FUNDAMENTAL AND APPLIED TOXICOLOGY 40, 175-184 (1997)ARllClE NO. F A972388

WORKSHOP OVERVIEW

William K. Boyes,.,3 Michael L. Dourson, t Jacqueline Patterson, t Hugh A. Tilson,. William F. Sette,:!:Robert C. MacPhail,. Abby A. Li,§ and John L. a'Donoghue'

*Neurotoxicology Division, National Health and Environmental Effects Re,\'earch Laboratory, U.S. Environmental Protection Agency,

Research Triangle Park, North Carolina; tToxicology Excellence for Risk Assessment, Cincinnati, Ohio; :l:USEPA, Washington, DC;§Monsanto Company, St. Louis, Missouri; and 'Eastman Kodak Company, Rochester, New York

Received October 2, 1997; accepted October 2, 1997

tests, measurements, and protocols used. Judgment of the adver-sity of an effect depends heavily on the amount and types of dataavailable. The attribution of a chemically induced effect to anaction on the nervous system depends on several factors such asthe quality of the study, the nature of the outcome, dose- responseand time-response relationships, and the possible involvement ofnonneural factors. The guidelines will also serve as a referencefor those conducting neurotoxicity testing, as well as establish aconsistent approach to neurotoxicity risk assessment by regulators.Extending this approach through international harmonizationwould be advantageous to the development of products for aworldwide market. Thus, both risk assessors and regulated indus-tries have a large stake in the guidelines to provide a frameworkthat wilt lead to accurate risk assessment decisions. ~ 1m Socidy of

Toxirology.

EP A's Neurotoxicity Risk Assessment Guidelines, Boyes, W. K.,Dourson, M. L., Patterson, J., Tilson, H. A., Sette, W. F., Mac-Phail, R. C., Li, A. A., and O'Donoghue, J. L. (1m), Fundam.Appl. Toxicol. 40, 175-184.

The proposed Neurotoxicity Risk Assessment Guidelines (U.S.EPA, 1995c Fed. Reg. 60(192), 52032-52056) of the U.S. Environ-mental Protection Agency (EPA) were the subject of a workshopat the 1m Meeting of the Society of Toxicology. The workshopconsidered the role of guidelines in the risk assessment process,the primary features, scientific basis, and implications of the guide-lines for EPA program offices, as well as for industrial neurotoxi-cologists from the perspectives of both pesticides and toxic sub-stances regulation. The U.S. National Academy of Sciences (NAS,1983, Risk Assessment in the Federal Government: Managing theProcess) established a framework for distinguishing risk manage-ment from risk assessment, the latter being the result of integratinghazard identification, hazard characterization, and exposure as-sessment data. The guidelines are intended to establish operatingprinciples that will be used when examining data in a risk assess-ment context. The proposed neurotoxicity risk assessment guide-lines provide a conceptual framework for deciding whether or nota chemically induced effect can be considered to be evidence ofneurotoxicity. Topics in the proposed guidelines include structuraland functional effects, dose-response and -duration considera-tions, and relationships between effects. Among the issues thatmust be considered are the multiplicity of chemical effects, thelevels of biological organization in the nervous system, and the

A workshop entitled EPA's Neurotoxicity Risk Assess-ment Guidelines was held at the 36th Annual Meeting ofthe Society of Toxicology (SOT) in Cincinnati, Ohio, in1997, which was jointly sponsored by the Neurotoxicologyand Risk Assessment Speciality Sections of the SOT. TheU.S. Environmental Protection Agency (EPA) publishedProposed Guidelines for Neurotoxicity Risk Assessment forpublic comment (EPA, 1995c). When final, these guidelineswill provide the scientific basis that the EP A will use tomake regulatory decisions based on neurotoxicity data. Be-cause the manner in which the EP A interprets data on neuro-toxicity could affect many members of the SOT, the work-shop was held to discuss the role, composition, and implica-tions of the proposed guidelines.

The overall goals of the workshop were: (1) to providean overview of the role risk assessment guidelines play inthe risk assessment process; (2) to present the major featuresand scientific basis of the proposed guidelines; and (3) todiscuss the unique perspectives of some of the major partiespotentially influenced by the guidelines, including EP A pro-gram offices and regulated industries. Topics included risk

1 Workshop held at the 36th Annual Meeting of the Society of Toxicology

(SOT), Cincinnati, OH, March 9-13,1997. Sponsored by the Neurotoxicol-ogy and Risk Assessment Speciality Sections of the SOT.

2 This manuscript has been reviewed by the National Health and Environ-

mental Effects Research Laboratory, USEPA, and approved for publication.Mention of trade names and commercial products does not constitute en-

dorsement or recommendation for use.3 To whom correspondence should be addressed at MD-74B, Neurotoxi-

cology Division, National Health and Environmental Effects Research Lab-oratory, U.S. Environmental Protection Agency, Research Triangle Park,

NC 27711.

17' 0272-0590/97 $25.00Copyright @ 1997 by the Society of Toxicology.All right.. of reproduction in any foml reserved.

176 BOYES ET AL.

assessment principles and guidelines, the scientific basis ofthe EPA's Neurotoxicity Risk Assessment Guidelines, howthe guidelines will be used by the EP A's Program Offices,and the implications of the guidelines for industrial neurotox-icologists from the perspectives of those regulated underboth the Federal Insecticide, Fungicide, and Rodenticide Act(FlFRA) and the Toxic Substances Control Act (TSCA).

RISK ASSESSMENT AND THE ROLE OF RISKASSESSMENT GUillELINES

(Michael L. Dourson and Jacqueline Patterson)

The practice and science of risk assessment in the U.S.was propelled forward by the National Academy of Sci-ences' (NAS) 1983 publication Risk Assessment in the Fed-eral Government: Managing the Process (NRC, 1983). Inthis publication, commonly referred to as "the Red Book,"the NAS outlined a paradigm with four distinct steps to therisk assessment process: hazard identification, dose-re-sponse assessment, exposure assessment, and risk character-ization. Risk management used the results summarized inthe risk characterization step to make decisions incorporatingother factors, including economics, technology, and publicpolicy. The EP A adopted many of the NAS recommenda-tions and institutionalized the risk assessment/risk manage-ment process throughout the Agency.

Before the NAS paradigm, risk assessment activities of thevarious EP A offices followed the procedures and practices ofeach office, sometimes leading to inconsistent conclusions.The distinction between the scientific judgments of risk as-sessment and the policy decisions of risk management wasfrequently not made clear. In addition, without clearly writ-ten documentation for the approaches used, those outsidethe process were not always able to understand the basis forthe results. With the NAS publication, the EPA adopted thedocument's definitions and framework and began in earnesta process to make its risk assessment practices consistentacross the Agency. Procedures for assessing risk for a num-ber of endpoints were published in five risk assessmentguidelines in 1986 in the areas of carcinogenicity, mutage-nicity, exposure assessment, chemical mixtures, and devel-opmental toxicity (EPA, 1987).

This first group of guidelines has been enhanced with thedevelopment of guidelines for reproductive toxicity (EP A,1994a), revised guidelines for developmental toxicity (EPA,1991 b), exposure assessment (EP A, 1992), and carcinoge-nicity (EP A, 1996). The Agency is also working on revisionsto the chemical mixtures guidelines. Current efforts to final-ize the Neurotoxicity Risk Assessment Guidelines (EP A,1995c) are the subject of this workshop. Although Agencyefforts to develop systemic toxicity guidelines in the 1980swere never completed, a number of papers have been pub-

lished on related topics, including reference dose (RfD)(Barnes and Dourson, 1988; Dourson, 1994), reference con-centration (RfC) (EPA, 1994b; Jarabek, 1994), and bench-mark dose (BPA, 1995a; Barnes et ai., 1995). Guidance isalso available on risk characterization (EPA, 1995b).

The objective of the EPA's risk assessment guidelinesis to provide guidance for evaluating risk to humans fromexposure to chemicals. Specifically, each set of guidelinesprovides definitions for terms, rationales for approachesto evaluating animal and human data, and overall consis-tency across the different endpoints in appropriate areas.These documents, and the publications on RiD, RfC, andbenchmark dose, also guide judgments on whether effectsare adaptive, compensatory, or adverse. Interpretation ofeffects or a syndrome of effects as the critical effect isalso discussed. Techniques for dose-response extrapola-tion are described. The EP A has been the world leader indeveloping guidelines and publishing them for both inter-nal and external use.

An additional role for risk assessment guidelines is toserve as a framework in which to identify the need for furtherresearch and its resulting potential for reduction of uncer-tainty. These guidelines are not static; they were designedto be changed as the science evolves. The EPA's revisionsto the developmental toxicity, carcinogenicity, and exposureguidelines demonstrate this clearly.

An additional EP A effort to enhance consistency in riskassessment activities across the Agency was the creation ofan internal peer review process to review the results of itsprogram-specific dose-response assessment deliberations.The resulting consensus opinion of these work groups isthen made available to the public through the Integrated RiskInformation System (IRIS). IRIS has been one of the mostsuccessful risk assessment resources ever developed. Its riskassessment values are used by public and private entitiesaround the world in their efforts to determine the humanhealth impacts from chemical exposures.

Within the guideline's development activities, the EPAhas had the opportunity to define areas of scientific uncer-tainty where further research can assist in reducing that un-certainty. Several hundred publications on a variety of topicsrelated to hazard identification, dose-response assessment,and exposure assessment techniques have been published bythe EP A alone, with many additional publications by othersin the field. The EP A's Office of Research and Developmentrecently restructured its office and research activities to ad-dress directly the risk assessment paradigm.

In summary, the 1983 NAS publication illuminated therole of risk assessment and risk management, clearly distin-guishing the two. This publication led to the developmentof risk assessment guidelines within the EP A which ensuredconsistency across the Agency, and provided others the abil-

177NEUROTOXICITY RISK ASSESSMENT

ity to understand and evaluate the Agency's efforts withina common framework. Development of these guidelines hasallowed risk assessors to utilize better the scientific informa-tion available in particular disciplines, such as neurotoxicol-ogy, and demonstrated how laboratory research can be usedin this important area. Additional research has focused onareas which can reduce uncertainty in assessments and theseimprovements have been used to update existing guidelines.

NEUROTOXICITY RISK ASSESSMENT GUIDELINESAND THEIR SCIENTIFIC BASIS

(Hugh A. Tilson)

The Proposed Neurotoxicity Risk Assessment Guidelines(EPA, 1995c), are intended to facilitate the assessment byAgency personnel of agents that are suspected to cause neu-rotoxicity in accordance with the policies and proceduresestablished in the statues by the EP A. The guidelines weredeveloped under the auspices of the Risk Assessment Forumby a work group composed of scientists from throughout theAgency. Selected drafts of the guidelines were peer-re-viewed internally and by experts from universities, environ-mental groups, industry, and other governmental agencies.A earlier draft underwent peer-review in a workshop heldin June, 1992, and was subsequently reviewed internallyby the Concordance and Oversight Committees of the RiskAssessment Forum. The Committee on the Environment andNatural Resources of the Office and Science TechnologyPolicy and the Science Advisory Board of the EP A alsoreviewed the guidelines in August 1995, and July 1996,

respectively.The guidelines describe several default assumptions to be

used in the risk assessment process as discussed in the Na-tional Research Council report on science and judgement inrisk assessment (NRC, 1994). Several assumptions concern-ing animal-to-human extrapolation and the presence of athreshold for neurotoxic effects are discussed. The guidelinesalso contain a number of working definitions of specificterms related to neurotoxicology, as well as a discussionconcerning crucial concepts related to reversible and irre-versible effects, direct and indirect effects, and reserve ca-pacity of the nervous system. The guidelines define the stepsinvolved in hazard identification to make a qualitative deci-sion concerning whether a chemical has neurotoxic potential.The section on dose-response assessment defines the quanti-tative relationship between dose and effect. The guidelinesalso provide guidance on exposure assessment, which pro-vides an estimate of human exposure levels for particularpopulations from all potential sources. The risk characteriza-tion section of the guidelines combines the hazard identifica-tion, dose-response assessment, and exposure assessmentcomponents to estimate some measure of the risk for neuro-

toxicity. As part of the risk characterization, a summary ofthe strengths and weaknesses of each component of the riskassessment is given along with major assumptions, scientificjudgments and, to the extent possible, qualitative and quanti-tative estimates of the uncertainties related to determining aRiD or RfC.

The draft neurotoxicity risk assessment guidelines re-quested public comment on several special issues, including(1) the issue of compensation and recovery of function inneurotoxicological studies and how to account for compen-sation in neurotoxicology risk assessment; (2) the use ofblood and/or brain acetylcholinesterase activity a... an indica-tion of neurotoxicity for risk assessment; (3) endpoints indic-ative of neurotoxicity that may not be covered by the guide-lines, i.e., endocrine-disruption or neuroendocrine-mediatedneurotoxicity; and (4) the possibility of no threshold forsome neurotoxic agents. The Agency received responsesfrom 25 separate groups or individuals, including chemicalcompanies and/or trade associations (7), environmental ad-vocacy groups (2), an animal rights advocacy group (1),governmental agencies (7), individuals in academia or pri-vate medical practice (4), and individual or corporate consul-tants or lobbyists for not-for-profit institutions (4).

The guidelines stressed that the risk assessor should notethat reversible neurotoxic changes should be of concern be-cause the nervous system, particularly cells in the centralnervous system, have a limited capacity for regeneration andthat the nervous system has the capacity to compensate fordamage up to a certain point. Therefore, reversibility of ef-fects may be indicative of this compensatory response orrepresent an activation of repair capacity, which could de-crease future potential adaptability. Public comment indi-cated general support for the discussion concerning compen-sation and reversible effects in the guidelines. It was indi-cated, however, that the concept of reserve capacity shouldbe included in the guidelines to help risk assessors under-stand the possible implications of reversible neurotoxic ef-fects. Behavioral and neurological functioning can be viewedas an adaptive process operating within some upper andlower limits, i.e., the functional reserve. Exposure to a neuro-toxicant alters the dynamic equilibrium of the organism'sfunctioning, which up to some point is maintained within anormal range by extant compensatory mechanisms. If it isassumed that a finite capability is built into the system, atsome point during exposure the reserve capacity of the sys-tem will be depleted and function will deteriorate (Tilsonand Mitchell, 1983).

Considerable questions have arisen within the EP A andelsewhere as to whether inhibition of cholinesterase activityconstitutes an adverse effect for defining hazard potential andevaluating risk. The neurotoxicity risk assessment guidelinesindicated that there is general agreement that clinical signs

178 BOYm ET AL.

for exploring other quantitative models for risk assessment,but most commentators were concerned that the BMD wasnot ready for use in neurotoxicity risk assessment. Othersindicated that if the use of the BMD were encouraged, thenthere was a need to include caveats for using the BMD inthe manner that was used for the NOAEULOAEL approach.

The Science Advisory Board (SAB) of the EPA reviewedthe neurotoxicity risk assessment guidelines in July 1996.The SAB identified many of the same issues raised duringthe public comment period. Once the SAB issues a finalreport, the guidelines will be revised in accordance with theirrecommendations and published in final form.

AN EPA PROGRAM OFFICE PERSPECTIVE ONNEUROTOXICITY RISK ASSESSMENT

(W. F. Sette and R. C. MacPhail)

In the EPA, the Office of Pesticide Programs (OPP), andits sister Office of Pollution Prevention and Toxic Substances(OPPT), are responsible for implementing the Federallnsec-ticide, Fungicide, and Rodenticide Act (FIFRA), whose fo-cus is the registxation and regulation of pesticides, and theToxic Substances Control Act (TSCA), whose focus is in-dustrial chemicals. Risk assessments under both TSCA andF1FRA focus on determinations of "unreasonable risk ofadverse effects on human health or the environment."

The Health Effects Division of OPP is responsible forreviewing all of the animal and human toxicology data usedto support a pesticide's registration. It is also responsiblefor performing risk assessments for potential health effects.These risk assessments are typically derived from referencedoses (RfDs) for chronic dietary exposures or short-termexposure limits for acute dietary exposures, and short-termdermal and/or inhalation occupational and residential expo-sures. There are currently two committees that meet weeklyto perform quality control assessments of the study reviewson a particular pesticide and to establish the chronic refer-ence doses and/or other exposure limits. Review of any neu-rotoxicity studies and other studies that include data relatedto neurotoxicity is one area considered in these deliberations.

Two critical steps in neurotoxicity risk assessments inOPP, then, are judging adversity and ascribing adversechemical-induced effects to neurotoxicity. These judgmentsmay vary as a function of the amount and types of dataavailable.

Neurotoxicity may be simply defined as any adverse effecton the structure or function of the nervous system (EP A,1991a). In contrast, consider the criteria for identification ofa human neurotoxicant provided by Spencer and Schaum-burg (1985): "I) a consistent pattern of neurological dys-function in humans; 2) comparable dysfunction in animals;and 3) reproducible lesions in animals and humans which

179NEUROTOXICnY RISK ASSESSMENT

There are some differences between the types of data inmaking judgments regarding both adversity and neurotoxic-ity (Sette and MacPhail, 1992). In general, histopathologicaleffects are measures of the physical integrity of the nervoussystem. These types of effects have been historically re-garded as adverse. But such damage may occur in the ab-sence of any demonstrable functional consequences. Neuro-physiological effects and neurochemical effects are also, forthe most part. reflections of a chemical's action on the ner-vous system, but judgement to their adversity is more depen-dent on their plausible relation to functional consequences,e.g., EEG changes and seizures, or cholinesterase inhibitionand blurred vision. Behavioral changes are generally consid-ered adverse (at some degree of change) in that they arealtered responses to the environment, but their relation tothe nervous system is often much less clear, and so theirconstruct validity as evidence of neurotoxicity is generallyless certain. Factors important to judging the neurotoxicityof behavioral effects include functional domains (concurrentchanges in multiple related endpoints) or physiological con-structs, correlative neurophysiological, neurochemical, orneuropathological effects, dose-response and time-depen-dent relations, and concurrent systemic toxicity. However,the relationship between systemic organ toxicity and behav-ior is generally not clear nor as well studied as the relation-ship between behavior and the nervous system.

In summary, the interpretation of data typically gatheredin neurotoxicity studies can be viewed as judgments of theiradversity and neurotoxicity. Adversity is defined in broadfunctional terms and depends, in Part. on psychological per-ceptions, the framing of the questions, and social values.Neuropathological effects are most clearly neurotoxic andgenerally seen as adverse, although the functional conse-quences may be unclear. Neurophysiological and neuro-chemical effects are generally considered neurotoxic, buttheir adversity depends more on their empirical or presump-tive functional consequences. Behavioral effects are gener-ally adverse, but their reflection of neurotoxicity is less clearand depends on a variety of constructs and other measures.

are related to the neurobehavioral dysfunction expressed."The degree to which a chemical's database satisfies thesecriteria may be evaluated in tefllls of several types of valid-ity. The concept of validity has been adapted from the litera-ture on human psychological testing, where it has long beenused to evaluate different tests of intelligence or other abili-ties and aptitudes (Sette, 1987). There are five principal ques-tions raised in the criteria and definition that may be de-scribed in tefllls of four types of validity: the extent to whicheffects can be viewed as consequences of exposure (contentvalidity); the correlation between measures of behavior,physiology, biochemistry, and morphology (concurrent va-lidity); whether effects in animal models are predictive ofwhat will happen in humans (predictive validity); whetherthe effects are adverse or toxicologically significant (con-struct validity); and whether the effects are neurotoxic (con-struct validity). The definition is focused on judgments of asingle effect. It is simple, perhaps deceptively so, and depen-dent on how one defines adverse, a question of constructvalidity. Describing the criteria in tefllls of validity, a consis-tent pattern of neurological dysfunction should provide con-current validity in teflllS of the pattern and reliability ofeffects, and construct validity to the extent that the testsidentify neurological dysfunctions. Comparable effects inanimals involve predictive validity between species and con-tent validity in that the animal models often establish a muchclearer relation between exposures and effects. Lesions re-lated to the dysfunctions involve concurrent validity betweendysfunctions at different levels of biological organization.While the first definition may be satisfied by a change in asingle endpoint, satisfaction of the general criteria for a hu-man neurotoxicant would require considerable data. In OPP,despite all of the animal studies that are often required forpesticides, the availability of sufficient data to satisfy allthese criteria is rare.

Adversity may be defined as "alterations from baselinethat diminish an organism's ability to survive, reproduce, oradapt to the environment" (EP A, 1995c). This third element,adaptation to the environment, places emphasis on functionaldeficits that are largely reflected as changes in behavior, butother types of effects, e.g., neurophysiological habituation,may also reflect adaptations. Some judgment is also requiredregarding the degree of any alteration that may be deemedsignificant. Adversity also has subjective psychological ele-ments in that people's views of the adversity of effects maydepend on their nature, the exposure situation (Slovic, 1987),and how the risk assessment is framed (Tversky and Kahne-man, 1981). There will always be tension between broadsocial views of what is adverse and the more rigorous de-mands of sound scientific conclusions about whether a chem-ical is a human neurotoxicant. (See also Ann. Am. A cad.Political Social Sci., May 1996).

A FIFRA-REGULATED INDUSTRIAL PERSPECTIVE OFTHE EPA'S DRAFT NEUROTOXICITY RISK

ASSESSMENT GUIDELINES(Abby A. Li)

Neurotoxicity risk assessment has become a driving forcefor regulatory actions and clean-up decisions. The TSCAMulti-substance Test Rule of Neurotoxicity (TSCA; 40 CFRPart 799, July 27,1993); the increasing number of data call-ins for neurotoxicity testing under FlFRA, and the recentinclusion of carbamates on the Resource Conservation andRecovery Act (RCRA) hazardous waste lists (60 FR 7824,

180 BOYES ET AL.

EP A states, "Neurotoxicity is an adverse change in thestructure or function of the central and/or peripheraJ nervoussystem following exposure to a chemical, physicaJ, or bio-Jogical agent. . . . Changes in function can also result fromtoxicity to other specific organ systems, and these indirectchanges may be considered adverse but not necessarily neu-rotoxic" (FR 60:52035).

The first principJe is that neurotoxicity is an adverse effect,not just any change. The second principle implies thatchanges in function can also result from toxicity to otherspecific organ systems, and these indirect changes may beconsidered adverse but not necessarily neurotoxic. TheEP A's neurotoxicity risk assessment guidelines can be im-proved if all the sections discussing specific neurotoxic end-points are made consistent with these two principles. Thereare some sections that are consistent, but several that arenot. Let me illustrate with examples using endpoints suchas motor activity and Jearning and memory which are intu-itively more famiJiar to most people.

The section discussing motor activity states that "neuro-toxic agents generaJJy decrease motor activity (FR 60:52045)" but does not baJance this statement by aJso statingthat many agents that are not specific neurotoxicants alsocause decreases in motor activity at higher doses. Withoutthis clarification, the reader couJd be Jeft with the incorrectimpression that any decrease in motor activity is evidenceof neurotoxicity. A simple clarification will make this sectionconsistent with the EP A's definition of neurotoxicity.

This section on motor activity does discuss different levelsof concern. "Agent-induced changes in motor activity asso-ciated with other overt signs of toxicity (e.g., loss of bodyweight, systemic toxicity) or occurring in non-dose-relatedfashion are of less concern than changes that are dose depen-dent, reJated to structuraJ or other functional changes in thenervous system, or occur in the absence of life-threateningtoxicity (FR 60: 52045)." However, guidance on how thesedifferent levels of concern will actually transJate into a mean-ingful difference in the final risk assessment process isneeded. As written now, changes of lesser and greater con-cern are both considered neurotoxic effects and will betreated identically to those of lesser concern in the final riskassessment process.

The example with motor activity illustrates the importanceof adhering to the principle outlined by the EP A's definitionof neurotoxicity that changes in function can be adverse butare not necessarily neurotoxic. The second principle is thatneurotoxicity is an adverse effect, not just any change frombaseJine. An adverse effect was defined by the EP A as achange from baseline that diminishes an organism's abilityto survive, reproduce, or adapt to the environment. We havethe abiJity to mea...ure both enhancements and deficits inmany functions and behaviors. For example we can measure

Feb. 9, 1995) are a few examples of recent regulatory deci-sions that are dependent upon neurotoxicity risk assessment.

The publication of the proposed U.S. EPA's NeurotoxicityRisk Assessment Guideline is a very important step forwardin making this risk assessment more consistent, transparent,and science based. Dr. Hugh Tilson and the many otherauthors at the EP A have done an outstanding job developinga guideline that provides a comprehensive understanding ofmany areas of neurotoxicology as well as practical and usefulguidance for those conducting and assessing studies for regu-latory purposes.

The purpose of the neurotoxicity risk assessment guide-lines is to protect humans from developing neurotoxic dis-eases as a result of exposure to chemicals. Our success inachieving this goal will depend, in part, upon the accuracywith which we define neurotoxic effects. In our eagernessto protect the public and environment, the temptation is tocover all the bases by defining any change that can possiblybe linked to the nervous system as a neurotoxic effect, in-cluding those that we do not fully understand and, hence,are most fearful of missing. While some degree of conserva-tism is necessary to compensate for gaps in scientific under-standing, an overly conservative definition of neurotoxicitycan have the harmful effect of focusing resources on issuesthat have little impact on reducing risk.

Neurotoxicity risk assessment is uniquely challenging be-cause of the large contribution of functional and behavioralmeasures. While inclusion of these measures elevate theimportance of functional behavioral changes, the potentialdanger is that all changes in behavior and function automati-cally will be viewed as neurotoxic effects. This is particularlyproblematic because studies must be conducted at the maxi-mum tolerated doses (Mill), which is a dose level that isoften orders of magnitude above expected exposure levels.At sufficiently high dose levels, all materials will producefunctional effects, many of which are likely to be nonspecificindicators of general intoxication and not neurotoxicity. Therisk assessment process already allows us to protect againstthese effects without having to classify them as neurotoxiceffects. Classifying effects as neurotoxic can lead to regula-tory enforcement actions (such as RCRA waste lists) someof which do not take exposure levels into account. Theseregulatory actions are better directed toward direct-actingneurotoxicants. Also, indiscriminately classifying adverseeffects on function as neurotoxic can compromise effectivehazard communication to the public.

Thus, it is important that risk assessors receive balancedguidance that will allow us to distinguish chemicals actingdirectly on the nervous system from those that produce ef-fects indirectly. The right balance can be achieved by makingthe entire Neurotoxicity Risk Assessment Guideline consis-tent with two very important principles outlined in EP A'sdefinition of neurotoxic effect.

181NEUROTOXICITY RISK ASSESSMENT

effects that was in the draft guideline provides a good guid-ance on how to distinguish between indirect and direct ef-fects. The EPA's definition of a neurotoxic or adverse effectshould encourage scientifically objective interpretation of thedata and should not be corrupted by the ill-defined concept of, 'unwanted" effects. Guidance should be included as to how

different levels of concern will make a difference in the riskassessment process. Finally, sufficient evidence of "reason-able certainty of no harm" should be defined in a manner thatacknowledges the value of a tiered approach to neurotoxicitytesting and of the value of well-conducted standard toxicitystudies.

IMPLICATIONS OF THE EPA'S NEUROTOXICITY RISKASSF$SMENT GUIDELINES FROM A TSCA-REGULATED INDUSTRIAL PERSPECTIVE

(John L O'Donoghue)

enhancements in cognition. Yet the guidelines appear to im-ply that all changes in learning and memory are adverse. Inother words, improvement in learning or memory could beconsidered a neurotoxic effect. Likewise, improved perfor-mance on an operant task which leads to more food rein-forcement and hence an improvement in the animal's abilityto adapt to the environment would also be a neurotoxiceffect. Yet these effects do not fit EPA's definition of anadverse effect. If the concern is that improvement in learningand memory or performance of a complex behavior mayindicate that the chemical is acting centrally to produce otherunknown adverse effects in the central nervous system, thenthis should be stated as a default assumption based on sci-ence policy and not presented as if it were objective scientificfact. It would be misleading to conclude that this chemicalproduces adverse effects on learning and memory.

Some of the previous speakers introduced a new definitionof an adverse effect that was not in the draft EP A Neurotox-icity Risk Assessment Guidelines; namely, an "unwantedeffect." This definition should not be included in theseguidelines. It is inconsistent with the EPA's Office of Pesti-cide Program (OPP) guidance on risk a.~sessments, whichstates that' 'risk assessments should be transparent, in thatthe conclusions drawn from the science are identified sepa-rately from policy judgments, and the use of default valuesor methods and the use of assumptions in the risk assessmentare clearly articulated (EPA, 1995e). Defining an adverseeffect as an "unwanted effect" will encourage interpreta-tions that are no longer scientifically objective or rigorousand hopelessly intertwined with policy decisions.

Adhering to the EP A's balanced definition of neurotoxic-ity is especially important in light of the EP A's discussionon characterizing the sufficiency of evidence for neurotoxiceffects. At present. the guidelines are written in a way thatmakes it very difficult to provide any reasonable certaintyof no harm: According to the guidelines, a series of negativestandard toxicity studies are not sufficient evidence of "noharm" to the nervous system. Negative screening studiesthat may measure general functional endpoints are not suffi-cient evidence of no harm for other specific neurotoxic ef-fects such a.~ cognition. This guidance dismisses a large bodyof potentially valuable data that should enter into the riskassessment evaluation, and it also ignores the value of atiered approach to neurotoxicity testing.

In summary, the draft EP A's Neurotoxicity Risk Assess-ment Guidelines is an important step forward in makingsure that there is a consistent approach to neurotoxicity riskassessment. This will be very beneficial to the registrationand review process of chemicals. Effective protection of thepublic will depend, in part. upon the ability to discriminatebetween indirect and direct adverse effects on the nervoussystem. The EP A's definition of neurotoxicity and adverse

The Toxic Substances Control Act (TSCA) regulates theimport, export, production, use, and disposal of nearly everyproduct and synthetic or processed natural chemical presentwithin the United States. These include household products,consumer products, bulk industrial chemicals, and chemicalintennediates, including site-limited intennediates. Eventhough pesticides, phannaceuticals, food and feed additives,and other regulated products are not considered TSCA-regu-lated materials, their ingredients, processing, or processwaste may be regulated under TSCA. Thus, the EPA's Neu-rotoxicity Risk Assessment Guidelines will have a verybroad impact not only on TSCA-regulated companies, butalso on a wide range of other commercial enterprises andthe American consumer.

The publication of the final Neurotoxicity Risk Assess-ment Guidelines will knit together a series of other guide-lines and processes that have been developed to detect andassess potentially neurotoxic substances. The various partsof the overall neurotoxicity risk assessment process includesgood laboratory practice guidelines, systemic toxicity testguidelines, neurotoxicity test guidelines, and neurotoxicityrisk assessment methods such as the safety factor and thebench-mark dose (BMD) methods for calculating risk levels.The Neurotoxicity Risk Assessment Guidelines deal witheach of these other processes and how they relate to eachother, and provide information on how to handle and inter-pret data from various sources to a risk assessor.

The primary target audience for the Neurotoxicity RiskAssessment Guidelines are EP A risk assessors and EP A con-tractors conducting risk assessments. The individuals usingthe Neurotoxicity Risk Assessment Guidelines in most in-stances will not be neurotoxicologists, but rather individualswith expertise in general principles of toxicology and riskassessment. The absence of a neurotoxicologist in the risk

BOYES ET AL.182

ship. The Neurotoxicity Risk Assessment Guidelines can beexpected to assist TSCA-regulated companies and help theEP A reach goals which go beyond neurotoxicity risk assess-ment and have more to do with proactive pollution preven-tion.

A potential outcome of publication of the NeurotoxicityRisk Assessment Guidelines is that it will set a standard fordevelopment of an international or global standard for doingneurotoxicity risk assessments. Currently, there is interestaround efforts to harmonize chemical classification and haz-ardous materials-labeling practices to replace the several,and sometimes conflicting, systems in place in the globalmarketplace. While the International Programme on Chemi-cal Safety (1986) has produced a well-received environmen-tal health criteria document on neurotoxicity assessment, theNeurotoxicity Risk Assessment Guidelines provides the firstwidely available process for regulatory agency use. Withpublication of the Neurotoxicity Risk Assessment Guide-lines, the EP A will have provided a model document uponwhich an international or global standard can be built.

While the above a.~pects may be seen as improvements inthe overall risk assessment process, there remain some seri-ous concerns with the Neurotoxicity Risk AssessmentGuidelines. Since these concerns have not been fully ad-dressed during the Neurotoxicity Risk Assessment Guide-lines review and comment, they bear attention during imple-mentation of the Neurotoxicity Risk Assessment Guidelinesor on a case-by-case basis during reviews on specific chemi-cals. A primary concern is the use of the Neurotoxicity RiskAssessment Guidelines to identify chemicals as neurotoxic.The concern about this issue has a number of different facetsto it. First. there is concern that the Neurotoxicity TestingGuidelines, which will be used to collect data for assessmentusing the Neurotoxicity Risk Assessment Guidelines, aredesigned such that the tests, especially the functional obser-vational battery (FOB), are likely to provide false-positiveresults. Because the Neurotoxicity Testing Guidelines in-clude many different endpoints, the probability of an end-point showing "an effect" is quite high. Perhaps more im-portant is that the Neurotoxicity Testing Guidelines and othertest guidelines that include !behavioral endpoints are typicallyconducted at toxic dose levels, which can cause nonspecificbehavioral manifestations that can be misinterpreted as neu-rotoxicity. To the EP A's credit, it has conducted researchlooking at such effects and the 1995 Annual Report of theNational Health and Environmental Effects Research Labo-ratory (EP A, 1995d) has concluded that' 'stress has beenshown to affect the manifestations of chemical-induced neu-rotoxicity and can have a significant affect on quantitativeand qualitative estimates of neurotoxic risks." The NHEERLannual report has also reported that' 'the FOB and motoractivity were used in a comparative study. . . to assess the

assessment process is of concern to TSCA-regulated compa-nies. However, since many risk assessments on TSCA-re-lated materials are currently being performed by individualswithout specialist training in neurotoxicology, the availabil-ity of a published, widely reviewed Neurotoxicity Risk As-sessment Guidelines should be seen as a significant improve-ment in the overall process. Publication of the NeurotoxicityRisk Assessment Guidelines could be looked upon as aninternal EP A harmonization process for risk assessmentmethodology. The Neurotoxicity Risk Assessment Guide-lines can be expected to make neurotoxicity risk a.~sessmentsmore consistent across the various EP A program officesdealing with TSCA-related issues.

While the Neurotoxicity Risk Assessment Guidelines areintended to be used by EP A risk assessors, it is abundantlyclear that they set a standard for neurotoxicity risk assess-ment for others as well. In addition to the use envisionedby the EPA, TSCA-related companies are likely to use theguidelines while working on pollution-prevention activities.Such use, however, will require a flexible interpretation ofthe guidelines. A literal interpretation of the NeurotoxicityRisk Assessment Guidelines would lead a risk assessor touse the guidelines as an "end of pipe approach," that is tosay they would be used when complete data sets are availablefor analysis. However, decisions about which chemicals touse in manufacturing processes are made early in the productdevelopment cycle, not at the immediate premanufacturingpoint when regulatory clearances are sought. Many compa-nies use product development processes that begin with aphase where products are first conceptualized. Successfulconcepts move on to a phase where the technology is devel-oped to bring the concept to a manufacturable product. Onlywhen the technology actually exists to create a product doesthe product development process move into high gear. Oncethe product development and process development neededto manufacture a product are in place, a new product canactually begin to have a marketplace presence. Companyand consulting toxicologists contribute to the product devel-opment cycle at earlier time points than ever before by col-lecting and interpreting toxicity data to meet the high expec-tations of the marketplace for ever-improving products. TheNeurotoxicity Risk Assessment Guidelines can playa rolein company product stewardship and pollution-preventionactivities by providing a guide to explain how the EPA ex-pects neurotoxicity data to be interpreted for TSCA pur-poses. Early assessment of neurotoxicity data against Neuro-toxicity Risk Assessment Guidelines standards means thatthe product development process can be adjusted to eliminateor improve the control of potentially neurotoxic substancesduring the product development process rather than at theend of the process when significant additional resourcesmight be required to ensure a high level of product steward-

NEUROTOXICITY RISK ASSESSMENT 183

intended to provide a conceptual framework for the interpre-tation of neurotoxicity data for the purpose of making riskassessment decisions.

The development and refinement of the proposed Neuro-toxicity Risk Assessment Guidelines, like other risk assess-ment documents, is a continuing and evolving process. Previ-ous drafts of these guidelines have undergone review by anumber of groups both within and outside of EP A, and theguidelines have been revised accordingly in an attempt toaddress issues raised. When published in final form, thedocument is intended to provide for more consistent andtransparent risk assessment decisions within and across themany offices of the EP A which must evaluate data on poten-tial neurotoxicity. In addition to its intended use, the docu-ment may be used outside of the EPA by companies duringproduct development and for product stewardship to antici-pate how the EPA will interpret product safety data submit-ted to the Agency, and may serve as a basis for internationalefforts to harmonize risk assessment as well.

Concerns of EP A program offices are that the documenthelps in determining whether reported outcomes should beconsidered as adverse, and whether those outcomes can beattributed to neurotoxicity. Concerns of regulated industriesinclude whether the document helps enough to clarify theinterpretation of nonspecific outcomes, in particular of func-tional changes occurring at high-dose levels. In addition,another concern was that an overly conservative tendencyin risk assessment prevents there from being a conclusionof sufficient evidence of no harm. These concerns shouldbe considered in the revision and/or implementation of theNeurotoxicity Risk Assessment Guidelines.

All the participants in the workshop felt that, on thewhole, the proposed Risk Assessment Guidelines wouldhelp provide more consistent, uniform, and appropriatedecisions regarding potential neurotoxicity of compoundsregulated by the EP A.

acute and subacute effects of 10 industrial and agriculturalchemicals.. . . Chemicals expected to have little or no neu-rotoxicity affected some behavioral measures at high doses."While sections of the Neurotoxicity Risk Assessment Guide-lines partially address the concern about nonspecific effectsbeing used to identify materials as neurotoxic, concern re-mains that the Neurotoxicity Risk Assessment Guidelineswill result in the incorrect identification of many chemicalsas neurotoxic.

While there is obvious concern that materials may beeasily and inappropriately classified as neurotoxic, there isalso concern that risk assessors using the Neurotoxicity RiskAssessment Guidelines may be reluctant or unable to identifymaterials which should be of no concern for neurotoxicity.That is, that it may be impossible to provide enough datato demonstrate that a material does not present a risk ofneurotoxicity under typical conditions of use. The Neurotox-icity Risk Assessment Guidelines provide guidance to therisk assessor on how to characterize the health-related data-base available for neurotoxicity risk assessment. This guid-ance is provided to identify chemicals with data which pro-vide sufficient or insufficient evidence for classification asa neurotoxic ant. However, well-studied chemicals that showno signs of neurotoxicity are left in regulatory limbo, notregarded as neurotoxic but never really identified as nonneu-rotoxic either. While it is understandable that the Agencyhas concerns about' 'the inherent difficulty in 'proving anynegative,' " it is not understandable why it has not created

a classification category identified as "chemicals of low con-cern for neurotoxicity." If the Agency agrees that its testguidelines are capable of identifying neurotoxic substances,it would seem to be a small matter to consider chemicalstested by the Neurotoxicity Testing Guidelines or equivalentprocesses to be of low or little concern for neurotoxicity.

While these concerns may be considered a criticism ofthe Neurotoxicity Risk Assessment Guidelines, they existedprior to their development. With the publication of the Neu-rotoxicity Risk Assessment Guidelines, it is possible thatsome of these concerns will lessen because the guidelinesprovide a common set of principles for the EPA risk asses-sors to use. Up until now, risk a.~sessors were without writtenguidance on the methods to be used for assessing chemicalsfor risk of neurotoxicity.

REFERENCES

CONCLUSIONS

Risk assessment, as conceived by the NAS, involves aseries of discrete steps that are intended to enhance the sepa-ration of the collection and interpretation of scientific infor-mation from the development of public policy decisionsbased on that information. The proposed Neurotoxicity RiskAssessment Guidelines fit into that process in that they are

Barnes, D. G., Daston, G. P., Evans J. S., Jarabek, A. M., Kavlock, R. J.,Kimmel, C. A., Park, C., and Spitzer, H. (1995). Benchmark dose work-shop: Criteria for use of a benchmark dose to estimate a reference dose.Regul. Toxicol. Pharmacal. 21, 296-306.

Barnes, D. G., and Dourson, M. L. (1988). Reference dose (RfD): Descrip-tion and u.~e in health risk assessments. Reg. Toxicol. Pharmacol. 8, 471-486.

Dourson, M. L. (1994). Methods for establishing oral reference doses(RfDs). In Risk Assessment of Essential Elements. (W. Mertz, C. O.Abernathy, and S. S. Olin, Eds.), pp. 51-61. ILSI Press, Washington,DC.

International Programme on Chemical Safety (1986). Environmental HealthCriteria 60: Principles and Methods for the Assessment of NeurotoxicityAssociated with Exposure to Chemicals. World Health Organization, Ge-neva.

184 BOYES ET AL

Jarabek, A. M. (1994). Inhalation RfC methodology: Dosimetric adjust-ments and dose-response estimation of noncancer toxicity in the upper

respircltory tract. Inhalation Toxicol. 6(Suppl.):301-325.

National Academy of Sciences, National Research Council (NRC). (1975).Principles for Evaluating Chemical,\" in the Environment. Nat!. Acad.

Press, Washington, DC.

National Academy of Sciences, National Research Council (NRC). (1983).Risk Assessment in the Federal Government: Managing the Process. Nat!.

Acad, Press, Washington, DC,

National Academy of Sciences, National Research Council (NRC). Scienceand judgement in risk assessment. Nat!. Acad. Pres..., Washington, DC.

Sette, W. F. (1987). Complexity of neurotoxicological assessment. Neuro-

toxic(}l. Teratol. 9,411-416.

Sette, W. F., and MacPhail, R. C, (1992). Qualitative and quantitative issuesin assessment of neurotoxic effects. Tn Neuroloxicology (H. Tilson and

C. Mitchell, Eds.), pp. 345-361. Raven Press, New York.

Slovic, P. (1987). Perception of risk, Science, 236, 280-285.

Spencer, P. S., and Schaumburg, H. H. (1985). Scand. J. Work Environ.

Heallh 11,53-60.

Tilson, H. A., and Mitchell, C. L. (1983). Neurotoxicants and adaptive re-

sponses of the nervous system. Fed. PrCK'. 42,3189-3190.

Tversky, A., and Kahneman, D. (1981). The framing of decisions and the

psychology of choice. Science 211, 453-458.

U.S. Environmental Protection Agency (EPA) (1987). The Ri,\"k As,~essmenl

Guidelines of 1986. Office of Health and Environmental Assessment,Washington, DC. [EPA/600/8-87/045, August]

U.S. Environmental Protection Agency (EPA) (1991a). OPPTS Tesl Guide-lines: Neurotoxicity (PB91-154617). National Technical Information Ser-

vice, Springfield, VA.

u.s. Environmental Protection Agency (EPA) (199lb). Guidelines for de-velopmental toxicity risk assessment. Fed. Regist. 56(234), 63798-

63826. [December 5]

U.S. Environmental Protection Agency (EPA) (1992). Guidelines for exp>-

sure assessment. Fed. Regist. 57,22888-22938. [May 29]

U.S. Environmental Protection Agency (EPA) (1994a). Guidelines for Re-productive Toxicity Risk Assessment, review draft. Office of Research

and Development, Washington, DC. [EPA/600/AP-94/OOI, February]

U.S. Environmental Protection Agency (EPA) (1994b). Methodsfor Deriva-tion of Inhalation Reference Concentrations and Application of Inhala-tion Dosimetry. Office of Health and Environmental Asse.~sment, Wash-

ington, DC. [EPA/600/8-90-066F, October]

U.S. Environmental Protection Agency (EPA) (1995a). The Use of theBenchmark Dose Approach in Health Risk Assessment. Risk Assessment

Forum. [EP A/630/R -94/007, February]

U.S. Environmental Protection Agency (EPA) (1995b). Policy and Guid-ance for Risk Characterization. [Memo from Caro1 M. Browner, Adminis-trator to Assistant Administrators, March 21]

U.S. Environmental Protection Agency (EPA) (1995c). Proposed guidelinesfor neurotoxicity risk assessment. Fed. Regist. 60, 52032-52056.

U.S. Environmental Protection Agency (EPA) (1995d). 1995 Annual Reportof the National Health and Environmental Effect.~ Re.~earch Laboratory.

Washington, DC. [EPA/600/R-96/100]

U.S. Environmental Protection Agency (EPA) (1995e). Draft Guidance on

Risk CharacterizatU}n.

U.S. Environmental Protection Agency (EPA) (1996). Proposed Guidelinesfor Carcinogen Risk Assessment. Office of Research and Development,

Washington, DC. [EPA/6OO/P-92/003C, April]