate, but does not preclude interpretation of behavioral changes as
evidence of neurotoxicity.
The panel felt that the draft guidelines language should be changed on pages 27 arid 32
(and elsewhere as necessary) to reflect the conclusions expressed in the above paragraph. This
could be accomplished by inserting the word "necessarily," for example:
» On page 27, line 17: ". .. data are not necessarily, interpreted. . ."
On page 32, line 13: ". . . not necessarily evidence.
5-19
-------
5.4.4 Data Interpretation
With respect to the interpretation of data, the panel concluded that interpretation of
FOB data should include an evaluation of the pattern of effects, consistency within functional
domains, degree of replication, severity of effects, and statistical considerations of multiple
statistical comparisons. Some panel members felt that these factors were reasonable
considerations for any neurotoxicology data set, but others felt that more discussion would be
required to reach consensus.
5.4.5 Conclusion
The panel considered behavior a major dimension comparable to the physiological or
morphologic dimensions. The feeling was not that one dimension was "better" or more
important, but that any could serve as an appropriate basis for categorization. The panel agreed
that behavior is a reflection of the status of the underlying physiology and morphology, but that
behavioral studies can reveal pathology that does not happen to be sampled by the other types
of studies in a given data set. An example would be an alteration of, neurotransmitter levels
that is reflected in behavior but could be difficult to capture in non-behavioral studies (e.g.,
using the TSCA neurotoxicity battery).
5.4.6 Categories for Classification as a Neurotoxicant
Some members of the panel had a concern about the possibility of overlabeling and
calling everything a neurotoxicant. Other panel members felt this was a consideration more
appropriate to a discussion of risk characterization than a discussion of hazard identification.
Panel members felt that there were a number of kinds of toxicologic information that
would help one decide whether a compound is neurotoxic at certain doses. The panel members
felt that the problem of categorization is difficult but not impossible in the presence of systemic
effects.
5-20
-------
As with other non-cancer ehdpoints (e.g., the EPA Guidelines for Developmental Risk
Assessment), there is difficulty in separating the hazard identification stage from the
dose-response assessment stage in neurotoxicity risk assessment. This occurs in part because
evidence of a dose-effect relationship may be part of the evidence that the effect is compound-
related. A second, perhaps more important reason is that the toxicity profiles of neuroactive
agents may be different at different dose levels or chronicities.
The panel emphatically did not agree with the concept that one should have
morphologic or physiological "confirmation" before categorizing an agent as a neurotoxicant.
On the other hand, they did not feel that any behavioral change is sufficient to necessarily
categorize an agent as a neurotoxicant. The opinion was that any behavioral change increases
the suspicion of potential neurotoxicity and that evaluations should be made in the light of all
available data.
5.4.7 Tangibility of Behavioral Data
In the plenary session, one observer complained that behavior was not tangible and that
he could not "touch" it. Another panel member asked if this individual would similarly question
the concept of blood pressure. This illuminates an important point. In casual and even learned
discussion "behavioral changes" are discussed as if changes in behavior are the object of
attention. These are indeed, if not directly observed by the speaker, intangible. However, the
phrase "behavioral changes" or "changes in behavior" is really a short form of saying "differences
in behavioral data. "The behavior may be ephemeral, but the data is as tangible as any other
data. A graph of behavioral performance is as real as any photomicrograph from a
histopathologic examination. Behavioral data from an appropriately designed experiment is
analogous to a well-prepared anatomic specimen and is just as tangible. The problem is not
with behavioral changes, per se, but with the assessment of their adversity.
5-21
-------
-------
APPENDIX A
AGENDA
-------
-------
U.S. Environmental Protection Agency
NEUROTOXICITY RISK ASSESSMENT GUIDELINES
PEER REVIEW WORKSHOP
June 2-3, 1992
Omni Georgetown Hotel
Washington, DC
FINAL AGENDA
Monday. June 1. 1992
7:30-9:OOPM Early Registration/Check-In
Tuesday. June 2. 1992
7:30-8:30AM
8:30AM
8:45AM
9:15AM
9:45AM
10:15AM
10:30AM
11:OOAM
11:30AM
12:00-1:15PM
.Registration
Opening Plenary Session: Preliminary Comments/
Definition Of Issues
Dr. William Greenlee, Purdue University
Welcome and Objectives
Dr. William Wood, USEPA, Risk Assessment Forum
Neurotoxicity as an Endpoint (Panel 1)
Dr. William Greenlee
Transient and Persistent Effects (Panel 2)
Dr. John O'Donoghue, Eastman Kodak Company
Direct and Indirect Effects (Panel 3)
Dr. Barry Wilson, University of California
Break
Animal-Human Extrapolation (Panel 4)
Dr. Shayne Gad, Becton Dickinson Research Center
Behavior (Panel 5)
Dr. John Orr, Southwest Research Institute
Observer Comments
Dr. William Greenlee
Lunch
(over)
A-l
-------
Tuesday. June 2. 1992 (continued)
Work Group Break-Out Sessions
1:15PM
3:OOPM
3:15PM
4.-30PM
5:30PM
Discussion Panels 2 and 4
Transient and Persistent Effects
Dr. John O'Donoghue
Animal-Human Extrapolation
Dr. Shayne Gad
Break
Panels 2 and 4 (continued)
Plenary Reports and Discussion
Transient and Persistent Effects
Dr. John O'Donoghue
Animal-Human Extrapolation
Dr. Shayne Gad
Adjourn
Wednesday. June 3, 1992
8:OOAM
10:OOAM
10.-15AM
11:15AM
12:15PM
Work Group Break-Out Sessions
Discussion Panels 3 and 5
Direct and Indirect
Dr. Barry Wilson
Behavior
Dr. John Orr
Break
Panels 3 and 5 (continued)
Plenary Reports and Discussion
Direct and Indirect
Dr. Barry Wilson
Behavior
Dr. John Orr
Lunch
(continued)
A-2
-------
Wednesday. June 3. 1992 (continued)
1:30PM Closing Plenary
Final Reports
Recommendations to EPA
3:OOPM Adjourn
A-3
-------
-------
APPENDIX B
LISTS OF PARTICIPANTS AND OBSERVERS
-------
-------
U.S. Environmental Protection Agency
NEUROTOXICITY RISK ASSESSMENT GUIDELINES
PEER REVIEW WORKSHOP
June 2-3, 1992
Omni Georgetown Hotel
Washington, DC
FINAL PARTICIPANTS LIST
Deborah Cory-Slechta
University of Rochester
School of Medicine & Dentistry
P.O. Box EHSC
Rochester, NY 14642
Wayne Daughtrey
Staff Toxicologist
Exxon Biomedical Sciences, Inc.
Mettlers Road - CN 2350
East Millstone, NJ 08875-2350
Shayne Gad
Director, Medical Affairs Technical Services
Becton Dickinson Research Center
21 Davis Drive - P.O. Box 12016
Research Triangle Park, NC 27709-2016
Michael Gill
Bushy Run Research Center
Union Carbide Corporation
6702 Mellon Road
Export, PA 15632
John Glowa
National Institute of Heajth
Building 14D - Room 311
LMC/NIDDK/NIH
Bethesda, MD 20892
William Greenlee
Professor and Head
Department of Pharmacology & Toxicology
Purdue University
1334 Robert E. Heine Pharmacy Building
Room202A
W. Lafayette, IN 47907-1334
G. Jean Harry
National Institute of Environmental
Health Services
P.O. Box 12233 (MD El-02)
Research Triangle Park, NC 27709
Donald McMillan
Department of Pharmacology & Toxicology
College of Medicine
University of Arkansas for Medical Sciences
4301 West Markham Street
Little Rock, AR 72205
John O'Donoghue
Director
Corporate Health &
Environment Laboratories.
Kodak Park, B-320
Eastman Kodak Company
Rochester, NY 14652-3615
JohnOrr
Principal Scientist
Department of Biosciences & Bioengineering
Southwest Research Institute
P.O. Drawer 28510
6220 CulebraRbad
San Antonio, TX 78228-0510
Thomas Sobotka
Neurobehavioral Toxicology Team
Center for Food Safety & Applied Nutrition
U.S. Food & Drug Administration/Mod 1
8301 Muirkirk Road (HFF162)
Laurel, MD 20708
B-l
-------
Peter Spencer
Director,
Center for Research on Occupational &
Environmental Toxicology
Oregon Health Sciences University
3181 Southwest Sam Jackson Park Road
Portland, OR 97201-3098
Barry Wilson
Professor
Department of Environmental Toxicology
3202 Meyer Hall
University of California
Davis, CA 95616
B-2
-------
U.S. Environmental Protection Agency
NEUROTOXIGITY RISK ASSESSMENT GUIDELINES
PEER REVIEW WORKSHOP
June 2-3,1992
Omni Georgetown Hotel
Washington, DC
FINAL OBSERVERS LIST
Martha Beauchamp
Principal
Sandus International
1616 P Street, NW - Suite 410
Washington, DC 20036
William Boyes
Chief, Neurophysiologjcal
Toxicology Branch (MD74B)
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Clark Carringtpn ,---..
Toxicologist ; -;
U.S. Food & Drug Administration
200 C Street, NW (HFS-156)
Washington, DC 20204
Hans Chen
Senior Research Pathologist
Haskell Laboratory
P.O. Box 50
Newark, DE 19714 ..
Greg Christoph ;
Senior Research Biologist
DuPont Company
Elkton Road - P.O. Box 50
Newark, DE 19714
Helen Cunny
Toxicologist
Rhone-Poulenc
P.O. Box 12014
Research Triangle Park, NC 27709
John Festa
Director, Chemical Control &
Health Program
American Paper Institute . ; ..;....
1250"Connecticut Avenue, NW - Suite 210
Washington, DC 20036
Theresa Fico .
Senior Toxicologist
FMC Corporation
P.O. Box 8
Princeton, NJ 08543
Joel Fischer
Senior Toxicologist
American Cyanamid Company
P.O. Box 400
Princeton, NJ 08543-0400 ,
John Foss
Senior Scientist , " ."'..-.
Argus Research Laboratories, Inc.
905 Sheehy Drive
Horsham, PA 19044
Christina Griffin
Delta Analytical Corporation
1414 Fenwick Lane
Silver Spring, MD 20910
James Griffiths
Associate Science Advisor
Distilled Spirits Council of the
United States
1250 Eye Street, NW - Suite 900
Washington, DC 20005-3998
B-3
-------
Jane Harris
Director of Toxicology
American Cyanamid Company
P.O. Box 400
Princeton, NJ 08543-0400
Alan Katz
Executive Director, Toxicology
Technical Assessment Systems, Inc.
The Flour Mill
1000 Potomac Street, NW - Suite 102
Washington, DC 20007
Carole Kimmel
Developmental Toxicologist
U.S. Environmental Protection Agency
401 M Street, SW (RD-689)
Washington, DC 20460
Elizabeth Lapadula
Senior Toxicologist
Texaco, Inc.
P.O. Box 509
Beacon, NY 12508
AbbyJJ
Neurotoxicology Team Leader
Monsanto Agricultural Company
645 South Newstead Avenue
St. Louis, MO 63110
Robert MacPhail
Chief Neurobehavioral Toxicologist
Health Effects Research Laboratory
Research Triangle Park, NC 27711
Amal Mahfouz
Senior Toxicologist
U.S. Environmental Protection Agency'
401 M Street, SW (WH-586)
Washington, DC 20460
Suzanne McMaster
U.S. Environmental Protection Agency
410 M Street, SW
Washington, DC 20460
Victor Miller
Toxicologist
U.S. Environmental Protection Agency
9300 Brookville Road
Silver Spring, MD
Gwen Moulton
Reporter
Daily Environment Reporter
The Bureau of National Affairs, Inc.
1231 25th Street, NW
Washington, DC 20037
Mary Janet Normandy
Toxicologist
. U.S. Department of Energy
(EH-422)
Washington, DC 20585
Tim Pastoor
Manager, Toxicology
Registration & Regulatory Affairs
Agricultural Products
ICI Americas
Murphy Road & Concord Pike (DCC-II)
Wilmington, DE 19897
Fred Pontius
Associate Director, Regulatory Affairs
American Waterworks Association
6666 West Quincy Avenue
Denver, CO 80235
Cooper Rees
Senior Staff Research &
Development Toxicologist
RJ. Reynolds Tobacco Company
Bowman Gray Technical Center
Scientific Affairs (611-12/Flash 204)
P.O. Box 2959
Winston-Salem, NC 27102
J. P. Rieth
Senior Toxicologist
Rhone-Poulenc
P.O. Box 12014
Research Triangle Park, NC 27709
B-4
-------
Bill Sette
Co-Chair, EPA Work Group
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
Brendan Steuble
Delta Analytical Corporation
1414 Fenwick Lane
Silver Spring, MD 20910
Clare Stine
Technical Liaison
Risk Assessment Forum
U.S. Environmental Protection Agency
401 M Street, SW (RD-672)
Washington, DC 20460
Greg Sykes
Staff Pathologist
Haskell Laboratory
DuPont Company
Elkton Road - P.O. Box 50
Newark, DE 19714
Sarah Thurin
Reporter
Bureau of National Affairs
1231 25th Street, NW
Washington, DC 20037
HughTilson
Director, Neurotoxicology Division
U.S. Environmental Protection Agency
(MD-74B)
Research Triangle Park, NC 27711
Sandra Tirey
Associate Director
Chemical Manufacturers Association
2501 M Street, NW
Washington, DC 20037
Alberto Tottme
Staff Scientist
Karch & Associates
1701 K Street, NW - Suite 1000
Washington, DC 20016
Kamala Tripathi
Toxicologist
U.S. Department of Agriculture
Annex Building #602
300 12th Street, SW
Washington, DC 20250
Molly Weiler
Study Director in Toxicology
Hazleton Wisconsin, Inc.
P.O. Box 7545
Madison, WI 53707
Amy Wilson
Scientific Assistant
Technical Assessment Systems, Inc.
The Flour Mill
1000 Potomac Street, NW - Suite 102
Washington, DC 20007
William Wood
Associate Director,
Risk Assessment Forum
U.S. Environmental Protection Agency
401 M Street, SW (RD-689)
Washington, DC 20460
Suzanne Wuerthele
Toxicologist
U.S. Environmental Protection Agency
999 18th Street - Suite 500
Denver, CO 80202-2405
B-5
-------
-------
APPENDIX C
PREMEEHNG COMMENTS
-------
-------
U.S. Environmental Protection Agency
Office of Research and Development
Risk Assessment Forum
v>EPA
PREMEETING COMMENTS FOR
PEER REVIEW OF NEUROTOXICITY
RISK ASSESSMENT GUIDELINES WORKSHOP
June 2-3, 1992
Omni Georgetown Hotel
Washington, DC
C-l
-------
-------
TABLE OF CONTENTS
Cory-Slechta, Deborah 1
Daughtrey, Wayne C. 8
Gad, Shayne C. 12
Gill, Michael 15
Glowa, John .. 22
Greenlee, William F. ....... 30
Harry, Jean G. ; 34
McMillan, Donald . . .... ..... 36
O'Donoghue, John L. 42
Orr, John L. ....... 50
Sobotka, Tom . . 53
Spencer, Peter . . . . 60
Wilson, Barry W. . ........ .. . .... 69
C-3
-------
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Deborah A. Cory-Slechta, Ph.D.
Environmental Health Sciences Center
P.O. BoxEHSC ~
University of Rochester
School of Medicine and Dentistry
Rochester, NY 14642
Comments
pp. 4-5. Although this may have been discussed in the context of defining the terms
neurotoxicity and neurotoxicant, I wonder if inclusion of the term "non-volitional" before the
term "exposure" would assist in differentiating neurotoxicity resulting from exposures to
chemicals, etc. from the side effects of non-therapeutic compounds taken by choice for their
beneficial effects upon the user. In the case of these chemical, physical, or biological agents
regulated by EPA, the human population is not the intended population for use or exposure,
another clear distinction from therapeutic compounds. Recognizing that the side effects of
therapeutic compounds could certainly be neurotoxic, the phrase "non-volitional exposure"
carries the connotation of environmental, occupational, etc., type exposures that are essentially
unintended and not by choice.
Some aspects of the definition of neurotoxicity itself are rather vague,, in particular, the
phrase "ability to adapt to the environment." When paired with the other descriptors included
in the definition of neurotoxicity, i.e., ability to survive and reproduce, "adapt to the
environment" carries the connotation that fairly serious consequences would be warranted to
constitute adverse effects. My own perception of the phrase in this context is that it relates
again to survival, since, failure to adapt to the environment would mean failure to survive,
particularly from an ecological or ethological point of view. Does this mean that subtle changes
in behavior, which might not be of sufficient magnitude to result in failure to survive, are not
considered adverse enough to warrant the term neurotoxicity? This would seem to be at odds
C-5
-------
with the conclusions to be discussed by Panel 2 members of the Workshop. If my ability to
reach my full potential is diminished but not completely decimated, is that neurotoxicity?
Moreover, adaptation to the environment is a long-term process, and this would further suggest
that adverse effects that may be relatively transient are therefore not problematic.
p. 5. There are additional reasons for considering reversible effects to be neurotoxic
when one considers behavioral changes. First, it has been demonstrated in the behavioral
pharmacology literature that past behavioral history is an important determinant of future
behavior. This is probably best exemplified by the numerous studies that have shown that
responding can be established and maintained by response-produced shock in animals with prior
training under shock-postponement or shock avoidance schedules (see Barrett, 1986). In
addition, past behavioral history can substantially alter the effects of drugs on behavior, even
when the effects of the different behavioral history are not evident in the current or ongoing
performance. Thus, even reversible or transient changes in behavior do become part of our
behavioral history. As such, these changes can subsequently impact our future behavior and
modify the nature of interactions with other chemical agents, since many of the compounds of
concern for risk assessment may be compounds which, like drugs, act on the central nervous
system.
p. 5. The third paragraph on p. 5 states that there are five principal questions that
should be addressed, but only four (content validity, construct validity, concurrent validity, and
predictive validity) are listed.
p. 7. It would seem that case studies, since they tend to focus on a relatively few
severely affected individuals, should be able to provide a rather good description of the signs
and symptoms of toxicity, rather than a poor characterization indicated by the text.
p. 8. The sentence indicating that positive epidemiological data are generally regarded
as the most convincing evidence of the potential neurotoxicity of a chemical seems overstated,
since these are correlational and not cause-effect studies. If this statement is based on the fact
that such studies involve human subjects, then perhaps the statement should be so qualified,
since it otherwise suggests that these studies constitute the strongest science.
C-6
-------
p. 9. Numerous questions still exist with regard to the issue of the sensitivity of the test
batteries being used to examine neurobehavioral toxicity in human populations. In particular,
questions remain as to whether the effect level, as determined from the tests represents the
lower limits of sensitivity of the test itself, or the actual LOAEL of the toxicant. As stated by
Guillon and Eckerman (1986), a neuropsychological or cognitive-abilities test may be reliable
and valid over a wide range of levels of the trait measured and yet not be sufficiently sensitive
to variation within the restricted range of subclinical effects to be useful for monitoring or
screening program. The data indicating the level of sensitivity of these tests are still by and
large not available. In fact, the sensitivity issue may be far more important than that of validity,
since the levels at which any effects are observed in these tests are very likely to receive primary
consideration in the determination of risk assessment. In such a case, levels of exposure may
then be based on the outcome of tests for which the sensitivities are unknown. In other words,
if the test is insufficiently sensitive, then effects are only detectable above a certain exposure
level, even though adverse effects actually may occur at still lower exposure levels. If the
assumption that the exposure level rather than test sensitivity is the determinant of the LOAEL
of the study, then, we end up setting exposure levels too high. This entire issue deserves further
consideration. .
The reason for suggesting that the sensitivity issue is in some ways more important than
the issue of validity is that it is clear that some behavioral function is being tapped by the tests,
and the exact nature of that function may be less important to EPA than the level of exposure
associated with that effect. Granted, one must also strive to find the specific types of behavioral
functions most impacted by the toxicant in order to fully delineate adverse effect levels.
However, different toxicants affect different behavioral functions to a greater or lesser degree,
inevitably requiring the use of a test battery that crosses functions and includes components with
documented sensitivity levels-
Recent studies also raise numerous questions about the reliability of many of the tests
typically included in these batteries (Arcia and Otto, 1992). In particular, tests measuring
learning/memory were found to be of low reliability. It may be premature to use these types of
studies at the current time as a basis of risk assessment.
C-7
-------
p. 15 and Table 2. These descriptions leave the misleading impression that lead (Pb)
causes its effects via myelinopathy. The table should be better qualified, particularly the title of
the table, which indicates "with specific neuronal targets," since this is not the sole target for
lead, nor presumably the most sensitive one.
p. 17. Is it really the case that glial fibrillary acidic protein increase actually represents a
uniform response to central nervous system (CNS) injury? The list of known neurotoxic
compounds or classes of compounds associated with such effects appears to be somewhat
restricted, and exhibits some rather notable exceptions, such as lead, carbon disulfides, and
pesticides as examples (O'Callaghan, 1992). Have these agents been examined with respect to
their impact on glial fibrillary acidic protein levels, or do they simply not induce any changes?
While it may be fine to conclude that changes in this protein are indicative of cell injury and
hence neurotoxicity, the converse, i.e., lack of effect on this protein indicates no neurotoxicity,
should not be assumed and should be clearly stated.
p. 23. Section 3.2.3. This section discusses the difficulty of deciding whether a
biochemical or neurochemical change is one of neurotoxicological significance. This, of course,
is where behavioral measures become extremely useful, since if such a change is of sufficient
biological magnitude and clinical relevance, then it should manifest itself in behavior, and
behavioral processes linked to that neurotransmitter system can be examined. Perhaps mention
of such possibilities could be made here.
pp. 30-31. The discussion of motor activity seems largely insufficient. For example, it
fails to point out how the difference devices used to measure motor activity (a global measure
of behavior) can actually measure quite different aspects of motor function. For example,
some may primarily measure ambulation, while others may also include measurements of
grooming, rearing, etc. This is bound to lead to confusion in the literature, as different
investigators note different effects on motor activity since they may actually be measuring
different aspects of motor function. Also, in the testing phases for new compounds, how can
one ensure that an appropriate device was used, i.e., one that measures "critical" aspects of
motor activity, whatever those may be for that compound. Perhaps it would be more
appropriate to differentiate the specific dependent variables and, their operational definitions
C-8
-------
rather than to simply refer to this as motor activity. It seems to refer to ambulation as a subset
of motor activity. This should also provide clarity with respect to any noted differences in
results across laboratories.
p. 31. The rationale for including the statement that most neurotoxicants decrease rates
of schedule-controlled operant responding at some doses is not clear. The reason this occurs is
artificial in that one necessarily wants to include a dose that produces overt toxicity or gross
behavioral manifestations to ensure that an adequate dose-range has been covered in a study.
But, aside from that, different toxicants have very different effects upon response rates per se.
Response rations are further differentiated by examination of different schedules of
reinforcement. Even within a toxicant, effects on response rates can be quite different for a
given schedule of reinforcement when one considers dose of toxicant or duration of treatment,
as has certainly been demonstrated in studies of lead effects on schedule-controlled behavior.
This discussion of schedule-controlled behavior also fails to note that these baselines can
be used not only to measure what is termed "steady-state performance," but also can be used in
a learning and memory context. With regard to learning, one can measure, for example, the
number of sessions to acquisition of the characteristic pattern of behavior associated with the
schedule. Moreover, one can continue to impose a change in a schedule parameter (length of
the fixed-interval, size of the fixed-ratio) and measure the time or number of previously utilized
parameters, one can ask questions about how the treatment affected "memory" since
reacquisition at a particular parameter value should be faster than the first acquisition curve.
The discussion of schedule-controlled behavior also does not refer to the similarity
across a wide range of species (including humans) of the characteristic patterns of
schedule-controlled behavior. This should be pointed out, since it has particular relevance to the
issue of risk assessment in which extrapolation across species" is an important consideration.
While this aspect does not address the probability that humans will likewise show changes in
schedule-controlled behavior in response to a particular toxicant (i.e., predictive validity), it does
attest to the fact that similar behavioral processes are being evaluated across species. Similar
evaluations across species are not being conducted for extrapolations that involve comparison of
some experimental animal paradigms which purport to measure a specific behavioral function to
C-9
-------
a computer or pen-and-paper based test which purports to measure the same function in
humans.
p. 34 and Table 5. What exactly is "discriminated conditioning?" This is not a standard
term from the behavioral literature and should be replaced with the appropriate terms. Does it
refer to discrimination learning?
p. 36. The definitions of learning and memory are not particularly satisfactory. For
example, with the definitions used, how does performance differ from memory? What does
"due to experience" mean?
pp. 46-47. What happens to the risk assessment process when the dose-effect or dose-
response curves are not linear, which is not unusual in neurotoxicology? Many of the dose-
effect curves obtained with lead show a U-shaped function. This is not restricted to behavioral
endpoints, but has been observed function. This is not restricted to behavioral endpoints, but
has been observed for other CNS effects of lead as well (Davis and Svendsgaard, 1990).
p. 48. The statement that "there also appears to be little biological justification for many
of the uncertainty factors" requires further explanation.
p. 50. The bottom line is that even with documents such as the one under consideration
here, a great deal of appropriate scientific judgment and expertise always will be needed to
make decisions on risk assessment even with the guidance provided by documents such as this.
Who will these people be; is the field of neurotoxicology generating sufficient personnel for this
purpose?
p. 58. The acronym MOE is never explained, or if it was, I missed it.
C-10
-------
References
Arcia, E. and Otto, DA. 1992. Reliability of selected tests from the neurobehavioral
evaluation system. Neurotoxicol and Teratol. 14, 103-110.
Barrrett, J.E. 1986. Behavior History: Residual influences on subsequent behavior and drug
effects. In: Developmental Behavioral Pharmacology, vol. 5, Advances in Behavioral
Pharmacology, N. Krasnegor, B. Gray and T. Thompson, eds., Lawrence Erlbaum
Associates, New Jersey, pp. 99-114.
Davis, J.M. and Svendsgaard, DJ. 1990. U-shaped dose-response curves: Their occurrence and
implications for risk assessment. J. Tbxicol. Environ. Health, 30, 71-83.
Gullion, CM. and Eckerman, D.A. 1986. Field Testing. In: Neurobehavioral Toxicology, Z.
Arniau, ed., The Johns Hopkins University Press, Baltimore, pp. 288-330.
O'Callaghan, J.P. Assessment of neurotoxicity using assays of neuron- and glia-localized -
proteins: Chronology and critique. In: Neurotoxicology, Target Organ Toxicology
Series, H.A. Tilson and C.L. Mitchell, eds., Raven Press, New York, pp. 83-100.
C-ll
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Wayne C. Daughtrey
Exxon Biomedical Sciences, Inc.
East Millstone, NJ
The guidelines being proposed by EPA fulfill a needed function in the area of data
interpretation for neurotoxicity studies carried out under a variety of regulatory programs. A
considerable amount of neurotoxicity data are being generated under TSCA, FIFRA, and other
regulatory initiatives that will need to be evaluated in a consistent and scientifically sound
manner. These guidelines hopefully will serve as a scientifically reasonable yardstick against
which newly generated data can be interpreted and appropriate conclusions drawn regarding
potential neurotoxic hazard and risk.
Below are comments on specific subject areas for consideration by EPA and the peer
review workgroup.
Categorization of Evidence for Neurotoxic Hazards (pp. 50-54)
EPA has proposed a scheme whereby data on a given chemical will be categorized as
"Sufficient Evidence" or "Insufficient Evidence" for characterizing neurotoxic hazards. This
scheme is identical in concept to the categorization scheme recently published by EPA for use
in developmental toxicity risk assessment. It is clearly desirable to have a conceptually similar
scheme for most noncancer endpoints, as the Agency is proposing.
In Table 7A, the "Sufficient Evidence" category is described as that which provides
enough information to judge whether or not a human neurotoxic hazard could exist. However,
under the two subcategory paragraphs of "Human Evidence" and "Experimental Animal
C-12
-------
Evidence," little attention is given to the or not side of the judgment. For example, under
Sufficient Human Evidence, it is stated^ that, "This category includes agents for which there is
sufficient evidence from epidemiologic studies..., to judge that some neurotoxic effect is
associated with exposure." Nothing more is stated about the situation in which there is
sufficient evidence to conclude that a neurotoxic hazard does not exist. For purposes of clarity
and understanding, the Agency's Guidelines would be improved if it were made more explicitly
clear that the category of Sufficient Evidence also includes those situations in which the data
allow a conclusion that a neurotoxic hazard does not exist. As presently written, one could
easily get the impression from Table 7A that this category was reserved only for those agents,
which had demonstrated frank neurotoxic effects.
Neurochemical Endpoints of Neurotoxicity (pp. 23-25)
The guidelines state that, "Many neuroactive agents can increase or decrease
neurotransmitter levels in the brain but such changes are not necessarily indicative of
neurotoxicity." This is a premise with which most individuals would agree. However, the
guidelines then go on to state, "However, agent-induced decreases in specific neurotransmitters
in the brain, or decreases in specific brain regions, especially when such changes are persistent,
are evidence of neurotoxicity." Persistent decreases in neurotransmitter levels are clearly
evidence of neurotoxicity; however, the beginning of this second sentence would seem to be at
odds with the earlier statement that changes are not necessarily indicative of neurotoxicity. Are
transient decreases in neurotransmitter levels in discrete regions to be considered neurotoxic?
It would appear that additional clarification of this matter is needed.
In discussion of NTE inhibition, no mention is made of threshold levels of inhibition for
the elicitation of clinical effects. It is generally accepted that levels of NTE inhibition in the
order of 60 to 70 percent are needed following an acute exposure for clinical neurotoxicity to be
manifested. For repeated dose studies, inhibition levels of 45 to 65 percent have been suggested
by M. K. Johnson as a threshold zone. EPA's proposed guidelines would be more informative
and valuable to the regulated community if this matter were addressed in more detail. While
any inhibition of NTE represents potential neurotoxic hazard, it is not at all clear what EPA's
C-13
-------
position is with regard to inhibition levels that pose a significant risk. Is there a level of NTE
inhibition that the Agency will assume to be without significant effect?
Developmental Neurotoxicity
In discussing the interpretation of developmental neurotoxicity data (p. 39), the Agency
notes the potential impact of maternal toxiciry on the developing organism. The guidelines
make a noteworthy distinction between "minimal" maternal toxicity and "excessive" maternal
toxicity. EPA states that at doses causing excessive maternal toxicity, information on
developmental effects may be difficult to interpret and of limited value. In contrast, at doses
that cause "minimal" maternal toxicity, the developmental effects are still considered to
represent neurotoxicity. Given the obvious importance of minimal vs. excessive maternal
toxicity, some general description of the two is needed. Although it is clearly not practical or
desirable to attempt to explicitly define all signs, symptoms, and findings characteristic of
excessive maternal toxicity, some general guidance would seem to be in order. Given the
importance of this issue, the proposed guidelines would be more informative if the terms
minimal and excessive were characterized.
Direct versus Indirect Effects
The draft guidelines conclude that chemicals acting through both indirect and direct
means can be considered neurotoxic. The Agency's Issues Paper lists a number of suppositions
and conclusions that were used to reach this position. While these suppositions are sound in
principle, this logic can lead to practical difficulties when testing is carried out in accord with
the neurotoxicity test guidelines established by EPA. The source of the difficulty is related to
the testing requirement that the high dose produce "significant neurotoxic effects or other
clearly toxic effects." As long as EPA's definition of neurotoxicity implicitly includes behavioral
changes, there is a significant probability that non-specific effects such as general sickness or
malaise will be operationally interpreted as representing neurotoxic effects. Thus testing in
accord with the established guidelines would seem to ensure a high likelihood of observing
indirect effects that will be interpreted as evidence of neurotoxicity.
C-14
-------
Elsewhere in the proposed Risk Assessment guidelines it is noted that agent-induced
changes in the FOB or motor activity that are associated with overt signs of toxicity (weight loss
or systemic toxicity) or that occur only at the high doses are not necessarily evidence of
neurotoxicity. It is encouraging to see this position espoused since it allows for an element of
reasoned scientific judgment to be brought to bear on the interpretation of the data.
Moreover, I would recommend that this phrasing also be incorporated into the "Definition"
section (pp. 4-5) of the proposed guidelines, where the issue of indirect effects is initially raised.
In the final analysis it would seem there is little to be gained by identifying as neurotoxic
a variety of chemical agents whose only effect on the nervous system occurs following .
administration of heroic doses and whose nervous system effects are comparatively insignificant
and a consequence of target organ effects produced elsewhere in the body.
Interpretation of Reversible Effects
The guidelines conclude that both reversible and irreversible effects of chemicals on the
nervous system should be considered adverse. An issue which might warrant further
consideration by EPA.and the Peer Review Group concerns terminology and it relates to effects
that have generally been considered pharmacological. No one would dispute the
characterization of excessive CNS depression following acute exposure to chemical agents as an
adverse effect. For many classes of compounds (aliphatic hydrocarbons for instance) this effect
is reversible upon cessation of exposure and recovery of function is typically complete. Rather
than refer to such pharmacologic effects as "neurotoxic" however, might it not be more accurate
and informative to refer to these effects using other terminology (such as neuroactive, for
example). For purposes of risk assessment and public protection, one would certainly want to
distinguish compounds that produce frank irreversible CNS damage from those that produce
acute, reversible effects. Referring to both types of endpoints as "neurotoxic" leads to a blurring
of these distinctions in terms of hazard and risk communication. It is suggested that
consideration be given to modifying the definition section on neurotoxicity in the guidelines to
more accurately reflect the differences between acute reversible pharmacologic effects and
irreversible nervous system damage.
C-15
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Shayne C. Gad
Becton Dickinson
21 Davis Drive
Research Triangle Park, NC
The proposed guidelines for neurotoxicity risk assessment are generally well written and
easy to read. My initial review of this April draft is divided into two sets of comments and
questions.
General Comments
2.
It must be noted that the authors have done an excellent job of addressing what this
reviewer believes to be valid issues of conditions effecting the relevance of findings from
animal studies. The guidelines now make clear that functional changes seen only at high
doses or in the presence of signs of marked acute systemic toxicity are not a priori
indicative of neurotoxicity.
If the purpose of the document is to provide for neurotoxicity evaluation and risk
assessment, then the authors have included an overly large amount of examples,
commentary, and justification for the need for such guidelines and the measures
provided. Though interesting and almost entirely accurate, the lack of reference
citations and the inclusion of a few cases that are not clear weakens the presentation's
standing as a background document. This is particularly of concern if the same effort
has taken away from addressing some issues of study design and interpretation for the
task at hand, as pointed out below.
C-16
-------
No guidance is provided on two key issues of design for preclinical neurotoxicity studies.
The issue of model selection (what species, age, sex^ or health status of animals should
be employed) is not addressed. This will be a critical point in the evaluation of new or
previously not evaluated chemical entities or mixtures and in evaluating the relevance of
existing findings in various animal models. Likewise, the issue of how doses are to be
selected is not addressed, other than by inference (i.e., the guidance that findings at
agonal doses are inappropriate). .
The only unaddressed question previously raised as to assessing the relevance of
functional findings in animals to neurotoxicity is that of "pharmacological" effects of
agents. That is, are agents to be considered neurotoxic if they effect some functional
components (such as motor activities) for an initial brief period of peak plasma levels
following dosing? I believe that purely pharmacological agents must be differentiated
from toxicological ones, and truly neurotoxic agents are those that have a persistent
(more than an hour after dosing/exposure) effect.
Specific Comments
1. P. 4 (second paragraph): "(2) any alteration from baseline that diminishes the ability to
survive, reproduce, or adapt to the environment." This begs the "pharmacologic" or time
course questionwhat about ethanol or vigorous exercise? If assessed immediately after
either of these, both would qualify as causing "adverse effects."
2. P. 7 (last sentence): "chronic solvent toxicity" is not currently a universally accepted
case, and therefore may not be compelling.'
3. P. 10 (end of first paragraph): Some good citations of recent U.S. cases would be very
useful here.
4. P. 14 (end of second paragraph): "An absolute loss of brain weight in adult animals
should be regarded as an indication of neurotoxicity." Is this without any finding of
C-17
-------
5.
6.
histopathological alteration or indication of functional change? A marginally statistically
significant finding here should be considered suspect. The strength of the approach
presented in these guidelines is that of broadly integrated measures.
P. 31 (line 9): Press (TYPO).
P. 34 (under incoordination): Why isn't righting reflex included as a technique? It is
simple, well established, and sensitive.
C-18
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Michael W. Gill, PhJD.
Bushy Run Research Center
Union Carbide Corporation
1602 Mellon Road
Export, PA 15632
The draft guidelines for neurotoxicity risk assessment are well written and reflect current
understanding of neurotoxicity testing. I appreciate the opportunity to review and comment on
the guidelines. My comments relate my experience with guideline neurotoxicity testing (TSCA
and FIFRA) to the principles espoused in the guidelines, discuss a few concerns, and suggest
ways that the guidance may be modified to improve consistency.
Interpreting Functional Observational Battery Data
Behavioral screening data are relatively imprecise compared to neurochemical,
anatomical, or electrophysiological data, and alterations in one or a few endpoints rarely lead to
a diagnosis of neurotoxicity. Guidance for interpreting functional observational battery data
should discuss the concept of functional domains of the nervous system since alterations in
functional domains form the basis for interpreting the absence or presence of neurotoxicity.
Grouping the data from the FOB and motor activity measurements into functional domains is a
generally accepted practice and has recently been adapted for. statistical analysis by investigators
in a number of laboratories. This grouping technique is useful when separating changes that
occur randomly or in conjunction with systemic toxicity from those treatment-related changes
that are indicative of gross alterations in nervous system function. For example, a number of
statistically significant findings from the isdpropanol 13-week vapor inhalation neurotoxicity
study were not considered to be exposure-related or neurotoxicologically significant based, in
part, on the lack of a demonstrated pattern of effects in one or more functional domains of the
C-19
-------
nervous system (Table 1). A discussion of functional domains and the importance of
establishing patterns of effects within domains should be included in the guidelines. This will
strengthen the guidance for data interpretation and minimize the potential for overemphasizing
nonspecific findings common to screening studies.
Interpreting Motor Activity Data
I strongly support the discussion on diagnosing neurotoxicity in the presence of systemic
toxicity and would like to relate it to motor activity measurements in light of the results of the
13-week triethylene glycol monomethyl ether (TGME) neurotoxicity study. A high dose of 4 g
TGME/kg/day in the study resulted in decreased mean body weight and food consumption
throughout the study and decreased motor activity during the latter half of the study (Table 2).
There were no FOB findings, clinical signs of toxicity, or neuropathology findings to support the
conclusion that the motor activity findings represented a direct effect of TGME on the nervous
system. Alternatively, the effects on motor activity may have been secondary to-the systemic
toxicity indicated by changes in body weight and food consumption. It would be inappropriate
to consider TGME to be a neurotoxicant in light of these confounding effects and the high
dosages used. These data underscore the importance of evaluating the data set for supporting
evidence of neurotoxicity as well as for signs of systemic toxicity.
General Comment on Interpreting Screening Data
High dose levels are required for the current screening studies to demonstrate either
toxicity or neurotoxicity. Indeed, a dose level above generally accepted limit doses was required
for the 13-week TGME neurotoxicity screening study. An ideal screen would be sensitive and
specific. Unfortunately, the high doses required for the neurotoxicity screens result in an
increase in apparent sensitivity while forfeiting specificity. A comment should be added to the
guidelines that recognizes the impact of toxic doses on test specificity.
C-20
-------
Reversibility and Adaptation versus Neurotoxicity
The determination of whether reversible functional effects reflect neurotoxicity should
consider the nature of the test substance, the effects observed, and knowledge of the potential
mechanisms for these effects. Reversible effects do not reflect neurotoxicity if the effects are
generally expected to be reversible at the biochemical level based on knowledge of the test
agent or class of test agents under investigation. The reversible apparent sedation of the central
nervous system following a single 6-hour exposure to high vapor concentrations of isopropanol is
one such example. The nature of the effects was consistent with'expected profiles for short
chain aliphatic alcohols. In addition, the time course of the effects paralleled the time course of
disappearance of isopropanol from blood following vapor inhalation.
Unfortunately, limited information will be available regarding the mechanism of action
for most test agents to be screened for neurotoxicity, and it will not be possible to conclude that
reversible functional effects reflect reversible biochemical events. A more conservative diagnosis
of neurotoxicity will be needed in these cases when significant reversible effects are detected.
Adaptation following repeated exposure may be evidence of permanent molecular and
structural changes in the nervous system unless information is available to support
pharmacokinetic alterations. Molecular or structural changes may alter the organism's ability to
respond to challenge and should be considered to represent potential neurotoxicity.
3.2.4.3 Schedule-Controlled Behavior. A rewording of the second sentence in the first
paragraph of this section is recommended since schedule-controlled behavior tests may be used
to measure learning, memory, or performance depending on the design of the test. Suggested
change: add "memory, and/or performance" after "learned behavior."
3.3.1 Statistical Considerations. The first two paragraphs of this section infer that a
minimum statistical power is necessary to detect a true effect of an agent and that the absence
of this minimum power will jeopardize the usefulness of the study. As indicated, power
C-21
-------
depends on the variability of the behavioral measure and the sample size. Further, it is
generally recognized that behavioral measures are inherently variable, and large sample sizes
may be necessary for some behavioral measures to satisfy a common power requirement of 0.8.
The authors of the recent FIFRA guidelines for neurotoxicity recognized this and omitted a
specific power requirement in order to limit the number of animals in these studies to an ethical
and manageable level. A discussion that recognizes the variability of behavioral measures, the
limits placed on the number of animals used in these studies, and the potential for decreases in
statistical power should be included in the guidelines.
3.3.4 In-vitro Data in Neurotoxicologv. The second sentence in the last paragraph of
this section should mention differences between the intact organism and the in-vitro system. I
suggest the following change in the text: "This validation process requires consideration in study
design, including defined end points of toxicity and an understanding of how a test agent would
be handled by a system in comparison to the intact organism."
5. Adequacy of The Evidence for Hazard Identification and Dose Response Assessment.
Page 52, line 2. The two sentences beginning with "Neurotoxicity..." and ending with "...well
conducted study," contradict the previous discussions on data interpretation and should be
modified. Taken literally, this passage says that a single statistically significant change reflects a
hazard and should be used to estimate the risk from the test agent. Since these sentences could
be interpreted out of context, this section must be changed in order to ensure consistent and
appropriate guidance on data interpretation.
Table 7A. Sufficient Human Evidence. The approach taken to categorize the amount
of information is not consistent throughout Table 7. This category also should include the cases
when epidemiology or experimental studies provide sufficient evidence to judge that there is no
neurotoxic effect associated with exposure.
Table 7B. This table argues to exclude the "weight of evidence" approach to interpreting
neurotoxicity and infers that any study that was not performed in compliance with the current
C-22
-------
guidelines would be inadequate to judge the neurotoxic potential of a test agent. However,
there are test agents for which there are well conducted subchronic and chronic toxicity studies,
and/or tbxicity studies modified to address neurotoxicity concerns. Categorizing all chemicals
that have not been through a guideline neurotoxicity screen as having "insufficient evidence" is
inappropriate given the large number of industrial chemicals in commerce that need to be tested
and our limited testing capacity. Clearly there are test agents with a sufficient weight of
evidence for assessing human risk without additional data from guideline neurotoxicity studies.
Screening methods must be allowed to evolve so that assessment of human risk can be
improved. The inflexible position presented in the section on "insufficient evidence" may hinder
the evolution of screening methods by occupying available laboratory resources with guideline
neurotoxicity studies for years to come.
In conclusion, the draft guidelines for neurotoxicity risk assessment are well written and
reflect current neurotoxicity knowledge. Thank you for the opportunity to review the guidelines,
and I hope that my comments and suggestions are useful.
C-23
-------
Table 1.
Functional Observational Battery Findings for Male and Female Rats Exposed
to Isopropanol for 13 Weeks
Group (ppm)
500
1,500
5,000
Pre-exposure
Grip Strength (hind)
Decreased
M
M
Study Week 2
Rearing Events
Decreased
Study Week 9
Tail Flick
Pupil Response
Decreased
Decreased
M
F, female; M, male; M or F notations indicate statistically significant differences from the
control group.
C-24
-------
Table 2.
Body Weight, Food Consumption, and Motor Activity for Male and Female Rats
Treated with 4 g/kg/day Triethylene Glycol Monomethyl Ether for 13 Weeks
- : -
Male
Female
*Mean of the absolute value
mean for the control group.
Study Week
1
4
9
13
1
4
9
13
of the measurement
Body
Weight
Food Motor
Consumption Activity
Percent of Control
92*
90*
82*
79*
100
96
95
92
was statistically
78*
88*
89*
83*
significantly
93
89
71*
82
88*
86*
91
85*
different
95
83
84
70*
from the
C-25
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
John R. Glowa, Ph.D.
LMC/NIDDK/NIH
Bldg 14D Rm 311
Bethesda, MD 20892
Panel 1: Neurotoxicity as an Appropriate Endpoint for Risk Assessment
The conclusion is correct, but perhaps could use more support in the body of the draft.
The inclusion of the term environmental is not fully understood.
Conclusions and Suppositions
The linking supposition, that neurotoxicity causes the failure of the normal functioning
of the nervous system, is omitted. In addition, the word "proper" is not defined. Attempts
within the draft to do so allude to the notion of a baseline, which then requires a notion of how
far the deviation of that baseline can be before normality is exceeded. Perhaps for the purposes
of the draft, methods that specify that (given the amount of information on hand) an event
exceeding the likelihood of p< 0.05 (two standard deviations, etc.) may be used to define a
deviation from normal functioning.
Many agents cause neurotoxicityfuller citation would help. The best evidence of a
need for neurotoxicity risk assessment is the historical record of neurotoxic events, perhaps
placed in context with other endpoints. There is a minimal effort in this regard on pp. 7-8.
Human exposure to neurotoxic agents either is or can be significant.
C-26
-------
Areas of Special Concern or Focus
The amount of newrotoxicological information is inadequate only in that enough is not
available to determine if most agents are safe at current exposure levels and conditions. There
is enough toxicological information available to suspect many agents will be neurotoxic,
especially in the developmental area.
Clearly standards have not been set. While this manuscript establishes a number of
reasonable endpoints, it does less to provide "information that will be useful -for the evaluation
of the data." (p. 1, last 4 lines).
Panel 2: Interpretation of Transient Data
The problem with the position statement is that it says that all effects are adverse (since
effects must either be reversible or irreversible). Further exclusion is required.
Conclusions and Suppositions
The first conclusion should be completely reworded. The nervous system is not
composed of wires, resilient complex patterns do not appear helpful in detecting effects, and the
processes of compensation and adaptation are too poorly described to use as a reason to study
transient effects.
The basis and linkage for this supposition are poorly established. It does not follow the
first supposition. The scientific basis for this statement, in particular reserve capacity, will have
to be provided.
C-27
-------
Areas of Special Concern or Focus
Given the concern with secondary effects, this supposition must be withdrawn. Acute
exposures can be lethal through primarily neuronal effects. These effects, therefore, must be
studied in lower dose ranges.
These effects, known as historical determinants in behavioral pharmacology, are
well-established and clearly reveal changes in function. Their neuronal basis is not
well-established.
This question is of concern, both because more data are needed to evaluate it and that
different agents may result in different life-time temporal patterns. As stated in the text, and as
applies throughout, each agent must be evaluated on a case by case basis.
Panel 3: Indirect and Direct Effects Should be Considered Neurotoxic
Agents produce neurotoxic effects through indirect as well as direct means. It seems
that the issue addressed here is whether the inability to adapt because of exposure-related
effects is mediated by neuronal processes. Clearly other processes can be affected. The
inability of baroreceptors to respond to change may be due to a non-neuronal toxic effect,
leading to neuronal damage (related to issues in the second area of special focus in Panel 2).
Conclusions and Suppositions
What distinctions sometimes exist between direct and indirect and primary and
secondary. It would seem that they are either the same or not.
The second supposition, while true, excludes many mechanisms employed even in the
example provided, not to mention an extremely wide range of other possibilities. Does the
occupation of receptors result in a toxic effect, or the subsequent actions at channels, within the
C-28
-------
cell, etc. Glutamate is a great example. Since it may function as a neurotransmitter at lower
levels, risk assessments will have to stay above these. However, it does not pass the blood brain
barrier.
If cell death is the functional equivalent, this can be true. However, this supposition
appears to be some sort of end run to lump all neurotoxic effects together, and will require
more than a little discussion.
There appear to be two logical exceptions to this supposition: (1) neurotoxins, as
defined in the text are different than neurotoxicants, and (2) dose, a compound can produce
neuro toxicity (at higher doses), but may have normal functioning, therapeutic, or other desirable
effects at lower doses. As we learn more about peptides and other modulators, dose-effect
functions not normally observed in pharmacology and toxicology have emerged, which require
further understanding rather than rule-governed simplification.
Again, this may not always be the case, although an example at the moment escapes me.
Essential minerals cannot be maintained at too low of a dose, without neurotoxiciry occurring
either. Also, for many agents, with increases in dose, the cause of death may not be related to
neurbtoxicologycardiac failure for example.
Areas of Special Concern or Focus (
This is true and why many of the endpoints chosen are practical.
Given that an agent is not regulated by systemic toxicity, and produces neurotoxiciry,
risks clearly should be assessed. It may be inappropriate to dismiss a candidate from regulation
because it produces primary toxicity on some other system. Risk managers may want to take
into account that risks across several categories of endpoint (cancer, systemic, etc.) increase the
likelihood of effect of any one endpoint.
C-29
-------
Panel 4: Extrapolation from Animal to Human
For neurotoxicity, where direct examination in humans frequently will not be possible,
the position reached is much too apologetic for the presumed adequacy of animal research.
The necessity to actively pursue characterization of neurotoxic agents in animals is based on a
lack of valid alternatives. These risk assessments are valid, period. On the other hand, the
discussion of validity in the draft seems wordy.
Conclusions and Suppositions
While this supposition is true, it opens the door for further questions. The relationships
between dose and effect are by and large similar, similar routes can be studied, while other
disciplines over-extend to use words like model, this supposition is much too shy.
True, more could be said. Since the nature of risk assessment is precision, and precision
cannot be obtained with epidemiological studies or case reports, it seems the only approach.
The same things can be measured in humans and (other) animals. One could even say
everything could be measured, but this might be a bit too strong. Again, this sounds too weak,
even at "many procedures." A similar point is made as supposition 3 in Panel 5. The use of the
word model is offensive. Neuronal degeneration in an animal is not a model of what happens in
a human, it is neuronal degeneration. Its effect may be used as a model of Parkinsonism, etc.
but if differences exist it will only be a partial model. If none exist it will be the same.
I suspect that the range of uncertainty factors will not always be the same. Part of this
concern is that little is actually known for the animal-to-human factors for particular agents.
One area in particular may be solvents, where uptake to steady-state tends to equilibrate
physical differences in size etc., and respiratory rates may actually shift things to the animal
being more sensitive. This is one area (neurotoxicology) where such issues can be directly
compared, if sufficient human data exist. Thus a basis for safety (fudge) factors can be
established from data rather than an assumption initiated long ago.
C-30
-------
Areas of Special Concern or Focus
While I concur with the spirit of the language (verbal behavior) comment, there are
clearly studies that show that chemical exposures in rodents and primates increase rates of
well-defined and communicative vocalizations. I question the relevance of the statement to risk
assessment. On the other hand, if an agent were shown to produce a cerebral stroke that
affected speech, and that agent had no effects on other species, a basis for the focus would be
substantiated. I know of no such agents.
<*« - - - . -
While this supposition is true on occasion, it should not negate the process of risk
assessment with animals. The possible exceptions are chemicals for which metabolism or
kinetics in humans are unique and toxic. Like the point above, specific examples must be
brought forth to hold these points up, otherwise speculation retards the ability to gather what
meaningful data exist. -
The most sensitive species exception also detracts from the process. There always will
be these exceptions. If metabolism in these species is unrepresentative (i.e., producing a unique
toxicant, etc.) then they shouldn't be used. If they represent different rate constants, levels, etc.,
then they should because then they are merely a biological extreme.
Panel 5: Interpretation of Behavioral Data
Behavioral change, or perhaps more importantly the lack of behavioral change, under
challenge conditions, as a result of chemical exposure can be a clear indication of neurdtoxicity.
However while behavioral change alone can provide evidence of neurotoxicity, transient
behavioral change is insufficient to conclude neurotoxic effects have occurred.
Conclusions and Suppositions
Behavior is not always the most sensitive indication of toxicity.
C-31
-------
While evaluation of behavior should play an important role in efforts to understand
brain function, it rarely is employed. There are many reasons for this. Many basic
neuroscientists are often not equipped or do not understand behavior. Others simply avoid the
issue of relevance. When behavior is studied, it is often at a primary screening level (i.e.
locomotor activity, classical conditioning effects).
Some behavioral evaluations have animal counterparts (i.e., startle) not most. Some are
verbal.
Probably false if extended to neurochemical effects. The issue rests with the ability to
measure physiology together with behavior, and the definition of "behavioral effect." The state
of the art is not advanced now, but may be by the time these guidelines are implemented. For
morphology, behavioral change may be more sensitive than gross "holes in the brain," but less
than growth of dendritic processes.
That the primary developmental effects of some chemicals may be behavioral seems
argumentative. Very little behavior is studied in utero. This may mean that correlates or
neurochemical determinants.of in utero effects have not been found or explored. However,
there is no question that there are developmental neurobehavioral effects of some chemicals,
and theses effects alone are a basis for regulation.
Areas of Special Concern or Focus
While true, "indirect" may be preferred over "non-specific." Examples may include
conditioned sickness, which can impair normal behavioral functioning without neurotoxic effects
(i.e., conditioned gastrointestinal effects).
A maximum tolerated dose for behavioral function rests on the ability to assess the
adaptability of the system as a function of dose.
C-32
-------
Although behavior can change due to physiological effects (see panel 3), this is not
necessarily a basis for excluding behavioral changes associated with more toxic effects (i.e., a
false positive).
C-33
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
William F. Greenlee, Ph.D.
Department of Pharmacology and Toxicology
Purdue University
1334 Robert £. Heine Building-Room 202A
West Lafiayette, IN 47907-1334
General Comments
The Proposed Guidelines for Neurotoxicity Risk Assessment contained an extensive review
of the actions of various classes of neurotoxicants in humans and in experimental animals. The
studies presented were organized according to the guidelines established by the National
Research Council. Data relevant to hazard identification, dose-response assessment and
exposure assessment were discussed and brought together in a concluding two page summary of
some of the issues to be addressed for the risk characterization of neurotoxicants.
Within an overall exposure-dose-response paradigm, there are key gaps in the current
knowledge of the pharmacokinetics and mechanisms of action of neurotoxicants that require
additional research with prototype agents in both experimental animals and appropriate in vitro
cell models. In vitro models should be used to support studies on the mechanisms of action and
not necessarily developed as short-term test systems for neurotoxicify. The strength of in vitro
approaches is the elucidation of specific molecular and biochemical events evoked in a surrogate
target cell by a potential neurotoxicant. Recent advances in molecular neurobiology have
resulted in the cloning of ion channel proteins, receptors for neurotransmitters and regulatory
proteins involved in signal transduction; the elucidation of the role of immediate early genes in
memory and responsiveness to environmental stimuli; and the implementation of cloning
strategies for genes involved in certain neurodegenerative diseases.
Using physiologically based pharmacokinetic approaches, models that incorporate
knowledge of the biological determinants of tissue dose for volatile agents, and certain
C-34
-------
chlorinated aromatic compounds have been developed and used to predict accurately tissue
concentrations at low exposure doses. Similar approaches are needed for ongoing
pharmacokinetic studies of neurotoxicants. Existing analytical techniques should permit
experimental validation of models that incorporate biological determinants relevant to the
distribution and/or metabolism of compounds in potential target cells within the nervous system.
Application of contemporary cell and molecular biology approaches to the study of
nervous system function provide the opportunity to gain new insights into the molecular
mechanisms of action of known neurotoxicants. Comparative analysis of the actions of these
compounds in experimental animals and in human and animal cell culture models should be
focused on elucidation of the-biological determinants of target cell- and species-specificity.
Integration of molecular, cellular and tissue dosimetry models for specific nervous system
endpoints with physiologically-based pharmocokinetic descriptions of tissue dose at _
environmentally relevant exposure levels are essential for the development of biologically-based
risk characterizations for neurotoxicants.
Specific Comments
Endpoints of Neurotoxicitv. Table 1 lists several endpoints of neurotpxicity within five
categories. The challenge is to develop a short list of quantifiable endpoints that represent
potential adverse changes in the structure or function of the nervous system likely to occur as a
result of chronic low level exposure to environmental agents of significant concern. It is
necessary to distinguish between reversible changes with no demonstrable adverse clinical
outcome versus those changes that are measurable, but based on solid experimental evidence,
not linked casually to a nervous system lesion. Molecular probes currently available and the
development of with well-characterized animal and cell culture models for the study of nervous
system function offer the potential for detailed analysis of the mechanisms of toxicity of certain
classes of neurotoxicants that act on specific protein targets within the nervous system (e.g.,
neurotransmitter receptors, and signal transduction proteins). The sensitivity of detection
.methods using DNA and/or antibody probes allow identification and quantitation of specific
changes in the level or function of these proteins in specific cell populations at low
C-35
-------
neurotoxicant exposure concentrations. Integration of studies in in vitro cell systems and animal
models are needed to determine the linkage of early events to nervous system dysfunction.
These systems also could be used to study the modulating influence of confounding factors on
neurotoxicants. However, like the nervous system itself, the interaction of these factors with a
gjven neurotoxicant is complex. For example, confounding factors such as alcohol consumption
can influence the metabolism and distribution of the neurotoxicant, as well as alter its action on
a target cell within the nervous system.
Pharmacokinetics. The pharmacokinetics of neurotoxicants, particularly at low exposure
doses, should be an area of increased emphasis. Advancements in physiologically based
pharmacokinetics provide a foundation for detailed study of the uptake, distribution and
metabolism of potential neurotoxicants. Models can be developed based on increased
knowledge of the xenobiotic metabolizing potential of various cell populations within the
nervous system and the various components that control circulating concentrations and tissue
localization of neurotoxicants; e.g., hepatic metabolism and elimination, blood flow to target
tissues, the blood-brain barrier, brain lipids, and high affinity binding to target proteins.
Comparative analysis of these parameters should be carried out across species, using both in
vivo and in vitro models. For studies focused on low level exposures, available data on the
metabolism and elimination of a given neurotoxicant in humans need to be considered in
deciding on relevant exposure levels in experimental animals. It is important that known
differences in the metabolism and elimination of xenobiotics in humans versus rodents be
incorporated into pharmacokinetic models for predicting the concentration of a given
neurotoxicant in a nervous system target tissue.
Risk CharacterizationConcluding Comments. The goal of risk characterization is to
develop a biological mechanisms-based risk assessment that incorporates knowledge of the
biology of the target organ(s) of interest. Application of contemporary molecular biology
approaches to the study of neurotoxicants should be focused in large part on elucidation of the
biological determinants of tissue- and species-specific responsiveness of nervous system targets.
Quantitative descriptions of low dose behavior of neurotoxicants will require the development of
physiologically based pharmacokinetic models. Linkage of physiologically-based tissue dosimetry
models with quantitative descriptions of relevant molecular and biochemical events elicited by
C-36
-------
the interaction of a neurotoxicant with a nervous system target tissue are key elements in a
biologically based risk assessment strategy. Given the complexity of the nervous system and the
large number of neurotoxicity endpoints, it is essential to prioritize efforts and focus on
endpoints that are quantifiable and linked to a clinical outcome relevant to environmental
exposure scenarios of concern.
C-37
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
G. Jean Harry
National Institute of Environmental Health Sciences
P.O. Box 12233 MD
EI-02 Research Triangle Park, NC 27709
Comments
In reference to structural endpoints of neurotoxicity, it needs to be made clear that such
perturbations may not be evident at early exposure times or at low doses with the general type
of neuropathological screening procedures employed. A more detailed evaluation may be
needed for detection. Cellular alterations such as cell death and cellular organelle restructuring
may be a late event in the toxicity response. This does not rule out the possibility that lower
exposure levels are producing neurotoxicity. The sensitivity of each detection method used must
be taken into consideration when evaluating site of structural perturbation.
Developmental exposure to a compound and alterations in structural endpoints must be
evaluated in the presence of normal cellular restructuring during the process of development.
The ability to detect a perturbation may be limited; however, the perturbation may indeed be
more detrimental. Such is the concern for evaluating GFAP reactivity in the nervous system
during development since an increase in GFAP is seen during the normal process of
development indicating a differential role for the protein in the developing versus mature
animal. Similar concerns exist with evaluations of neurochemical and electrophysiological
endpoints during the developing process. Evaluation of the mature system following
developmental exposure would not be limited by such concerns but would examine the
long-term consequences of such exposure. For these reasons, it is felt that the problems of
evaluating alterations in the developing organism must be fully understood when the data are to
be used to evaluate risk.
C-38
-------
One model of exposure that should be considered is the situation where the person is
removed from the exposure environment to allow for reported symptoms to subside and then
placed in a similar exposure environment. Evaluation of increased sensitivity following
re-exposure could be used to determine the long-term compromise of the nervous system and
may offer information to be used in evaluating risk.
C-39
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
Donald E. McMillan
Department of Pharmacology of Toxicology
College of Medicine
University of Arkansas for Medical Sciences
4301 W. Markham Street
Little Rock, AR 72205
The document uses the Nuclear Regulatory Commission (NRC) four-stage approach to
risk assessment: hazard identification, dose-response assessment, exposure assessment, and risk
characterization. The emphasis of the document is on hazard identification of neurotoxicity
with some reference to dose-response assessment, but very little discussion about exposure
assessment and risk characterization. Perhaps the major weakness of the document is that it
lacks specificity in its recommendations. Although there is general discussion of how different
kinds of data would be interpreted by the U.S. Environmental Protection Agency (EPA), the
information is not presented'in any detail. Although many measures of central nervous system
(CNS) function and integrity are discussed, it is never clear exactly what tests or methods would
actually be recommended or required by EPA when a company wished to market a new
chemical. Nevertheless, the document is a good starting point for more detailed discussion of
neurotoxicity testing.
An immediate problem arises from the decision to pool all functional and anatomical
changes produced by chemicals into a single class, which they define as neurotoxins. This
decision forces EPA to classify chemicals that produce frank irreversible CNS lesions in the
same category as short-acting chemicals, whose effects may be serious in some situations, but
whose effects are readily reversible upon discontinuation of exposure. By defining all chemicals
that affect either the morphology or the function of the CNS as neurotoxins, it makes it likely
that a chemical with modest short-term behavioral effects will be inappropriately branded as a
neurotoxin. In my opinion, EPA should consider at least two categories of definitions. For
example, definitions might be made on the basis of chronicity of effects (e.g., chronic
C-40
-------
neurotoxins versus acute neurotoxins, or reversible versus irreversible neurotoxins), or
anatomical versus functional effects (e.g., neurotoxins versus behavioral toxins). This would
prevent a short-term behavioral change based on the odor of a chemical from being considered
with the same level of concern as a chemical that produces widespread lesions in the brain.
Table 1 presents an impressive list of potential endpoints for the measurement of
neurotoxicity. Any chemical making it through the battery of tests listed in this table would
certainly have a high probability of receiving a clean bill of health, provided that the studies
performed were of high quality. Is the purpose of the table to suggest a model test battery for
neurobehavioral toxicity testing or is it merely listing a range of tests that could provide useful
data about the potential neurotoxicity of a chemical? Perhaps I don't understand the purpose
of the document, but someone in charge of setting up a neurotoxicity test battery for a company
would not get much guidance about what kinds of testing should be conducted based on this
document.
A related problem is that the document does not define what constitutes an acceptable
study and whether or not all kinds of information will be treated identically. Will animal studies
require placebo controls? Should active placebos be used? Should FOB be done by "blind"
scorers? Are behavioral endpoints to be weighted the same as neurochemical endpoints?
These and similar questions are not addressed.
A major strength of the document is the attention that it gives to functional testing and
to the transient effects of chemicals. It is very important to recognize that transient effects of
chemicals on behavior are important indices of toxicity, even though they may be reversible on
discontinuation of exposure. Although it is difficult to argue against giving our most serious
concern to those chemicals that cause permanent lesions in the CNS and produce profound
functional consequences, the document recognizes, perhaps better than any similar document,
the great importance that chemicals can have in producing transient functional effects. A
chemical that slows reaction time, affects intellectual functions, or has other behavioral effects
during and perhaps for a short period after exposure, can have devastating consequences for
someone driving an automobile or operating dangerous machinery. It is very important that
EPA has recognized this and is attempting to screen for these effects.
C-41
-------
EPA also should be congratulated for choosing a broad range of functional endpoints on
which to base reference doses. This is particularly true with respect to behavioral endpoints.
Unfortunately, there are certain issues concerning behavioral endpoints that have been raised
repeatedly by those who resist this type of testing. Although I have discussed some of these
issues previously (McMillan, 1986; 1990), a few of them bear mentioning here. An issue certain
to be raised is the interpretation of behavioral changes following exposure to a chemical. For
example, does a change in motor activity during or following chemical exposure represent
toxicity? Does a decrease in reaction time, or an improvement in memory represent toxicity?
My opinion is that any change in a behavioral baseline from "normal values" represents a
behavioral toxicity. This is especially true when it is recognized that the exposed population has
not chosen to have their behavior altered by chemicals, but rather the population has
involuntarily been exposed to a chemical that changes behavior.
A variation on this theme is that given enough of a chemical, all chemicals affect
behavior. Therefore, all chemicals become "neurotoxins" at some dose and for this reason
testing for behavioral toxicity is not very useful. Although the axiom that all chemicals can
produce behavioral effects at some dose is probably true, this hardly constitutes a reason for
challenging the importance and validity of behavioral testing. The same issues can be raised
about many types of toxicological data, such as chemical-induced changes in the immune
response or induction of P 450. Since all of these effects can represent responses to the stress
from a chemical, perhaps with the consequence of a limited capacity for responses to further
stressors, they all denote toxicity. The dosage issue can be handled adequately by other
components of the NRC approach, such as dose-response assessment, exposure assessment, and
risk characterization. If a chemical produces behavioral effects at doses far above those that
affect other endpoints, the behavioral effects will be of little importance in establishing
reference doses.
Yet another variation on this theme concerns the specificity of neurobehavioral effects.
For example, a chemical may produce liver damage, leading to illness, which can be manifested
in behavioral changes. It is argued that such effects should not be labeled as neurotoxic effects.
I agree; however, it seems likely that indirect behavioral changes produced by chemicals can
serve as markers for the general well being of the animal. It would seem that the onus of
C-42
-------
proving that these indirect changes in behavior do not represent primary behavioral toxicity, or
primary neurotoxicity, is upon the company wishing to market the chemical. It is interesting to
note that the writers of the current document appear to be reaching this same conclusion (p. 37)
for behavioral measurements, yet they deny that it applies for neurochemical measurements (p.
23). Why should neurochemical and behavioral changes be treated differentially?
There are several minor points in this section that need to be reconsidered. At the end
of paragraph 2 on p. 31, the final statement reads that although schedule-controlled behavior
has been used to study drugs in humans its use in "toxicology is limited." The verb should be
changed to has been. Granted there are ethical, problems with exposing humans to most
neurotoxins to study their effects on behavior prospectively. However, the techniques may be
quite useful in the study of oceupatkmally or environmentally exposed populations. This
remains to be explored. :
On p. 32, paragraph 1, it is stated that behavioral changes may indicate neurotoxicology
if they are not producing concurrent alterations in motivation, or overt signs of toxicity. I see no
reason for these exclusions. Clearly motivation is an important aspect of behavior (e.g., the
purported amotivational syndrome produced by marijuana). Most psychologists would consider
motivation an important determinant of behavior and a function of the CNS. The exclusion of
neurotoxicity when there are "overt signs of systemic toxicity" also seems unreasonable. Is it not
possible to observe important neurobehavioral toxicity concurrent with toxicity in other systems?
Again I believe that the task of determining whether behavioral toxicity is a primary effect, or is
a secondary effect from other toxic effects should fall on whomever is responsible for providing
the toxicity test data. Similarly, on p. 32, paragraph 3, the statement appears to rule out any
interpretation of the data as showing neurobehavioral toxicity when body weight changes, or
other signs of systemic toxicity occur. This is inappropriate. For example, hypothalamic effects
of a chemical may reduce appetite and food intake, resulting in a weight loss. This is
neurobehavioral toxicity. When behavioral effects occur concurrently with other kinds of
toxicity, the task becomes one of determining the relationship among these effects. It does not
mean that behavioral forms of toxicity should be ignored, just because other forms of toxicity
can be documented at the same dose or exposure level.
C-43
-------
The final paragraph on p. 31 makes a confusing assumption about the relationship
among variability, sensitivity, and specificity (the latter term is not discussed). The implication
of the paragraph seems to be that the lower the variability of a behavioral test, the lower the
sensitivity of the test to disruption by extraneous variables (e.g., chemicals). This is not
necessarily true, especially when chemicals produce specific effects on a behavior. For example,
one might train animals to respond under a multiple schedule with components A and B.
Behavior in component A might show much lower variability than behavior in component B, but
the behavior in component A might be much more sensitive to the effects of a given chemical,
especially when the chemical specifically affects the behavior maintained in component A. I
would suggest elimination of this paragraph.
The section on developmental neurotoxicity is a welcome addition to a document of this
type. Generally, this section is well developed, but some discussion about control groups would
be a useful addition. It is now state of the art in developmental toxicology to use pair-fed
controls and to use the technique of cross fostering to control for possible post-partum maternal
effects. Are such controls needed in developmental toxicity studies? It could be argued that
these techniques are not needed to establish reference doses, but are needed for more
mechanism-oriented studies., This is a debatable point. The issue of appropriate ages for
testing in developmental studies also might be worth discussing.
The document appropriately emphasizes the importance of dose-effect data in
neurobehavioral toxicity testing; however, it is not clear when dose-effect data should be
collected in relationship to the collection of other toxicity data. Should neurobehavioral toxicity
testing be independent of and proceed in parallel to other types of toxicity testing? Should
neurobehavioral data be used to establish dose levels for other types of toxicity testing or vice
versa? Does one proceed through a neurobehavioral screen in a hierarchical order, or should
an entire screen be conducted at one time? The issue of when testing does occur, what gets
tested in what order is an important issue.
The sections on exposure assessment and risk characterization are perfunctory and might
be expanded. Are there any special issues that need to be considered with regard to
neurobehavioral toxicity testing as regards exposure assessment and risk characterization? The
C-44
-------
goal of replacing the current unscientific "safety factor" approach in establishing a reference
dose with a more scientific approach based on pharmacokinetic and pharmacodynamic
observations is strongly encouraged. The present method for calculation of reference doses
based on the use of arbitrary safety factors is not based on good science and should be replaced
as soon as possible. The assumptions in using a "factor of 10" as a margin of safety have not
been adequately documented, although the data base for doing so is almost certainly available in
the literature for some chemicals.
With respect to issues related to the extrapolation of animal data to humans, there is a
growing literature on species scaling that, should be consulted. It might be argued that EPA
should make some specific recommendations about the use of different species in toxicity
testing. For example, are data on neurotoxic effects in fish suitable for establishing reference
doses? Should all neurobehavioral toxicity testing require that some tests be performed in at
least two mammalian species to increase the generality of the findings?
The perspective of the document on in vitro testing is quite appropriate. Although such
tests can provide clues as to the possible toxicity of chemicals, such methods cannot model the
metabolic activation of toxins and probably never will be able to model the complex interactions
of the nervous system that result in behavior. This js especially true when one considers how
behavior develops from "experiences" of an organism interacting with its environment. Similarly,
computer models are not useful at this time in predicting neurobehavioral effects of chemicals.
Even if such models develop sufficiently to make some useful predictions, these predictions will
have to be validated by animal testing. Neurobehavioral toxicity testing (and in fact all of
toxicity testing) will have to rely on whole animal experiments for the foreseeable future.
REFERENCES
McMillan, D.E. (1986). Risk assessment for neurobehavioral toxicity. Envir. Health
Perspect. 76: 155-161.
McMillan, D.E. (1990). The pigeon as a model for comparative behavioral
pharmacology and toxicology. Neurotoxicol. Teratol. 12: 523-529.
C-45
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
John L. O'Donoghue, V.M.D., Ph.D.
Corporate Health and Environment Laboratories
Eastman Kodak Company
1100 Ridgeway Avenue
Rochester, NY 14652-3615
Neurotoxicify as an Appropriate Endpoint for Environmental Risk Assessment
The neurotoxicity risk assessment process should include the concept of developing an
efficient method for reasonably assuring the public that a chemical does not present a human
health concern at environmentally relevant exposure levels. The public should not be led to
believe, however, that any evaluation method can absolutely ensure that a chemical can or
cannot be neurotoxic under all circumstances of exposure. As the draft guidelines indicate,
most of the data that will be available for analysis will be from test animals, not humans.
Specific neurotoxicity data for the species of concern (humans) will, therefore, most often not
be available. The public also should not be led to believe that large numbers of neurotoxicants
are present in the environment. The available data do not support such a conclusion. Such a
position leads to unnecessary and potentially harmful public anxiety.
Reports about the number of neurotoxic chemicals frequently overestimate the number
of chemicals that need specific testing. For example, such reports frequently cite the number of
chemicals on the TSCA Inventory or the number of Pre-Manufacture Notices as evidence that
there are vast numbers of chemicals that could be neurotoxic. But inspection of such lists shows
that the number of chemicals available that could be neurotoxic or result in environmental
exposure are much smaller. Some reports cite the number of chemicals that do not have
specific neurotoxicity tests as evidence that there are many undiscovered neurotoxicants. These
presentations usually ignore the likelihood for potential contact with chemicals at significant
exposure levels and the role that routine screening tests play in identifying a variety of chemical
toxicities, including neurotoxicity. Often these reports suggest that most or many neurotoxicants
C-46
-------
were first discovered because of human poisonings; such assertions are not supported by the
data and become even weaker if the neurotoxicants discovered before the advent of modern
scientific practices are eliminated from consideration.
Current estimates of the number of potential chemicals that might be neurotoxic range
from approximately 5 to 30 percent. If the upper limit of these estimates is accepted as
accurate, then the testing of all chemicals in a large series of specific neurdtoxiciry tests would
be inefficient because at least 70 percent of the chemicals would not be found to be
neurotoxicants. Thus the use of screening tests to indicate which chemicals need specific
neurotoxicity testing and formal risk assessments becomes very important and could save :
significant resources that could be used to control other more significant environmental risks.
Interpretation of Neurotoxicity Data When Effects Are Transient
In considering whether or not reversible or irreversible effects should be considered
adverse it is important first to consider whether or not the endpoint itself should be considered
evidence of neurotoxicity. When considering whether or not behavioral changes are adverse, it
is important to acknowledge that behavior is the end result of an organism's interaction with its
environment. Observation of a behavioral change is not necessarily a signal that something
adverse has occurred, rather it is an indication that the organism has reacted to a change in its
environment. The response of the organism may be considered positive, neutral, or negative
depending on other external factors that need to be considered when evaluating the observed -"'-
response. Therefore, interpretation of behavioral changes, many of which will be reversible^
needs to be considered very critically. Effects that are considered trivial should not be given the
same weight in assessing risk as those effects that are considered serious. -
While neurons that are damaged severely are considered to have limited capacity for
regeneration, there is no reason to believe that neurons that are involved in readily reversible
functional changes suffer a similar fate. Likewise, recovery from readily reversible functional
changes would not be expected to (1) represent activation of reserve neural capacity, (2)
C-47
-------
decrease the potential of the nervous system to adapt to future challenges, or (3) result in
changes that would later be revealed through an environmental or pharmacological challenge.
Agents Acting through Indirect and Direct Means Can Be Considered Neurotoxic
The significance of direct versus indirect effects is a complicated issue that is closely
associated with identifying what is neurotoxic and how the neurotoxicity testing guidelines
actually are implemented. When chemicals have a direct effect on the nervous system, there is
usually little disagreement about identifying them as neurotoxic. Some chemicals, like carbon
monoxide, can have both direct and indirect effects on the nervous system for example by
interfering with delivery of oxygen to the nervous system and interfering with respiratory
enzymes in the nervous system. Such materials are routinely considered neurotoxic and are not
confused with simple asphyxiants, such as nitrogen at normal atmospheric pressures, which are
usually not considered neurotoxic even though exposure to them can cause behavioral changes.
Of much greater concern is how to interpret data derived from studies conducted at
exposure levels that are systemically toxic and many orders of magnitude higher than anticipated
environmental exposure levels. It is quite clear that chemicals given in large doses can cause
behavioral changes by disrupting non-neural target tissues. Without invoking concerns about
hepatic encephalopathy, it is an ordinary experience of most lexicologists to note that animals
change their behavior when a chemical "makes them sick" by damaging the liver or other
parenchymal organs. Chemicals that cause behavioral changes, even adverse ones, at dose levels
that result in systemic toxicity should be controlled as systemic toxicants and not as
neurotoxicants. To do otherwise would waste precious resources devoted to establishing
chemical control procedures. In a clinical setting, we would chastise a practitioner who would
warn a patient to be on the lookout for signs of convulsions and not jaundice when he/she
knows that the patient will be nearly dead from liver failure before the onset of convulsions.
Why in an environmental setting would we warn people about neurotoxicity when we know that
liver toxicity is the critical endpoint and that if we prevent hepatotoxicity, we will prevent
behavioral changes from occurring?
C-48
-------
Much of the controversy about direct and indirect effects could be eliminated by
recognizing that neurotoxicity tests should not be conducted at excessively high dose levels or
dose levels that result in systemic toxicity. The practice of deliberately confounding
neurotoxicity studies by using massive dose levels is one which should be discouraged. It only
prevents the risk assessor from having a clear picture of the risks associated with exposure to a
particular chemical. : _ ;
Extrapolation of Neurotoxicity Data from Laboratory Animals to Humans
Uncertainty factors are used to express the degree of confidence that the no-effect levels
that are determined in test animal studies are sound for determining safe human exposure.
Generally, the less confidence one has in the data, the greater the uncertainty factor. The
converse also should be considered when determining uncertainty factors: the uncertainty
factors should be decreased as one becomes more confident in the data. As more and more
sensitive endpoints are used to improve the reliability of the no-effect level determination,
smaller uncertainty factors should be considered. In this way, there is an incentive to collect
more and better data and thus provide better identification of materials that present potential
human health problems.
There should be a significant concern about extrapolation of nonspecific effects in
animal studies to humans. For example, in Table 1 of the draft, hemorrhage in nerve tissue,
GFAP increases, increases or decreases in motor activity, and changes in brain weight are listed
as examples of potential endpoints for neurotoxicity. Minor hemorrhage in nerve tissue is quite
common as an agonal change in animals dying for a variety of reasons and ordinarily should not
be regarded as evidence of neurotoxicity. Quantitative changes in GFAP should riot be shown
in Table 1 as an interpretable endpoint because there are insufficient data and experience with
GFAP measurements to allow an understanding of what changes in GFAP mean. Motor
activity also is an endpoint that can change because of nonspecific illness in an animal and
therefore should not be considered as indicative of neurotoxicity. . ''-:-:-
C-49
-------
The issue of brain weight is discussed in more detail on p. 14 of the draft guidelines.
The draft states without support that statistically significant changes in absolute brain weight
should not be discounted because relative brain weights appear normal. The draft also indicates
that "the brain is usually protected in weight loss," although it doesn't indicate under what
conditions it is not protected. Since most neurotoxicity studies are conducted during the growth
phase of the test animals, concern about brain weight differences will be due to growth
retardation, which can affect the ultimate size of the entire animal including the brain. There
seems to be no basis for the draft to state without equivocation that a change in brain weight
should be considered as an adverse neurotoxic effect.
Page 17 of the draft states: "An alteration in the structure of the nervous system is
regarded as evidence of neurotoxicity." Alterations in structure, like other endpoints, should be
interpreted carefully. While most alterations in structure may be considered adverse, all
changes are not adverse. For example, antioxidants have been given to laboratory animals to
reduce the deposition of lipofuscin in the brain. Such changes are usually interpreted as
positive or useful changes and not neurotoxicity.
Page 45 discusses the interpretation of in vitro data in neurotoxicity risk assessment. The
draft indicates that demonstrated neurotoxicity in vitro in the absence of in vivo data, should be
regarded as suggestive evidence of neurotoxicity. This conclusion is not consistent with the
preceding discussion in the draft about the difficulties of interpreting in vitro studies. The draft
also indicates that in vitro data confirmed by in vivo data are convincing evidence of
neurotoxicity. The draft should indicate that for in vivo data to provide confirmation of in vitro
data there should be a plausible biological association between the in vitro and the in vivo
endpoints. In the absence of biological plausibility, the data sets should not be regarded as
complementary.
On page 52, second paragraph, the draft seems to establish a new standard for
determining that a chemical is neurotoxic in spite of finding that multiple individual studies are
negative. The draft states: "In some cases, while no individual study may be judged sufficient to
establish a hazard, the total available data may support such a conclusion." The rationale for
this new standard appears to be given in the example on p. 52, which suggests that greater
C-50
-------
concern should be felt for chemicals we know something about rather than those we know
nothing about. The use of marginal data in this way does not appear to be appropriate.
On page 53, the draft appears to discount all studies that have not been conducted
according to the EPA neurotoxicity guidelines. There is no basis for such a determination.
Neurotoxicity studies have been conducted for many years in the absence of EPA guidelines,
and there should be no scientific basis for the Agency to summarily dismiss the entire scientific
literature developed either before or after the development of EPA guidelines. A better
approach might be to describe the attributes of adequately conducted and reported studies and
use these criteria as a test for the acceptability of data to be used in the risk assessment process.
Interpretation of Behavioral Data
Behavioral changes in toxicology studies often are seen as nonspecific endpoints, which
generally require correlation with other endpoints before they can be considered evidence of
neurotoxicity. Behavior is the end result of the many interactions the nervous system has with
the environment. Increases or decreases in behavioral signs frequently indicate a response to a
stimulus which allows an animal to adapt to its environment. In common with other endpoints
that are considered sensitive, there should be a concern that these endpoints may be nonspecific
and error prone, and their use in isolation might lead to incorrect regulatory classification of
chemicals for neurotoxicity.
While behavior has been regarded by some as the most sensitive indicator of
neurotoxicity, there is little evidence to support this premise, and there are substantial reasons
to consider the issue irrelevant, because it is difficult to imagine how behavior could be the most
sensitive endpoint for all types of neurotoxicity.
G-51
-------
Sufficient Evidence to Categorize Neurotoxic Hazards
The criteria for classification in the "Sufficient Human Evidence" category are too broad
and will force many nonneurotoxic chemicals to be improperly categorized. Before an
epidemiologic study is used to determine that there is sufficient human evidence of
neurotoxicity, there should be a rigorous review of all available data (including negative data)
that should reveal (1) that the study meets the criteria being developed for Good Epidemiologic
Practice and (2) there is biologic plausibility that explains why the observed effects should be
considered causally related to a chemical exposure. The use of simple association with exposure
as a criterion for the acceptability of epidemiologic data is inappropriate. Use of a case study to
determine sufficient evidence is even more likely to produce errors in classification, particularly
if the adequacy of "supporting data" is not rigorously defined. The "supporting data" for case
studies should provide evidence of a biological plausibility between the effects observed and the
observations made in the case studies. In vitro data and nonspecific findings in animal studies
do not provide the types of data needed to support the validity of case study reports. The main
support for the use of these types of data is based on the unsupported statement that "most
neurotoxicants have been 'discovered' in humans," with the implication that chemicals,
particularly man-made chemicals, are creating widespread neurotoxic disease.
Sufficient Experimental Animal Evidence
The discussion about what constitutes sufficient experimental animal evidence for
classification includes the criteria for determining that a chemical is not a neurotoxicant. The
risk assessment document should provide a clear category for such materials that are found to
be not neurotoxic.
Insufficient Evidence Category
This category would contain materials for which data would not provide sufficient
evidence for classification, but the text of the document also should discuss the lack of need for
C-52
-------
extensive neurotoxicity studies for all chemicals. Chemicals that should be considered for
formal risk assessment should be those that potentially present concerns to human health.
Many chemicals are tested in routine toxicology tests that provide a first screen for
neurotoxicity. Chemicals that are negative for neurotoxicity on routine screening tests should
not be considered as insufficiently tested for neurotoxicity but rather they should be placed into
a category indicating a low degree of concern.
Categorization in General
In order to address the needs of the public, the Agency should present factual
information and minimize the potential for such information to be misused. The proposed
classification schemes facilitate such misuse. Although the Agency has attempted to address this
issue in the text, describing the "Sufficient" and "Insufficient" categories, the Agency's efforts fall
short of what is needed. Many, particularly at the state level, and those pursuing legislation
through the initiative process, will interpret the "Sufficient" category as meaning a hazard exists,
rather than that "sufficient information exists to judge whether or not a human neurotoxic
hazard could exist." I strongly suggest the title of the categories be changed to reflect the
Agency's interest. A potential change might be "Sufficient/Insufficient Data to Proceed with
Hazard Characterization." Ideally, categorization schemes should be eliminated. In the
meantime, the Agency has an obligation to minimize the potential for misuse of the scheme.
C-53
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
John L. Orr, Ph.D., D.A.B.T.
Southwest Research Institute
P.O. Drawer 28510
San Antonio, TX 78228-0510
These comments are presented according to sections of the guidelines document.
2.1 Neurotoxicity
The definition of adverse effect is usefully broad. The listing of behavioral changes,
however, should include: (1) motivational changes (2) degradation of skilled performance, and
(3) degradation of decision quality as examples of adverse effects.
2.2 Neurotoxicant
There is a discrepancy between the number of types of validity and the number listed.
Nonetheless, the comments on the different types of validity deserve a separate heading and
consideration as a device to summarize the status of Sections 3.1 and 3.2. One could visualize
the types of validity as a series of stages to the conclusion that a material is a neurotoxicant:
Predictive Validity
Concurrent Validity
Construct Validity
Content Validity
C-54
-------
3.2.4.3 Schedule Controlled Behavior \
I tend to think of simple schedules as the fundamental units from which the complex
instrumental behavioral systems such as those used for tests of sensory systems, memory, and
learning ability are synthesized.
Sections 3.2.4.3 and 3.2.4.4 could well be subcategories of a category called performance.
3.2.5 Developmental Neurotoxicity
Several of the desirable features listed on pp. 37 and 38 such as replicate study design
and pharmacologic challenge also should apply to other neurotoxicity studies.
33.1 Pharmacokinetics.
Eventually issues will need to be addressed in neurotoxicology that are similar to those
addressed in the area of carcinogenesis risk assessment. Factors such as fraction of life span
and relevant scaling parameters will need to be studied.
Statistical Considerations
The discussion of power appears to have something missing. To obtain an alpha of 0.05
and power of 0.8, you need to specify an effect size!
The issues related to repeated-measures statistics and the use of corrections for multiple
comparisons deserve technical attention.
C-55
-------
3.3.4 In-vitro Data in Neurotoxicology
This is a good discussion, however, the analog of the power issue is not addressed. Is
lack of neurotoxicity in vitro to be considered evidence that a compound is not neurotoxic in
vivol
5. Adequacy. ...
This section could be strengthened by working through a couple of examples for which
some data are available.
C-56
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Thomas J. Sobotka, Ph.D.
Neurobehavioral Toxicology Team, HFF-162
Center for Food Safety and Applied Nutrition
Food and Drug Administration
8301 Muirkirk Road
Laurel, MD 20708
Sections Hazard Identification
The document does not discuss the application of tiered testing or the differences in
informational value obtained at each level of testing in the assessment of neurotoxicity. There is
no distinction between stage 1 screening and stage 2 or 3 in-depth neurotoxicity testing. Is it
the intent to use or allow the use of screening (FOB) information as the basis for determining
NOAEL or LOEL and/or for making risk assessment determinations?
Section 3.2.1 Structural Endpoints of Neurotoxicity
With regards to reductions in brain size, the absolute association of decreased brain size
(particularly with reference to whole brain size) with neurotoxicity, regardless of the body size, is
not warranted. While it may generally be true that "most of the body weight reduction reflects a
loss of body fat" and that "dieters do not lose brain tissue," it is equally true to say that the
brains of small people are not abnormally small. Body size must be taken into consideration.
While an absolute reduction in (whole) brain weight in adult animals may certainly be regarded
as a possible indication of neurotoxicity, a decrease in brain weight m-and-of-itself should not be
taken as definitive evidence of neurotoxicity. However, if one considers the size of discrete :
brain regions (weight, width, or length), there may be more of a reason to argue that absolute
reductions (or increases) in the size of specific brain regions may be associated with
neurotoxicity regardless of body size.
C-57
-------
Section 3.2.1 Structural Endpoints of Neurotoxicity
This section also should describe the nonneuronal components of the nervous system
that also may be involved in neuropathological effects of chemical substances, i.e., the glial
elements.
Section 3.2.1 Structural Endpoints of Neurotoxicity
Page 17, lines 21-25: "Since increases in GFAP may be an early indicator of neuronal
injury in adults, treatment-dependent increases in GFAP are considered neurotoxic. Changes in
GFAP levels have been observed in immature animals, but have not been conclusively linked to
neuronal injury."
The meaning of the latter sentence is unclear. Does this mean that treatment-related
changes in GFAP have been observed in immature animals but not conclusively associated with
neurotoxicity? Furthermore, if changes in GFAP are considered neurotoxic in adults but not in
immature animals, it seems that the document should make some statement as to when this age-
related transition in significance of GFAP occurs from the immature to adult animal.
Section 3.2.2 Neurophysiological Endpoints of Neurotoxicity
As an endpoint of neurotoxicity, neurophysiology encompasses two broad areas: (1)
electrophysiology, which deals primarily with the electrical activity associated with the nervous
system, and (2) general physiology, which involves the functioning of peripheral organs that are
controlled or modulated by the nervous system. In general, the use of electrophysiological
techniques (e.g., EEG, sensory-evoked responses, nerve-conduction velocity) provides a means
of directly assessing neuronal function, whereas general physiological status (e.g., blood pressure,
lacrimation, salivation, body temperature) provides an indirect means of assessing neuronal
function. One of the key roles played by the nervous system is to orchestrate the general
physiological functions of the body to help maintain homeostasis. To this end, the nervous
C-58
-------
system and many of the peripheral organ systems are integrated and functionally
interdependent. Since many peripheral organ functions involve neuronal components, changes
in such physiological endpoints as blood pressure, heart rate, EKG, body temperature,
respiration, lacrimination, or salivation may indirectly reflect possible treatment-related effects
on the functional integrity of the nervous system. However, since physiological endpoints also
depend on the integrity of the related peripheral organ itself, changes in physiological function
also may reflect systemic toxicity involving that organ. Consequently, the neurotoxicological
significance of a physiological change must be interpreted within the context of other signs of
toxicity. When performed properly, neurophysiological techniques provide information on the
integrity of defined portions and/or functional operations of the nervous system.
Section 3.2.2.3 Convulsions
The following two statements are made in this section: (1) "Behavioral convulsions that
occur only at legal or near lethal dose levels do not necessarily constitute evidence of
neurotoxicity," and (2) "Convulsions that occur in the presence of systemic toxicity are not
necessarily evidence of neurotoxicity." Both of these statements seem to be in contradiction
with the statement on p. 5/1.2-3 that "Chemicals may produce neurotoxicity effects by either
direct or indirect means." In the instance of convulsion, it seems that the endpoint of
neurotoxicity is the convulsion. The mechanism whereby the convulsion is produced is not
necessarily the determining issue. The presence of convulsion, by whatever mechanism,
indicates that neurotoxicity is a component of the toxicological profile for that chemical
treatment. It may be of little significance in terms of the overall toxicity of that chemical,
particularly if convulsions only occur at lethal or near lethal doses, but none the less the
convulsion indicates a treatment-related neurotoxic effect. On the other hand, the convulsion
may be the most significant part of the toxicological profile, including the systemic toxicity, for
that particular chemical treatment.
C-59
-------
Section 3.2.23 Convulsions
The statement is made that "Convulsions that occur in the presence of systemic toxicity
are not necessarily evidence of neurotoxicity." But then the next sentence states that "In such
cases, neurophysiologjcal recordings of electrical activity in the brain that is indicative of
seizures provides evidence of neurotoxicity." Does this mean that only electrical recordings are
to be accepted as reliable measures of seizures and as evidence of neurotoxicity? Why isn't
observation of convulsion accepted as evidence of neurotoxicity? Are electrical recordings
accepted only when there is systemic toxicity present? This needs some clarification.
Section 3.23 Neurochemical Endpoints of Neurotoxicity
This section makes some apparent contradictory statements. On p. 23/lines 26-30, the
statement is made that "By themselves [neurochemical effects], demonstrated does-related
effects on these endpoints are not evidence of neurotoxiciry. Many neuroactive agents can
increase or decrease neurotransmitter levels in the brain but such changes are not necessarily
indicative of neurotoxicity.". Yet, on p. 24/lines 1-3, it states rather definitively that "...agent-
induced increases in specific neurotransmitters in the brain, or decreases in specific brain
regions, especially when such changes are persistent, are evidence of neurotoxicity." This section
needs some clarification.
Section 3.2.4 Behavioral Endpoints of Neurotoxicity
The point of the examples in the second paragraph is unclear. The statement is made
that "...toxicant-induced changes in behavior can result from a variety of physiological changes in
addition to effects on the nervous system." But it is not very clear how this statement relates to
the examples given, i.e., "changes in relative and absolute organ weights may be signs of systemic
toxicity" and "Food and water consumption data are necessary in determining the relative
contribution of general toxicity..." It might be worth reiterating that indirect toxicant-related
effects on behavior (e.g., behavioral changes elicited via toxicant-induced effects on physiological
C-60
-------
systems) are regarded as neurotoxic, but then stating that it is still important to discern,
whenever possible, whether systemic toxicity may be involved in the resulting behavioral or
neurotoxic effects. Several common endpoints used as signs of systemic toxicity include relative
and absolute organ weight changes, altered food and water consumption, etc.
Section 3.2.4.1 Functional Observational Batteries
Page 27, lines 7-10. The test results from the FOB should certainly be judged, as the
document states, according to the number of signs affected, the dose at which neurotoxic signs
are observed, and the persistence of the effects. But, consideration should also be given to the
nature of the effects observed as well as their severity.
Section 3.2.4.1 Functional Observational Batteries
Section 3.2.4.2 Motor Activity
Section 3.2.4.3 Schedule-Controlled Behavior
Section 3.2.4.4 Specialized Tests for Neurotoxicity
Page 27, lines 13-18; p. 31, lines 1-3; p. 32, lines 9-14; and p. 32, 3rd paragraph. Some
discussion at the workshop should be centered around the issue of whether neurological signs
occurring at the high dose in conjunction with other overt signs of toxicity should or should not
be considered neurotoxicity. Why is this not considered a possible indirect neurotoxic effect
associated with treatment? It should be clear that dose is a very significant factor in the
neurotoxic potential of any chemical substance. For some chemicals, neurotoxicity may only
occur at high doses which produce general systemic toxicity. It is important to consider also, as
is alluded to on p. 39 with reference to maternal/pup toxicity, that the systemic toxic effects may
be reversible but the neurotoxic effects may not be. Should exceptions be made? If such effects
are not considered neurotoxic, then it seems that the position that "Chemicals may produce
JC-61
-------
neurotoxicity effects by either direct or indirect means" (p. 5) should be reconsidered or, at
least, restated.
Section 3.2.4.1 Functional Observational Batteries
Page 27, line 22. The developing organism is referred to in this section. Does this
means that the FOB is to be applied in developmental studies, as well as adult studies?
Section 3.2.4.3 Schedule-Controlled Behavior
Page 32, line 5. Define quarter life.
Page 34, Table 5. The word "pyrethroids" is in the wrong column.
Section 3.2.4.4 Specialized Test for Neurotoxicity: Cognitive Function
Page 37, lines 5-8. The sentence "...it is not necessarily the case that a decrease in
responding on a learning memory task is adverse...," as written, may be misinterpreted to mean
that decreased learning/memory responding may not be considered adverse.
Section 3.2.5 Developmental Neurotoxicity
Pages 38-39. There are a number of indented sentences. What are these supposed to
be? Are they factors to consider in evaluating developmental neurotoxicity data as part of the
risk assessment process? Some introductory statement should be inserted to explain what these
indented sentences represent.
C-62
-------
Page 40, lines 7-16. The issues presented in this paragraph are generally applicable for
the interpretation of all types of neurotoxicity data. It would be appropriate to expand the
discussion of these ideas and to include this in one of the introductory sections of this document
or in the guidelines for interpretation in Section 5.
Section 3.3.3 Statistical Considerations
If power is defined as one minus the probability of a Type II error (1.13), then how can
power increase with sample size if the probabilities of Type I and Type II errors are held constant
(1.21-22)? '
SectionS Adequacy of Evidence...
Page 51, lines 11-12. The statement that "most...neurobehavioral changes are regarded
as adverse" is too unqualified and too general and does not address the critical issue of how
adverse should be defined. Obviously, there are a number of factors that must be considered in
determining whether any particular behavioral change should be considered adverse, including
severity and nature of the effect. In making this determination of adversity, consideration
should.be given to Section 2: Definitions (p. 4 of this document) which states that adverse
effects include: (1) unwanted effects and/or (2) any alteration from baseline that diminishes the
ability to survive, reproduce, or adapt to the environment.
Sections. Adequacy of Evidence -
Page 52, bottom paragraph. This paragraph brings out an excellent point and expresses
it very wellthe potential for neurotoxic hazard may be very dependent on exposure route and
exposure level.
C-63
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3, 1992
Peter S. Spencer
Center for Research on Occupational and Environmental Toxicology
Oregon Health Sciences University
Portland, OR 97201
The following comments are presented according to section of the proposed guidelines
document.
2.1 Neurotoxicity
Expand definition to: "Any acute and chronic adverse change in the structure or function
of the developing, mature, or aged central and/or peripheral neuro-muscular system (including
sensory and special sensory organs) attributable to a chemical, physical, or biological agent."
This represents the true scope of the guidelines and introduces the specific notion of cause and
effect. It explicitly includes : acute and chronic effects; changes at all stages of life; actions on
brain, spinal cord, nerve, muscle, sense organs, and special sensory organs. Still unclear,
however, is whether the definition is intended to include adverse effects on the neuroendocrine
system as implied by the statement: an "alteration from baseline that diminishes the ability to
reproduce." Failure to "adapt to the environment" could result from many perturbations
unassociated with an action on the developing or mature nervous system.
If such an all-encompassing definition is selected, it will be important to discuss the long-
term consequences of agent-induced changes. On the basis of current understanding (always
limited), most agent-induced disorders of the adult nervous system improve after exposure has
ceased. In some case, improvement may not occur until the disorder has played out for some
weeks after the exposure terminated (a phenomenon known as coasting). The degree of
recovery (reduction of the signs of the disorder) varies with the agent and the exposure
condition. All of the following are possible: complete recovery with no functional or anatomical
residua or susceptibility to re-exposure to the agent; apparent complete recovery with subclinical
C-64
-------
anatomical or functional residua and increased susceptibility upon re-exposure to the agent;
partial recovery with overt but stable anatomical and functional abnormalities (which may slowly
increase in severity with the advance of age). Some conditions may not recover (e.g., OP-
induced spasticity), and others may even be progressive (e.g., tardive malignant neurological
degeneration following acute "exposure to carbon monoxide). The problem of long-latency
effects is treated in Section 2.2.
Paragraph 2: Change "chemical" to "biochemical" or "molecular and cellular." Add to
the end of the paragraph: "neurological syndromes spanning a wide variety of usually mixed
behavioral abnormalities." ; ,.
2.2 Neurotoxicant : . -
The tautologous distinction of "neurotoxicant" and "neurotoxin" has been dropped in
most professional circles. The distinction almost always causes confusion: "neurotoxicant is a..
. biological " and Neurotoxins are naturally occurring* .." More important is to state that
"many if not all chemicals have neurotoxic potential in certain dose and exposure conditions,
including chemicals which are required for normal physiological function." Use of glutamate as an
example of a neurotoxicant/neurotoxin (?) then makes sense.
The issue of long-latency neurotoxicity is not addressed in this section, although it is
alluded to on p. 55, lines 14-16. While available data are too meager to impose regulatory
standards, it is important to mention this concept. There are two ides, neither of which is
inconsistent with the other: (1) chemical exposure may elicit a subclinical lesion which, as a
result of age-related attrition of the same cell population, may surface years or decades later as
a progressive disorder (e.g., MPTP and subclinical Parkinsonism); and (2) certain agents may act
as "slow toxins" which trigger clinically silent events 'that come to clinical significance weeks
(organophosphates), months (carbon monoxide), years/decades (cycad toxins and western Pacific
amyotrophic lateral sclerosis/Parkinsonism-dementia complex) later/Concern for the existence
of long-latency effects justifies inclusion of subclinical anatomical and functional changes in the
guidelines. --'..- '
C-65
-------
3 Hazard Identification
An explicit hierarchy of studies by which to judge relevance and attribute weight is
proposed. For example, if the goal is to evaluate hazard to humans from specific agent(s), then
the following sources of data should be given descending weight in the evaluation process:
1. Controlled human studies. Studies of humans exposed to an identified
agent/mixture for a defined period at a known dose. (Most of these are to be
found in the literature dealing with the side effects of therapeutic drugs, a
literature often ignored by neurotoxicologists.)
2. Controlled animal studies. Experimental studies of a species exposed to an
identified agent/mixture for a defined period at a known dose.
3. Uncontrolled human studies. Non-epidemiological reports of humans exposed to
an identified agent/mixture.
4. Uncontrolled epidemiological studies. Studies that offer unproven associations
between agents and health effects in humans and animals.
5. Uncontrolled animal studies. Reports of animals exposed to an identified
agent/mixture. (Mostly from veterinary studies, another source of literature
commonly ignored by neurotoxicologists.)
3,1.1 Epidemiological Studies
The classification in #3 gives less weight to reports of epidemiological associations
between agents and effect than a series of well performed case reports or controlled animal
studies. Indeed, epidemiological studies are usually unable to demonstrate causation and must
rely on other studies (i.e., controlled studies with experimental animals) for this purpose.
Regulating agents on the basis of epidemiological conjecture is inappropriate.
3.1.13 Outcome Measurement
Neurotoxicants that cause central-peripheral distal axonopathies (a neologism I am guilty
of introducing!) are hardly "the most prominent categories. . . ." It seems to be a largely
stereotyped response of the nervous systems to a wide variety of chemicals of disparate
structure, but so are many other types of chemically induced neurological abnormalities of
equally great or greater clinical significance.
C-66
-------
Far too little weight is given to the neurological examination. This has been the
cornerstone of diagnosis of nervous system disease and deserves to be represented as such.
Interpretation of the results of the sensory-motor examination vis-a-vis overall neurological
status is far superior than anything presently offered by behavioral techniques. However, the
neurological examination is weak on the precise assessment of mental state.
The discussion of the electrophysiological methods is poor: e.g., peripheral neuropathy
requires assessment of nerve conduction properties (nor velocity alone). There is no mention of
contemporary human brain examination/assessment techniques, such as computerized tomograhy
(CT), nuclear magnetic resonance spectroscopy (NMR),, single photon emission tomography
(SPET), or positron emission tomography (PET).
3.1.1.4 Confounding Variables
The document seems to treat age as a confounding variable. Presumably, the guidelines
are being developed for the protection of individuals of all ages.
Table 1. Spell out acronyms. Change "Neurophysiolbgical" to "Neurophysiological and
Functional" and add endpoints relating reproducible changes identified by contemporary
techniques, such as "change in the neurotransmitier marker by PET." ,
Add a new table listing the rich variety of neurological endpoints following exposure to
chemicals. Include chemical/physical agents with direct/indirect effects on the brain, spinal cord,
special sensory organs, sensory motor and autonomic nerve fibers/end organs, muscle and (?)
neurpendocrine system.
The number of examples listed under "Behavior Endpoints" is disproportionately high.
3.2 Animal Studies
This section should begin with a description of gross physical/behavioral changes,
including convulsions (move from Section 3.2.2.3).
C-67
-------
3.2.1 Structural Endpoints of Neurotoxicity
Brain edema may be seen grossly as a loss of the normal pattern of convolutions. Brain
weight: Reduce discussion to 1-2 lines; it is rarely used or useful. It is unclear that breakdown
of cells may be preceded by accumulation, proliferation, or rearrangement of structural
elements. Part of the classification used was introduced by P.S. Spencer and H.H. Schaumburg
(1980), Experimental and Clinical Neurotoxicology. Williams and Wilkins, Baltimore, MD, and
not as cited.
"Neurodegeneration" is not used to refer to changes in nerve terminals; most commonly,
the word refers to neuronal loss seen in progressive neurodegenerative diseases.
Developing nervous system: The key vulnerability of the nervous system to exposures at
particular sensitive developmental times needs to be emphasized (an agent that is refractory at
some points during development may cause devastative abnormalities if exposure occurs at
certain states of organogenesis).
The aged appear to be uncommonly vulnerable to chemicals with neurotoxic potential
because: liver and kidney metabolism may be compromised; body weight may be lower; and
certain groups of nerve cells and their processes may have undergone 'normally occurring' age-
related compromise.
The section dealing with neuropathology needs to address the issue of artefactual
changes (induced by preparative trauma, fixatives, dehydration steps, etc.) which all too often
are mistakenly taken as evidence of neurotoxicity.
Table 2. This is very poor and reflects much misunderstanding. Buckthorn toxin
probably causes an axonopathy, not as listed under myelinopathy. MPTP induces degeneration
of nerve terminals and substantia nigra neurons. An organophosphate anticholinesterase agent
is an example of a chemical that acts at (certain) nerve terminals. Lead and peripheral
neuropathy are adequate examples, although diphtheria toxin (a biological agent) might be a
useful addition. Acrylamide, hexacarbons and carbon disulfide are all associated with
C-68
-------
"Peripheral Neuropathy," not "Vitamin Deficiency." Proximal axonopathy and motor neuron
disease are not equivalent.
3.2.2.1 Nerve Conduction Studies
Refractory period, chronaxy, and rheobase are rarely used and difficult to interpret in
relation to neurotoxicity.
3.2.3 Neurochemical Endpoints of Neurotoxicity
All endpoints, but most especially neurochemical endpoints, acquire validity and standing
when the magnitude of the change is shown to vary as a function of the dose of the agent
"administered and, if possible, the duration of chemical exposure.
Agents that perturb axonal transport function (fast or slow anterograde, retrograde)
should be included as neurochemical endpoints of neurotoxicity.
3.2.4 Behavioral Endpoints
This section has received much more detailed treatment than sections dealing with
morphology, neurophysiology, neurochemistry. Does it reflect a bias that neurobehavioral
methods are more important than other methods for the detection of neurotoxicity? Little
guidance is offered as to the difficulty, reliability, and reproducibility of the techniques, or their
relevance to human neurological dysfunction, Even more troubling is the recognition on p. 51
that "there are adverse behavioral effects that may not reflect a direct action of the nervous
system."
Paragraph 2. Define "general toxicity."
3.2.4.1 Functional Observational Battery
The statement on p. 27 regarding the distinction between evidence of neurotoxicity and
lack thereof is crucially important. The discussion as written is unclear. A concrete example of
what EPA has in mind would be more helpful.
C-69
-------
3.2.4.2 Motor Activity
It is doubtful that "neurotoxic agents generally decrease motor activity." Many
compounds increase the central nervous system (CNS) excitation and, at certain dosages,
increase motor activity (see also p. 31, last two lines). EPA is advised to avoid these types of
generalizations, especially if they are untrue.
3.2.4.4 Specialized Tests for Neurotoxicity
Ataxia may result from a motor (cerebellar) or sensory (vestibular, proprioceptive)
defect.
3.2.5 Developmental Neurotoxicity
Emphasize critical role of the timing of agent exposure (see Section 3.2.1). This thought
is currently buried in the text. Table 6 reinforces the notion that the agent, rather than the
agent and the timing of exposure, are factors dictating developmental neurotoxicity.
Replicate (or triplicate) study design adds confidence to any study.
3.3 Other Considerations
Parts of the CNS (circumventricular organs) and peripheral nervous system (dorsal root
ganglia and sympathetic ganglia) normally lack a blood-brain regulatory interface. Access of
blood-borne chemicals to these regions is immediate and total.
Statistical Considerations
This discussion should be placed after Section 3.3.4.
33.4 In vitro Data in Neurotoxicity
The second paragraph is very negative. The use of organotypic explants has been
extensively explored. They reproduce the structural and functional features of nervous tissue in
vivo, and they respond to chemical, physical and biological agents in a manner that reproduces
changes in vivo. They have been useful in demonstrating the site of chemical action and the
resulting effect, and they are also predictive of neurotoxicological effects in animals.
C-70
-------
4 Dose Response Assessment , .
It is important to recognize that a single agent may act at multiple points in the nervous
system at a single dose/time level, and may attack different sites at other dose/time levels. The
discussion ignores the problem of extrapolation between acute and chronic effects on the
nervous system and behavior. A NOAEL may be critically important to avoid a chronic
neurological effect (e.g., n-hexane neuropathy) but of little relevance to short-term exposures
where much greater levels may be tolerated without harm.
5 Adequacy of the Evidence for the Determination of Hazard
(And in Table 7) An explicit hierarchy of studies by which to judge relevance and
attribute weight is proposed. For example, if the goal is to evaluate hazard to humans from
specific agent(s), then the following sources of data should be given descending weight in the
evaluation process: .
1. Controlled human studies. Studies of humans exposed to an identified
agent/mixture for a defined period at a known dose. (Most of these are to be
found in the literature dealing with the side effects of therapeutic drugs.)
2. Controlled animal studies. Experimental studies of a species exposed to an
identified agent/mixture for a defined period at a known dose.
3. Uncontrolled human studies. Non-epidemiological reports of humans exposed to
an identified agent/mixture.
4. Uncontrolled epidemiological studies. Studies that offer unproven associations
between agents and health effects in humans and animals.
5. Uncontrolled animal studies. Reports of animals exposed to an identified
agent/mixture. (Mostly from veterinary studies.)
Use of structure-activity considerations should be limited to those classes of chemicals for which
an understanding exists.
The notion that "correlations support a coherent and logical link between behavioral
effects and biochemical mechanisms" is very valuable. Another important concept is the issue of
replicability. Another omitted concept is that of dose-dependent changes which allow the
C-71
-------
researcher to distinguish spurious effects from those related specifically to the agent under
study.
7 Risk Characterization
There is an implicit understanding that guidelines will be developed for the protection of
normal healthy individuals (of differing ages or ethnicity), rather than those with some inherent
susceptibility (e.g., diabetics with subclinical neuropathy who work with acrylamide or «-hexane).
Should this issue be addressed?
Some ethnic groups are more susceptible to certain chemical agents because of specific
genetic phenotypes that dictate metabolic capacity. The best known example is the regulation of
acetylation for drugs like isoniazid: genetically slow acetylators metabolize (acetylate) the agent
slowly and are more susceptible than fast acetylators to isoniazid neuropathy and
encephalopathy.
C-72
-------
Premeeting Comments for
Peer Review of Neurotoxicity Risk Assessment Guidelines Workshop
Washington, DC
June 2-3,1992
Barry W. Wilson
Department of Avian Sciences
University of California
Davis, California 95616
My comments on the proposed guidelines deal first with the panel outlines and then with
the draft guidelines themselves.
Panel 1: Neurotoxicity as an Appropriate Endpoint for Environmental Risk Assessment
I agree with the general statement that neurotoxicity is an appropriate endpoint for
assessing risk. I also concur that, the opinion is properly founded on the importance of the
nervous system to human health, the large number of agents known to cause neurotoxicity, and
the possibility that there may be significant exposure of humans to neurotoxic agents. I also
agree with the caveats that inadequate information is available on many potentially harmful
chemicals and that better standards for evaluating neurotoxic potential are needed, especially
with regard to low-level chronic exposures and possible synergistic effects.
One issue that touches upon many parts of the report is deciding what constitutes a no-
observed adverse-effect level (NOAEL). Phrased another way, when do behavioral, physiological
and/or biochemical effects that are acceptable as markers that exposure to a chemical has
occurred become signals of an adverse effect? The definition of an adverse effect on p. 4, "(1)
unwanted effects, and/or (2) any alteration from baseline that diminishes the ability to survive,
reproduce, or adapt to the environment" specifically in the "structure and function of the central
and/or peripheral nervous system...," is generally a reasonable one. But converting it into
practice requires detailed knowledge of exposure levels and effects for probably both humans !'
and experimental animals.
C-73
-------
Panel 2: Interpretation of Neurotoxicity Data When Effects Are Transient
The statements that "nerve cells have limited capacity for regeneration" and that
"apparent recovery ...represents activation of reserve capacity..." need clarification and perhaps
some delimiting. For example, there is rapid regrowth of axons and reinnervation of denervated
motor neurons, especially in young animals, presumably without loss of reserve capacity. Is
there evidence that recovery from exposure to compounds like acrylamide and n-hexane that
require repeated exposures to produce major neural damage uses up some "capacity" of the
cells? Once a cell body dies, the axon and other cell processes will die too, but the synthetic
capacity of nerve cells often would permit eventual full recovery if the cell body was not
destroyed. Perhaps different kinds of "recoveries" could be considered. For example, some
recoveries may be due to the destruction of the chemical by metabolism or its loss by excretion;
in other cases recovery may require synthesis of new proteins and other tissue constituents.
Whether or not pharmacological effects are traditionally considered short-acting and
toxicants long-acting, many naturally occurring toxicants such as curare and eserine are acute in
their actions, and some nerve gases that can hardly be considered "pharmacological," act acutely
and have been shown to have little long-term effects after repeated exposures (Wilson et al.,
1992b).
The proposal that some effects of exposure to a chemical may only appear following
environmental or pharmacological stresses is an interesting one, since it points towards different
paradigms than are currently used in testing for neurotoxicity. This proposal should be
buttressed by examples.
Panel 3: Agents Acting through Indirect and Direct Means Can Be Considered Neurotoxic
I am not sure that I agree that "effectsproduced through direct or indirect means are
functionally equivalent... ." Few chemicals are so specific in their action as to have only one
major effect, and events that occur downstream from an initial chemical-receptor interaction
may be multiple. Indeed, most chemicals affect the body at more than a single site of action.
C-74
-------
Tetrodqtoxin, and other similar natural channel blocking toxins, are some of the few examples
compared to the many "dirty" neuroactive drugs.
Perhaps the report could make more of the phenomenon of bioactivation-and less of
whether an effect is primary or secondary with respect to risk assessment. The report itself
makes the point that "knowledge of exact mechanisms of action is not...necessary to reach the
conclusion that a chemical produces neurotoxicity." Setting reliable NOAELS may require more
knowledge and study. Risk assessment should proceed from clearly established endpoints that
occur at specific levels of exposure regardless of the mechanisms of the toxicity. One issue to
discuss is how one establishes whether direct and indirect actions are the most sensitive
indicators of an effect. It may be important to determine whether or not specific damage occurs
to the nervous system at levels below that which causes damage to other parts of the body. This
determination may be important in defining a chemical as a neurotoxicant and thereby assessing
its risk from its ability to injure the nervous system rather than do damage to other parts of the
body.
Panel 4: Extrapolation of Neurotoxicity Data from Laboratory Animals to Humans
The idea that a neurotoxicant in one species is a neurotoxicant in others, including the
human, neglects examples of neurotoxicants that differ in their actions from species to species
presumably due to differences in bioactivation, metabolic detoxification, excretion, induction of
oxidases, etc. Given the present state of knowledge of the action mechanisms of many
chemicals and their effects on humans the use of experimental animals in establishing the
scientific bases for risk assessment seems unavoidable. The more is learned about the
mechanisms of action of an agent in humans and experimental animals the more likely arbitrary
uncertainty factors can be replaced by factors based on solid data. The fact that other animals
do not have the full repertoire of behavior of the human has a flip side if one considers that
many animals "see" into the ultraviolet, "hear" above the human range, and some even sense
magnetic frequencies. One issue is whether speech, or some other feature of human behavior
(or physiology or biochemistry), constitutes the most sensitive endpoint for detecting a specific
chemical's effect.
C-75
-------
In vitro systems played little role in the report, even though they have been much studied
as possible primary screens for neurotoxicants in the hopes that cell level effects could be
revealed in cultured cell systems and followed up in experimental animals and humans.
Panel 5: Interpretation of Behavioral Data
What is the evidence that behavior is "one of the most sensitive indicators of toxicity?" Is
it not also often the most readily altered by experimental conditions? Can the statement that
behavioral changes appear prior to physiological or morphological endpoints be documented?
The concerns expressed for maximum tolerated doses and chemicals that produce neurotoxic
and/or behavioral changes at high doses may not be as important as worrying about establishing
the lowest doses that yield detectable effects and the systems and species that are the most
sensitive.
If the primary effect of a toxicant is defined as the first biochemical action of the
activated chemical, usually at the receptor level, then it follows that no primary event can be a
behavioral one. Behavior is, a consequence of biochemical, molecular, usually multiple events in
the nervous system. One advantage of emphasizing behavioral endpoints is that alterations in
them may reflect more than one molecular event; one disadvantage is that important
biochemical level events could be masked from expression elsewhere in the nervous system.
Another question is whether a behavioral change, by definition, is always an "adverse
effect," in contrast to biochemical or physiological events that may not be harmful, in and of
themselves?
How many behavioral endpoints have been validated as screens for neurotoxicants?
C-76
-------
Acetylcholinesterase Determinations
The report recommends that "inhibition of "...AChE..." is evidence of neurotoxicity." It is
safe to say that decreases in specific acetylcholinesterases (AChE, E.G. 3.1.1.7) and nonspecific
cholinesterases (BChE, E.G. 3.1.1.8) are generally accepted as bipmarkers of exposure to
organophosphate (OP) and organocarbamate anti-ChE esters, providing that disorders leading
to decreased levels of cholinesterases (ChEs) are not present, regardless of the tissues .
concerned. One issue is whether such decreases constitute an adverse effect in and of
themselves. An SAB/SAP Joint Panel on Cholinesterases did not recommend using statistically
significant decreases in plasma and red blood cell ChEs as indicators of adverse effects (Weiss,
1990). The panel, like the present report, supported using decreases in brain AChE levels as an
adverse effect. However, a recent EPA workshop on choUnesterase methodologies (Wilson et
al., 1992a) found there were large differences between laboratories in the determinations of
tissue ChEs, raising doubts on the use of such data for regulatory purposes. (Variances on the
order of 20 percent were commonplace.) Problems in assay methods were especially important
with carbamates, since many inhibited carbamylated-ChE complexes are readily reactivatable.
Steps are under way to standardize the procedures used by the laboratories, reducing both inter-
and intra- laboratory variances. Until standards are set, one approach would be to recommen4
that AChE measurements be considered on a case by case basis, and referred to EPA's Office
of Pesticide Programs and the panel of scientists they convened.
Another issue is what level of enzyme decrease is acceptable as "adverse." There have
been proposals to make "a statistically significant" decrease in enzyme activity the baseline for
an adverse effect. At first glance, such a criterion seems clear-cut and readily applicable by
trained personnel. But there are several major problems:
So long as there are no standard operating procedures that specify reasonable
confidence limits and variances, the company with the sloppier methods, and thus
the higher variances, will be given higher permissible levels of residues. In other
words, the criterion of a statistically significant difference only works when there
is an even playing field based on standard procedures and internal controls.
Whether or not the criterion of a statistically significant difference is accepted
and adhered to by all laboratories, scientists do not agree about what level of
decrease of brain AChE produces an adverse effect. '
C-77
-------
3. There are large differences in AChE levels between different parts of the brain,
and dissection and sampling procedures must be strictly adhered to for brain
AChE assays to be meaningful.
Organophosphate Induced Delayed Neurotoxicity
The statement in the report that "inhibition of neurotoxic esterase...has been associated
with agents that produce OPIDN...and is considered evidence of neurotoxicity," needs
clarification to accurately reflect current thinking and data. Presently, Organophosphate-
induced Delayed Neuropathy (OPIDN) is screened by specific animal tests, and determination
of neuropathy target esterase (neurotoxic esterase [NTE]) activity is a part of the testing.
Inhibition of NTE is not evidence per se that a compound is neuropathic. For example,
phenylmethanesulfonyl fluoride (PMSF), carbamates (e.g., physostigmine), and phosphinates
inhibit NTE activity but do not cause OPIDN. There is a body of evidence (Meredith and
Johnson, 1989; Johnson, 1990) to indicate that most OPs that cause OPIDN not only inhibit
NTE, but also "age" (shift an alkyl group from the phosphorylated active site to another part of
the NTE molecule). The precise role of "aging" is still unclear; indeed recent experiments have
led Johnson and others to question the role of aging in the onset of the neuropathy (Johnson et
al, 1991; Io>tti, 1991). Regardless, overt symptoms of OPIDN will not appear with acute
exposures unless inhibition of brain NTE exceeds 70 to 80 percent (Johnson, 1990). (Levels of
lymphocyte NTE activity and brain NTE activity after repeated exposures have not been as
reliable predictors). It is safe to say that it is generally agreed that, even though inhibition of
NTE alone is not sufficient to brand a chemical as neuropathic, or to use the NTE assay, by
itself, as a screen for the disorder, OPs that inhibit NTE activity are possible neuropathic
compounds and deserve further investigation.
References
Johnson M.K., Vilanova E., and Read DJ. 1991. Anomalous biochemical responses in tests of
the delayed neuropathic potential of methamidophos (O,S-dimethyl
phosphorothioamidate), its resolved isomers and of some higher O-alkyl homologues.
Archives of Toxicology, 65(8):618-24.
C-78
-------
Johnson M.K. 1990. Organophosphates and delayed neuropathyis NTE alive and well?
Toxicology and Applied Pharmacology, 102(3):385-99.
Meredith C, Johnson M.K. 1989. Species distribution of paraoxon-resistant brain polypeptides
radiolabelled. with diisopropyl phosphorofluoridate ([3H]DiPF): electrophoretic assay
for the aged polypeptide of [3H]DiPF-labelled neuropathy target esterase. Journal of
Neurochemistry, 52(4):1248-52.
Lotti M. 1991. -The pathogenesis of organophosphafe polyneuropathy. Critical Reviews in
Toxicology, 21(6):465-87.
Weiss, B. (Chair). 1990. Review of cholinesterase inhibition and its effects. Report of the
SAB/SAP Joint Study Group on Cholinesterase, U.S. Environmental Protection Agency.
Wilson, B.W., Jaeger, B. and Baetke, K. (Editors). 1992a. Proceedings of the U.S. EPA
Workshop on Cholinesterase Methodologies, Dec. 4-5, 1991, Office of Pesticide
Programs.
Wilson, B.W., Hooper, M.J., Hansen, M.E. and Nieberg, P.S. 1992b. Reactivation of
organophosphate inhibited AChE with oximes. In "Organophosphates, Chemistry, Fate
and Effects," Chambers, J.E. and J^evi, P.E. (Eds.) Academic Press, pp. 107-137.
C-79
-------
-------
APPENDIX D
POSTMEETING COMMENTS
-------
-------
APPENDIX D
Postmeeting Comments
Dr. Barry Wilson, Ph.D., Workgroup Chair of Direct and Indirect Effects Panel
One group that has considered the definition of an "adverse health effect" is the
"Scientific Assembly for Environmental and Occupational Health of the American Thoracic
Society" with reference to air pollution and the Clean Air Act (Am. Rev. Respir. Dis. 131:666-
668, 1985). The concerns discussed in their report are similar to those discussed by the
neurotoxicology panels. Health effects were classed in an ascending triangle with mortality,
morbidity and pathophysiologic changes being classed as adverse health effects at the top, and
physiologic changes of uncertain significance and pollutant burdens at the bottom. The
approach was epidemiologjcal to the extent that the proportion of the population affected
decreased as one progressed from the bottom (physiologic changes) to the top (mortality) of the
pyramid. Factors singled out as most important in considering when a physiologic change
should be considered an adverse effect included: (a) differences between "statistical significance
and medical or biological significance," and (b) the idea that "not all changes (e.g. physiologic)
are necessarily adverse." The reaction of carbon monoxide with hemoglobin was used as an
example; (c) the problem of reversible effects, and (d) when an effect occurs in the lifetime of
an individual were also considered.
The report defines "adverse respiratory health effects" as medically significant physiologic
i*
or pathologic changes generally evidenced by one or more of the following: (1) interference with
the normal activity of the affected person or persons; (2) episodic respiratory illness; (3)
incapacitating illness; (4) permanent respiratory injury; and/or (5) progressive respiratory
dysfunction. The report stressed human epidemiological, clinical and human exposure studies
more than animal studies, (Parenthetically, the area of neurotoxicology has much better
molecular, cellular and organismal models for human toxicities than does that of respiratory
pollution.) Nevertheless, it is striking that both panels independently generated hierarchal
schemes for establishing adverse health effects and physiologic events of "uncertain significance."
D-l
ft U.S. GOVERNMENT PRINTING OFFICE: 1992750-002/60,106
-------
------- |