[Federal Register Volume 60, Number 192 (Wednesday, October 4, 1995)]
[Notices]
[Pages 52032-52056]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 95-24652]




[[Page 52031]]

_______________________________________________________________________

Part II





Environmental Protection Agency





_______________________________________________________________________



Proposed Guidelines for Neurotoxicity Risk Assessment; Notice

Federal Register / Vol. 60, No. 192 / Wednesday, October 4, 1995 / 
Notices 

[[Page 52032]]


ENVIRONMENTAL PROTECTION AGENCY

[FRL-5306-2]


Proposed Guidelines for Neurotoxicity Risk Assessment

AGENCY: Environmental Protection Agency (EPA).

ACTION: Proposed guidelines for Neurotoxicity Risk Assessment and 
request for comments.

-----------------------------------------------------------------------

SUMMARY: The U.S. Environmental Protection Agency (EPA; Agency) is 
today issuing proposed guidelines for assessing the risks for 
neurotoxicity from exposure to environmental agents. As background 
information for this guidance, this notice describes the scientific 
basis for concern about exposure to agents that cause neurotoxicity and 
outlines the general process for assessing potential risk to humans 
because of environmental contaminants.
    These proposed Guidelines for Neurotoxicity Risk Assessment 
(hereafter ``Guidelines'') are intended to guide Agency evaluation of 
agents that are suspected to cause neurotoxicity in line with the 
policies and procedures established in the statutes administered by the 
EPA. The Guidelines were developed as part of an interoffice guidelines 
development program under the auspices of the Risk Assessment Forum, 
within EPA's Office of Research and Development. Draft Guidelines were 
developed by an Agency work group composed of scientists from 
throughout the Agency, and selected drafts were peer reviewed 
internally and by experts from universities, environmental groups, 
industry, and other governmental agencies. A subsequent draft has 
undergone peer review in a workshop held on June 2-3, 1992, and has 
received internal review by the Concordance and Oversight Subcommittees 
of the Risk Assessment Forum. Most recently, the Committee on the 
Environment and Natural Resources of the Office of Science and 
Technology Policy reviewed the guidelines at a meeting held on August 
15, 1995. The proposed Guidelines are based, in part, on 
recommendations derived from these reviews and on those made at various 
scientific meetings and workshops on neurotoxicology.
    The public is invited to comment, and public comments will be 
considered in EPA decisions in formulating the final Guidelines. 
Commenters are asked to focus on several special issues, particularly, 
(1) the issue of compensation and recovery of function in 
neurotoxicological studies and how to account for compensation in 
neurotoxicology risk assessment; (2) the use of blood and/or brain 
acetylcholinesterase activity as an indication of neurotoxicity for 
risk assessment; (3) endpoints indicative of neurotoxicity that may not 
be covered by these guidelines, i.e., endocrine disruption or 
neuroendocrine-mediated neurotoxicity; and (4) the possibility of no 
threshold for some neurotoxic agents.
    The EPA Science Advisory Board (SAB) also will review these 
proposed Guidelines at a meeting to be announced in a future Federal 
Register. Agency staff will prepare summaries of the public and SAB 
comments, analyses of major issues presented by commenters, and Agency 
responses to those comments. Appropriate comments will be incorporated, 
and the revised Guidelines will be submitted to the Risk Assessment 
Forum for review. The Agency will consider comments from the public, 
the SAB, and the Risk Assessment Forum in its recommendations to the 
EPA Administrator.

DATES: The Proposed Guidelines are being made available for a 120-day 
public review and comment period. Comments must be in writing and must 
be postmarked by February 1, 1996. Please submit one unbound original 
with pages consecutively numbered, and three copies. If there are 
attachments, include an index numbered consecutively with comments, and 
three copies.

FOR FURTHER INFORMATION CONTACT: Dr. Hugh A. Tilson, Tel: 919-541-2671; 
Fax: 919-541-4849.

ADDRESSES: Comments on the proposed Guidelines may be mailed or 
delivered to: Dr. Hugh A. Tilson, Neurotoxicology Division (MD-74B), 
National Health and Environmental Effects Research Laboratory, U.S. 
Environmental Protection Agency, Research Triangle Park, NC 27711. 
Please note that all comments received in response to this notice will 
be placed in a public record. Commenters should not send any item of 
personal information, such as medical information or home address, if 
they do not wish it to be part of the public record.

SUPPLEMENTARY INFORMATION: In its 1983 book, Risk Assessment in the 
Federal Government: Managing the Process, the National Academy of 
Sciences recommended that Federal regulatory agencies establish 
``inference guidelines'' (1) to promote consistency and technical 
quality in risk assessment, and (2) to ensure that the risk assessment 
process is maintained as a scientific effort separate from risk 
management. A task force within EPA accepted that recommendation and 
requested that Agency scientists begin to develop such guidelines.
    In 1984, EPA scientists began work on risk assessment guidelines 
for carcinogenicity, mutagenicity, suspect developmental toxicants, 
chemical mixtures, and exposure assessment. Following extensive 
scientific and public review, these first five guidelines were issued 
on September 24, 1986 (51 FR 33992-34054). Since 1986, additional risk 
assessment guidelines have been proposed for male and female 
reproductive risk (53 FR 24834-847; 53 FR 24850-869), and two of the 
1986 guidelines, suspect developmental toxicants (56 FR 63798-826) and 
exposure assessment (57 FR 22888-938), have been revised, reproposed, 
and finalized.
    The Guidelines proposed today continue the guidelines development 
process initiated in 1984. These Guidelines set forth principles and 
procedures to guide EPA scientists in the conduct of Agency risk 
assessments and to inform Agency decision makers and the public about 
these procedures. In particular, the Guidelines emphasize that risk 
assessments will be conducted on a case-by-case basis, giving full 
consideration to all relevant scientific information. This case-by-case 
approach means that Agency experts study scientific information on each 
chemical under review and use the most scientifically appropriate 
interpretation to assess risk. The Guidelines also stress that this 
information will be fully presented in Agency risk assessment 
documents, and that Agency scientists will identify the strengths and 
weaknesses of each assessment by describing uncertainties, assumptions, 
and limitations, as well as the scientific basis and rationale for each 
assessment.
    The Guidelines are formulated in part to bridge gaps in risk 
assessment methodology and data. By identifying these gaps and the 
importance of the missing information to the risk assessment process, 
EPA wishes to encourage research and analysis that will lead to new 
risk assessment methods and data.

    Dated: September 25, 1995.
Carol M. Browner,
Administrator.

Proposed Guidelines for Neurotoxicity Risk Assessment Contents

I. Introduction
    A. Organization of These Guidelines
    B. The Role of Environmental Agents in Neurotoxicity 

[[Page 52033]]

    C. Neurotoxicity Risk Assessment
    D. Assumptions
II. Definitions and Critical Concepts
III. Hazard Characterization
    A. Neurotoxicological Studies: End Points and Their 
Interpretation
    1. Human Studies
    a. Clinical Evaluations
    b. Case Reports
    c. Epidemiologic Studies
    (1) Cross-sectional studies
    (2) Case-control (retrospective) studies
    (3) Cohort (prospective, follow-up) studies
    d. Human Laboratory Exposure Studies
    2. Animal Studies
    a. Structural End Points of Neurotoxicity
    b. Neurophysiological End Points of Neurotoxicity
    (1) Nerve conduction studies
    (2) Sensory, motor, and other evoked potentials
    (3) Seizures/convulsions
    (4) Electroencephalography (EEG)
    c. Neurochemical End Points of Neurotoxicity
    d. Behavioral End Points of Neurotoxicity
    (1) Functional observational battery
    (2) Motor activity
    (3) Schedule-controlled operant behavior
    (4) Convulsions
    (5) Specialized tests for neurotoxicity
    (a) Motor function
    (b) Sensory function
    (c) Cognitive function
    e. Developmental Neurotoxicity
    3. Other Considerations
    a. Pharmacokinetics
    b. Comparisons of Molecular Structure
    c. Statistical Considerations
    d. In Vitro Data in Neurotoxicology
    B. Dose-Response Evaluation
    C. Characterization of the Health-Related Data Base
IV. Dose-Response Analysis
    A. LOAEL/NOAEL and Benchmark Dose (BMD) Determination
    B. Determination of the Reference Dose or Reference 
Concentration
V. Exposure Assessment
VI. Risk Characterization
    A. Overview
    B. Integration of Hazard Characterization, Dose-Response 
Analysis, and Exposure Assessment
    C. Quality of the Data Base and Degree of Confidence in the 
Assessment
    D. Descriptors of Neurotoxicity Risk
    1. Estimation of the Number of Individuals
    2. Presentation of Specific Scenarios
    3. Risk Characterization for Highly Exposed Individuals
    4. Risk Characterization for Highly Sensitive or Susceptible 
Individuals
    5. Other Risk Descriptors
    E. Communicating Results
    F. Summary and Research Needs
VII. References

List of Tables

Table 1. Examples of possible indicators of a neurotoxic effect
Table 2. Neurotoxicants and diseases with specific neuronal targets
Table 3. Examples of neurophysiological measures of neurotoxicity
Table 4. Examples of neurotoxicants with known neurochemical 
mechanisms
Table 5. Summary of measures in a representative functional 
observational battery, and the type of data produced by each
Table 6. Examples of specialized behavioral tests to measure 
neurotoxicity
Table 7. Examples of developmental neurotoxicants
Table 8. Characterization of the Health-Related Database

I. Introduction

    These proposed Guidelines describe the principles, concepts, and 
procedures that the U.S. Environmental Protection Agency (EPA; Agency) 
would follow in evaluating data on potential neurotoxicity associated 
with exposure to environmental toxicants. The Agency's authority to 
regulate substances that have the potential to interfere with human 
health is derived from a number of statutes that are implemented 
through multiple offices within the EPA. The procedures outlined here 
are intended to help develop a sound scientific basis for neurotoxicity 
risk assessment, promote consistency in the Agency's assessment of 
toxic effects on the nervous system, and inform others of the 
approaches used by the Agency in those assessments.

A. Organization of These Guidelines

    This Introduction (section I) summarizes the purpose of these 
proposed Guidelines within the overall framework of risk assessment at 
the EPA. It also outlines the organization of the guidance and 
describes several default assumptions to be used in the risk assessment 
process as discussed in the recent National Research Council report 
``Science and Judgment in Risk Assessment (NRC, 1994).''
    Section II sets forth definitions of particular terms widely used 
in the field of neurotoxicology. These include ``neurotoxicity'' and 
``behavioral alterations.'' Also included in this section are 
discussions concerning reversible and irreversible effects and direct 
versus indirect effects.
    Risk assessment is the process by which scientific judgments are 
made concerning the potential for toxicity to occur in humans. The 
National Research Council (NRC, 1983) has defined risk assessment as 
including some or all of the following components (paradigm): hazard 
identification, dose-response assessment, exposure assessment, and risk 
characterization. In its 1994 report ``Science and Judgment in Risk 
Assessment'' the NRC extended its view of the paradigm to include 
characterization of each component (NRC, 1994). In addition, it noted 
the importance of an approach that is less fragmented and more 
holistic, less linear and more interactive, and one that deals with 
recurring conceptual issues that cut across all stages of risk 
assessment. These Guidelines propose a more interactive approach by 
organizing the process around components that focus on evaluation of 
the toxicity data (hazard characterization), the quantitative dose-
response analysis, the exposure assessment, and the risk 
characterization. This is done because, in practice, hazard 
identification for neurotoxicity and other noncancer health effects is 
usually done in conjunction with an evaluation of dose-response 
relationships in the studies used to identify the hazard. Determining a 
hazard often depends on whether a dose-response relationship is present 
(Kimmel et al., 1990). Thus, the hazard characterization provides an 
evaluation of a hazard within the context of the dose, route, duration, 
and timing of exposure. This approach combines the information 
important in comparing the toxicity of a chemical to potential human 
exposure scenarios (Section V). Secondly, it avoids the potential for 
labeling chemicals as ``neurotoxicants'' on a purely qualitative basis. 
This organization of the risk assessment process is similar to that 
discussed in the Guidelines for Developmental Toxicity Risk Assessment 
(56 FR 63798), the main difference being that the quantitative dose-
response analysis is discussed under a separate section in these 
guidelines.
    Hazard characterization involves examining all available 
experimental animal and human data and the associated doses, routes, 
timing, and durations of exposure to determine if an agent causes 
neurotoxicity in that species and under what conditions. From the 
hazard characterization and criteria provided in these Guidelines, the 
health-related data base can be characterized as sufficient or 
insufficient for use in risk assessment (section III.C). Combining 
hazard identification and some aspects of dose-response evaluation into 
hazard characterization does not preclude the evaluation and use of 
data when quantitative information for setting reference doses (RfDs) 
and reference concentrations (RfCs) are not available.
    The next step, the dose-response analysis (section IV) is the 
quantitative analysis, and includes determining the no-observed-
adverse-effect-level (NOAEL) and/or the lowest-observed-adverse-effect-
level (LOAEL) for each study and type of effect. Because of the 
limitations associated with the use of the NOAEL, the Agency is 
beginning to use an additional approach, i.e., the 

[[Page 52034]]
benchmark dose approach (Crump, 1984; U.S. EPA, 1995a), for more 
quantitative dose-response evaluation when sufficient data are 
available. The benchmark dose approach takes into account the 
variability in the data and the slope of the dose-response curve, and 
provides a more consistent basis for calculation of the RfD or RfC. If 
data are considered sufficient for risk assessment, and if 
neurotoxicity is the effect occurring at the lowest dose level (i.e., 
the critical effect), an oral or dermal RfD or an inhalation RfC, based 
on neurotoxic effects, is then derived. This RfD or RfC is derived 
using the NOAEL or benchmark dose divided by uncertainty factors to 
account for interspecies differences in response, intraspecies 
variability and other factors of study design or the data base. A 
statement of the potential for human risk and the consequences of 
exposure can come only from integrating the hazard characterization and 
dose-response analysis with the human exposure estimates in the final 
risk characterization.
    The section on exposure assessment (section V) identifies human 
populations exposed or potentially exposed to an agent, describes their 
composition and size, and presents the types, magnitudes, frequencies, 
and durations of exposure to the agent. The exposure assessment 
provides an estimate of human exposure levels for particular 
populations from all potential sources.
    In risk characterization (section VI), the hazard characterization, 
dose-response analysis, and the exposure assessment for given 
populations are combined to estimate some measure of the risk for 
neurotoxicity. As part of risk characterization, a summary of the 
strengths and weaknesses of each component of the risk assessment is 
given along with major assumptions, scientific judgments, and, to the 
extent possible, qualitative and quantitative estimates of the 
uncertainties. This characterization of the health-related data base is 
always presented in conjunction with information on the dose, route, 
duration and timing of exposure as well as the dose-response analysis 
including the RfD or RfC. If human exposure estimates are available, 
the exposure basis used for the risk assessment is clearly described, 
e.g., highly exposed individuals or highly sensitive or susceptible 
individuals. The NOAEL may be compared to the various estimates of 
human exposure to calculate the margin(s) of exposure (MOE). The 
considerations for judging the acceptability of the MOE are similar to 
those for determining the appropriate size of the uncertainty factor 
for calculating the RfD or RfC.
    The Agency recently issued a policy statement and associated 
guidance for risk characterization (U.S. EPA, 1995b, 1995c), which is 
currently being implemented throughout EPA. This policy statement is 
designed to ensure that critical information from each stage of a risk 
assessment is used in forming conclusions about risk and that this 
information is communicated from risk assessors to risk managers 
(policy makers), from middle to upper management, and from the Agency 
to the public. Additionally, the policy provides a basis for greater 
clarity, transparency, reasonableness, and consistency in risk 
assessments across Agency programs. Final neurotoxicity risk assessment 
guidelines may reflect additional changes in risk characterization 
practices resulting from implementation activities.
    Risk assessment is just one component of the regulatory process and 
defines the potential adverse health consequences of exposure to a 
toxic agent. The other component, risk management, combines risk 
assessment with statutory directives regarding socioeconomic, 
technical, political, and other considerations, to decide whether to 
control future exposure to the suspected toxic agent and, if so, the 
nature and level of control. One major objective of these risk 
assessment Guidelines is to help the risk assessor determine whether 
the experimental animal or human data indicate the potential for a 
neurotoxic effect. Such information can then be used subsequently to 
categorize evidence to identify and characterize neurotoxic hazards as 
described in section III.3.C, Characterization of the Health-Related 
Data Base, and Table 8 of these Guidelines. Risk management is not 
dealt with directly in these Guidelines because the basis for decision 
making goes beyond scientific considerations alone, but the use of 
scientific information in this process is discussed. For example, the 
acceptability of the MOE is a risk management decision, but the 
scientific bases for establishing this value are discussed here.

B. The Role of Environmental Agents in Neurotoxicity

    Chemicals are an integral part of life, with the capacity to 
improve as well as endanger health. The general population is exposed 
to chemicals with neurotoxic properties in air, water, foods, 
cosmetics, household products, and drugs used therapeutically or 
illicitly. Naturally occurring neurotoxins, such as animal and plant 
toxins, present additional hazards. During daily life, a person 
experiences a multitude of exposures, both voluntary and unintentional, 
to neuroactive substances, singly and in combination. Levels of 
exposure vary and may or may not pose a hazard depending on dose, 
route, and duration of exposure.
    A link between human exposure to some chemical substances and 
neurotoxicity has been firmly established (Anger, 1986; OTA, 1990). 
Because many natural and synthetic chemicals are present in today's 
environment, there is growing scientific and regulatory interest in the 
potential for risks to humans from exposure to neurotoxic agents. If 
sufficient exposure occurs, the effects resulting from such exposures 
can have a significant adverse impact on human health. It is not known 
how many chemicals may be neurotoxic in humans (Reiter, 1987). The 
EPA's inventory of toxic chemicals is greater than 65,000 and 
increasing yearly. An overwhelming majority of the materials in 
commercial use have not been tested for their neurotoxic potential 
(NRC, 1984). Estimates of the number of chemicals with neurotoxic 
properties have been made for subsets of substances. For instance, a 
large percentage of the more than 500 registered active pesticide 
ingredients are neurotoxic to varying degrees. Of 588 chemicals listed 
by the American Conference of Governmental Industrial Hygienists, 167 
affected the nervous system or behavior at some exposure level (Anger, 
1984). Anger (1990) estimated that of the approximately 200 chemicals 
to which one million or more American workers are exposed, more than 
one-third may have adverse effects on the nervous system, if sufficient 
exposure occurs. Anger (1984) also recognized neurotoxic effects as one 
of the 10 leading workplace disorders. A number of therapeutic 
substances, including some anticancer and antiviral agents and abused 
drugs, can cause adverse or neurotoxicological side effects at 
therapeutic levels (OTA, 1990). Thus, estimating the risks of exposure 
to chemicals with neurotoxic potential is of concern with regard to the 
overall impact of these exposures on human health.

C. Neurotoxicity Risk Assessment

    In addition to its primary role in cognitive functions, the nervous 
system controls most, if not all, other bodily processes. It is 
sensitive to perturbation from various sources and has limited ability 
to regenerate. There is evidence that even small anatomical, 
biochemical, or physiological insults to the nervous system may result 
in 

[[Page 52035]]
adverse effects on human health. Therefore, there is a need for 
consistent guidance on how to evaluate data on neurotoxic substances 
and assess to what degree, if any, they have the potential to cause 
transient or persistent, direct or indirect effects on human health.
    To help address these needs, these Guidelines develop principles 
and concepts in several areas. First, these Guidelines outline the 
scientific basis for evaluating effects due to exposure to 
neurotoxicants and discuss principles and methods for evaluating data 
from human and animal studies on behavior, neurochemistry, 
neurophysiology, and neuropathology. This guidance document also 
discusses adverse effects on neurological development and function in 
infants and children following prenatal and perinatal exposure to 
chemical agents. Other sections of these Guidelines outline the method 
for calculating reference doses or reference concentrations when 
neurotoxicity is the critical effect, discuss the availability of 
alternative mathematical approaches to dose-response analyses, 
characterize the health-related data base for neurotoxicity risk 
assessment, and discuss integration of exposure information with the 
results of the dose-response assessment to characterize risks of 
exposures of concern. These Guidelines do not advocate developing 
reference doses specific for neurotoxicity, but rather the use of 
neurotoxicity as one possible end point to develop reference doses.
    EPA offices have published guidelines for neurotoxicity testing in 
animals (U.S. EPA, 1986, 1987, 1988a, 1991a). The testing guidelines 
address the development of new data for use in risk assessment. These 
proposed neurotoxicity risk assessment Guidelines provide the Agency's 
first comprehensive guidance on the use and interpretation of 
neurotoxicity data. These proposed Guidelines are part of the Agency's 
risk assessment guidelines development process, which was initiated in 
1984. As part of its neurotoxicity guidelines development program, the 
EPA has sponsored or participated in several conferences on relevant 
issues (Tilson, 1990); these and other sources (see references) provide 
the scientific basis for these proposed risk assessment Guidelines. 
This guidance is intended for use by Agency risk assessors and is 
separate and distinct from the recently published document on 
principles of neurotoxicity risk assessment (U.S. EPA, 1993). The 
document on principles was prepared under the auspices of the 
Subcommittee on Risk Assessment of the Federal Coordinating Council for 
Science, Engineering, and Technology and was not intended to provide 
specific directives for how neurotoxicity risk assessment should be 
performed.
    It is expected that, like other EPA risk assessment guidelines 
(U.S. EPA, 1991b), this document will encourage research and analysis 
leading to new risk assessment methods and data, which in turn would be 
used to revise and improve the Guidelines and better guide Agency risk 
assessors.

D. Assumptions

    There are a number of unknowns in the extrapolation of data from 
animal studies to humans. Therefore, a number of default assumptions 
are made that are generally applied in the absence of data on the 
relevance of effects to potential human risk. Default assumptions 
should not be applied indiscriminantly. First, all available 
mechanistic and pharmacokinetic data should be considered. If these 
data indicate that an alternative assumption is appropriate or obviate 
the need for applying an assumption, such information should be used in 
the risk assessment of that agent. The following default assumptions 
form the basis of the approaches taken in these Guidelines.
    It is assumed that an agent that produces detectable adverse 
neurotoxic effects in experimental animal studies will pose a potential 
hazard to humans. This assumption is based on the comparisons of data 
for known human neurotoxicants (Anger, 1990; Kimmel et al., 1990; 
Spencer and Schaumburg, 1980), which indicate that experimental animal 
data are frequently predictive of a neurotoxic effect in humans.
    It is assumed that behavioral, neurophysiological, neurochemical, 
and neuroanatomical manifestations are of concern. In the past, the 
tendency has been to consider only neuropathological changes as end 
points of concern. Based on the data on agents that are known human 
neurotoxicants (Anger, 1990; Kimmel et al., 1990; Spencer and 
Schaumberg, 1980), there is usually at least one experimental species 
that mimics the types of effects seen in humans, but in other species 
tested, the type of neurotoxic effect may be different or absent. Thus, 
a biologically significant increase in any of the manifestations is 
considered indicative of an agent's potential for disrupting the 
structure or function of the human nervous system.
    It is assumed that the types of neurotoxic effects seen in animal 
studies may not always be the same as those produced in humans. 
Therefore, it may be difficult to determine which will be the most 
appropriate species in terms of predicting the specific types of 
effects seen in humans. The fact that every species may not react in 
the same way is probably due to species-specific differences in 
maturation of the nervous system, differences in timing of exposure, 
metabolism, or mechanisms of action.
    It is assumed that the most appropriate species will be used when 
data are available to estimate human risk. In the absence of such data, 
the most sensitive species is used, based on the fact that for the 
majority of known human neurotoxicants, humans are as sensitive or more 
so than the most sensitive animal species tested.
    In general, a threshold is assumed for the dose-response curve for 
most neurotoxicants. This is based on the known capacity of the nervous 
system to compensate for or to repair a certain amount of damage at the 
cellular, tissue, or organ level. In addition, because of the 
multiplicity of cells in the nervous system, multiple insults at the 
molecular or cellular level may be required to produce an effect on the 
whole organism.
    These assumptions are ``plausibly conservative'' (NRC, 1994) in 
that they are protective of public health and are also well-founded in 
scientific knowledge about the effects of concern.

II. Definitions and Critical Concepts

    This section defines the key terms and concepts that the EPA will 
use in the identification and evaluation of neurotoxicity. The various 
health effects that fall within the broad classification of 
neurotoxicity are described and examples are provided.
    Adverse effects include alterations from baseline that diminish an 
organism's ability to survive, reproduce, or adapt to the environment. 
Neurotoxicity is an adverse change in the structure or function of the 
central and/or peripheral nervous system following exposure to a 
chemical, physical, or biological agent (Tilson, 1990). Neurotoxic 
effects include changes in somatic/autonomic, sensory, motor, and/or 
cognitive function. Structural effects are defined as neuroanatomical 
changes occurring at any level of nervous system organization; 
functional changes are defined as neurochemical, neurophysiological, or 
behavioral alterations. Changes in function can also result from 
toxicity to other specific organ systems, and these indirect changes 
may be considered adverse but not necessarily neurotoxic. 

[[Page 52036]]

    The risk assessor also should know that there are different levels 
of concern based on the magnitude of effect and reversibility of some 
neurotoxic effects. Neurotoxic effects may be irreversible, i.e., 
cannot return to the state prior to exposure, resulting in a permanent 
change in the organism, or reversible, i.e., can return to the pre-
exposure condition, allowing the organism to return to its state prior 
to exposure. Clear or demonstrable irreversible change in either the 
structure or function of the nervous system causes greater concern than 
do reversible changes. If neurotoxic effects are observed at some time 
during the life span of the organism but are slowly reversible, the 
concern is also high. There is lesser concern for effects that are 
rapidly reversible or transient, i.e., measured in minutes, hours, or 
days, and appear to be associated with the pharmacokinetics of the 
causal agent and its presence in the body. Reversible changes that 
occur in the occupational setting or environment, however, may be of 
high concern if, for example, exposure to a short-acting solvent 
interferes with operation of heavy equipment in an industrial plant. 
The context of the exposure should be considered in evaluating 
reversible effects. The risk assessor should note that once damaged, 
neurons, particularly in the central nervous system, have a limited 
capacity for regeneration. Reversibility of effects resulting from cell 
death or from the destruction of cell processes may represent an 
activation of repair capacity, decreasing future potential 
adaptability. Therefore, even reversible neurotoxic changes should be 
of concern. Evidence of progressive effects, i.e., those that continue 
to worsen even after the causal agent has been removed; or delayed 
effects, i.e., those that occur at a time distant from the last contact 
with the causal agent; or residual effects, i.e., those that persist 
beyond a recovery period; or latent effects, i.e., those that become 
evident only after an environmental challenge or aging, have a high 
level of concern. Environmental challenges can include stress, 
increased physical or cognitive workload, pharmacological 
manipulations, and nutritional deficiency or excess.
    Neurotoxic effects can be observed at various levels of 
organization of the nervous system, including neurochemical, 
anatomical, physiological, or behavioral. At the neurochemical level, 
for example, an agent that causes neurotoxicity might inhibit 
macromolecule or transmitter synthesis, alter the flow of ions across 
cellular membranes, or prevent release of neurotransmitter from the 
nerve terminals. Anatomical changes may include alterations of the cell 
body, the axon, or the myelin sheath. At the physiological level, a 
chemical might change the thresholds for neural activation or reduce 
the speed of neurotransmission. Behavioral alterations can include 
significant changes in sensations of sight, hearing, or touch; 
alterations in simple or complex reflexes and motor functions; 
alterations in cognitive functions such as learning, memory or 
attention; and changes in mood, such as fear or rage, disorientation as 
to person, time, or place, or distortions of thinking and feeling, such 
as delusions and hallucinations. At present, relatively few neurotoxic 
syndromes have been thoroughly characterized in terms of the initial 
neurochemical change, structural alterations, physiological 
consequence, and behavioral effects. Knowledge of exact mechanisms of 
action is not, however, necessary to conclude that a chemically induced 
change is a neurotoxic effect.
    Neurotoxic effects can be produced by chemicals that do not require 
metabolism prior to interacting with their target sites in the nervous 
system, i.e., primary neurotoxic agents, or those that require 
metabolism prior to interacting with their target sites in the nervous 
system, i.e., secondary neurotoxic agents. Chemically induced 
neurotoxic effects can be direct, i.e., due to an agent or its 
metabolites acting directly on target sites in the nervous system, or 
indirect, i.e., due to agents or metabolites that produce their effects 
primarily by interacting with target sites outside the nervous system, 
which subsequently affect target sites in the nervous system. 
Excitatory amino acids such as domoic acid damage specific neurons 
directly by activating excitatory amino acid receptors in the nervous 
system, while carbon monoxide decreases oxygen availability, which 
indirectly kills neurons. Other examples of indirect effects of 
chemicals that could lead to altered structure and/or function of the 
nervous system include cadmium-induced spasms in blood vessels 
supplying the nervous system, dichloroacetate-induced perturbation of 
metabolic pathways, and chemically induced alterations in 
skeletomuscular function or structure and effects on the endocrine 
system. Professional judgment may be required in making determinations 
about direct versus indirect effects.
    The interpretation of data as indicative of a potential neurotoxic 
effect involves the evaluation of the validity of the data base. This 
approach and these terms have been adapted from the literature on human 
psychological testing (Sette, 1987; Sette and MacPhail, 1992) where 
they have long been used to evaluate the level of confidence in 
different measures of intelligence or other abilities, aptitudes, or 
feelings. There are four principal questions that should be addressed: 
whether the effects result from exposure (content validity); whether 
the effects are adverse or toxicologically significant (construct 
validity); whether there are correlative measures among behavioral, 
physiological, neurochemical, and morphological end points (concurrent 
validity); and whether the effects are predictive of what will happen 
under various conditions (predictive validity). Addressing these issues 
can provide a useful framework for evaluating either human or animal 
studies or the weight of evidence for a chemical (Sette, 1987; Sette 
and MacPhail, 1992). The next sections indicate the extent to which 
chemically induced changes can be interpreted as providing evidence of 
neurotoxicity.

III. Hazard Characterization

A. Neurotoxicological Studies: End Points and Their Interpretation

    Identification and characterization of neurotoxic hazard can be 
based on either human or animal data (Anger, 1984; Reiter, 1987; U.S. 
EPA, 1993). Such data can result from accidental, inappropriate, or 
controlled experimental exposures. This section describes many of the 
general and some of the specific characteristics of human studies and 
reports of neurotoxicity. It then describes some features of animal 
studies of neuroanatomical, neurochemical, neurophysiological, and 
behavioral effects relevant to risk assessment. The process of 
characterizing the sufficiency or insufficiency of neurotoxic effects 
for risk assessment is described in section III.C. Additional sources 
of information relevant to hazard characterization, such as comparisons 
of molecular structure among compounds and in vitro screening methods, 
are also discussed.
    The hazard characterization should:
    a. Identify strengths and limitations of the database:

--Epidemiological studies (case reports, cross-sectional, case-control, 
cohort, or human laboratory exposure studies);
--Animal studies including (structural or neuropathological, 
neurochemical, neurophysiological, behavioral or neurological, or 
developmental end points).


[[Page 52037]]

    b. Evaluate the validity of the database:

--Content validity (effects result from exposure);
--Construct validity (effects are adverse or toxicologically 
significant);
--Concurrent validity (correlative measures among behavioral, 
physiological, neurochemical, or morphological end points);
--Predictive validity (effects are predictive of what will happen under 
various conditions).
    c. Identify and describe key toxicological studies.
    d. Describe the type of effects:

--Structural (neuroanatomical alternations);
--Functional (neurochemical, neurophysiological, behavioral 
alterations).

    e. Describe the nature of the effects (irreversible, reversible, 
transient, progressive, delayed, residual, or latent effects).
    f. Describe how much is known about how (through what biological 
mechanism) the chemical produces adverse effects.
    g. Discuss other health end points of concern.
    h. Comment on any non-positive data in humans or animals.
    i. Discuss the dose-response data (epidemiological or animal) 
available for further dose-response analysis.
    j. Discuss the route, level, timing, and duration of exposure in 
studies demonstrating neurotoxicity as compared to expected human 
exposures.
    k. Summarize the hazard characterization:

--Confidence in conclusions;
--Alternative conclusions also supported by the data;
--Significant data gaps; and
--Highlight of major assumptions.
1. Human Studies
    It is well established that information from the evaluation of 
human exposure can identify neurotoxic hazards (Anger and Johnson, 
1985; Anger, 1990). Prominent among historical episodes of 
neurotoxicity in human populations are the outbreaks of methylmercury 
poisoning in Japan and Iraq and the neurotoxicity seen in miners of 
metals, including mercury, manganese, and lead (Carson et al., 1987; 
Silbergeld and Percival, 1987; OTA, 1990). In the last decade, lead 
poisoning in children has been a prominent issue of concern (Silbergeld 
and Percival, 1987). Neurotoxicity in humans has been studied and 
reviewed for many pesticides (Hayes, 1982; NRDC, 1989; Ecobichon and 
Joy, 1982; Ecobichon et al., 1990). Organochlorines, organophosphates, 
carbamates, pyrethroids, certain fungicides, and some fumigants are all 
known neurotoxicants. They may pose occupational risks to manufacturing 
and formulation workers, pesticide applicators and farm workers, and 
consumers through home application or consumption of residues in foods. 
Families of workers may also be exposed by transport into the home from 
workers' clothing. Data on humans can come from a number of sources, 
including clinical evaluations, case reports, and epidemiologic 
studies. A more extensive description of issues concerning human 
neurotoxicology and risk assessment has been published elsewhere (U.S. 
EPA, 1993).
    a. Clinical Evaluations. Clinical methods are used extensively in 
neurology and neuropsychology to evaluate patients suspected of having 
neurotoxicity. An extensive array of examiner-administered and paper-
and-pencil tasks are used to assess sensory, motor, cognitive, and 
affective functions and personality states/traits. Neurobehavioral data 
are synthesized with information from neurophysiologic studies and 
medical history to derive a working diagnosis. Brain imaging techniques 
based on magnetic resonance imaging or emission tomography may also be 
useful in helping diagnose neurodegenerative disorders following 
chemical exposures in humans (Omerod et al., 1994; Callender et al., 
1994). Clinical diagnostic approaches have provided a rich conceptual 
framework for understanding the functions (and malfunctions) of the 
central and peripheral nervous systems and have formed the basis for 
the development of methods for measuring the behavioral expression of 
nervous system disorders. Human neurobehavioral toxicology has borrowed 
heavily from neurology and neuropsychology for concepts of nervous 
system impairment and functional assessment methods. Neurobehavioral 
toxicology has adopted the neurologic/neuropsychologic model, using 
adverse changes in behavioral function to assist in identifying 
chemically or drug-induced changes in nervous system processes.
    Neurologic and neuropsychologic methods have long been employed to 
identify the adverse health effects of environmental workplace 
exposures (Sterman and Schaumburg, 1980). Peripheral neuropathies (with 
sensory and motor disturbances), encephalopathies, organic brain 
syndromes, extrapyramidal syndromes, demyelination, autonomic changes, 
and dementia are well-characterized consequences of acute and chronic 
exposure to chemical agents. The range of exposure conditions that 
produce clinical signs of neurotoxicity also has been defined by these 
clinical methods. It is very important to make external/internal dose 
measurements in humans to determine the actual dose(s) that can cause 
unwanted effects.
    Aspects of the neurologic examination approach limit its usefulness 
for neurotoxicologic risk assessment. Information obtained from the 
neurologic exam is mostly qualitative and descriptive rather than 
quantitative. Estimates of the severity of functional impairment can be 
reliably placed into only three or four categories (for example, mild, 
moderate, severe). Much of the assessment depends on the subjective 
judgment of the examiner. For example, the magnitude and symmetry of 
muscle strength are often judged by having the patient push against the 
resistance of the examiner's hands. The end points are therefore the 
absolute and relative amount of muscle load sensed by the examiner in 
his or her arms.
    Compared with other methods, the neurologic exam may be less 
sensitive in detecting early neurotoxicity in peripheral sensory and 
motor nerves. While clinicians' judgments are equal in sensitivity to 
quantitative methods in assessing the amplitude of tremor, tremor 
frequency is poorly quantified by clinicians. Thus, important aspects 
of the clinical neurologic exam may be insufficiently quantified and 
lack sufficient sensitivity for detecting early neurobehavioral 
toxicity produced by environmental or workplace exposure conditions. 
However, a neurologic evaluation of persons with documented 
neurobehavioral impairment would be helpful for identifying nonchemical 
causes of neurotoxicity, such as diabetes and cardiovascular 
insufficiency.
    Administration of a neuropsychological battery also requires a 
trained technician, and interpretation requires a trained and 
experienced neuropsychologist. Depending on the capabilities of the 
patient, 2 to 4 hours may be needed to administer a full battery; 1 
hour may be needed for the shorter screening versions. These practical 
considerations may limit the usefulness of neuropsychological 
assessment in large field studies of suspected neurotoxicity.
    In addition to logistical problems in administration and 
interpretation, neuropsychological batteries and neurologic exams share 
two disadvantages with respect to neurotoxicity risk assessment. First, 


[[Page 52038]]
neurologic exams and neuropsychological test batteries are designed to 
confirm and classify functional problems in individuals selected on the 
basis of signs and symptoms identified by the patient, family, or other 
health professionals. Their usefulness in detecting low base-rate 
impairment in workers or the general population is generally thought to 
be limited, decreasing the usefulness of clinical assessment approaches 
for epidemiologic risk assessment.
    Second, neurologic exams and neuropsychologic test batteries were 
developed to assess the functional correlates of the most common forms 
of nervous system dysfunction: brain trauma, focal lesions, and 
degenerative conditions. The clinical tests were validated against 
these neurologic disease states. With a few notable exceptions, 
chemicals are not believed to produce impairment similar to that from 
trauma or lesions; neurotoxic effects are more similar to the effects 
of degenerative disease. There has been insufficient research to 
demonstrate which tests designed to assess functional expression of 
neurologic disease are useful in characterizing the modes of central 
nervous system impairment produced by chemical agents and drugs.
    b. Case reports. The first type of human data available is often 
the case report or case series, which can identify cases of a disease 
and are reported by clinicians or discerned through active or passive 
surveillance, usually in the workplace. However, case reports where 
exposure involved a single neurotoxic agent, although informative, are 
rare in the literature; for example, farmers are likely to be exposed 
to a wide variety of potentially neurotoxic pesticides. Careful case 
histories assist in identifying common risk factors, especially when 
the association between the exposure and disease is strong, the mode of 
action of the agent is biologically plausible, and clusters occur in a 
limited period of time.
    Case reports are inexpensive compared with epidemiologic studies 
and can be obtained more quickly than more complex studies. However, 
they provide little information about disease frequency or population 
at risk, but their importance has been clearly demonstrated, 
particularly in accidental poisoning or acute exposure to high levels 
of toxicant. They remain an important source of index cases of new 
diseases and for surveillance.
    c. Epidemiologic Studies. Epidemiology has been defined as ``the 
study of the distributions and determinants of disease and injuries in 
human populations'' (Mausner and Kramer, 1985). Knowing the frequency 
of illness in groups and the factors that influence the distribution is 
the tool of epidemiology that allows the evaluation of causal inference 
with the goal of prevention and cure of disease (Friedlander and Hearn, 
1980). Epidemiologic studies are a means of evaluating the effects of 
neurotoxic substances on human populations, but such studies are 
limited because they must be performed shortly after exposure if the 
effect is acute. Most often these effects are suspected to be a result 
of occupational exposures due to the increased opportunity for exposure 
to industrial and other chemicals. Frequently, determining the precise 
dose or exposure concentration can be difficult in epidemiological 
studies.
    (1) Cross-sectional studies. In cross-sectional studies or surveys, 
both the disease and suspected risk factors are ascertained at the same 
time, and the findings are useful in generating hypotheses. A group of 
people are interviewed, examined, and tested at a single point in time 
to ascertain a relationship between a disease and a neurotoxic 
exposure. This study design does not allow the investigator to 
determine whether the disease or the exposure came first, rendering it 
less useful in estimating risk. These studies are intermediate in cost 
and time required to complete compared with case reports and more 
complex analytical studies but should be augmented with additional 
data.
    (2) Case-control (retrospective) studies. Last (1986) defines a 
case-control study as one that ``starts with the identification of 
persons with the disease (or other outcome variable) of interest, and a 
suitable control population (comparison, reference group) of persons 
without the disease.'' He states that the relationship of an 
``attribute'' to the disease is measured by comparing the diseased with 
the nondiseased with regard to how frequently the attribute is present 
in each of the groups. The cases are assembled from a population of 
persons with and without exposure, and the comparison group is selected 
from the same population; the relative distribution of the potential 
risk factor (exposure) in both groups is evaluated by computing an odds 
ratio that serves as an estimate of the strength of the association 
between the disease and the potential risk factor. The statistical 
significance of the ratio is determined by calculating a p-value and is 
used to approximate relative risk.
    The case-control approach to the study of potential neurotoxicants 
in the environment provides a great deal of useful information for the 
risk assessor. In his textbook, Valciukas (1991) notes that the case-
control approach is the strategy of choice when no other environmental 
or biological indicator of neurotoxic exposure is available. He further 
states: ``Considering the fact that for the vast majority of neurotoxic 
chemical compounds, no objective biological indicators of exposure are 
available (or if they are, their half-life is too short to be of any 
practical value), the case-control paradigm is a widely accepted 
strategy for the assessment of toxic causation.'' The case-control 
study design, however, can be very susceptible to bias. The potential 
sources of bias are numerous and can be specific to a particular study. 
Many of these biases also can be present in cross-sectional studies. 
For example, recall bias or faulty recall of information by study 
subjects in a questionnaire-based study can distort the results of the 
study. Analysis of the case-comparison study design assumes that the 
selected cases are representative persons with the disease--either all 
cases with the disease or a representative sample of them have been 
ascertained. It further assumes that the control or comparison group is 
representative of the nondiseased population (or that the prevalence of 
the characteristic under study is the same in the control group as in 
the general population). Failure to satisfy these assumptions may 
result in selection bias, but violation of assumptions does not 
necessarily invalidate the study results.
    An additional source of bias in case-control studies is the 
presence of confounding variables, i.e., factors known to be associated 
with the exposure and causally related to the disease under study. 
These must be controlled either in the design of the study by matching 
cases to controls on the basis of the confounding factor or in the 
analysis of the data by using statistical techniques such as 
stratification or regression. Matching requires time to identify an 
adequate number of potential controls to distinguish those with the 
proper characteristics, while statistical control of confounding 
factors requires a larger study.
    The definition of exposure is critical in epidemiologic studies. In 
occupational settings, exposure assessment often is based on the job 
assignment of the study subjects, but can be more precise if detailed 
company records allow the development of exposure profiles. Positive 
results from a properly controlled retrospective 

[[Page 52039]]
study should weigh heavily in the risk assessment process.
    (3) Cohort (prospective, followup) studies. In a prospective study 
design, a healthy group of people is assembled and followed forward in 
time and observed for the development of disease. Such studies are 
invaluable for determining the time course for development of disease 
(e.g., followup studies performed in various cities on the effects of 
lead on child development). This approach allows the direct estimate of 
risks attributed to a particular exposure since disease incidence rates 
in the cohort can be determined. Prospective study designs also allow 
the study of chronic effects of exposure. One major strength of the 
cohort design is that it allows the calculation of rates to determine 
the excess risk associated with an exposure. Also, biases are reduced 
by obtaining information before the disease develops. This approach, 
however, can be very time-consuming and costly.
    In cohort studies information bias can be introduced when 
individuals provide distorted information about their health because 
they know their exposure status and may have been told of the expected 
health effects of the exposure under study.
    A special type of cohort study is the retrospective cohort study in 
which the investigator goes back in time to select the study groups and 
traces them over time, often to the present. The studies usually 
involve specially exposed groups and have provided much assistance in 
estimating risks due to occupational exposures. Occupational 
retrospective cohort studies rely on company records of past and 
current employees that include information on the dates of employment, 
age at employment, date of departure, and whether diseased (or dead in 
the case of mortality studies). Workers can then be classified by 
duration and degree of exposure. Positive results from a properly 
controlled prospective study should weigh heavily in the risk 
assessment process.
    d. Human Laboratory Exposure Studies. Neurotoxicity assessment has 
an advantage not afforded the evaluation of other toxic end points, 
such as cancer or reproductive toxicity, in that the effects of some 
chemicals are short in duration and reversible. This makes it ethically 
possible to perform human laboratory exposure studies and obtain data 
relevant to the risk assessment process. Information from experimental 
human exposure studies has been used to set occupational exposure 
limits, mostly for organic solvents that can be inhaled. Laboratory 
exposure studies have contributed to risk assessment and the setting of 
exposure limits for several solvents and other chemicals with acute 
reversible effects.
    Human exposure studies sometime offer advantages over epidemiologic 
field studies. Combined with appropriate sampling of biologic fluids 
(urine or blood), it is possible to calculate body concentrations, 
examine toxicokinetics, and identify metabolites. Bioavailability, 
elimination, dose-related changes in metabolic pathways, individual 
variability, time course of effects, interactions between chemicals, 
and interactions between chemical and environmental/biobehavioral 
processes (stressors, workload/respiratory rate) are factors that are 
generally easier to collect under controlled conditions.
    Other goals of laboratory studies include the indepth 
characterization of effects, the development of new assessment methods, 
and the examination of the sensitivity, specificity, and reliability of 
neurobehavioral assessment methods across chemical classes. The 
laboratory is the most appropriate setting for the study of 
environmental and biobehavioral variables that affect the action of 
chemical agents. The effects of ambient temperature, task difficulty, 
rate of ongoing behavior, conditioning variables, tolerance/
sensitization, sleep deprivation, motivation, and so forth are 
sometimes studied.
    From a methodologic standpoint, human laboratory studies can be 
divided into two categories--between-subjects and within-subjects 
designs. In the former, the neurobehavioral performance of exposed 
volunteers is compared with that of nonexposed participants. In the 
latter, preexposure performance is compared with neurobehavioral 
function under the influence of the chemical or drug. Within-subjects 
designs have the advantage of requiring fewer participants, eliminating 
individual differences as a source of variability, and controlling for 
chronic mediating variables, such as caffeine use and educational 
achievement. A disadvantage of the within-subjects design is that 
neurobehavioral tests must be administered more than once. Practice on 
many neurobehavioral tests often leads to improved performance that may 
confound the effect of the chemical/drug. There should be a sufficient 
number of test sessions in the pre-exposure phase of the study to allow 
performance on all tests to achieve a relatively stable baseline level.
    Participants in laboratory exposure studies may have been recruited 
from populations of persons already exposed to the chemical/drug or 
from naive populations. Although the use of exposed volunteers has 
ethical advantages, can mitigate against novelty effects, and allows 
evaluation of tolerance/sensitization, finding an accessible exposed 
population in reasonable proximity to the laboratory is difficult. 
Naive participants are more easily recruited but may differ 
significantly in important characteristics from a representative sample 
of exposed persons. Naive volunteers are often younger, healthier, and 
better educated than the populations exposed environmentally, in the 
workplace, or pharmacotherapeutically.
    Compared with workplace and environmental exposures, laboratory 
exposure conditions can be controlled more precisely, but exposure 
periods are much shorter. Generally only one or two relatively pure 
chemicals are studied for several hours while the population of 
interest may be exposed to multiple chemicals containing impurities for 
months or years. Laboratory studies are therefore better at identifying 
and characterizing effects with acute onset and the selective effects 
of pure agents.
    Neurobehavioral test methods may have been selected according to 
several strategies. A test battery that examines multiple 
neurobehavioral functions may be more useful for screening and the 
initial characterization of acute effects. Selected neurobehavioral 
tests that measure a more limited number of functions in multiple ways 
may be more useful for elucidating mechanisms or validating specific 
effects.
    Both chemical and behavioral control procedures are valuable for 
examining the specificity of the effects. A concordant effect among 
different measures of the same neurobehavioral function (e.g., reaction 
time) and a lack of effect on some other measures of psychomotor 
function (e.g., untimed manual dexterity) would increase the confidence 
in a selective effect on motor speed and not on attention or 
nonspecific motor function. Likewise, finding concordant effects among 
similar chemical or drug classes along with different effects from 
dissimilar classes would support the specificity of chemical effect. 
For example, finding that the effects of a solvent were similar to 
those of ethanol but not caffeine would support the specificity of 
solvent effects on a given measure of neurotoxicity.
2. Animal Studies
    This section provides an overview of the major types of end points 
that may be evaluated in animal neurotoxicity 

[[Page 52040]]
studies, describes the kinds of effects that may be observed and some 
of the tests used to detect and quantify these effects, and provides 
guidance for interpreting data. Compared with human studies, animal 
studies are more often available for specific chemicals, provide more 
precise exposure information, and control environmental factors better 
(Anger, 1984). For these reasons, risk assessments tend to rely heavily 
on animal studies.
    Many tests that can measure some aspect of neurotoxicity have been 
used in the field of neurobiology in the last 50 years. The Office of 
Prevention, Pesticides and Toxic Substances (OPPTS) has published 
animal testing guidelines that were developed in cooperation with the 
Office of Research and Development (U.S. EPA, 1991a). While the test 
end points included serve as a convenient focus for this section, there 
are many other end points for which there are no current EPA 
guidelines. The goal of this document is to provide a framework for 
interpreting data collected with tests frequently used by 
neurotoxicologists.
    Five categories of end points will be described: Structural or 
neuropathological, neurophysiological, neurochemical, behavioral, and 
developmental end points. Table 1 lists a number of end points in each 
of these categories.

Table 1.--Examples of Possible Indicators of a Neurotoxic Effect

I. Structural or Neuropathological End Points
    1. Gross changes in morphology, including brain weight
    2. Hemorrhage in nerve tissue
    3. Breakdown of neurons, glial cells
    4. Accumulation, proliferation, or rearrangement of structural 
elements
    5. Glial fibrillary acidic protein increases (in adults)
II. Neurochemical End Points
    1. Alterations in synthesis, release, uptake, degradation of 
neurotransmitters
    2. Alterations in second messenger associated signal 
transduction
    3. Alterations in membrane-bound enzymes regulating neuronal 
activity
    4. Inhibition of neuropathy target enzyme (40%)
III. Neurophysiological End Points
    1. Change in velocity, amplitude, or refractory period of nerve 
conduction
    2. Change in latency or amplitude of sensory-evoked potential
    3. Change in electroencephalographic pattern
IV. Behavioral and Neurological End Points
    1. Increases or decreases in motor activity
    2. Changes in touch, sight, sound, taste, or smell sensations
    3. Changes in motor coordination, weakness, paralysis, abnormal 
movement or posture, tremor, ongoing performance
    4. Absence or decreased occurrence, magnitude, or latency of 
sensorimotor reflex
    5. Altered magnitude of neurological measurement, including grip 
strength, hindlimb splay
    6. Seizures
    7. Changes in rate or temporal patterning of schedule-controlled 
behavior
    8. Changes in learning, memory, intelligence, attention
V. Developmental End Points
    1. Chemically induced changes in the time of appearance of 
behaviors during development
    2. Chemically induced changes in the growth or organization of 
structural or neurochemical elements.
    a. Structural End Points of Neurotoxicity. Structural end points 
are typically defined as neuropathological changes measured through 
gross observation or with the aid of a microscope. Gross changes in 
morphology can include discrete or widespread lesions in nerve tissue. 
Changes in brain size (weight, width, or length) are considered to be 
indicative of neurotoxic events. This is true regardless of changes in 
body weight, because brain size is generally protected during 
undernutrition or weight loss, unlike many other organs or tissues. It 
is inappropriate to express brain weight changes as a ratio of body 
weight and thereby dismiss changes in absolute brain weight. The risk 
assessor should be aware that a unit of measurement that is 
biologically meaningful should be used for analysis. Brain length 
measurements, for example, expressed to 1 or 10 micron units is 
biologically meaningless. The same is true for brain width.
    Neurons are composed of a neuronal body, axon, and dendritic 
processes. Various types of neuropathological lesions may be classified 
according to the site where they occur (WHO, 1986; Krinke, 1989; 
Griffin, 1990). Neurodegenerative lesions in the central or peripheral 
nervous system may be classified as a neuronopathy (changes in the 
neuronal cell body), axonopathy (changes in the axons), myelinopathy 
(changes in the myelin sheaths), or terminal degeneration. For 
axonopathies, a more precise location of the changes may also be 
described (i.e., proximal, central, or distal axonopathy). In the case 
of some developmental exposures, a neurotoxic chemical might delay or 
accelerate the differentiation or proliferation of cells or cell types. 
Alteration in the axonal termination site might also occur with 
exposure. In an aged population, exposure to some neurotoxicants might 
accelerate the normal loss of neurons associated with aging (Reuhl, 
1991). In rare cases, neurotoxic agents have been reported to produce 
neuropathic conditions resembling neurodegenerative disorders in humans 
such as Parkinson's disease (WHO, 1986). Table 2 lists examples of such 
neurotoxic chemicals, their putative site of action, the type of 
neuropathology produced, and the disease or condition that each 
typifies.

                      Table 2.--Neurotoxicants and Diseases With Specific Neuronal Targets                      
----------------------------------------------------------------------------------------------------------------
                                                                                             Corresponding      
       Site of action                Neuropathology               Neurotoxicant        neurodegenerative disease
                                                                                             or condition       
----------------------------------------------------------------------------------------------------------------
Neuron cell body............  Neuronopathy................  Methylmercury Quinolinic  Minamata disease,         
                                                             acid 3-Acetylpyridine.    Huntington's disease,    
                                                                                       Cerebellar ataxia.       
Nerve terminal..............  Terminal destruction........  1-Methyl-4-phenyl-        Parkinson's disease.      
                                                             1,2,3,6-                                           
                                                             tetrahydropyridine                                 
                                                             (dopaminergic).                                    
Schwann cell Myelin.........  Myelinopathy................  Hexachlorophene.........  Congenital                
                                                                                       hypomyelinogenesis.      
Central-peripheral distal     Distal axonopathy...........  Acrylamide Carbon         Peripheral neuropathy.    
 axon.                                                       disulfide n-Hexane.                                
Central axons...............  Central axonopathy..........  Clioquinol..............  Subacute                  
                                                                                       myeloopticoneuropathy.   
Proximal axon...............  Proximal axonopathy.........  B,B'-                     Motor neuron disease.     
                                                             iminodipropionitrile.                              
----------------------------------------------------------------------------------------------------------------

    Alterations in the structure of the nervous system (i.e., 
neuronopathy, axonopathy, myelinopathy, terminal degeneration) are 
regarded as evidence of a neurotoxic effect. The risk assessor should 
note that pathological changes in many cases require time for the 
perturbation to become observable, especially with evaluation at the 
light 

[[Page 52041]]
microscopic level. Neuropathological studies should control for 
potential differences in the area(s) and section(s) of the nervous 
system sampled, in the age, sex, and body weight of the subject, and in 
fixation artifacts (WHO, 1986). Concern for the structural integrity of 
nervous system tissues derives from its functional specialization and 
the lack of regenerative capacity in the central nervous system.
    In general, chemical effects can lead to two types of structural 
alteration at the cellular level: the breakdown of cells, in whole or 
in part, or the accumulation, proliferation, or rearrangement of 
structural elements (e.g., intermediate filaments, microtubules) or 
organelles (e.g., mitochondria). Some changes may be associated with 
regenerative processes that reflect adaptive changes associated with 
exposure to a toxicant.
    Chemically induced injury to the central nervous system may be 
associated with astrocytic hypertrophy. Such changes may be seen using 
immunocytochemical techniques visualized by light microscopy or 
quantified more precisely by radioimmunoassay (RIA) procedures. Assays 
of glial fibrillary acidic protein (GFAP), the major intermediate 
filament protein of astrocytes, have been proposed as a biomarker of 
this response (O'Callaghan, 1988). The interpretation of a chemical-
induced change in GFAP is facilitated by corroborative data from the 
neuropathology or neuroanatomy evaluation. A number of chemicals known 
to injure the central nervous system, including trimethyltin, 
methylmercury, cadmium, 3-acetylpyridine, and 
methylphenyltetrahydropyridine (MPTP), have been shown to increase 
levels of GFAP. Measures of GFAP are now included in the Neurotoxicity 
Test Battery testing guidelines (U.S. EPA, 1991a).
    Increases in GFAP above control levels may be seen at dosages below 
those necessary to produce damage seen by standard microscopic or 
histopathological techniques. Because increases in GFAP reflect an 
astrocyte response in adults, treatment-related increases in GFAP are 
considered to be evidence that a neurotoxic effect has occurred. 
Decreases in GFAP are not clearly interpretable as indicative of 
neurotoxicity. The absence of a change in GFAP following exposure does 
not necessarily mean that the chemical is devoid of neurotoxic 
potential. Known neurotoxicants such as cholinesterase-inhibiting 
pesticides, for example, would not be expected to increase brain levels 
of GFAP. Interpretation of GFAP changes prior to weaning is confounded 
by the possibility that chemically induced increases in GFAP may be 
masked by changes in the concentration of this protein associated with 
maturation of the central nervous system, and these data may be 
difficult to interpret.
    b. Neurophysiological End Points of Neurotoxicity. 
Neurophysiological studies are those that measure the electrical 
activity of the nervous system. The term ``neurophysiology'' is often 
used synonymously with ``electrophysiology'' (Dyer, 1987). 
Neurophysiological techniques provide information on the integrity of 
defined portions of the nervous system. Several neurophysiological 
procedures are available for application to neurotoxicological studies. 
Examples of neurophysiological measures of neurotoxicity are listed in 
Table 3. They range in scale from procedures that employ 
microelectrodes to study the function of single nerve cells or 
restricted portions of them, to procedures that employ macroelectrodes 
to perform simultaneous recordings of the summed activity of many 
cells. Microelectrode procedures typically are used to study mechanisms 
of action and are frequently performed in vitro. Macroelectrode 
procedures are generally used in studies to detect or characterize the 
potential neurotoxic effects of agents of interest because of potential 
environmental exposure. The present discussion concentrates on 
macroelectrode neurophysiological procedures because it is more likely 
that they will be the focus of decisions regarding critical effects in 
risk assessment. All of the procedures described below for use in 
animals also have been used in humans to determine chemically induced 
alterations in neurophysiological function.

   Table 3.--Examples of Neurophysiological Measures of Neurotoxicity   
------------------------------------------------------------------------
    System/function             Procedure          Representative agents
------------------------------------------------------------------------
Retina.................  Electroretinography      Developmental lead.   
                          (ERG).                                        
Visual pathway.........  Flash evoked potential   Carbon disulfide.     
                          (FEP).                                        
Visual function........  Pattern evoked           Carbon disulfide.     
                          potential (PEP)                               
                          pattern size and                              
                          contrast).                                    
Auditory pathway.......  Brain stem auditory      Aminoglycoside,       
                          evoked potential         Antibiotics, Toluene,
                          (BAER) (clicks).         styrene.             
Auditory function......  BAER (tones)...........  Aminoglycoside,       
                                                   Antibiotics, Toluene,
                                                   styrene.             
Somatosensory pathway..  Somatosensory evoked     Acrylamide, n-Hexane. 
                          potential (SEP)                               
                          (shocks).                                     
Somatosensory function.  SEP (tactile)..........  Acrylamide n-Hexane.  
Spinocerebellar pathway  SEP recorded from        Acrylamide n-Hexane.  
                          cerebellum.                                   
Mixed nerve............  Peripheral nerve         Triethyltin.          
                          compound action                               
                          potential (PNAP).                             
Motor axons............  PNAP isolate motor       Triethyltin.          
                          components.                                   
Sensory axons..........  PNAP isolate sensory     Triethyltin.          
                          components.                                   
Neuromuscular..........  Electromyography (EMG),  Dithiobiuret.         
                          H-reflex, M-response.                         
General central nervous  Electroencephalography   Anesthetics.          
 system/level of          (EEG).                                        
 arousal.                                                               
------------------------------------------------------------------------

    (1) Nerve conduction studies. Nerve conduction studies, generally 
performed on peripheral nerves, can be useful in investigations of 
possible peripheral neuropathy. Most peripheral nerves contain mixtures 
of individual sensory and motor nerve fibers, which may or may not be 
differentially sensitive to neurotoxicants. It is possible to 
distinguish sensory from motor effects in peripheral nerve studies by 
measuring activity in purely sensory 

[[Page 52042]]
nerves such as the sural nerve or by measuring the muscle response 
evoked by nerve stimulation to measure motor effects. While a number of 
end points can be recorded, the most critical variables are (1) nerve 
conduction velocity, (2) response amplitude, and (3) refractory period.
    Nerve conduction measurements are influenced by a number of 
factors, the most important of which is temperature. An adequate nerve 
conduction study will either measure the temperature of the limb under 
study and mathematically adjust the results according to well-
established temperature factors or control limb temperature within 
narrow limits. Studies that measure peripheral nerve function without 
regard for temperature are not adequate for risk assessment.
    In well-controlled studies, statistically significant decreases in 
nerve conduction velocity are indicative of a neurotoxic effect. While 
a decrease in nerve conduction velocity is indicative of demyelination, 
it frequently occurs later in the course of axonal degradation because 
normal conduction velocity may be maintained for some time in the face 
of axonal degeneration. For this reason, a measurement of normal nerve 
conduction velocity does not rule out peripheral axonal degeneration if 
other signs of peripheral nerve dysfunction are present.
    Decreases in response amplitude reflect a loss of active nerve 
fibers and may occur prior to decreases in conduction velocity in the 
course of peripheral neuropathy. Hence, changes in response amplitude 
may be more sensitive measurements of axonal degeneration than 
conduction velocity. Measurements of response amplitude, however, can 
be more variable and require careful application of experimental 
techniques, a larger sample size, and greater statistical power than 
measurements of velocity to detect changes. The refractory period 
refers to the time required after stimulation before a nerve can fire 
again and provides a measure reflecting the functional status of nerve 
membrane ion channels. Chemically induced changes in refractory periods 
in a well-controlled study indicate a neurotoxic effect.
    In summary, alterations in peripheral nerve response amplitude and 
refractory period in studies that are well controlled for temperature 
are indicative of a neurotoxic effect. Alterations in peripheral nerve 
function are frequently associated with clinical signs such as 
numbness, tingling, or burning sensations or with motor impairments 
such as weakness. Examples of compounds that alter peripheral nerve 
function in humans or experimental animals include acrylamide, carbon 
disulfide, n-hexane, lead, and some organophosphates.
    (2) Sensory, motor, and other evoked potentials. Evoked potential 
studies are electrophysiological procedures that measure the response 
elicited from a defined stimulus such as a tone, a light, or a brief 
electrical pulse. Evoked potentials reflect the function of the system 
under study, including visual, auditory, or somatosensory; motor 
involving motor nerves and innervated muscles; or other neural pathways 
in the central or peripheral nervous system (Rebert, 1983; Dyer, 1985; 
Mattsson and Albee, 1988; Mattsson et al., 1992; Boyes, 1992, 1993). 
Evoked potential studies should be interpreted with respect to the 
known or presumed neural generators of the responses, and their likely 
relationships with behavioral outcomes, when such information is 
available. Such correlative information strengthens the confidence in 
electrophysiological outcomes. In the absence of such supportive 
information, the extent to which evoked potential studies provide 
convincing evidence of neurotoxicity is a matter of professional 
judgment on a case-by-case basis. Judgments should consider the nature, 
magnitude, and duration of such effects, along with other factors 
discussed elsewhere in this document.
    Data are in the form of a voltage record collected over time and 
can be quantified in several ways. Commonly, the latency (time from 
stimulus onset) and amplitude (voltage) of the positive and negative 
voltage peaks are identified and measured. Alternative measurement 
schemes may involve substitution of spectral phase or template shifts 
for peak latency and spectral power, spectral amplitude, root-mean-
square, or integrated area under the curve for peak amplitude. Latency 
measurements are dependent on both the velocity of nerve conduction and 
the time of synaptic transmission. Both of these factors depend on 
temperature, as discussed in regard to nerve conduction, and similar 
caveats apply for sensory evoked potential studies. In studies that are 
well controlled for temperature, increases in latencies or related 
measures can reflect deficits in nerve conduction, including 
demyelination or delayed synaptic transmission, and are indicators of a 
neurotoxic effect.
    Decreases in peak latencies, like increases in nerve conduction 
velocity, are unusual, but the neural systems under study in sensory 
evoked potentials are complex, and situations that might cause a peak 
measurement to occur earlier are conceivable. Two such situations are a 
reduced threshold for spatial or temporal summation of afferent neural 
transmission and a selective loss of cells responding late in the peak, 
thus making the measured peak occur earlier. Decreases in peak latency 
should not be dismissed outright as experimental or statistical error, 
but should be examined carefully and perhaps replicated to assess 
possible neurotoxicity. A decrease in latency is not conclusive 
evidence of a neurotoxic effect.
    Changes in peak amplitudes or equivalent measures reflect changes 
in the magnitude of the neural population responsive to stimulation. 
Both increases and decreases in amplitude are possible following 
exposure to chemicals. Whether excitatory or inhibitory neural activity 
is translated into a positive or negative deflection in the sensory 
evoked potential is dependent on the physical orientation of the 
electrode with respect to the tissue generating the response, which is 
frequently unknown. Comparisons should be based on the absolute change 
in amplitude. Therefore, either increases or decreases in amplitude may 
be indicative of a neurotoxic effect.
    Within any given sensory system, the neural circuits that generate 
various evoked potential peaks differ as a function of peak latency. In 
general, early latency peaks reflect the transmission of afferent 
sensory information. Changes in either the latency or amplitude of 
these peaks are considered convincing evidence of a neurotoxic effect 
that is likely to be reflected in deficits in sensory perception. The 
later-latency peaks, in general, reflect not only the sensory input but 
also the more nonspecific factors such as the behavioral state of the 
subject, including such factors as arousal level, habituation, or 
sensitization (Dyer, 1987). Thus, changes in later-latency evoked 
potential peaks must be interpreted in light of the behavioral status 
of the subject and would generally be considered evidence of a 
neurotoxic effect.
    (3) Seizures/convulsions. Neurophysiological recordings of brain 
electrical activity that demonstrate seizure-like activity are 
indicative of a neurotoxic effect. Occasionally, behaviors resembling 
convulsions might follow actions outside the nervous system, such as 
direct effects on muscle. When convulsion-like behaviors are observed, 
as described in the behavioral section, neurophysiological recordings 
can determine if these behaviors 

[[Page 52043]]
originate from seizure activity in the brain.
    In addition to producing seizures directly, neurotoxicants also may 
alter the frequency, severity, duration, or threshold for eliciting 
seizures produced through other means. Such changes can occur after 
acute exposure or after repeated exposure to dose levels below the 
acute threshold and are considered to be neurotoxic effects. Examples 
of agents that produce convulsions include lindane, DDT (dichloro-
diphenyl-trichloroethane), pyrethroids, and trimethyltin.
    (4) Electroencephalography (EEG). EEG analysis is used widely in 
clinical settings for the diagnosis of neurological disorders and less 
often for the detection of subtle toxicant-induced dysfunction (WHO, 
1986; Eccles, 1988). The basis for using EEG in either setting is the 
relationship between specific patterns of EEG waveforms and specific 
behavioral states. Because states of alertness and stages of sleep are 
associated with distinct patterns of electrical activity in the brain, 
it is generally thought that arousal level can be evaluated by 
monitoring the EEG.
    Dissociation of EEG activity and behavior can, however, occur after 
exposure to certain chemicals. Normal patterns of transition between 
sleep stages or between sleeping and waking states are known to remain 
disturbed for prolonged periods of time after exposure to some 
chemicals. Changes in the pattern of the EEG can be elicited by stimuli 
producing arousal (e.g., lights, sounds) and anesthetic drugs. In 
studies with toxicants, changes in EEG pattern can sometimes precede 
alterations in other objective signs of neurotoxicity (Dyer, 1987).
    EEG studies must be done under highly controlled conditions, and 
the data must be considered on a case-by-case basis. Chemically induced 
seizure activity detected in the EEG pattern is evidence of a 
neurotoxic effect.
    c. Neurochemical End Points of Neurotoxicity. Many different 
neurochemical end points have been measured in neurotoxicological 
studies, and some have proven useful in advancing the understanding of 
mechanisms of action of neurotoxic chemicals (Bondy, 1986; Mailman, 
1987; Morell and Mailman, 1987; Costa, 1988). Normal functioning of the 
nervous system depends on the synthesis and release of specific 
neurotransmitters and activation of their receptors at specific 
presynaptic and postsynaptic sites. Chemicals can interfere with the 
ionic balance of a neuron, act as a cytotoxicant after transport into a 
nerve terminal, block reuptake of neurotransmitters and their 
precursors, act as a metabolic poison, overstimulate receptors, block 
transmitter release, and inhibit transmitter synthetic or catabolic 
enzymes. Table 4 lists several chemicals that produce neurotoxic 
effects at the neurochemical level (Bondy, 1986; Mailman, 1987; Morell 
and Mailman, 1987; Costa, 1988).

Table 4.--Examples of Neurotoxicants With Known Neurochemical Mechanisms
------------------------------------------------------------------------
           Site of action                          Examples             
------------------------------------------------------------------------
1. Neurotoxicants Acting on Ionic                                       
 Balance:                                                               
    A. Inhibit sodium entry........  Tetrodotoxin.                      
    B. Block closing of sodium       p,p'-DDT, pyrethroids.             
     channel.                                                           
    C. Increase permeability to      Batrachotoxin.                     
     sodium.                                                            
    D. Increase intracellular        Chlordecone.                       
     calcium.                                                           
2. Cytotoxicants--Depend on uptake   MPTP.                              
 into nerve terminal.                                                   
3. Uptake blockers.................  Hemicholinium.                     
4. Metabolic poisons...............  Cyanide.                           
5. Hyperactivation of receptors....  Domoic acid.                       
6. Blocks transmitter release        Botulinum toxin.                   
 (Acetylcholine [ACh]).                                                 
7. Inhibition of transmitter         Pesticides of the organophosphate  
 degradation (ACh).                   and carbamate classes.            
8. Blocks axonal transport.........  Acrylamide.                        
------------------------------------------------------------------------

    As stated previously, any neurochemical change is potentially 
neurotoxic, but each determination requires professional judgment. 
Persistent or irreversible chemically induced neurochemical changes are 
indicative of neurotoxicity. Because the ultimate functional 
significance of some biochemical changes is not known at this time, 
neurochemical studies should be interpreted with reference to the 
presumed neurotoxic consequence(s) of the neurochemical changes. For 
example, many neuroactive agents can increase or decrease 
neurotransmitter levels, but such changes are not necessarily 
indicative of a neurotoxic effect. If, however, these neurochemical 
changes may be expected to have neurophysiological, neuropathological, 
or neurobehavioral correlates, then the neurochemical changes could be 
classified as neurotoxic effects.
    Some neurotoxicants, such as the organophosphate and carbamate 
pesticides, are known to inhibit the activity of a specific enzyme, 
acetylcholinesterase (for a review see Costa, 1988), which hydrolyzes 
the neurotransmitter acetylcholine. Inhibition of the enzyme prolongs 
the action of the acetylcholine at the neuron's synaptic receptors and 
is responsible for the autonomic stimulation and death that these 
agents cause.
    Within EPA and elsewhere, questions have arisen as to whether 
inhibition of cholinesterase activity constitutes an adverse effect for 
defining hazard potential and evaluating risk. There is agreement among 
scientists that statistically significant inhibition of cholinesterase 
activity in multiple organs and tissues accompanied by clinical effects 
constitutes a hazard. However, there is scientific uncertainty and 
related controversy about the risk assessment implications of data 
describing inhibition of cholinesterase enzyme activity in the absence 
of observable clinical effects. While there is agreement that such 
inhibition is a biomarker of exposure, there is continued disagreement 
over whether cholinesterase inhibition, especially in blood, 
constitutes an adverse effect.
    At this point, it can be stated that there is general agreement 
among scientists that objective clinical measures of dysfunction/
impairment can be overt manifestations of inhibition of cholinesterase 
in the nervous system. On the basis of clinical manifestations, e.g., 
muscle weakness, tremor, blurred vision, one should be able to evaluate 
dose-response and dose-effect relationships and define the presence and 
absence of given effects. A relationship between the effect and 
cholinesterase inhibition should be 

[[Page 52044]]
confirmed by biochemical measures of reduced cholinesterase activity.
    In addition, a reduction in brain cholinesterase activity may or 
may not be accompanied by clinical manifestations. Most experts in the 
field acknowledge that when significant reductions in brain 
cholinesterase activity alone occur, reduced cholinesterase levels 
either are themselves toxic or would lead to a neurotoxic effect if 
exposure were to persist over time or increase in magnitude. Therefore, 
statistically significant decreases in brain cholinesterase could be 
considered to be a biologically significant effect.
    A reduction in RBC and/or plasma cholinesterase activity also may 
or may not be accompanied by clinical manifestations. At this time, 
there is general agreement that the observation of inhibition of RBC 
and/or plasma cholinesterase contributes to the overall hazard 
identification of cholinesterase inhibiting agents by serving as 
biomarkers. As such, these enzyme parameters can provide information 
that will help scientists evaluate whether reported clinical effects 
are associated with cholinesterase inhibition. There remains, however, 
a lack of consensus as to whether RBC and/or plasma cholinesterase 
represent biologically significant events. Discussions on this topic 
are continuing within the Agency.
    A subset of organophosphate agents also produces organophosphate-
induced delayed neuropathy (OPIDN) after acute or repeated exposure. 
Prolonged inhibition (i.e., aging) of neurotoxic esterase (or 
neuropathy target enzyme) has been associated with agents that produce 
OPIDN (Johnson, 1990), a clear neurotoxic effect.
    d. Behavioral End Points of Neurotoxicity. EPA's testing guidelines 
developed for the Toxic Substances Control Act and the Federal 
Insecticide, Fungicide and Rodenticide Act describe the use of 
functional observational batteries (FOB), motor activity, and schedule-
controlled behavior for assessing neurotoxic potential (U.S. EPA, 
1991a). There are many other measures of behavior, including 
specialized tests of motor and sensory function and of learning and 
memory (Tilson, 1987; Anger, 1984). Examples of behavioral end points 
that have been used to detect neurotoxicity are included in Table 1. 
The risk assessor should know that the literature is clear that a 
number of other behaviors besides those listed in Tables 1 and 5 could 
be affected by chemical exposure. For example, alterations in food and 
water intake, reproduction, sleep, temperature regulation, and 
circadian rhythmicity are controlled by specific regions of the brain 
and chemical-induced alterations in these behaviors could be indicative 
of neurotoxicity. It is reasonable to assume that a NOAEL or LOAEL 
could be based on one or more of these end points.

    Table 5.--Summary of Measures in a Representative Functional Observational Battery, and the Type of Data    
                                                Produced by Each                                                
----------------------------------------------------------------------------------------------------------------
       Home cage and open field                      Manipulative                         Physiologic           
----------------------------------------------------------------------------------------------------------------
Posture (D)                             Ease of removal (R)                    Body temperature (I).            
Convulsions, tremors (D)                Handling reactivity (R)                Body weight (I).                 
Palpebral closure (R)                   Palpebral closure (R).                 .................................
Lacrimation (R)                         Approach response (R).                                                  
Piloerection (Q)                        Click response (R).                                                     
Salivation (R)                          Touch response (R).                                                     
Vocalizations (Q)                       Tail pinch response (R).                                                
Rearing (C)                             Righting reflex (R).                                                    
Urination (C)                           Landing foot splay (I).                                                 
Defecation (C)                          Forelimb grip strength (I).                                             
Gait (D, R)                             Hindlimb grip strength (I).                                             
Arousal (R)                             Pupil response (Q).                                                     
Mobility (R).                                                                                                   
Stereotypy (D).                                                                                                 
Bizarre behavior (D)                                                                                            
----------------------------------------------------------------------------------------------------------------
D--descriptive data; R--rank order data; Q--quantal data; I--interval data; C--count data.                      

    Behavior is an indication of the overall well-being of the 
organism. Changes in behavior can arise from a direct effect of a 
toxicant on the nervous system or indirectly from its effects on other 
physiological systems. Understanding the interrelationship between 
systemic toxicity and behavioral changes is extremely important (e.g., 
the relationship between liver damage and motor activity). The presence 
of systemic toxicity may complicate, but does not necessarily preclude, 
interpretation of behavioral changes as evidence of neurotoxicity. In 
addition, a number of behaviors (e.g., schedule-controlled behavior) 
may require a motivational component for successful completion of the 
task. In such cases, experimental paradigms designed to assess the 
motivation of an animal during behavior might be necessary to interpret 
the meaning of some chemical-induced changes in behavior.
    The following sections describe in general behavioral tests and 
their uses and offer guidance on interpreting data.
    (1) Functional observational battery. A functional observational 
battery is designed to detect and quantify major overt behavioral, 
physiological, and neurological signs (Gad, 1982; O'Donoghue, 1989; 
Moser, 1989). A number of batteries have been developed, each 
consisting of tests generally intended to evaluate various aspects of 
sensorimotor function (Tilson and Moser, 1992). Many FOB tests are 
essentially clinical neurological examinations that rate the presence 
or absence, and in many cases the severity, of specific neurological 
signs. Some FOBs in animals are similar to clinical neurological 
examinations used with human patients. Most FOBs have several 
components or tests. A typical FOB is summarized in Table 5 and 
evaluates several functional domains, including neuromuscular (i.e., 
weakness, incoordination, gait, and tremor), sensory (i.e., audition, 
vision, and somatosensory), and autonomic (i.e., pupil response and 
salivation) function. FOB data may be in the form of interval, ordinal, 
or continuous measurements.
    The relevance of statistically significant test results from an FOB 
is judged according to the number of signs affected, the dose(s) at 
which effects are observed, and the nature, severity, and persistence 
of the effects and their 

[[Page 52045]]
incidence in relation to control animals. If only a few unrelated 
measures in the FOB are affected, or the effects are unrelated to dose, 
the results are not considered evidence of a neurotoxic effect. If 
several neurological signs are affected but only at the high dose and 
in conjunction with other overt signs of toxicity, including systemic 
toxicity, large decreases in body weight, decreases in body 
temperature, or debilitation, there is no conclusive evidence of a 
direct neurotoxic effect. In cases where several related measures in a 
battery of tests are affected and the effects appear to be dose 
dependent, the data are considered to be evidence of a neurotoxic 
effect, especially in the absence of systemic toxicity. Recently, it 
was proposed that data from FOB studies be grouped into several 
neurobiological domains, including neuromuscular (i.e., weakness, 
incoordination, abnormal movements, gait), sensory (i.e, auditory, 
visual, somatosensory), and autonomic functions (Tilson and Moser, 
1992). This statistical technique is useful when separating changes 
that occur on the basis of chance or in conjunction with systemic 
toxicity from those treatment-related changes indicative of neurotoxic 
effects. In the case of the developing organism, chemicals may alter 
the maturation or appearance of sensorimotor reflexes. Significant 
alterations in or delay of such reflexes is evidence of a neurotoxic 
effect.
    Examples of chemicals that affect neuromuscular function are 3-
acetylpyridine, acrylamide, and triethyltin. Organophosphate and 
carbamate insecticides produce autonomic dysfunction, while 
organochlorine and pyrethroid insecticides increase sensorimotor 
sensitivity, produce tremors, and in some cases, cause seizures and 
convulsions (Spencer and Schaumberg, 1980).
    (2) Motor activity. Motor activity represents a broad class of 
behaviors involving coordinated participation of sensory, motor, and 
integrative processes. Assessment of motor activity is noninvasive and 
has been used to evaluate the effects of acute and repeated exposure to 
neurotoxicants (MacPhail et al., 1989). An organism's level of activity 
can, however, be affected by many different types of environmental 
agents, including nonneurotoxic agents. Motor activity measurements 
also have been used in humans to evaluate disease states, including 
disorders of the nervous system (Goldstein and Stein, 1985).
    Motor activity is usually quantified as the frequency of movements 
over a period of time. The total counts generated during a test period 
will depend on the recording mechanism and size and configuration of 
the testing apparatus. Effects of agents on motor activity can be 
expressed as absolute activity counts or as a percentage of control 
values. In some cases, a transformation (e.g., square root) may be used 
to achieve a normal distribution of the data. The frequency of motor 
activity within a session usually decreases and is reported as the 
average number of counts occurring in each successive block of time. 
The EPA's Office of Prevention, Pesticides and Toxic Substances 
guidelines (U.S. EPA, 1991a), for example, call for test sessions of 
sufficient duration to allow motor activity to approach steady-state 
levels during the last 20 percent of the session for control animals. A 
sum of the counts in each epoch will add up to the total number of 
counts per session.
    In the adult, neurotoxic agents generally decrease motor activity 
(MacPhail et al., 1989). Examples include many pesticides (e.g., 
carbamates, chlorinated hydrocarbons, organophosphates, and 
pyrethroids), heavy metals (lead, tin, and mercury), and other agents 
(3-acetylpyridine, acrylamide, and 2,4-dithiobiuret). Some 
neurotoxicants (e.g., toluene, xylene, triadimefon) produce transient 
increases in activity by presumably stimulating neurotransmitter 
release, while others (e.g., trimethyltin) produce persistent increases 
in motor activity by destroying specific regions of the brain (e.g., 
hippocampus).
    Following developmental exposures, neurotoxic effects are often 
observed as a change in the developmental profile or maturation of 
motor activity patterns. Frequently, developmental exposure to 
neurotoxic agents will produce an increase in motor activity that 
persists into adulthood or that results in changes in other behaviors. 
This type of effect is evidence of a neurotoxic effect. Like other 
organ systems, the nervous system may be differentially sensitive to 
toxicants in groups such as the young. For example, toxicants 
introduced to the developing nervous system may kill stem cells and 
thus cause profound effects on adult structure and function. Moreover, 
toxicants may have greater access to the developing nervous system 
before the blood-brain barrier is completely formed or before metabolic 
detoxifying systems are functional.
    Motor activity measurements are typically used with other tests 
(e.g., FOB) to help detect neurotoxic effects. Agent-induced changes in 
motor activity associated with other overt signs of toxicity (e.g., 
loss of body weight, systemic toxicity) or occurring in non-dose-
related fashion are of less concern than changes that are dose 
dependent, related to structural or other functional changes in the 
nervous system, or occur in the absence of life-threatening toxicity.
    (3) Schedule-controlled operant behavior. Schedule-controlled 
operant behavior (SCOB) involves the maintenance of behavior (e.g., 
performance of a lever-press or key-peck response) by reinforcement. 
Different rates and patterns of responding are controlled by the 
relationship between response and subsequent reinforcement. SCOB 
provides a measure of performance of a learned behavior (e.g., lever 
press or key peck) and involves training and motivational variables 
that must be considered in evaluating the data. Agents may interact 
with sensory processing, motor output, motivational variables (i.e., 
related to reinforcement), training history, and baseline 
characteristics (Rice, 1988; Cory-Slechta, 1989). Rates and patterns of 
SCOB display remarkable species and experimental generality.
    In laboratory animals, SCOB has been used to study a wide range of 
neurotoxicants, including methylmercury, many pesticides, carbon 
disulfide, organic and inorganic lead, and triethyl and trimethyltin 
(MacPhail, 1985; Tilson, 1987; Rice, 1988). The primary SCOB end points 
for evaluation are response rate and the temporal pattern of 
responding. These end points may vary as a function of the contingency 
between responding and reinforcement presentation (i.e., schedule of 
reinforcement). While most chemicals decrease the efficiency of 
responding at some dose, some agents may increase response efficiency 
on schedules requiring high response rates due to a stimulant effect or 
an increase in central nervous system excitability. Agent-induced 
changes in responding between reinforcements (i.e., the temporal 
pattern of responding) may occur independently of changes in the 
overall rate of responding. Chemicals may also affect the reaction time 
to respond following presentation of a stimulus. Agent-induced changes 
in response rate or temporal patterning associated with other overt 
signs of toxicity (e.g., body weight loss, systemic toxicity, or 
occurring in a non-dose-related fashion) are of less concern than 
changes that are dose dependent, related to structural or other 
functional changes in the nervous system, or occur in the absence of 
life-threatening toxicity.
    (4) Convulsions. Observable convulsions in animals are indicative 
of an adverse effect. These events can 

[[Page 52046]]
reflect central nervous system activity comparable to that of epilepsy 
in humans and could be defined as neurotoxicity. Occasionally, other 
toxic actions of compounds, such as direct effects on muscle, might 
mimic some convulsion-like behaviors. In some cases, convulsions or 
convulsion-like behaviors may be observed in animals that are otherwise 
severely compromised, moribund, or near death. In such cases, 
convulsions might reflect an indirect effect of systemic toxicity and 
are less clearly indicative of neurotoxicity. As discussed in the 
section on neurophysiological measures, electrical recordings of brain 
activity could be used to determine specificity of effects on the 
nervous system.
    (5) Specialized tests for neurotoxicity. Several procedures have 
been developed to measure agent-induced changes in specific 
neurobehavioral functions such as motor, sensory, or cognitive function 
(Tilson, 1987; Cory-Slechta, 1989). Table 6 lists several well-known 
behavioral tests, the neurobehavioral functions they were designed to 
assess, and agents known to affect the response. Many of these tests in 
animals have been designed to assess neural functions in humans using 
similar testing procedures.

      Table 6.--Examples of Specialized Behavioral Tests to Measure     
                              Neurotoxicity                             
------------------------------------------------------------------------
        Function                Procedure          Representative agents
------------------------------------------------------------------------
Neuromuscular:                                                          
    Weakness...........  Grip strength; swimming  n-Hexane, methl n-    
                          endurance; suspension    butylketone,         
                          rod; discriminative      carbaryl.            
                          motor function.                               
    Incoordination.....  Rotorod, gait            3-Acetylpyridine,     
                          measurements; righting   ethanol.             
                          reflex.                                       
    Tremor.............  Rating scale, spectral   Chlordecone, Type I   
                          analysis.                pyrethroids, DDT.    
    Myoclonia spasms...  Rating scale, spectral   DDT, Type II          
                          analysis.                pyrethroids.         
Sensory:                                                                
    Auditory...........  Discrimination           Toluene, trimethyltin.
                          conditioning Reflex                           
                          modification.                                 
    Visual.............  Discrimination           Methylmercury.        
                          conditioning.                                 
    Somatosensory......  Discrimination           Acrylamide.           
                          conditioning.                                 
    Pain sensitivity...  Discrimination           Parathion.            
                          conditioning                                  
                          (titration);                                  
                          functional                                    
                          observational battery.                        
    Olfactory..........  Discrimination           3-Methylindole,       
                          conditioning.            methylbromide.       
Learning/Memory:                                                        
    Habituation........  Startle reflex.........  Diisopropyl-          
                                                   fluorophosphate (DFP)
                                                   Pre/neonatal         
                                                   methylmercury.       
    Classical            Nictitating membrane...  Aluminum.             
     conditioning.       Conditioned flavor       Carbaryl.             
                          aversion.               Trimethyltin, IDPN.   
                         Passive avoidance......  Neonatal trimethyltin.
                         Olfactory conditioning.                        
    Operant              One-way avoidance......  Chlordecone.          
     conditioning.       Two-way avoidance......  Pre/neonatal lead.    
                         Y-maze avoidance.......  Hypervitaminois A.    
                         Biel water maze........  Styrene.              
                         Morris water maze......  DFP.                  
                         Radial arm maze........  Trimethyltin.         
                         Delayed matching to      DFP.                  
                          sample.                 Carbaryl.             
                         Repeated acquisition...  Lead.                 
                         Visual discrimination..                        
------------------------------------------------------------------------

    A statistically significant chemically induced change in any 
measure in Table 6 is presumptive evidence of adverse effect. Judgments 
of neurotoxicity may involve not only the analysis of changes seen but 
the structure and class of the chemical and other available 
neurochemical, neurophysiological, and neuropathological evidence. In 
general, behavioral changes seen across broader dose ranges indicate 
more specific actions on the systems underlying those changes, i.e., 
the nervous system. Changes that are not dose dependent or that are 
confounded with body weight changes and/or other systemic toxicity may 
be more difficult to interpret as neurotoxic effects.
    (a) Motor function: Neurotoxicants commonly affect motor function. 
These effects can be categorized generally into (1) weakness or 
decreased strength, (2) tremor, (3) incoordination, and (4) spasms, 
myoclonia, or abnormal motor movements (Tilson, 1987; Cory-Slechta, 
1989). Specialized tests used to assess weakness include measures of 
grip strength, swimming endurance, suspension from a hanging rod, and 
discriminative motor function. Rotarod and gait assessments are used to 
measure incoordination, while rating scales and spectral analysis 
techniques can be used to quantify tremor and other abnormal movements.
    (b) Sensory function: Gross perturbations of sensory function can 
be observed in simple neurological assessments such as the FOB. 
However, these tests may not be sufficiently sensitive to detect subtle 
sensory changes. Psychophysical procedures that study the relationship 
between a physical dimension (e.g., intensity, frequency) of a stimulus 
and behavior may be necessary to quantify agent-induced alterations in 
sensory function. Examples of psychophysical procedures include 
discriminated conditioning and startle reflex modification.
    (c) Cognitive function: Alterations in learning and memory in 
experimental animals must be inferred from changes in behavior 
following exposure when compared with that either seen prior to 
exposure or with a nonexposed control group. Learning is defined as a 
relatively lasting change in behavior due to experience, and memory is 
defined as the persistence of a learned behavior over time. Table 6 
lists several examples of learning and memory tests and representative 
neurotoxicants known to affect these tests. Measurement of changes in 
learning and memory must be separated from other changes in behavior 
that do not involve cognitive or associative processes (i.e., motor 
function, sensory capabilities, motivational factors). In addition, any 
apparent toxicant-induced change in learning or memory should ideally 
be 

[[Page 52047]]
demonstrated over a range of stimulus and response conditions and 
testing conditions. In developmental exposures, it should be shown that 
the animals have matured enough to perform the specified task. 
Developmental neurotoxicants can accelerate or delay the ability to 
learn a response or interfere with cognitive function at the time of 
testing. Older animals frequently perform poorly on some types of 
tests, and it must be demonstrated that control animals in this 
population are capable of performing the procedure. Neurotoxicants 
might accelerate age-related dysfunction or alter motivational 
variables that are important for learning to occur. Further, it is not 
necessarily the case that a decrease in responding on a learning task 
is adverse while an increase in performance on a learning task is not. 
It is well known that lesions in certain regions of the brain can 
facilitate the acquisition of certain types of behaviors by removing 
preexisting response tendencies (e.g., inhibitory responses due to 
stress) that moderate the rate of learning under normal circumstances. 
Examples of learning and memory procedures include simple habituation, 
classical conditioning, and operant (or instrumental) conditioning, 
including tests for spatial learning and memory.
    e. Developmental Neurotoxicity. Although the previous discussion of 
various neurotoxicity end points and tests applies to studies in which 
developmental exposures are used, there are particular issues of 
importance in the evaluation of developmental neurotoxicity studies. 
Exposure to chemicals during development can result in a spectrum of 
effects, including death, structural abnormalities, altered growth, and 
functional deficits (U.S. EPA, 1991b). Children are often 
differentially sensitive to chemical exposure. A number of agents have 
been shown to cause developmental neurotoxicity when exposure occurred 
during the period between conception and sexual maturity (e.g., Riley 
and Vorhees, 1986; Vorhees, 1987). Table 7 lists several examples of 
agents known to produce developmental neurotoxicity in experimental 
animals. Animal models of developmental neurotoxicity have been shown 
to be sensitive to several environmental agents known to produce 
developmental neurotoxicity in humans, including lead, ethanol, x-
irradiation, methylmercury, and polychlorinated biphenyls (PCBs) 
(Kimmel et al., 1990; Needleman, 1990; Jacobson et al., 1985; 
Needleman, 1986). In many of these cases, functional deficits are 
observed at dose levels below those at which other indicators of 
developmental toxicity are evident or at minimally toxic doses in 
adults. Such effects may be transient, but generally are considered to 
be adverse effects.

            Table 7.--Examples of Developmental Neurotoxicants          
------------------------------------------------------------------------
                                                                        
------------------------------------------------------------------------
Alcohols...........................  Methanol, ethanol.                 
Antimitotics.......................  X-radiation, azacytidine.          
Insecticides.......................  DDT, kepone.                       
Metals.............................  Lead, methylmercury, cadmium.      
Polyhalogenated hydrocarbons.......  PCBs, PBBs.                        
Solvents...........................  Carbon disulfide, toluene.         
------------------------------------------------------------------------

    Testing for developmental neurotoxicity has not been required 
routinely by regulatory agencies in the United States, but is required 
by the EPA when other information indicates the potential for 
developmental neurotoxicity (U.S. EPA, 1986, 1988a, 1988b, 1989, 1991a, 
1991b). Useful data for decision making may be derived from well-
conducted adult neurotoxicity studies, standard developmental toxicity 
studies, and multigeneration studies, although the dose levels used in 
the latter may be lower than that in studies with shorter term 
exposure.
    Important design issues to be evaluated for developmental 
neurotoxicity studies are similar to those for standard developmental 
toxicity studies (e.g., a dose-response approach with the highest dose 
producing minimal overt maternal or perinatal toxicity, number of 
litters large enough for adequate statistical power, randomization of 
animals to dose groups and test groups, litter generally considered as 
the statistical unit). In addition, the use of a replicate study design 
provides added confidence in the interpretation of data. A 
pharmacological/physiological challenge may also be valuable in 
evaluating neurologic function and ``unmasking'' effects not otherwise 
detectable. For example, a challenge with a psychomotor stimulant such 
as d-amphetamine may unmask latent developmental neurotoxicity (Hughes 
and Sparber, 1978; Adams and Buelke-Sam, 1981; Buelke-Sam et al., 
1985).
    Direct extrapolation of developmental neurotoxicity to humans is 
limited in the same way as for other end points of toxicity, i.e., by 
the lack of knowledge about underlying toxicological mechanisms and 
their significance (U.S. EPA, 1991b). However, comparisons of human and 
animal data for several agents known to cause developmental 
neurotoxicity in humans showed many similarities in effects (Kimmel et 
al., 1990). Comparisons at the level of functional category (sensory, 
motivational, cognitive, and motor function and social behavior) showed 
close agreement across species for the agents evaluated, even though 
the specific end points used to assess these functions varied 
considerably across species (Stanton and Spear, 1990). Thus, it can be 
assumed that developmental neurotoxicity effects in animal studies 
indicate the potential for altered neurobehavioral development in 
humans, although the specific types of developmental effects seen in 
experimental animal studies will not necessarily be the same as those 
that may be produced in humans. Therefore, when data suggesting adverse 
effects in developmental neurotoxicity studies are encountered for 
particular agents, they should be considered in the risk assessment 
process.
    Functional tests with a moderate degree of background variability 
(e.g., a coefficient of variability of 20 percent or less) may be more 
sensitive to the effects of an agent on behavioral end points than are 
tests with low variability that may be impossible to disrupt without 
using life-threatening doses. A battery of functional tests, in 
contrast to a single test, is usually needed to evaluate the full 
complement of nervous system functions in an animal. Likewise, a series 
of tests conducted in animals in several age groups may provide more 
information about maturational changes and their persistence than tests 
conducted at a single age.
    It is a well-established principle that there are critical 
developmental periods for the disruption of functional competence, 
which include both the prenatal and postnatal periods to the time of 
sexual maturation, and the effect of a toxicant is likely to vary 
depending on the time and degree of exposure (Rodier, 1978, 1990). It 
is also important to consider the data from studies in which postnatal 
exposure is included, as there may be an interaction of the agent with 
maternal behavior, milk composition, pup suckling behavior, as well as 
possible direct exposure of pups via dosed food or water (Kimmel et 
al., 1992).
    Agents that produce developmental neurotoxicity at a dose that is 
not toxic to the maternal animal are of special concern. However, 
adverse developmental effects are often produced at doses that cause 
maternal toxicity (e.g., <20 percent reduction in weight gain during 
gestation and lactation). In these cases, the 

[[Page 52048]]
developmental effects are still considered to represent neurotoxicity 
and should not be discounted as being secondary to maternal toxicity. 
At doses causing moderate maternal toxicity (i.e., 20 
percent reduction in weight gain during gestation and lactation), 
interpretation of developmental effects may be confounded. Current 
information is inadequate to assume that developmental effects at doses 
causing minimal maternal toxicity result only from maternal toxicity; 
rather, it may be that the mother and developing organism are equally 
sensitive to that dose level. Moreover, whether developmental effects 
are secondary to maternal toxicity or not, the maternal effects may be 
reversible while the effects on the offspring may be permanent. These 
are important considerations for agents to which humans may be exposed 
at minimally toxic levels either voluntarily or involuntarily, because 
several agents are known to produce adverse developmental effects at 
minimally toxic doses in adult humans (e.g., alcohol) (Coles et al., 
1991).
    Although interpretation of developmental neurotoxicity data may be 
limited, it is clear that functional effects must be evaluated in light 
of other toxicity data, including other forms of developmental toxicity 
(e.g., structural abnormalities, perinatal death, and growth 
retardation). For example, alterations in motor performance may be due 
to a skeletal malformation rather than nervous system change. Changes 
in learning tasks that require a visual cue might be influenced by 
structural abnormalities in the eye. The level of confidence that an 
agent produces an adverse effect may be as important as the type of 
change seen, and confidence may be increased by such factors as 
reproducibility of the effect either in another study of the same 
function or by convergence of data from tests that purport to measure 
similar functions. A dose-response relationship is an extremely 
important measure of a chemical's effect; in the case of developmental 
neurotoxicity both monotonic and biphasic dose-response curves are 
likely, depending on the function being tested. The EPA Guidelines for 
Developmental Toxicity Risk Assessment (U.S. EPA, 1991b) may be 
consulted for more information on interpreting developmental toxicity 
studies. The endpoints frequently used to assess developmental 
neurotoxicity in exposed children was recently reviewed by Winneke 
(1995).
3. Other Considerations
    a. Pharmacokinetics. Extrapolation of test results between species 
can be aided considerably by data on the pharmacokinetics of a 
particular agent in the species tested and, if possible, in humans. 
Information on a toxicant's half-life, metabolism, absorption, 
excretion, and distribution to the peripheral and central nervous 
system may be useful in predicting risk. Of particular importance for 
the pharmacokinetics of neurotoxicants is the blood-brain barrier, 
which ordinarily excludes ionic and nonlipid soluble chemicals from the 
central nervous system. The brain contains circumventricular organs 
whose purpose seems to be to sense the chemical composition of the 
peripheral circulation and activate mechanisms to bring the composition 
of the blood back to equilibrium if disturbed. These areas are 
technically inside the brain, but they lie outside of the blood-brain-
barrier. Therefore, chemicals from the periphery can pass directly into 
the brain at these sites. The majority of these structures are located 
within or near the hypothalamus, an area that is crucial for 
maintenance of neuroendocrine function. Pharmacokinetic data may be 
helpful in defining the dose-response curve, developing a more accurate 
basis for comparing species sensitivity (including that of humans), 
determining dosimetry at target sites, and comparing pharmacokinetic 
profiles for various dosing regimens or routes of administration. The 
correlation of pharmacokinetic parameters and neurotoxicity data may be 
useful in determining the contribution of specific pharmacokinetic 
processes to the effects observed.
    b. Comparisons of Molecular Structure. Comparisons of the chemical 
or physical properties of an agent with those of known neurotoxicants 
may provide some indication of the potential for neurotoxicity. Such 
information may be helpful for evaluating potential toxicity when only 
minimal data are available. The structure-activity relationships (SAR) 
of some chemical classes have been studied, including hexacarbons, 
organophosphates, carbamates, and pyrethroids. Therefore, class 
relationships or SAR may help predict neurotoxicity or interpret data 
from neurotoxicological studies. Under certain circumstances (e.g., in 
the case of new chemicals), this procedure is one of the primary 
methods used to evaluate the potential for toxicity when little or no 
empirical toxicity data are available. It should be recognized, 
however, that effects of chemicals in the same class can vary widely. 
Moser (1994), for example, reported that the behavioral effects of 
prototypic cholinesterase-inhibiting pesticides differed qualitatively 
in a battery of behavioral tests.
    c. Statistical Considerations. Properly designed studies on the 
neurotoxic effects of compounds will include appropriate statistical 
tests of significance. In general, the likelihood of obtaining a 
significant effect will depend jointly on the magnitude of the effect 
and the variability obtained in control and treated groups. A number of 
texts are available on standard statistical tests (e.g., Siegel, 1956; 
Winer, 1971; Sokal and Rohlf, 1969; Salsburg, 1986; Gad and Weil, 
1988).
    Neurotoxicity data present some unique features that must be 
considered in selecting statistical tests for analysis. Data may 
involve several different measurement scales, including categorical 
(affected or not), rank (more or less affected), and interval and ratio 
scales of measurement (affected by some percentage). For example, 
convulsions are usually recorded as being present or absent 
(categorical), whereas neuropathological changes are frequently 
described in terms of the degree of damage (rank). Many tests of 
neurotoxicity involve interval or ratio measurements (e.g., frequency 
of photocell interruptions or amplitude of an evoked potential), which 
are the most powerful and sensitive scales of measurement. In addition, 
measurements are frequently made repeatedly in control and treated 
subjects, especially in the case of behavioral and neurophysiological 
end points. For example, OPPTS guidelines for FOB assessment call for 
evaluations before exposure and at several times during exposure in a 
subchronic study (U.S. EPA, 1991a).
    Descriptive data (categorical) and rank order data can be analyzed 
using standard nonparametric techniques (Siegel, 1956). In some cases, 
if it is determined that the data fit the linear model, the categorical 
modeling procedure can be used for weighted least-squares estimation of 
parameters for a wide range of general linear models, including 
repeated-measures analyses. The weighted least-squares approach to 
categorical and rank data allows computation of statistics for testing 
the significance of sources of variation as reflected by the model. In 
the case of studies assessing effects in the same animals at several 
time points, univariate analyses can be carried out at each time point 
when the overall dose effect or the dose-by-time interaction is 
significant. 

[[Page 52049]]

    Continuous data (e.g., magnitude, rate, amplitude), if found to be 
normally distributed, can be analyzed with general linear models using 
a grouping factor of dose and, if necessary, repeated measures across 
time (Winer, 1971). Univariate analyses of dose, comparing dose groups 
to the control group at each time point, are performed when there is a 
significant overall dose effect or a dose-by-time interaction. Post hoc 
comparisons between control and treatment groups can be made following 
tests for overall significance. In the case of multiple end points 
within a series of evaluations, some type of correction for multiple 
observations is warranted (Winer, 1971).
    d. In Vitro Data in Neurotoxicology. Methods and procedures that 
fall under the general heading of short-term tests include an array of 
in vitro tests that have been proposed as alternatives to whole-animal 
tests (Goldberg and Frazier, 1989). In vitro approaches use animal or 
human cells, tissues, or organs and maintain them in a nutritive 
medium. Various types of in vitro techniques produce data for 
evaluating potential and known neurotoxic substances, including primary 
cell cultures, cell lines, and cloned cells. While such procedures are 
important in studying the mechanism of action of toxic agents, their 
use in hazard identification in human health risk assessment has not 
been explored to any great extent.
    Data from in vitro procedures are generally based on simplified 
approaches that require less time to yield information than do many in 
vivo techniques. However, in vitro methods generally do not take into 
account the distribution of the toxicant in the body, the route of 
administration, or the metabolism of the substance. It also is 
difficult to extrapolate in vitro data to animal or human neurotoxicity 
end points, which include behavioral changes, motor disorders, sensory 
and perceptual disorders, lack of coordination, and learning deficits. 
In addition, data from in vitro tests cannot duplicate the complex 
neuronal circuitry characteristic of the intact animal.
    Many in vitro systems are now being evaluated for their ability to 
predict the neurotoxicity of various agents seen in intact animals. 
This validation process requires considerations in study design, 
including defined end points of toxicity and an understanding of how a 
test agent would be handled in vitro as compared to the intact 
organism. Demonstrated neurotoxicity in vitro in the absence of in vivo 
data is suggestive but inadequate evidence of a neurotoxic effect. In 
vivo data supported by in vitro data enhance the reliability of the in 
vivo results.

B. Dose-Response Evaluation

    Dose-response evaluation is a critical part of hazard 
characterization and involves the description of the dose response 
relationship in the available data. Human studies covering a range of 
exposures are rarely available and therefore animal data are typically 
used for estimating exposure levels likely to produce adverse effects 
in humans. Evidence for a dose-response relationship is an important 
criterion in establishing a neurotoxic effect, although this analysis 
may be limited when based on standard studies using three dose groups 
or fewer. The evaluation of dose-response relationships includes 
identifying effective dose levels as well as doses associated with no 
increase in incidence of adverse effects when compared with controls. 
Much of the focus is on identifying the critical effect(s) observed at 
the lowest-observed-adverse-effect-level and the no-observed-adverse-
effect-level associated with that effect. The NOAEL is defined as the 
highest dose at which there is no statistically or biologically 
significant increase in the frequency of an adverse neurotoxic effect 
when compared with the appropriate control group in a data base 
characterized as having sufficient evidence for use in a risk 
assessment (see section C). Although a threshold is assumed for 
neurotoxic effects, the existence of a NOAEL in an animal study does 
not prove or disprove the existence or level of a biological threshold. 
Alternatively, mathematical modeling of the dose-response relationship 
may be performed to determine a quantitative estimate of responses in 
the experimental range. This approach can be used to determine a BMD, 
which may be used in place of the NOAEL (Crump, 1994) (see Dose-
Response Analysis, Section IV).
    In addition to identifying the NOAEL/LOAEL or BMD, the dose-
response evaluation defines the range of doses that are neurotoxic for 
a given agent, species, route of exposure, and duration of exposure. In 
addition to these considerations, pharmacokinetic factors and other 
aspects that might influence comparisons with human exposure scenarios 
should be taken into account. For example, dose-response curves may 
exhibit not only monotonic but also U-shaped or inverted U-shaped 
functions (Davis and Svendsgaard, 1990). Such curves are hypothesized 
to reflect multiple mechanisms of action, the presence of homeostatic 
mechanisms, and/or activation of compensatory or protective mechanisms. 
In addition to considering the shape of the dose-response curve, it 
should also be recognized that neurotoxic effects vary in terms of 
nature and severity across dose or exposure level. At high levels of 
exposure, frank lesions accompanied by severe functional impairment may 
be observed. Such effects are widely accepted as adverse. At 
progressively lower levels of exposure, however, the lesions may become 
less severe and the impairments less obvious. At levels of exposure 
near the NOAEL and LOAEL, the effects will often be mild, possibly 
reversible, and inconsistently found. In addition, the end points 
showing responses may be at levels of organization below the whole 
organism (e.g., neurochemical or electrophysiological end points). The 
adversity of such effects can be contentious (e.g., cholinesterase 
inhibition), yet it is such effects that are likely to be the focus of 
risk assessment decisions. To the extent possible, this document 
provides guidance on determining the adversity of neurotoxic effects. 
However, the identification of a critical adverse effect often requires 
considerable professional judgment and should consider factors such as 
the biological plausibility of the effect, the evidence of a dose-
effect continuum, and the likelihood for progression of the effect with 
continued exposure.

C. Characterization of the Health-Related Data Base

    This section describes a scheme for characterizing the sufficiency 
of evidence for neurotoxic effects. This scheme defines two broad 
categories: sufficient and insufficient (Table 8). Categorization is 
aimed at providing certain criteria for the Agency to use to define the 
minimum evidence necessary to define hazards and to conduct dose-
response analyses. It does not address the issues related to 
characterization of risk, which requires analysis of potential human 
exposures and their relation to potential hazards to estimate the risks 
of those hazards from anticipated or estimated exposures.

Table 8.--Characterization of the Health-Related Database

Sufficient Evidence

    The sufficient evidence category includes data that collectively 
provide enough information to judge whether or not a human 
neurotoxic hazard could exist. This category may include both human 
and experimental animal evidence.

Sufficient Human Evidence

    This category includes agents for which there is sufficient 
evidence from 

[[Page 52050]]
epidemiologic studies, e.g., case control and cohort studies, to judge 
that some neurotoxic effect is associated with exposure. A case 
series in conjunction with other supporting evidence may also be 
judged ``sufficient evidence.'' Epidemiologic and clinical case 
studies should discuss whether the observed effects can be 
considered biologically plausible in relation to chemical exposure. 
(Historically, much often has been made of the notion of causality 
in epidemiologic studies. Causality is a more stringent criterion 
than association and has become a topic of scientific and 
philosophical debate. See Susser [1986], for example, for a 
discussion of inference in epidemiology.)

Sufficient Experimental Animal Evidence/Limited or No Human Data

    This category includes agents for which there is sufficient 
evidence from experimental animal studies and/or limited human data 
to judge whether a potential neurotoxic hazard may exist. Generally, 
agents that have been tested according to current test guidelines 
would be included in this category. The minimum evidence necessary 
to judge that a potential hazard exists would be data demonstrating 
an effect in a single appropriate, well-executed study in a single 
experimental animal species, whereas the minimum evidence needed to 
judge that a potential hazard does not exist would include data from 
appropriate, well-executed laboratory animal studies that evaluated 
a variety of the potential manifestations of neuroxtoxicity and 
showed no effects at doses that were at least minimally toxic. 
Information on pharmacokinetics, mechanisms, or known properties of 
the chemical class may also strengthen the evidence.

Insufficient Evidence

    This category includes agents for which there is less than the 
minimum evidence sufficient for identifying whether or not a 
neurotoxic hazard exists, such as agents for which there are no data 
on neurotoxicity or agents with data bases from studies in animals 
or humans that are limited by study design or conduct (e.g., 
inadequate conduct or report of clinical signs). Many general 
toxicity studies, for example, are considered insufficient in terms 
of the conduct of clinical neurobehavioral observations or the 
number of samples taken for histopathology of the nervous system. 
Thus, a battery of negative toxicity studies with these shortcomings 
would be regarded as providing insufficient evidence of the lack of 
a neurotoxic effect of the test material. Further, most screening 
studies based on simple observations involving autonomic and motor 
function provide insufficient evaluation of many sensory or 
cognitive functions. Data, which by itself would likely fall in this 
category, would also include information on structure-activity 
relationships or data from in vitro tests. While such information 
would be insufficient by itself to proceed further in the assessment 
it could be used to support the need for additional testing.
    Data from all potentially relevant studies, whether indicative 
of potential hazard or not, should be included in this 
characterization. The primary sources of data are human studies and 
case reports, experimental animal studies, other supporting data, 
and in vitro and/or structure-activity relationship data. Because a 
complex interrelationship exists among study design, statistical 
analysis, and biological significance of the data, a great deal of 
scientific judgment, based on experience with neurotoxicity data and 
with the principles of study design and statistical analysis, is 
required to adequately evaluate the data base on neurotoxicity. In 
many cases, interaction with scientists in specific disciplines 
either within or outside the field of neurotoxicology (e.g., 
epidemiology, statistics) may be appropriate.
    The adverse nature of different neurotoxicity end points may be 
a complex judgment. In general, most neuropathological and many 
neurobehavioral changes are regarded as adverse. However, there are 
adverse behavioral effects that may not reflect a direct action on 
the nervous system. Neurochemical and electrophysiological changes 
may be regarded as adverse as a function of their known or presumed 
relation to neuropathological and/or neurobehavioral consequences. 
In the absence of supportive information, a professional judgment 
must be made regarding the adversity of such outcomes, considering 
factors such as the nature, magnitude, and duration of the effects 
reported. Thus, correlated measures of neurotoxicity strengthen the 
evidence for a hazard. Correlations between functional and 
morphological effects, such as the correlation between leg weakness 
and paralysis and peripheral nerve damage from exposure to tri-
ortho-cresyl phosphate, are the most common and striking example of 
this form of validity. Correlations support a coherent and logical 
link between behavioral effects and biochemical mechanisms. 
Replication of a finding also strengthens the evidence for a hazard. 
Some neurotoxicants cause similar effects across most species. Many 
chemicals shown to produce neurotoxicity in laboratory animals have 
similar effects in humans. Some neurologic effects may be considered 
adverse even if they are small in magnitude, reversible, or the 
result of indirect mechanisms.
    Because of the inherent difficulty in ``proving any negative,'' 
it is more difficult to document a finding of no apparent adverse 
effect than a finding of an adverse effect. Neurotoxic effects (and 
most kinds of toxicity) can be observed at many different levels, so 
that only a single end point needs to be found to demonstrate a 
hazard, but many end points need to be examined to demonstrate no 
effect. For example, to judge that a hazard for neurotoxicity could 
exist for a given agent, the minimum evidence sufficient would be 
data on a single adverse end point from a well-conducted study. In 
contrast, to judge that an agent is unlikely to pose a hazard for 
neurotoxicity, the minimum evidence would include data from a host 
of end points that revealed no neurotoxic effects. This may include 
human data from appropriate studies that could support a conclusion 
of no evidence of a neurotoxic effect. With respect to clinical 
signs and symptoms, human exposures can reveal far more about the 
absence of effects than animal studies, which are confined to the 
signs examined.
    In some cases, it may be that no individual study is judged 
sufficient to establish a hazard, but the total available data may 
support such a conclusion. Pharmacokinetic data and structure-
activity considerations, data from other toxicity studies, as well 
as other factors may affect the strength of the evidence in these 
situations. For example, given that gamma diketones are known to 
cause motor system neurotoxicity, a marginal data set on a candidate 
gamma diketone, e.g., \1/10\ animals affected, might be more likely 
to be judged sufficient than equivalent data from a member of a 
chemical class about which nothing is known.
    A judgment that the toxicology data base is sufficient to 
indicate a potential neurotoxic hazard is not the end of analysis. 
The circumstances of expression of hazard are essential to 
describing human hazard potential. Thus, reporting should contain 
the details of the circumstances under which effects have been 
observed, e.g., ``long-term oral exposures of adult rodents to 
compound X at levels of roughly 1 mg/kg have been associated with 
ataxia and peripheral nerve damage.''

IV. Dose-Response Analysis

    This section describes several approaches (including the LOAEL/
NOAEL and BMD) for determining the reference dose or reference 
concentration. The NOAEL or BMD/uncertainty factor approach results in 
a RfD or RfC, which is an estimate (with uncertainty spanning perhaps 
an order of magnitude) of a daily exposure to the human population 
(including sensitive subgroups) that is likely to be without an 
appreciable risk of deleterious effects during a lifetime.
    The dose-response analysis characterization should:
    a. Describe how the RfD/RfC was calculated;
    b. Discuss the confidence in the estimates;
    c. Describe the assumptions or uncertainty factors used; and
    d. Discuss the route and level of exposure observed, as compared to 
expected human exposures. (Specifically, are the available data from 
the same route of exposure as the expected human exposures? How many 
orders of magnitude do you need to extrapolate from the observed data 
to environmental exposures?)

A. LOAEL/NOAEL and Benchmark Dose (BMD) Determination

    As indicated earlier, the LOAEL and NOAEL are determined for 
endpoints that are seen at the lowest dose level (so-called critical 
effect). Several limitations in the use of the NOAEL have been 
identified and described (e.g., Barnes and Dourson, 1988; Crump, 1984). 
For example, the NOAEL is derived from a single end point from a single 
study (the critical study) and 

[[Page 52051]]
ignores both the slope of the dose-response function and baseline 
variability in the end point of concern. Because the baseline 
variability is not taken into account, the NOAEL from a study using 
small group sizes may be higher than the NOAEL from a similar study in 
the same species that uses larger group sizes. The NOAEL is also 
directly dependent on the dose spacing used in the study. Finally, and 
perhaps most importantly, use of the NOAEL does not allow estimates of 
risk or extrapolation of risk to lower dose levels.
    Because of these and other limitations in the NOAEL approach, 
mathematical curve-fitting techniques (Crump, 1984; Gaylor and Slikker, 
1990; Glowa, 1991; U.S. EPA, 1995a) are beginning to be used with, or 
as an alternative to, the NOAEL in calculating the RfD or RfC. The 
Agency is in the process of implementing these newer techniques and 
strongly encourages the calculation of BMDs for neurotoxicity and other 
health effect end points. These techniques typically apply a 
mathematical function that describes the dose-response relationship and 
then interpolate to a level of exposure associated with a small 
increase in effect over that occurring in the control group or under 
baseline conditions. The BMD has been defined as a lower confidence 
limit on the effective dose associated with some defined level of 
effect, e.g., a 5 percent or 10 percent increase in response (i.e., a 
BMD05 or BMD10 for a particular effect). Because the model is 
only used to interpolate within the dose range of the study, no 
assumptions about the existence (or nonexistence) of a threshold are 
needed. Thus, any model that fits the data well is likely to provide a 
reasonable estimate of the BMD.
    Many neurotoxic end points provide continuous measures of response, 
such as response speed, nerve conduction velocity, IQ score, degree of 
enzyme inhibition, or the accuracy of task performance. Although it is 
possible to impose a dichotomy on a continuous effects distribution and 
to classify some level of response as ``affected'' and the remainder as 
``unaffected,'' it may be very difficult and inappropriate to establish 
such clear distinctions, because such a dichotomy would misrepresent 
the true nature of the neurotoxic response. Alternatively, quantitative 
models designed to analyze continuous effect variables may be 
preferable. Other techniques that allow this approach, with 
transformation of the information into estimates of the incidence or 
frequency of affected individuals in a population, have been proposed 
(Crump, 1984; Gaylor and Slikker, 1990). Categorical regression 
analysis has been proposed since it can evaluate different types of 
data and derive estimates for short-term exposures (Rees and Hattis, 
1994). Decisions about the most appropriate approach require 
professional judgment, taking into account the biological nature of the 
continuous effect variable and its distribution in the population under 
study.
    Although dose-response functions in neurotoxicology are generally 
linear or monotonic, curvilinear functions, especially U-shaped or 
inverted U-shaped curves, have been reported as noted earlier (Section 
III B). Dose-response analyses should consider the uncertainty that U-
shaped dose-response functions might contribute to the estimate of the 
NOAEL/LOAEL or BMD. Typically, estimates of the NOAEL/LOAEL are taken 
from the lowest part of the dose-response curve associated with 
impaired function or adverse effect.

B. Determination of the Reference Dose or Reference Concentration

    Since the availability of dose-response data in humans is limited, 
extrapolation of data from animals to humans usually involves the 
application of uncertainty factors to the NOAEL/LOAEL or BMD. The NOAEL 
or BMD/uncertainty factor approach results in a RfD or RfC, which is an 
estimate (with uncertainty spanning perhaps an order of magnitude) of a 
daily exposure to the human population (including sensitive subgroups) 
that is likely to be without an appreciable risk of deleterious effects 
during a lifetime. The oral RfD and inhalation RfC are applicable to 
chronic exposure situations and are based on an evaluation of all the 
noncancer health effects, including neurotoxicity data. RfDs and RfCs 
in the Integrated Risk Information System (IRIS-2) data base for 
several agents are based on neurotoxicity end points and include a few 
cases in which the RfD or RfC is calculated using the BMD approach 
(e.g., methylmercury, carbon disulfide). The size of the final 
uncertainty factor used will vary from agent to agent and will require 
the exercise of scientific judgment, taking into account interspecies 
differences, the shape of the dose-response curve, and the 
neurotoxicity end points observed. Default uncertainty factors are 
typically multiples of 10 and are used to compensate for human 
variability in sensitivity, the need to extrapolate from animals to 
humans, and the need to extrapolate from less than lifetime (e.g., 
subchronic) to lifetime exposures. An additional factor of up to 10 may 
be included when only a LOAEL (and not a NOAEL) is available from a 
study, or depending on the completeness of the data base, a modifying 
factor of up to 10 may be applied, depending on the confidence one has 
in the data base. Barnes and Dourson (1988) provide a more complete 
description of the calculation, use, and significance of RfDs in 
setting exposure limits to toxic agents by the oral route. Jarabek et 
al. (1990) provide a more complete description of the calculation, use, 
and significance of RfCs in setting exposure limits to toxic agents in 
air. Neurotoxicity can result from acute, shorter term exposures, and 
it may be appropriate in some cases, e.g., for air pollutants or water 
contaminants, to set shorter term exposure limits for neurotoxicity as 
well as for other noncancer health effects.

V. Exposure Assessment

    Exposure assessment describes the magnitude, duration, frequency, 
and routes of exposure to the agent of interest. This information may 
come from hypothetical values, models, or actual experimental values, 
including ambient environmental sampling results. Guidelines for 
exposure assessment have been published separately (U.S. EPA, 1992) and 
will, therefore, be discussed only briefly here.
    The exposure assessment should include an exposure characterization 
that:
    a. Provides a statement of the purpose, scope, level of detail, and 
approach used in the exposure assessment;
    b. Presents the estimates of exposure and dose by pathway and route 
for individuals, population segments, and populations in a manner 
appropriate for the intended risk characterization;
    c. Provides an evaluation of the overall level of confidence in the 
estimate of exposure and dose and the conclusions drawn; and
    d. Communicates the results of the exposure assessment to the risk 
assessor, who can then use the exposure characterization, along with 
the characterization of the other risk assessment elements, to develop 
a risk characterization.
    A number of considerations are relevant to exposure assessment for 
neurotoxicants. An appropriate evaluation of exposure should consider 
the potential for exposure via ingestion, inhalation, and dermal 
penetration from relevant sources of exposures, including multiple 
avenues of intake from the same source. On-going Agency activities that 
support neurotoxicity exposure 

[[Page 52052]]
assessment include characterizing cumulative risk and revising the 
Guidelines for the Health Risk Assessment of Chemical Mixtures.
    In addition, neurotoxic effects may result from short-term (acute), 
high-concentration exposures as well as from longer term (subchronic), 
lower level exposures. Neurotoxic effects may occur after a period of 
time following initial exposure or be obfuscated by repair mechanisms 
or apparent tolerance. The type and severity of effect may depend 
significantly on the pattern of exposure rather than on the average 
dose over a long period of time. For this reason, exposure assessments 
for neurotoxicants may be much more complicated than those for long-
latency effects such as carcinogenicity. It is rare for sufficient data 
to be available to construct such patterns of exposure or dose, and 
professional judgment may be necessary to evaluate exposure to 
neurotoxic agents.

VI. Risk Characterization

A. Overview

    Risk characterization, the culmination of the risk assessment 
process, consists of an integrative analysis and a risk 
characterization summary. The integrative analysis (a) involves 
integration of the toxicity information from the hazard 
characterization and dose-response analysis with the human exposure 
estimates, (b) provides an evaluation of the overall quality of the 
assessment and the degree of confidence in the estimates of risk and 
conclusions drawn, and (c) describes risk in terms of the nature and 
extent of harm. The risk characterization summary communicates the 
results of the risk assessment to the risk manager.
    This summary should include but is not limited to a discussion of 
the following elements:
    a. Quality of and confidence in the available data;
    b. Uncertainty analysis;
    c. Justification of defaults or assumptions;
    d. Related research recommendations;
    e. Contentious issues and extent of scientific consensus;
    f. Effect of reasonable alternative assumptions on conclusions and 
estimates;
    g. Highlight reasonable plausible ranges;
    h. Reasonable alternative models; and
    i. Perspective through analogy.
    The risk manager can then use the risk assessment, along with other 
risk management elements, to make public health decisions.
    An effective risk characterization must fully, openly, and clearly 
characterize risks and disclose the scientific analyses, uncertainties, 
assumptions, and science policies that underlie decisions throughout 
the risk assessment and risk management processes. The risk 
characterization must feature values such as transparency in the 
decision-making process; clarity in communicating with each other and 
the public regarding environmental risk and the uncertainties 
associated with assessments of environmental risk; and consistency 
across program offices in core assumptions and science policies, which 
are well grounded in science and reasonable.
    The following sections describe these four aspects of the risk 
characterization in more detail.

B. Integration of Hazard Characterization, Dose-Response Analysis and 
Exposure Assessment

    In developing the hazard characterization, dose-response analysis 
and exposure portions of the risk assessment, the assessor must take 
into account many judgments concerning human relevance of the toxicity 
data, including the appropriateness of the various animal models for 
which data are available and the route, timing, and duration of 
exposure relative to expected human exposure. These judgments should be 
summarized at each stage of the risk assessment process (e.g., the 
biological relevance of anatomical variations may be established in the 
hazard characterization process, or the influence of species 
differences in metabolic patterns in the dose-response analysis). In 
integrating the information from the assessment, the risk assessor must 
determine if some of these judgments have implications for other 
portions of the assessment and whether the various components of the 
assessment are compatible.
    The risk characterization should not only examine the judgments but 
also explain the constraints of available data and the state of 
knowledge about the phenomena studied in making them, including (1) the 
qualitative conclusions about the likelihood that the chemical may pose 
a specific hazard to human health, the nature of the observed effects, 
under what conditions (route, dose levels, time, and duration) of 
exposure these effects occur, and whether the health-related data are 
sufficient to use in a risk assessment; (2) a discussion of the dose-
response characteristics of the critical effects(s), data such as the 
shapes and slopes of the dose-response curves for the various end 
points, the rationale behind the determination of the NOAEL and LOAEL 
and calculation of the benchmark dose, and the assumptions underlying 
the estimation of the RfD or RfC; and (3) the estimates of the 
magnitude of human exposure; the route, duration, and pattern of the 
exposure; relevant pharmacokinetics; and the number and characteristics 
of the population(s) exposed.
    If data to be used in a risk characterization are from a route of 
exposure other than the expected human exposure, then pharmacokinetic 
data should be used, if available, to make extrapolations across routes 
of exposure. If such data are not available, the Agency makes certain 
assumptions concerning the amount of absorption likely or the 
applicability of the data from one route to another (U.S. EPA, 1992).
    The level of confidence in the hazard characterization should be 
stated to the extent possible, including the appropriate category 
regarding sufficiency of the health-related data. A comprehensive risk 
assessment ideally includes information on a variety of end points that 
provide insight into the full spectrum of potential neurotoxicological 
responses. A profile that integrates both human and test species data 
and incorporates a broad range of potential adverse neurotoxic effects 
provides more confidence in a risk assessment for a given agent.
    The ability to describe the nature of the potential human exposure 
is important to predict when certain outcomes can be anticipated and 
the likelihood of permanence or reversibility of the effect. An 
important part of this effort is a description of the nature of the 
exposed population and the potential for sensitive, highly susceptible, 
or highly exposed populations. For example, the consequences of 
exposure to the developing individual versus the adult can differ 
markedly and can influence whether the effects are transient or 
permanent. Other considerations relative to human exposures might 
include the likelihood of exposures to other agents, concurrent 
disease, and nutritional status.
    The presentation of the integrated results of the assessment should 
draw from and highlight key points of the individual characterizations 
of component analyses performed under these Guidelines. The overall 
risk characterization represents the integration of these component 
characterizations. If relevant risk assessments on the agent or an 
analogous agent have been done by EPA or other Federal agencies, these 
should 

[[Page 52053]]
be described and the similarities and differences discussed.

C. Quality of the Data Base and Degree of Confidence in the Assessment

    The risk characterization should summarize the kinds of data 
brought together in the analysis and the reasoning on which the 
assessment is based. The description should convey the major strengths 
and weaknesses of the assessment that arise from availability of data 
and the current limits of our understanding of the mechanisms of 
toxicity.
    Health risk is a function of the hazard characterization, dose-
response analysis, and exposure assessment. Confidence in the results 
of a risk assessment is, thus, a function of confidence in the results 
of the analysis of these elements. Each of these elements should have 
its own characterization as a part of the assessment. Within each 
characterization, the important uncertainties of the analysis and 
interpretation of data should be explained, and the risk manager should 
be given a clear picture of consensus or lack of consensus that exists 
about significant aspects of the assessment. Whenever more than one 
view is supported by the data and choosing between them is difficult, 
all views should be presented. If one has been selected over the 
others, the rationale should be given; if not, then all should be 
presented as plausible alternative results.

D. Descriptors of Neurotoxicity Risk

    There are a number of ways to describe risks. Several ways that are 
relevant to describing risks for neurotoxicity are as follows:
1. Estimation of the Number of Individuals
    The RfD or RfC is taken to be a chronic exposure level at or below 
which no significant risk occurs. Therefore, presentation of the 
population in terms of those at or below the RfD or RfC (``not at 
risk'') and above the RfD or RfC (``may be at risk'') may be useful 
information for risk managers. This method is particularly useful to a 
risk manager considering possible actions to ameliorate risk for a 
population. If the number of persons in the at-risk category can be 
estimated, then the number of persons removed from the at-risk category 
after a contemplated action is taken can be used as an indication of 
the efficacy of the action.
2. Presentation of Specific Scenarios
    Presenting specific scenarios in the form of ``what if?'' questions 
is particularly useful to give perspective to the risk manager, 
especially where criteria, tolerance limits, or media quality limits 
are being set. The question being asked in these cases is, at this 
proposed limit, what would be the resulting risk for neurotoxicity 
above the RfD or RfC?
3. Risk Characterization for Highly Exposed Individuals
    This measure is one example of the just-discussed descriptor. This 
measure describes the magnitude of concern at the upper end of the 
exposure distribution. This allows risk managers to evaluate whether 
certain individuals are at disproportionately high or unacceptably high 
risk.
    The objective of looking at the upper end of the exposure 
distribution is to derive a realistic estimate of a relatively highly 
exposed individual or individuals. This measure could be addressed by 
identifying a specified upper percentile of exposure in the population 
and/or by estimating the exposure of the highest exposed individual(s). 
Whenever possible, it is important to express the number of individuals 
who comprise the selected highly exposed group and discuss the 
potential for exposure at still higher levels.
    If population data are absent, it will often be possible to 
describe a scenario representing high-end exposures using upper 
percentile or judgment-based values for exposure variables. In these 
instances caution should be used not to compound a substantial number 
of high-end values for variables if a ``reasonable'' exposure estimate 
is to be achieved.
4. Risk Characterization for Highly Sensitive or Susceptible 
Individuals
    This measure identifies populations sensitive or susceptible to the 
effect of concern. Sensitive or susceptible individuals are those 
within the exposed population at increased risk of expressing the toxic 
effect. All stages of nervous system maturation might be considered 
highly sensitive or susceptible, but certain subpopulations can 
sometimes be identified because of critical periods for exposure, for 
example, pregnant or lactating women, infants, children.
    In general, not enough is understood about the mechanisms of 
toxicity to identify sensitive subgroups for all agents, although 
factors such as nutrition, personal habits (e.g., smoking, alcohol 
consumption, illicit drug abuse), or preexisting disease (e.g., 
diabetes, sexually transmitted diseases) may predispose some 
individuals to be more sensitive to the neurotoxic effects of various 
agents.
5. Other Risk Descriptors
    In risk characterization, dose-response information and the human 
exposure estimates may be combined either by comparing the RfD or RfC 
and the human exposure estimate or by calculating the margin of 
exposure (MOE). The MOE is the ratio of the NOAEL from the most 
appropriate or sensitive species to the estimated human exposure level. 
If a NOAEL is not available, a LOAEL may be used in calculating the 
MOE. Alternatively, a benchmark dose may be compared with the estimated 
human exposure level to obtain the MOE. Considerations for the 
evaluation of the MOE are similar to those for the uncertainty factor 
applied to the LOAEL/NOAEL or the benchmark dose. The MOE is presented 
along with a discussion of the adequacy of the data base, including the 
nature and quality of the hazard and exposure data, the number of 
species affected, and the dose-response information.
    The RfD or RfC comparison with the human exposure estimate and the 
calculation of the MOE are conceptually similar but are used in 
different regulatory situations. The choice of approach depends on 
several factors, including the statute involved, the situation being 
addressed, the data base used, and the needs of the decision maker. The 
RfD or RfC and the MOE are considered along with other risk assessment 
and risk management issues in making risk management decisions, but the 
scientific issues that must be taken into account in establishing them 
have been addressed here.
    If the MOE is equal to or more than the uncertainty factor  x  any 
modifying factor used as a basis for an RfD or RfC, then the need for 
regulatory concern is likely to be small. Although these methods of 
describing risk do not actually estimate risks per se, they give the 
risk manager some sense of how close the exposures are to levels of 
concern.

E. Communicating Results

    Once the risk characterization is completed, the focus turns to 
communicating results to the risk manager. The risk manager uses the 
results of the risk characterization along with other technological, 
social, and economic considerations in reaching a regulatory decision. 
Because of the way in which these risk management factors 

[[Page 52054]]
may affect different cases, consistent but not necessarily identical 
risk management decisions must be made on a case-by-case basis. These 
Guidelines are not intended to give guidance on the nonscientific 
aspects of risk management decisions.

F. Summary and Research Needs

    These Guidelines summarize the procedures that the U.S. 
Environmental Protection Agency would use in evaluating the potential 
for agents to cause neurotoxicity. These Guidelines discuss the general 
default assumptions that should be made in risk assessment for 
neurotoxicity because of gaps in our knowledge about underlying 
biological processes and how these compare across species. Research to 
improve the risk assessment process is needed in a number of areas. For 
example, research is needed to delineate the mechanisms of 
neurotoxicity and pathogenesis, provide comparative pharmacokinetic 
data, examine the validity of short-term in vivo and in vitro tests, 
elucidate the functional modalities that may be altered, develop 
improved animal models to examine the neurotoxic effects of exposure 
during the premating and early postmating periods and in neonates, 
further evaluate the relationship between maternal and developmental 
toxicity, provide insight into the concept of threshold, develop 
approaches for improved mathematical modeling of neurotoxic effects, 
improve animal models for examining the effects of agents given by 
various routes of exposure, and address the synergistic or antagonistic 
effects of mixtures of chemicals and neurotoxic response. Such research 
will aid in the evaluation and interpretation of data on neurotoxicity 
and should provide methods to assess risk more precisely.

VII. References

Adams, J.; Buelke-Sam, J. (1981) Behavioral testing of the postnatal 
animal: testing and methods development. In: Kimmel, C.A.; Buelke-
Sam, J., eds. Developmental toxicology. New York: Raven Press, pp. 
233-238.
Anger, W.K. (1984) Neurobehavioral testing of chemicals: impact on 
recommended standards. Neurobehav. Toxicol. Teratol. 6:147-153.
Anger, W.K. (1986) Workplace exposures. In: Annau, Z.A. ed., 
Neurobehavioral toxicology. Baltimore: Johns Hopkins University 
Press, pp. 331-347.
Anger, W.K. (1990) Worksite behavioral research: results, sensitive 
methods, test batteries, and the transition from laboratory data to 
human health. Neurotoxicology 11:627-718.
Anger, K.; Johnson, B.L. (1985) Chemicals affecting behavior. In: 
O'Donoghue, J., ed. Neurotoxicity of industrial and commercial 
chemicals. Boca Raton, FL: CRC Press.
Barnes, D.G.; Dourson, M. (1988) Reference dose (RfD): description 
and use in health risk assessments. Regul. Toxicol. Pharmacol. 
8:471-486.
Bondy, S.C. (1986) The biochemical evaluation of neurotoxic damage. 
Fundam. Appl. Toxicol. 6:208-216.
Boyes, W.K. (1992) Testing visual system toxicity using visual 
evoked potential technology. In: Isaacson, R.L.; Jensen, K.F., eds. 
The vulnerable brain and environmental risks, Vol. 1: Malnutrition 
and hazard assessment. New York: Plenum, pp. 193-222.
Boyes, W.K. (1993) Sensory-evoked potentials: measures of 
neurotoxicity. In: Erinoff, L., ed. Assessing the toxicity of drugs 
of abuse. NIDA Research Monograph 136. National Institute on Drug 
Abuse, Alcohol, Drug Abuse and Mental Health Administration, U.S. 
Department of Health and Human Services, pp. 63-100.
Buelke-Sam, J., Kimmel, C.A.; Adams, J. (1985) Design considerations 
in screening for behavioral teratogens: results of the collaborative 
teratology study. Neurobehav. Toxicol. Teratol. 7:537-589.
Callender, T.J.; Morrow, L.; Subramanian, K. (1994) Evaluation of 
chronic neurological sequelae after acute pesticide exposure using 
SPECT brain scans. J. Toxicol. Environ. Hlth. 41:275-284.
Carson, B.L.; Stockton, R.A.; Wilkinson, R.R. (1987) Organomercury, 
lead, tin compounds in the environment and the potential for human 
exposure. In: Tilson, H.A.; Sparber, S.B., eds. Neurotoxicants and 
neurobiological function: effects of organoheavy metals. New York: 
J. Wiley, pp. 1-80.
Coles, C.D.; Brown, R.T.; Smith, I.E.; Platzman, K.A.; Erickson, S.; 
Falek, A. (1991) Effects of prenatal alcohol exposure at school age 
I. Physical and cognitive development. Neurotoxicol. Teratol. 
13:357-367.
Cory-Slechta, D.A. (1989) Behavioral measures of neurotoxicity. 
Neurotoxicology 10:271-296.
Costa, L.G. (1988) Interactions of neurotoxicants with 
neurotransmitter systems. Toxicology 49:359-366.
Crump, K.S. (1984) A new method for determining allowable daily 
intakes. Fundam. Appl. Toxicol. 4:854-871.
Davis, J.M.; Svendsgaard, D.J. (1990) U-shaped dose-response curves: 
their occurrence and implication for risk assessment. J. Toxicol. 
Environ. Health 30:71-83.
Dyer, R.S. (1985) The use of sensory evoked potentials in 
toxicology. Fundam. Appl. Toxicol. 5:24-40.
Dyer, R.S. (1987) Macrophysiological assessment of organometal 
neurotoxicity. In: Tilson, H.A.; Sparber, S.B., eds. Neurotoxicants 
and neurobiological function effects of organoheavy metals. New 
York: J. Wiley, pp. 137-184.
Eccles, C.U. (1988) EEG correlates of neurotoxicity. Neurotoxicol. 
Teratol. 10:423-428.
Ecobichon, D.J.; Joy, R.M. (1982) Pesticides and neurological 
diseases. Boca Raton, FL: CRC Press, pp. 151-203.
Ecobichon, D.J.; Davies, J.E.; Doull, J.; Ehrich, M.; Joy, R.; 
McMillan, D.; MacPhail, R.; Reiter, L.W.; Slikker, W., Jr.; Tilson, 
H. (1990) Neurotoxic effects of pesticides. In: Baker, S.R.; 
Wilkinson, C.F., eds. The effect of pesticides on human health. 
Princeton, NJ: Princeton Scientific Publishing Co., Inc., pp. 131-
199.
Friedlander, B.R.; Hearn, H.T. (1980) Epidemiologic considerations 
in studying neurotoxic disease. In: Spencer, P.S.; Schaumberg, H.H., 
eds. Experimental and clinical neurotoxicology. Baltimore: Williams 
and Wilkins, pp. 650-662.
Gad, S.C. (1982) A neuromuscular screen for use in industrial 
toxicology. J. Toxicol. Environ. Health 9:691-704.
Gad, S.; Weil, C.S. (1988) Statistics and experimental design for 
toxicologists, 2nd ed. Caldwell, NJ: Telford Press.
Gaylor, D.W.; Slikker, W. (1990) Risk assessment for neurotoxic 
effects. Neurotoxicology 11:211-218.
Glowa, J.R. (1991) Dose-effect approaches to risk assessment. 
Neurosci. Biobehav. Rev. 15:153-158.
Goldberg, A.M.; Frazier, J.M. (1989) Alternatives to animals in 
toxicity testing. Sci. Am. 261:24-30.
Goldstein, M.K.; Stein, G.H. (1985) Ambulatory activity in chronic 
disease. In: Tryon, W.H., ed. Behavioral assessment in behavioral 
medicine. New York: Springer Publishing Co., pp. 160-162.
Griffin, J.W. (1990) Basic pathologic processes in the nervous 
system. Toxicol. Pathol. 18:83-88.
Hayes, W.J. (1982) Pesticides studied in man. Baltimore: Williams 
and Wilkins.
Hughes, J.A.; Sparber, S.B. (1978) d-Amphetamine unmasks postnatal 
consequences of exposure to methylmercury in utero: methods for 
studying behavioral teratogenesis. Pharmacol. Biochem. Behav. 8:365-
375.
Jacobson, S.W.; Fein, G.G.; Jacobson, J.L.; Schwartz, P.M.; Dowler, 
J.K. (1985) The effect of intrauterine PCB exposure on visual 
recognition memory. Child Dev. 56:853-860.
Jarabek, A.M.; Menache, M.G.; Overton, J.H.; Dourson, M.L.; Miller, 
F.J. (1990) The U.S. Environmental Protection Agency's inhalation 
RfD methodology: risk assessment for air toxics. Toxicol. Ind. 
Health 6:279-301.
Johnson, M.K. (1990) Organophosphates and delayed neuropathy--Is NTE 
alive and well? Toxicol. Appl. Pharmacol. 102:385-399. 

[[Page 52055]]

Kimmel, C.A.; Rees, D.C.; Francis, E.Z., eds. (1990) Qualitative and 
quantitative comparability of human and animal developmental 
neurotoxicity. Neurotoxicol. Teratol. 12:175-292.
Kimmel, C.A.; Kavlock, R.J.; Francis, E.Z. (1992) Animal models for 
assessing developmental toxicity. In: Guzelian, P.S.; Henry, C.J.; 
Olin, S.S. eds. Similarities and differences between children and 
adults: implications for risk assessment. Washington, DC: ILSI 
Press, pp. 43-65.
Krinke, G.J. (1989) Neuropathologic screening in rodent and other 
species. J. Am. Coll. Toxicol. 8:141-155.
Last, J.M. (1986) Epidemiology and health information. In: Last, 
J.M., ed. Maxcy-Rosenau public health and prevention medicine. New 
York: Appleton-Century-Crofts.
MacPhail, R.C. (1985) Effects of pesticides on schedule-controlled 
behavior. In: Seiden, L.S.; Balster, R.L., eds. Behavioral 
pharmacology: the current status. New York: A.R. Liss, pp. 519-535.
MacPhail, R.C.; Peele, D.B.; Crofton, K.M. (1989) Motor activity and 
screening for neurotoxicity. J. Am. Coll. Toxicol. 8:117-125.
Mailman, R.B. (1987) Mechanisms of CNS injury in behavioral 
dysfunction. Neurotoxicol. Teratol. 9:417-426.
Mattsson, J.L.; Albee, R.R. (1992) Sensory evoked potentials in 
neurotoxicology. Neurotoxicol. Teratol. 10:435-443.
Mattsson, J.L.; Boyes, W.K.; Ross, J.F. (1992) Incorporating evoked 
potentials into neurotoxicity test schemes. In: Tilson, H.A.; 
Mitchell, C.L., eds. Target organ toxicology series: 
neurotoxicology. New York: Raven Press Ltd., pp. 125-145.
Mausner, J.S.; Kramer, S. (1985) Epidemiology--an introductory text, 
2nd ed. Philadelphia: W.B. Saunders.
Morell, P.; Mailman, R.B. (1987) Selective and nonselective effects 
of organometals on brain neurochemistry. In: Tilson, H.A.; Sparber, 
S.B., eds. Neurotoxicants and neurobiological function: effects of 
organoheavy metals. New York: Wiley, pp. 201-230.
Moser, V.C. (1989) Screening approaches to neurotoxicity: a 
functional observational battery. J. Am. Coll. Toxicol. 8:85-93.
Moser, V.C. (1994) Comparisons of the acute effects of 
cholinesterase inhibitors using a neurobehavioral screening battery 
in rats. The Toxicologist 14:241.
National Research Council (NRC). (1983) Risk assessment in the 
federal government. Managing the process. Washington, DC: National 
Academy of Sciences.
National Research Council (NRC). (1984) Toxicity testing: strategies 
to determine needs and priorities. Washington, DC: National Academy 
of Sciences.
National Research Council (NRC). (1994) Science and judgment in risk 
assessment. Washington, DC: National Academy of Sciences.
National Resources Defense Council (NRDC). (1989) Intolerable risk: 
pesticides in our children's food. New York: Natural Resources 
Defense Council.
Needleman, H. (1986) Epidemiological studies. In: Annau, Z.A., ed. 
Neurobehavioral toxicology. Baltimore: Johns Hopkins University 
Press, pp. 279-287.
Needleman, H.L. (1990) Lessons from the history of childhood 
plumbism for pediatric neurotoxicology. In: Johnson, B.L.; Anger, 
W.K.; Durao, A.; Xintaris, C., eds. Advances in neurobehavioral 
toxicology: application in environmental and occupational health. 
Chelsea, MI: Lewis Publishers, Inc., pp. 331-337.
O'Callaghan, J.P. (1988) Neurotypic and gliotypic proteins as 
biochemical markers of neurotoxicity. Neurotoxicol. Teratol. 10:445-
452.
O'Donoghue, J.L. (1989) Screening for neurotoxicity using a 
neurologically based examination and neuropathology. J. Am. Coll. 
Toxicol. 8:97-115.
Office of Technology Assessment (OTA). (1990) Neurotoxicity: 
identifying and controlling poisons of the nervous system. U.S. 
Congress. Office of Technology Assessment (OTA-BA-436). Washington, 
DC: U.S. Government Printing Office.
Omerand, I.E.; Harding, A.E.; Miller, D.H.; Johnson, G.; MacManus, 
D., duBoulay, E.P.G.H.; Kendall, B.E.; Moseley, I.F.; McDonald, W.I. 
(1994) Magnetic resonance imaging in degenerative atoxic disorders. 
J. Neurol. Neurosurg. Psychiatry. 57:51-57.
Rebert, C.S. (1983) Multisensory evoked potentials in experimental 
and applied neurotoxicology. Neurobehav. Toxicol. Teratol. 5:659-
671.
Reiter, L.W. (1987) Neurotoxicology in regulation and risk 
assessment. Dev. Pharmacol. Ther. 10:354-368.
Rees, D.C.; Hattis, D. (1994) Developing quantitative strategies for 
animal to human extrapolation. In: Hayes, A.W., ed. Principles and 
methods of toxicology, 3rd ed. New York: Raven Press, pp. 275-315.
Reuhl, K.R. (1991) Delayed expression of neurotoxicity: the problem 
of silent damage. Neurotoxicol. 12:341-346.
Rice, D.C. (1988) Quantification of operant behavior. Toxicol. Lett. 
43:361-379.
Riley, E.P.; Vorhees, C.V., eds. (1986) Handbook of behavioral 
teratology. New York: Plenum Press.
Rodier, P.M. (1978) Behavioral teratology. In: Wilson, J.G.; Fraser, 
F.C., eds. Handbook of teratology, vol. 4. New York: Plenum Press, 
pp. 397-428.
Rodier, P. (1990) Developmental neurotoxicology. Toxicol. Pathol. 
18:89-95.
Salsburg, D.S. (1986) Statistics for toxicologists. New York: Marcel 
Dekker, Inc.
Sette, W.F. (1987) Complexity of neurotoxicological assessment. 
Neurotoxicol. Teratol. 9:411-416.
Sette, W.F.; MacPhail, R.C. (1992) Qualitative and quantitative 
issues in assessment of neurotoxic effects. In: Tilson, H.; 
Mitchell, C., eds. Target organ toxicity series: neurotoxicology, 
2nd ed. New York: Raven Press, pp. 345-361.
Siegel, S. (1956) Nonparametric statistics for the behavioral 
sciences. New York: McGraw-Hill.
Silbergeld, E.K.; Percival, R.V. (1987) The organometals: impact of 
accidental exposure and experimental data on regulatory policies. 
In: Tilson, H.A.; Sparber, S.B., eds. Neurotoxicants and 
neurobiological function: effects of organoheavy metals. New York: 
J. Wiley, pp. 328-352.
Sokal, R.R.; Rohlf, F.J. (1969) Biometry. San Francisco: W.H. 
Freeman and Company.
Spencer, S.; Schaumburg, H.H. (1980) Experimental and clinical 
neurotoxicology. Baltimore: Williams and Wilkins.
Stanton, M.E.; Spear, L.P. (1990) Workshop on the qualitative and 
quantitative comparability of human and animal developmental 
neurotoxicity. Workgroup I report: Comparability of measures of 
developmental neurotoxicity in humans and laboratory animals. 
Neurotoxicol. Teratol. 12:261-267.
Sterman, A.B.; Schaumberg, H.H. (1980) The neurological examination. 
In: Spencer, P.S.; Schaumberg, H.H., eds. Experimental and clinical 
neurotoxicology. Baltimore: Williams and Wilkins, pp. 675-680.
Susser, M. (1986) Rules of inference in epidemiology. Regul. 
Toxicol. Pharmacol. 6:116-128.
Tilson, H.A. (1987) Behavioral indices of neurotoxicity: what can be 
measured? Neurotoxicol. Teratol. 9:427-443.
Tilson, H.A. (1990) Neurotoxicology in the 1990s. Neurotoxicol. 
Teratol. 12:293-300.
Tilson, H.A.; Moser, V.C. (1992) Comparison of screening approaches. 
Neurotoxicology 13:1-14.
U.S. Environmental Protection Agency. (1986) Triethylene glycol 
monomethyl, monoethyl, and monobutyl ethers; proposed test rule. 
Federal Register 51:17883-17894.
U.S. Environmental Protection Agency. (1987, May 20) Toxic 
Substances Control Act testing guidelines. 50 FR 39397, September 
27, 1985 as amended. 40 CFR 798.6050. Federal Register 52:19082.
U.S. Environmental Protection Agency. (1988a) Diethylene glycol 
butyl ether and diethylene glycol butyl ether acetate; final test 
rule. Federal Register 53:5932-5953.
U.S. Environmental Protection Agency. (1988b) Proposed guidelines 
for assessing male reproductive risk. Federal Register 53:24850-
24869.
U.S. Environmental Protection Agency. (1989) FIFRA accelerated 
reregistration phase 3 technical guidance, Appendix D. Office of 
Prevention, Pesticides and Toxic Substances, Washington, DC. EPA No. 
540/09-90-078. Available from: NTIS, Springfield, VA. PB-90-161530. 

[[Page 52056]]

U.S. Environmental Protection Agency. (1991a) Pesticide assessment 
guidelines, subdivision F. Hazard evaluation: human and domestic 
animals. Addendum 10: Neurotoxicity, series 81, 82, and 83. Office 
of Prevention, Pesticides and Toxic Substances, Washington, DC. EPA 
540/09-91-123. Available from: NTIS, Springfield, VA. PB91-154617.
U.S. Environmental Protection Agency. (1991b) Guidelines for 
developmental toxicity risk assessment. Federal Register 56:63798-
63826.
U.S. Environmental Protection Agency. (1992) Guidelines for exposure 
assessment. Federal Register 57:22888-22938.
U.S. Environmental Protection Agency. (1994) Final report: 
principles of neurotoxicology risk assessment. Federal Register 
59:42360-42404.
U.S. Environmental Protection Agency. (1995a) The Use of the 
Benchmark Dose Approach in Health Risk Assessment. Office of 
Research and Development, Washington, DC. EPA/630/R-94/007.
U.S. Environmental Protection Agency. (1995b) Policy for Risk 
Characterization. Office of the Administrator, Washington, DC.
U.S. Environmental Protection Agency. (1995c) Guidance for Risk 
Characterization. Science Policy Council, Washington, DC.
Valciukas, J.A. (1991) Foundations of environmental and occupational 
neurotoxicology. New York: Van Nostrand Reinhold.
Vorhees, C.V. (1987) Reliability, sensitivity and validity of 
indices of neurotoxicity. Neurotoxicol. Teratol. 9:445-464.
Winer, B.J. (1971) Statistical principles in experimental design. 
New York: McGraw-Hill.
Winneki, G. (1995) Endpoints of developmental neurotoxicity in 
environmentally exposed children. Toxicol. Lett. 77:127-136.
World Health Organization. (1986) Principles and methods for the 
assessment of neurotoxicity associated with exposure to chemicals. 
In: Environmental Health Criteria Document 60. Geneva: World Health 
Organization.

[FR Doc. 95-24652 Filed 10-3-95; 8:45 am]
BILLING CODE 6560-50-P