Diseases of environmental origin result from exposures to synthetic and naturally occurring chemical toxicants encountered in the environment, ingested with foods, or administered as pharmaceutical agents. They are, by definition, preventable: they can be prevented by eliminating or reducing exposures to toxicants. The fundamental purpose of testing chemical substances for neurotoxicity is to prevent disease by identifying toxic hazards before humans are exposed. That approach to disease prevention is termed "primary prevention." In contrast, "secondary prevention" consists of the early detection of disease or dysfunction in exposed persons and populations followed by prevention of additional exposure. (Secondary prevention of neurotoxic effects in humans is discussed in Chapter 5.)
In the most effective approach to primary prevention of neurotoxic disease of environmental origin, a potential hazard is identified through premarket testing of new chemicals before they are released into commerce and the environment. Identifying potential neurotoxicity caused by the use of illicit substances of abuse or by the consumption of foods that contain naturally occurring toxins is less likely. Disease is prevented by restricting or banning the use of chemicals found to be neurotoxic or by instituting engineering controls and imposing the use of protective devices at points of environmental release.
Each year, 1,200–1,500 new substances are considered for premarket review by the Environmental Protection Agency (EPA) (Reiter, 1980), and several hundred compounds are added to the 70,000 distinct chemicals and the more than 4 million mixtures, formulations, and blends already in commerce. The proportion of the new chemicals that could be neurotoxic if exposure were sufficient is not known (NRC, 1984) and cannot be estimated on the basis of existing information (see Chapter 1). However, of the 588 chemicals used in substantial quantities by American industry in 1982 and judged to be of toxicologic importance by the American Conference of Governmental Industrial Hygienists (ACGIH), 28% were recognized to have adverse effects on the nervous system; information on the effects was part of the basis of the exposure limits recommended by ACGIH (Anger, 1984).
Given the absence of data on neurotoxicity of most chemicals, particularly industrial chemicals, it is clear that comprehensive primary prevention would require an extensive program of toxicologic evaluation. EPA has regulatory mechanisms to screen chemicals coming into commerce, but overexposures continue to occur, and episodes of neurotoxic illness have been induced by chemicals, such as Lucel-7 (Horan et al., 1985), that have slipped through the regulatory net. Serious side effects of pharmaceutical agents also continue to surface, such as the 3 million cases of tardive dyskinesia that developed in patients on chronic regimens of antipsychotic drugs (Sterman and Schaumburg, 1980). It is now possible to identify only a small fraction of neurotoxicants solely on the basis of chemical structure through analysis of structure-activity relationships (SARs), so in vivo and in vitro tests will be needed for premarket evaluation until greater understanding of SARs permits them to be used with confidence.
If neurotoxic disease is to be prevented, public policy must be formulated as though all chemicals are potential neurotoxicants; a chemical cannot be regarded as free of neurotoxicity merely because data on its toxicity are lacking. Prudence dictates that all chemical substances, both old and new, be subjected to at least basic screening for neurotoxicity in the light of expected use and exposure. However, the sheer number of untested chemicals constitutes a practical problem of daunting magnitude for neurotoxicology. Given the number of untested chemicals and current limitations on resources, they cannot all be tested for neurotoxicity in the near future. Testing procedures designed for neurotoxicologic evaluation that have been developed so far might be reasonably effective, but are so resource-intensive that they could not be applied to all untested chemicals.
A rational approach to neurotoxicity testing must contain the following elements:
- Sensitive, replicable, and cost-effective neurotoxicity tests with explicit guidelines for evaluating and interpreting their results.
- A logical and efficient combination of tests for screening and confirmation.
- Procedures for validating a neurotoxicologic screen and for guiding appropriate confirmatory tests.
- A system for setting priorities for testing.
This chapter discusses systematic assessment of chemicals for neurotoxic hazards. It begins by describing biologic and public-health issues that are peculiar to neurotoxicology. It then presents a review of current techniques for neurotoxicologic assessment to address the question: "How effective are current test procedures for identify neurotoxic hazards?" Concomitantly, the review seeks to identify gaps and unmet needs in current neurotoxicity test procedures. Finally, the chapter describes how the regulatory agencies currently evaluate potential neurotoxicity and how the tests reviewed could be combined in an improved neurotoxicity testing strategy.
A byproduct of the testing strategy outlined in this chapter will be the development of an array of biologic markers of neurotoxicity. These markers can be used in future studies of experimental animals, as well as in clinical and epidemiologic studies of humans exposed to neurotoxicants, as was proposed in Chapter 3.
Approach to Neurotoxicity Testing
Difficulties in Neurotoxicity Testing
Neurotoxicity testing is relatively new. Although its rapid development is noteworthy, its progress has been constrained by several factors that complicate neurotoxicologic assessment. Some of the complexities, such as sex- or age-related variability in response, are common to all branches of toxicology. Neurotoxicology, however, faces unique difficulties, because of several characteristics that make the nervous system particularly vulnerable to chemically induced damage (Chapter 2). Those characteristics include the limited ability of the nervous system to repair damage, because of the absence of neurogenesis in adults; the precarious dependence of axons and synaptic boutons on long-distance intracellular transport; the system's distinct metabolic requirements; the system's highly specialized cellular subsystems; the use of large numbers of chemical messengers for interneuronal communication; and the complexity of the nervous system's structural and functional integration. The nervous system exhibits a greater degree of cellular, structural, and chemical heterogeneity than other organ systems. Toxic chemicals potentially can affect any functional or structural component of the nervous system—they can affect sensory and motor functions, disrupt memory processes, and cause behavioral and neurologic abnormalities. The large number of unique functional subsystems suggests that a great diversity in test methods is needed to ensure assessment of the broad range of functions susceptible to toxic impairment. The special vulnerability of the nervous system during its long period of development is also a critical issue for neurotoxicology.
Despite the inherent difficulties of neurotoxicity testing, some validated tests have been developed and implemented. Testing strategies must take those facets of the nervous system into account, and they must consider a number of variables known to modify responses to neurotoxic agents, such as the developmental stage at which exposure occurs and the age at which the response is evaluated. The issue of timing is complex. During brain development, limited damage to cell function—even reversible inhibition of transmitter synthesis, for instance—can have serious, long-lasting effects, because of the trophic functions of neurotransmitters during neuronal development and synaptogenesis (Rodier, 1986). Kellogg (1985) and others have shown the striking differences between the effects of perinatal exposure of animals to some neuroactive drugs (e.g., diazepam) and the effects of exposure of adult animals to the same drugs. Other agents (e.g., methylmercury and lead) are toxic at every age, but are toxic at lower doses in developing organisms. In addition, in developing animals, the blood-brain barrier might not be sufficiently developed to exclude toxicants. Stresses later in development might lead to the expression of relatively sensitive effects that were latent or unchallenged at earlier stages of development. During senescence, the CNS undergoes further change, including a loss of nerve cells in some regions. CNS function in senescence could be vulnerable to cytotoxic agents that, if encountered earlier in development, might have been protected against by redundant networks or compensated for by "rewiring" of networks (Edelman, 1987). To address those complex issues, testing paradigms that incorporate both exposures and observations during development and during aging need to be considered. The disorders might be acute and reversible or might lead to progressive disorders over the course of chronic exposure. More sensitive biologic markers as early indicators of neurotoxicity are urgently needed. Opportunities should be exploited for detecting neurotoxicity when chronic lifetime bioassays are conducted for general toxicity or carcinogenicity.
In the design of neurotoxicity screens, no test can be used to examine all aspects of the nervous system. The occurrence of an effect of a chemical on one function of the nervous system will not necessarily predict an effect on another function. Therefore, it is important to use a variety of initial tests that measure different chemical, structural, and functional changes to maximize the probability of detecting neurotoxicity or to use tests that sample many functions in an integrated fashion.
The Testing Strategy
Efficient identification of potential hazards warrants a tiered testing strategy. The first tier of testing (the screen) need not necessarily be specifically predictive of the neurotoxicity likely to occur in humans, unless regulatory agencies are to use the results for direct risk-management decisions. The tests in later tiers are essential to assess specificity and confirm screening results and are appropriate for defining dose-response relationships and mechanism of action (Tilson, 1990a). Screening tests would be followed, as appropriate, by more specific assessments of particular functions. Such an approach permits a decision to be made about whether to continue testing at each step of the progression. In the case of chemicals already in use (in which case people are already being exposed and financial consequences might be considerable), detailed testing to determine mechanisms of toxicity would be pursued when screening tests revealed neurotoxic effects; positive findings on a chemical undergoing commercial development might trigger its abandonment with no further testing.
Testing at the first tier is intended to determine whether a chemical has the potential to produce any neurotoxic effects—i.e., to permit hazard identification, the first step of the NRC (1983) risk-assessment paradigm. The next tier is concerned with characterization of neurotoxicity, such as the type of structural or functional damage produced and the degree and location of neuronal loss. During hazard characterization (the second step of the NRC paradigm), tests are used to study quantitative relationships between exposure (applied dose) and the dose at the target site of toxic action (delivered dose) and between dose and biologic response. The third and final tier of neurotoxicity testing is the study of mechanism(s) of action of chemical agents.
The decision to characterize a chemical through second-tier testing might be motivated by structure-activity relationships, existing data that suggest a chemical is neurotoxic, or reports of neurotoxic effects in humans exposed to the chemical, in addition to the results of first-tier testing. Testing at this second tier can help to resolve several issues, including whether the nervous system is the primary target for the chemical and what the dose-effect and time-effect relationships are for relatively sensitive end points. Such tests can also be useful in the determination of a no-observed-adverse-effect level (NOAEL) or lowest-observed-adverse-effect level (LOAEL). Experiments done at the third tier can also examine the mechanism of action associated with a neurotoxic agent; they will often involve neurobehavioral, neurochemical, neurophysiologic, or neuropathologic measures. They might also suggest biologic markers of neurotoxicity for validation and use in new toxicity tests and in epidemiologic studies.
Characteristics of Tests Useful for Screening
Like any other toxicity test, screening tests for neurotoxicity should be sensitive, specific, and valid. Sensitivity is a test's ability to detect an effect when it is produced (the ability to register early or subtle effects is especially desirable). Sensitivity depends on inherent properties of the test and on study design factors, such as the numbers of animals studied and the amount and duration of exposure. Specificity is a test's ability to respond positively only when the toxic end point of interest is present. Specificity and sensitivity are aspects of accuracy. An inaccurate test fails to identify the hazardous potential of some substances and incorrectly identifies as hazardous other substances that are not. In statistical terms, the failure to identify a hazard is a false-negative result, and the mistaken classification of a safe substance as hazardous is a false-positive result. Increasing test specificity reduces the incidence of false positives, but often has the unwanted consequence of increasing the incidence of false negatives (decreasing sensitivity); increasing test sensitivity reduces the incidence of false negatives, but often increases the incidence of false positives.
An ideal screen would have broad specificity, so that it could detect all aspects of nervous system dysfunction. In practice, no screening procedure is likely to provide the desired coverage without producing false positives. The sensitivity of the first tier might be maximized by a battery of screening tests that are individually quite specific. Several durations of exposure, long postexposure observation periods, and lifetime tests might all be necessary to cover the possible manifestations of neurotoxicity. In a tiered test system, very high sensitivity of the screening tier is usually considered essential; the loss of specificity must be compensated for in later tiers, to reveal the false positives. If, however, product development will be aborted on any indication of neurotoxicity, false positives could have high social costs. In the later tiers, narrower specificity is appropriate to characterize a suspected toxicant or to establish its mechanism of action.
Validation is the process by which the credibility of a test is established for a specific, purpose (Frazier, 1990). It entails demonstrating the reliability of the test's performance in giving reproducible results within a laboratory and in different laboratories and giving appropriate results for a control panel of substances of known toxicity. The usefulness of the results of a neurotoxicity screening test depends on a positive outcome's being strongly correlated with a neurotoxic effect that is actually caused by exposure to the test substance. Direct mechanistic causality is not essential for their interpretation, although direct insight into mechanism would be valuable, as is sought when developing a biologic marker of effect.
Validation of a test system should also include demonstration that positive test results indicate that neurotoxic effects would occur if humans were exposed to the substance. Predicting an influence on human affect or cognition with nonhuman test systems is challenging, but possible. Many animal testing models do produce effects that correspond to those seen in humans exposed to the same substance, but a result would also be considered valid if it correctly predicted any neurotoxic effect in the human population. For example, consider a screening test for an effect produced only at high doses that is consistently correlated with a milder, presumably precursor effect that occurs at lower doses. The easily detectable effect occurring at high doses in experimental animals might never be observed in humans, whose exposure would never be extreme, but its occurrence in the animal model might indicate that low-level exposure in humans could produce a more subtle toxic effect. O'Donoghue (1986) has noted that chemicals that damage axons are often metabolic poisons that produce a retardation of weight gain without decreasing food intake—a more easily observed end point. It is important to recognize that chemicals can adversely affect multiple organ systems, and an effect in other organs might influence some measures of neurotoxicity. In vitro assay systems often measure end points that appear unrelated to neurologic functioning in whole animals, but detect alterations to crucial underlying mechanistic processes.
Application of the Screen and Later Tiers
Given the enormous number of substances that have not been tested for neurotoxicity (or any form of toxicity), some characteristics of chemicals, such as their structure or production volume, must contribute to the determination of priority for screening. Confidence in the combined sensitivity of tests composing a screening battery should be high enough for negative findings to be reliably regarded as acceptable evidence that a substance is unlikely to have neurotoxic activity without the need for additional testing. If a new chemical gives positive results, there might nevertheless be academic or commercial interest in pursuing a risk assessment, or its development for use might be abandoned without further testing. Detailed testing to characterize neurotoxicity revealed by screening procedures will be more common for existing substances with commercial value or wide exposure. In either case, the path followed through the tiers of testing would be contingent on a chemical's unfolding toxicity profile, including types of toxicity other than neurotoxicity. A feasible screening system must strike a balance among the amount of time and expense that society is willing to expend in testing, its desire for certainty that neurotoxic substances are being kept out of or removed from the environment, and its interest in gaining benefits from various types of chemicals.
Current Methods Based on Structure-Activity Relationships
Given the overwhelming lack of epidemiologic or toxicologic data on most chemical substances and the need to develop rational strategies to prevent adverse health effects of both new and existing chemicals, attempts have been made in many subfields of toxicology to generate predictive strategies based primarily on chemical structure. The basis for inference from structure-activity relationships (SARs) can be either comparison with structures known to have biologic activity or knowledge of structural requirements of a receptor or macromolecular site of action. However, given the complexity of the nervous system and the lack of information on biologic mechanisms of neurotoxic action, SARs should be regarded as, at best, providing information that might be useful to identify potentially neuroactive substances. SARs are clearly insufficient to rule out all neurotoxic activity; it is not prudent to use them as a basis for excluding potential neurotoxicity. Caution is warranted in interpreting SARs in anything other than the most preliminary analyses. An intelligent use of SARs requires detailed knowledge not only of structure, but also of each critical step in the pathogenetic sequence of neurotoxic injury. Such knowledge is still generally unavailable. SAR evaluations form a major basis for EPA and FDA decisions on whether to pursue full neurotoxicity assessments, so it must be concluded that the approach of these agencies cannot be expected to provide an adequate screen for identifying neurotoxicants.
SAR approaches are more successful when the range of possible sites of action or mechanisms of action is narrow. Thus, SARs have had more use in relation to carcinogenicity and mutagenicity than in other kinds of toxicity. The SAR approaches used in the development of novel neuropharmacologic structures deserve consideration in neurotoxicology, but they must rest on fuller understanding of neurotoxic mechanisms.
SARs have not been used extensively in neurotoxicology, for several reasons. Many agents are neurotoxic—from elements, such as lead and mercury, to complex molecules, such as MPTP and neuroactive drugs. Neurotoxicants have so many potential targets that it is difficult to rule out any chemical a priori as not likely to be neurotoxic. Finally, for most neurotoxicants, even those which are well characterized, data on the mechanism of action at the target site are insufficient for the elucidation of useful SARs.
SARs have proved useful in some cases—usually, within particular classes of compounds on which some mechanistic data are available. Extensive studies of anticholinesterase organophosphates and carbamates, for example, have led to a mechanistic model that requires a structure that binds to a specific site on the cholinesterase molecule and to the establishment of an SAR for cholinesterase inhibition. The SAR is relatively straightforward, in that enzyme inhibition is the primary action that leads to various symptoms of poisoning. Another series of studies has been conducted in an attempt to develop an SAR for the pyrethroid insecticides. Although the nerve membrane sodium channel has been identified as the critical target site of pyrethroids, studies have failed to establish a clear-cut SAR, despite the large number of pyrethroids synthesized and tested. However, the importance of the d-cyano group in distinguishing two behavioral manifestations (choreoathetosis and salivation versus tremor) and the importance of the ester moiety in pyrethroid SAR are well documented (Aldridge, 1990). A more comprehensive SAR for pyrethroids is lacking partly became most studies were performed with insecticidal action as a measure of activity, not the interaction with the sodium channel. Action at the target site must be known to establish an SAR. However, insecticidal potency involves factors other than recognition at the target site, including absorption and metabolism. As the mechanism of action of n-hexane and methyl n-butylketone neurotoxicity becomes more clear, so does the capacity to look at an alkane or ketone and predict its neurotoxic potential. The critical issue in the neurotoxicity of such a chemical is whether it will be metabolized to a γ-diketone; α-, β-, and δ-diketones are less likely to be neurotoxic (Krasavage et al., 1980; Spencer et al., 1978), but most γ-diketones and γ-diketone precursors cause a specific type of neurotoxicity, a motor-sensory peripheral neuropathy caused by neurofilament abnormalities. The gamma spacing of the two ketone groups allows formation of the five-membered heterocyclic ring, the pyrrole, except where steric hindrance is present (Genter et al., 1987).
Current in Vitro Procedures
In vitro procedures for testing have practical advantages, but studies must be done to correlate their results with responses in whole animals. One advantage of validated in vitro tests is that they minimize the use of live animals. Some of the more developed in vitro tests might be simple and might not have to be conducted by highly trained personnel, but, as with many in vivo tests, the analysis and interpretation of results is likely to require expertise. Some in vitro tests lend themselves to computer-controlled automated operation, as do some well-developed, highly sophisticated behavioral tests (see, for example, Evans, 1989), and that results in savings in time and expense and allows testing of large numbers of substances. Experience with the Ames test for mutagenesis confirms the advantages of in vitro procedures, but also illustrates the problems that arise when an assay is used to predict an end point that is not exactly what it measures (e.g., carcinogenicity, rather than specific sorts of genotoxicity).
Biochemical Assays
Biochemical assays epitomize the advantages of in vitro tests. Their current usefulness, however, is limited to substances in a very few chemical classes. Their usefulness will no doubt increase as the molecular basis of action of other classes of neurotoxicants becomes known.
Neurotoxic chemicals exert their effects via specific molecular interactions with biologic targets. For a few toxicants, the molecule that is affected is well established, and it is possible to investigate the potential toxicity of other substances in the same class with a biochemical assay. An example of this approach is the enzyme-inhibition assay of organophosphorus (OP) esters. Some OP esters cause a distal polyneuropathy that is not evident until several days after exposure (OP-induced delayed neuropathy, or OPIDN). There is now good evidence (Abou-Donia, 1981) that the capacity of a given OP ester to cause OPIDN correlates strongly with the relative inhibition of a CNS enzyme activity called neurotoxic esterase. The enzyme activity is easily measured in a test-tube assay, and the addition of OP esters and OP-like compounds to the assay allows one to screen for ability to inhibit the enzyme and thus for its potential to produce OPIDN (Johnson, 1977). The ease of this assay is complicated, however, by the precautions necessary to protect the technicians performing the assay from exposure to potentially highly neurotoxic substances. Many OP compounds also produce an immediate toxic effect via inhibition of another enzyme, acetylcholinesterase. A test-tube assay of that enzyme's activity can serve as a screen for such acute neurotoxicity.
Tissue Culture
The study of in vitro systems has provided much fundamental information of value in understanding the nervous system in vivo, and in vitro investigations have sometimes been invaluable in guiding in vivo neurotoxicologic research. The physiologically excitatory actions of some amino acids (for example, glutamate and aspartate) on neurons can become pathologic, and brain lesions can be produced by amino acid analogues, such as kainate or quisqualate. In vitro systems are being used extensively to study the mechanisms of the neuronal injury produced by those agents and to identify antagonists that might be useful in reducing excitotoxic brain damage. In vitro studies can also identify compounds with a high neurotoxic potential that might merit study in intact organisms. An example is β-N-methylamino-L-alanine (BMAA), a constituent of a dietary plant, the false sago palm. BMAA was initially identified as an excitotoxic amino acid in cultures of tissue from spinal cord and cerebral cortex. That led to experiments with primates that showed that BMAA produces motoneuronal lesions in the cortex and spinal cord.
A broad range of tissue-culture systems are available for assessing the neurologic impact of environmental agents. Although those systems are not now used for hazard detection, they can be used to characterize chemical-induced effects. They can be classified according to their increasing complexity, from cell lines to organ cultures.
- Cell lines. Neuronal and glial cell lines have been used in many neurobiologic studies and are valuable in neurotoxicology. They consist of populations of continuously dividing cells that, when treated appropriately, stop dividing and exhibit differentiated neuronal or glial properties. Various neuronal lines, for instance, develop electric excitability, chemosensitivity, axon formation, transmitter synthesis and secretion, and synapse formation. Large quantities of cells can be generated routinely to develop extensive dose-response or other quantitative data. For example, the dose range over which a group of chemicals affect cell differentiation and proliferation was established in neuroblastoma cells (Stark et al., 1989). Those are tumor cells, however, so the interpretation of such data with regard to toxicity in the intact nervous system must be guarded. A culture system for nontransformed neural cells was recently announced (Ronnett et al., 1990).
- Dissociated cell cultures. When neural tissue, typically from fetal animals, is dissociated into a suspension of single cells, and the suspension is inoculated into tissue-culture dishes, the neurons and glia survive, grow, and establish functional neuronal networks. Such preparations have been made from most regions of the CNS and exhibit highly differentiated, site-specific properties that constitute an in vitro model of different portions of the CNS. Most of the neuronal transmitter and receptor phenotypes can be demonstrated, and a variety of synaptic interactions can be studied. Glial cells are also present, and neuroglial interactions are a prominent feature of the cultures. A substantial battery of assays (neurochemical and neurophysiologic) is now available to assess the development of the cultures and to indicate toxic effects of test agents added to the culture medium. Relatively pure populations of different cell types can be isolated and cultured, so that effects on specific cell types can be assessed independently. Pure glial cells or neurons, or even specific neural categories, can be prepared in this way and studied separately, or interaction between neurons and glial cells can be studied at high resolution. The neurobiologic measures used to assess the effect of any agent can be very specific (for example, activity of a neurotransmitter-related enzyme or binding of a receptor ligand) or global (for example, neuron survival or concentration of glial fibrillary acidic protein). The two-dimensional character of the preparations makes them particularly suited for morphologic evaluation, and detailed electrophysiologic studies are readily performed. The toxic effects and mechanisms of anticonvulsants, excitatory amino acids, and various metals and divalent cations have been assessed with these preparations. The cerebellar granule cell culture system, for example, has been exploited recently in studies of the mechanism of alkyllead toxicity (Verity et al., 1990).
- Reaggregate cultures. A related but distinct preparation made from single-cell suspensions of neural tissue is the reaggregate culture. Instead of being placed in culture dishes and allowed to settle onto the surface of the dishes, the cells are kept in suspension by agitation; under appropriate conditions, they stick to one another and form aggregates of controllable size and composition. Typically, the cells in an aggregate organize themselves and exhibit intercellular relations that are a function of and bear some resemblance to the brain region that was the source of the cells. The cells establish a three-dimensional, often laminated structure, perhaps approximating the in vivo nervous system more closely than do the dissociated cultures grown on the surface of a dish. Reaggregate cultures lend themselves to large-scale, quantitative experiments in which neurobiologic variables can be examined, although morphologic and ligand-binding studies are performed less readily than with surface cultures.
- Explant cultures. Organotypic explant cultures are even more closely related to the intact nervous system. Small pieces or slices of neural tissue are placed in culture and can be maintained for long periods with substantial maintenance of structural and cell-cell relations of intact tissue. Specific synaptic relations develop and can be maintained and evaluated, both morphologically and electro-physiologically. Because all regions of the nervous system are amenable to this sort of preparation, it is possible to analyze toxic agents that are active only in specific regions of the central or peripheral nervous system. Explants can be made from relatively thin slices of neural tissue, so detailed morphologic and intracellular electrophysiologic studies are possible. Their anatomic integrity is such that they capture many of the cell-cell interactions characteristic of the intact nervous system while allowing a direct, continuing evaluation of the effects of a potentially neurotoxic compound added to the culture medium. The process of myelination has been studied extensively in explant cultures, and considerable neurotoxicologic information has been gained. As noted above, the pathogenetic actions of excitatory amino acids normally active in the nervous system, as well as such analogues as the neurotoxin BMAA, have been revealed by experiments with organotypic cultures.
- Organ Cultures. A preparation similar to an explant culture is the organ culture, in which an entire organ, such as the inner ear or a ganglion, rather than slices or fragments, is grown in vitro. Obviously, only structures so small that their viability is not compromised can be treated in this way.
The advantages of the various types of in vitro systems are summarized in Table 4-1. Most in vitro preparations are made from young, usually prenatal animals. (But cultures derived from human neural tissue have been the object of a number of studies.) Typically, a period of rapid change and development occurs immediately after the cultures are initiated, and conditions become much more stable if the cultures are maintained for weeks or months. Thus, the preparations can be used to study neurotoxic effects that might be specific to developing nervous tissue and to compare the effects of agents in developing stable tissue.
TABLE 4-1
In Vitro Neurobiologic Test Systems.
In general, the technical ease of maintaining a culture varies inversely with the degree to which it captures a spectrum of in vivo characteristics of nervous system behavior. The problem of biotransformation of potentially neurotoxic compounds is shared by all, although the more complete systems (explant or organ cultures) might alleviate this problem in specific instances. In many culture systems, complex and ill-defined additives—such as fetal calf serum, horse serum, and human placental serum—are used to promote cell survival. A number of thoroughly described synthetic media are now available, however, and such fully defined culture systems can be used where necessary. Indicators of neuronal and glial function, and hence indicators of neurotoxicity, are outlined in Table 4-2.
TABLE 4-2
Markers for Assessing Neurotoxicity in In Vitro Systems.
Plausibility of an In Vitro Screening Battery
A broad range of in vitro systems are now available for studying development of the nervous system and the normal function of neurons and glial cells. The possible neurotoxic impact of any chemical on any specific neurobiologic variable could, in principle, be screened with an appropriate set of in vitro tests. In practice, of course, because the number of potential neurobiologic end points to be measured is so large, screening for the effects of any agent on all of them would be prohibitively expensive in time and money. The question, then, is whether a feasible battery of tests will pick up an acceptably large percentage of toxic chemicals (while generating an acceptably low percentage of false positives). Ideally, the screen would provide direction for the more intensive study of substances that it identifies as having neurotoxic potential.
To what extent could an in vitro system provide such a screening instrument? Two basic questions are associated with the use of in vitro tests for that purpose:
- What indicators of neuronal (or glial) damage would be sufficiently general to be useful?
- What specific test systems or combinations would be adequate to cover a number of different and differentially site-specific neurotoxic agents?
The first question might be answered by a combination of assays that would include general indicators of neuronal and glial survival and a few more specific indicators. Counts of numbers of surviving neurons (or glia) and biochemical measurements of tetanus-toxin-specific or sodium-channel-specific ligand binding could be used as general indicators. Uptake of γ-aminobutyric acid (GABA), benzodiazepine binding, and cholinergic and aminergic markers could be used, depending on the neural system chosen, to get some indication of neurotransmitter-related functions. If a chemical has a single very specific neurologic target, this would in general be missed, but it might be anticipated that such a specific effect would be accompanied by more general secondary neuropathologic consequences. For instance, if spinal-cord or cerebral cortical cell cultures are exposed to the specific voltage-dependent sodium channel blocker tetrodotoxin for 4–5 days, the decrease in electric activity kills about half the neurons.
As to the second question, the choice of culture systems to be used is difficult. It is axiomatic that no cell culture represents a normal nervous system. Even in cocultured explants, the normal connections among cells are disrupted. Specific neurobiologic properties have been shown not to be expressed in various in vitro preparations where they might be expected. For instance, in dissociated hippocampal pyramidal cells, serotonergic responses cannot be demonstrated; the responses appear during development in vivo, but appropriate signals that induce their expression are evidently lacking in vitro. It is impossible to predict how such departures from normality will influence the screen's ability to detect the effects of a test substance. Some compromise between comprehensiveness and fiscal feasibility would have to be made. Neural and glial cell lines are available and relatively straightforward technically. Robust versions of cerebral-cortex, spinal-cord, and subcortical systems are available either as dissociated preparations or in explant form. Indeed, essentially all regions of the nervous system are being grown in vitro in one form or another. If a candidate neurotoxic material has properties that suggest site-specific activity, obviously one would include its putative target in a test battery. If no such information is available, then some arbitrary panel of test systems would have to be used.
Even if these two questions are answered, cell-culture techniques have several remaining disadvantages when used as screens. A given toxicant may require metabolism outside the nervous system to produce a toxic metabolite, so exposure to the toxicant in vitro may give a false-negative result; conversely, the chemical might be detoxified before reaching the nervous system. A related disadvantage might be low solubility of a given toxicant in an aqueous culture medium, which could limit the quantity of toxicant to which the cells are actually exposed. A related problem concerns the lack of a blood-brain barrier in in vitro experiments. Toxicity could be attributed to a compound that would not reach the brain, either at all or in sufficient concentration to cause toxicity, in in vivo exposure. Last, and most important, the more complex functions of the nervous system are properties of assemblies of neurons. Learning, memory, emotions, the coordination of movement, and homeostatic regulation, for instance, cannot be studied in vitro.
Despite those disadvantages, some applications of cell cultures are nearly ready for use in neurotoxicity screening. Several indicators of glial and neuronal function in vitro can be used as biologic markers of effect (Table 4-2). In vitro systems have been extremely useful in identifying and analyzing excitotoxic materials. Also, in vitro demonstrations of neurotoxicity of anticonvulsants based on both general and specific indicators have been corroborated by clinical studies of IQ decrements related to phenobarbital treatment in children. Many neurotoxicants attack the glial cells that form myelin sheaths around axons, and cell cultures can demonstrate chemically induced myelin abnormalities; this has been shown with the myelin abnormalities produced by methylmercury (Kim, 1971), thallium. (Spencer et al., 1985), triethyltin (Graham et al., 1975), and several other substances. Tests on cultures can also detect toxic damage to neurons themselves. For example, aluminum neuronopathy is reproduced in cultures of dorsal root ganglion cells (Seil et al., 1969), and the axonal degeneration characteristic of γ-diketone toxicity is reproduced in organotypic cocultures (Veronesi et al., 1978). Furthermore, as noted in Chapter 2, the cells of the CNS are uniquely dependent on a high rate of glucose metabolism; many neurotoxicants impair neural glucose metabolism (Damstra and Bondy, 1980), and this toxic mechanism could be reflected in a decrease in the viability of cultured cells.
A somewhat arbitrary, but specific, protocol for developing a set of in vitro tests for screening is presented, for illustrative purposes, in Table 4-3. It is an empirical question whether such a battery of in vitro tests could contribute importantly to neurotoxicologic screening of a broad range of environmental agents. Research needs to be done with known neurotoxic agents and related compounds to see whether culture systems can reliably identify dangerous substances and distinguish them from related, but inactive or less active, substances. Analyses of the contributions of short-term in vitro tests to neurotoxicity are for less developed and for fewer in number than analyses of in vitro tests in other fields, e.g., carcinogenesis (Lave and Omenn, 1986).
TABLE 4-3
Proposed Protocol for Developing an In Vitro Neurotoxicity Screening System.
It should be possible to define the goals of such research to be finite and realizable. The question may be framed as follows: What is the smallest set of tests that gives positive indicators of neurotoxicity for all of a control panel of compounds known to have diverse neurotoxic properties? Of equal importance, and perhaps more difficult, would be the evaluation of presumably nonneurotoxic substances to preclude a high rate of false positives. One might start with a fairly inclusive set of test cultures, such as of cerebral cortex, brain stem, cerebellum, spinal cord, sympathetic ganglion neurons, and a continuous cell line, such as PC12 cells. Global assays, such as cell counts and tetanus-toxin binding, would be initial indicators, and specific neurotransmitter-related probes could be used as appropriate. The essential feature of such an effort is the reliability of the culture systems. Reliability typically is achieved only after a fairly long period of use in a given laboratory.
It will be difficult to cover all end points of concern with a modest number of in vitro assays. Whole-animal tests have the potential of exploiting the integrative nature of behavior, thereby covering a diversity of adverse end points that might result from exposure to neurotoxic agents. It might be more efficient to screen with in vivo tests and use an in vitro approach when concentrating on specific mechanisms of action. A goal of research directed at development of adequate neurotoxicologic methods should be to combine behavioral testing on intact animals with selected in vitro tests. Can some behavioral tests be replaced by less expensive in vitro tests without loss of diagnostic power? Can individual behavioral tests function as well as or better than groups of in vitro tests? One might start with the study of Tilson (1990a) as a base and complement it with in vitro study of the same chemicals. The results might lead to a choice of tests that will yield an optimal combination of broad coverage and low cost.
Current In Vivo Procedures
Neurotoxic effects on complex integrative functions—such as motor performance, sensory acuity, memory, and cognitive processes—can be detected only in vivo. Moreover, the neural activities that mediate integrative function involve large numbers of neurons relatively distant from each other. Integration may be disrupted by the removal and isolation of neural tissue that is necessary for some in vitro techniques, but neurotoxic effects of relatively low exposures involving integrative functions can be detected in whole animals.
Species differ in their susceptibility to various toxic agents and the degree to which their nervous systems resemble that of humans. Tests using species closest to humans would logically yield the data about which we could feel most confident; however, various considerations—including cost, ethics, and the extent of pre-existing data bases—favor the use of small laboratory rodents, such as mice and rats, for in vivo hazard identification and characterization. Table 4-4 lists commonly reported neurotoxic effects of several classes of toxic chemicals in humans and animals. It is apparent from the list that motor disturbances, mood alteration, and sensory abnormalities are especially common in humans, but that the findings in animals are only sometimes comparable. Species differences might account for some of these discrepancies; however, it is not clear that animal tests always are intended to be or can be exactly comparable with human tests.
TABLE 4-4
Neurotoxic Effects of Representative Agents in Humans and Animals.
Testing for neurotoxicity in humans implies that exposure has occurred. Such testing is therefore considered a means of secondary prevention, so the testing methods specifically for humans are discussed in Chapter 5.
Behavioral Assessment
Chemical-induced functional alterations of the nervous system are often assessed with behavioral techniques. Perhaps the greatest scientific challenge to neurotoxicology is to integrate observations of behavior with other aspects of neurobiology—such as morphology, neurochemistry, and neurophysiology—to develop a unified theory not only of toxicity, but also of the nervous system. By itself, behavior is an important end point, even if its biologic substrates have not been clearly identified. Numerous behavioral techniques are available to measure chemical-induced alterations in sensory, motor, autonomic, and cognitive function (Table 4-5).
TABLE 4-5
Examples of Behavioral Measures of Functional Neurotoxicity.
Behavioral methods differ greatly in their complexity and specificity. At one extreme, some methods (e.g., some observational tests) can be applied broadly and routinely to assess the neurotoxicity of a wide array of chemicals and chemical exposures; that is, they are useful in hazard identification. Such tests typically incorporate responses already well established in an organism. Other methods require instrumentation or training of the animals before chemical exposure; although they might not be appropriate for the routine screening (tier 1) of chemicals for neurotoxicity, they could be used for characterization of toxicant-induced effects (tier 2).
Functional Observational Batteries
Functional observational batteries (FOBs) are designed to detect and measure major overt neurotoxic effects. Several have been used, each consisting of tests generally intended to evaluate various aspects of sensorimotor function (EPA, 1985; Haggerty, 1989; Kulig, 1989; Moser, 1989; O'Donoghue, 1989). FOB tests are essentially clinical examinations that detect the presence or absence, and in some cases the relative degree, of specific neurologic signs.
Screening neuroactive chemicals with an FOB is well established. Irwin (1968), for example, described a series of tests for evaluating the effects of drugs in mice and showed how different drugs produced different patterns of effects that could be easily recognized. Gad (1982) described a battery of tests for assessing the neuromuscular effects of industrial chemicals in rats. More recently, Moser and colleagues (Moser et al., 1988; Moser, 1989) developed a similar battery for assessing the neurobehavioral effects of a broad range of industrial and pesticidal chemicals in rats.
FOBs are sets of observations and tests each made on individual experimental animals; e.g., the FOB suggested by Moser (1989) is presented in Table 4-6. It is assumed that many of the individual observational components overlap in the neurologic functions that they assess (autonomic function, motor function, equilibrium, excitability, and sensorimotor reflexes). Therefore, if several unrelated observed end points in an entire FOB were affected, there would be little concern about a chemical's neurotoxicity. If several unrelated neurologic functions were affected, but only at high doses and in conjunction with other overt signs of toxicity, including death or debilitation, there would be more concern. If several related functions were affected and the effects appeared to be dose- and time-dependent, there would be still more concern. As the number of chemicals tested for neurotoxic potential increases, many different combinations of affected functions will emerge. Deciding which combinations of positive findings indicate the need to continue testing will not be trivial.
TABLE 4-6
End Points That Might be Included in a Functional Observational Battery.
From many standpoints, FOBs have shortcomings. Most of their observations are semiquantitative. The sensitivity, reliability, and reproducibility of some have not been well documented. This deficiency can often be overcome by using more quantitative methods. For example, extensor thrust and grip strength can be measured with simple devices that use a strain gauge (Cabe and Tilson, 1978; Meyer et al., 1979). Hindlimb splay can be measured by inking the animal's paws, dropping it onto a sheet of paper, and measuring the distance between the footprints (Edwards and Parker, 1977). Equilibrium and muscle coordination can be measured by the length of time a rat can maintain itself on a rotating rod (Bogo et al., 1981). Tremor can be characterized with automated devices (Gerhart et al., 1985; Newland, 1988). Many studies have shown that the observations commonly included in FOBs can detect the effects of known neurotoxicants (Haggerty, 1989; Kulig, 1989; Moser, 1989; O'Donoghue, 1989).
Motor Activity
Motor activity includes a broad class of behaviors involving coordinated participation of sensory, motor, and integrative processes (MacPhail et al., 1989). Motor activity has several advantages for testing: it is noninvasive; motivational procedures, such as food deprivation, are not needed to produce it; and its recording is usually automated, and that reduces experimenter-animal interactions (Maurissen and Mattsson, 1989). Many studies have shown that motor activity can be affected by psychoactive and neuroactive chemicals (Reiter and MacPhail, 1979; Tilson, 1987; MacPhail et al., 1989). MacPhail et al. (1989) evaluated motor-activity measures and found them highly consistent across replications. The sensitivity of motor-activity measures in detecting reproducible neurotoxic effects is generally comparable with that of more sophisticated measures of neurobehavioral function.
Although motor-activity measures are often used to identify neurotoxic chemicals, they have disadvantages, particularly with regard to specificity of adverse effects on the nervous system. It has been argued that the results of motor-activity tests alone lack specificity and do not often provide information useful for later testing or characterization (Maurissen and Mattsson, 1989). However, specificity might not be as important in hazard identification as consistency and sensitivity. Motor-activity measurements are commonly used in conjunction with FOBs.
Schedule-Controlled Operant Behavior
Schedule-controlled operant behavior (SCOB) involves the maintenance of behavior (performance) by intermittent reinforcement. Different patterns of behavior and response rates are controlled by the relationship between response and later reinforcement. SCOB affords a measure of learned behavior and is useful for studying chemical-induced effects on motor, sensory, and cognitive function.
The primary end points for evaluation are agent-induced changes in response rate or frequency and the temporal pattern of responding. Response rate is usually related to an objective response, such as a lever press or key peck, and differs according to the schedule of reinforcement. Response rates are expressed per unit of time. For some classes of chemicals, the direction of an effect on response rate can differ between low and high doses. Agent-induced changes in temporal pattern of responding can occur independently of changes in rate and require analysis of the distribution of responses relative to the reinforcement schedule.
SCOB has been used to study the effects of psychoactive drugs on behavior and is sensitive to many neurotoxicants, including methylmercury, solvents, pesticides, acrylamides, carbon monoxide, and organic and inorganic lead (see, for example, MacPhail, 1985; Tilson, 1987; and Rice, 1988). The experimental animal often serves as its own control, and the procedure provides an opportunity to study a few animals extensively over a relatively long period. SCOB typically requires motivational procedures, such as food deprivation; and training sessions are usually required to establish a stable baseline of responding. Because of its sensitivity to neuroactive chemicals, SCOB has greater potential for use in hazard characterization than in hazard identification.
Many of the behavioral tests, and particularly the FOB, have been developed and validated with well-characterized neurotoxicants. It is much easier to interpret the results of behavioral tests when a large body of information already exists on a particular substance; one can use this information to help interpret the results. It is much more difficult to interpret the results of screening tests on a new product on which very little information is available.
Specialized Tests of Neurologic Function
Neurotoxicants produce a wide array of functional deficits, including motor, sensory, and learning-memory dysfunction. Many procedures have been devised to assess relatively subtle changes in those functions; hence their applicability to hazard characterization. Specialized tests and agents that affect them have been reviewed recently (WHO, 1986; Tilson, 1987) and are discussed only briefly below.
Motor Function
Motor dysfunction is a common neurotoxic effect, and many different types of tests have been devised to measure time-and dose-dependent effects. Anger (1984) reported 14 motor effects of 89 substances, which could be classified into four categories: tremor, convulsions, weakness, and incoordination. Chemical-induced changes in motor function can be determined with relatively simple techniques and may be used as a component of an FOB.
Several procedures have been used to characterize chemical-induced motor dysfunction. An example has been described by Newland (1988), who trained squirrel monkeys to hold a bar within specified limits (i.e, displacement) to receive positive reinforcement. The bar was also attached to a rotary device, which allowed measurement of chemical-induced tremor. Spectral analysis was used to characterize the tremor, which was found to be similar to those seen in humans exposed to neurotoxicants or with such neurologic diseases as Parkinson's disease.
Incoordination and performance changes can be assessed with procedures that measure chemical-induced alterations in force (Fowler, 1987). Animal are trained to receive positive reinforcement by applying force to a fixed lever. Training can also include maintenance of an appropriate force for a specified period. The accuracy of performance is sensitive to many psychoactive drugs (Walker et al., 1981; Newland, 1988). Gait has been measured in rats under standardized conditions and can be a sensitive indication of specific damage to the basal ganglia and motor cortex (Hruska et al., 1979) as well as damage to the spinal cord and peripheral nervous system.
Procedures to characterize chemical-induced motor dysfunction have not been used extensively in neurotoxicology. Most of them require pre-exposure training (including alterations of motivational state) of experimental animals. However, such tests might be useful, inasmuch as similar procedures are often used in assessing humans.
Sensory Function
Alterations in sensory processes (e.g., paresthesias and visual or auditory impairments) are frequently reported signs or symptoms in humans exposed to toxicants (Anger, 1984). Several approaches have been devised to measure sensory deficits. Data from tests of sensory function must be interpreted within the context of changes in body weight and body temperature. Furthermore, many tests assess the behavioral response of an animal to a specific stimulus; the response is usually a motor movement that could be directly affected by chemical exposure. Thus, care must be taken to determine whether proper controls were included to eliminate the possibility that changes in response to a stimulus were related to agent-induced motor dysfunction.
Several testing procedures have been devised to screen for sensory deficits. Many rely on orientation or the response of an animal to a stimulus. Such tests are usually included in the FOB used in screening (e.g., tail-pinch or click responses). Responses are usually recorded as being either present, absent, or changed in magnitude (Moser, 1989; O'Donoghue, 1989). The tests would not be suitable to characterize chemical-induced changes in acuity or fields of perception. Sensory deficits are usually characterized with psychophysical methods, which study the relationship between the physical dimensions of a stimulus and the behavioral response it generates (Maurissen, 1988).
One approach to the characterization of sensory function involves the use of reflex-modification techniques (Crofton and Sheets, 1989). A stimulus of varying intensity is presented before a stimulus that elicits a defined sensorimotor reflex. If the time between the two stimuli is appropriate, the response to the eliciting stimulus can be significantly inhibited (i.e., prepulse inhibition). The observation of inhibition is contingent on the ability of the animal to perceive the presence of the first stimulus. Agent-induced changes in the frequency or threshold required to inhibit the reflex are taken as possible agent-induced changes in sensory function. Changes in the ability of the first stimulus to inhibit the reflex must be interpreted within the context of changes in the response to the eliciting stimulus, i.e., a sensory change is inferred primarily on the basis of an agent's ability to alter the degree of inhibition of the reflex in the absence of any related change in sensorimotor function. Control runs should be performed to determine the basal response without the initial stimulus. Prepulse inhibition has been used only recently in neurotoxicology (Fechter and Young, 1983) and can be used to assess sensory function in humans, as well as in experimental animals.
Various behavioral procedures require that a learned response occur only in the presence of a specific stimulus. Chemical-induced changes in sensory function are determined by altering the physical characteristics of the stimulus (e.g., magnitude or frequency) and measuring the alteration in response rate or accuracy. In an example of the use of a discriminated conditional response to assess chemical-induced sensory dysfunction, Maurissen et al. (1983) trained monkeys to report the presence of a vibratory or electric stimulus applied to the fingertip. Repeated dosing with acrylamide produced a persistent decrease in vibration sensitivity; sensitivity to electric stimulation was unimpaired. That pattern of sensory dysfunction corresponded well to known sensory deficits in humans. Discriminated conditional response procedures have been used to assess the ototoxicity produced by toluene (Pryor et al., 1983) and the visual toxicity produced by acrylamide (Merigan et al., 1982).
Procedures to characterize chemical-induced sensory dysfunction have been used often in neurotoxicology. As in the case of most procedures designed to characterize toxicity, training and motivational factors can be confounding factors. Many tests designed for laboratory animals can be applied to humans.
Learning and Memory
Learning and memory disorders are neurotoxic effects of great importance. Impairment of memory is reported fairly often by adult humans as a consequence of toxic exposure. Behavioral deficits in children have been caused by lead exposure (Smith et al., 1989). And it is hypothesized (Calne et al., 1986) that chronic low-level exposure to toxic agents can have a role in the pathogenesis of senile dementia.
Learning is defined as a lasting change in behavior, memory as the persistence of learned behavior. Alterations in learning and memory must be inferred from changes in behavior. However, changes in learning and memory must be separated from other changes in behavior that do not involve cognitive or associative processes (e.g., motor function, sensory capabilities, and motivational factors), and an apparent toxicant-induced change in learning or memory should be demonstrated over a range of stimuli and conditions. Before it is concluded that a toxicant alters learning and memory, effects should be confirmed in a second learning procedure. It is well known that lesions in some regions of the brain can facilitate some types of learning by removing behavioral tendencies (e.g., inhibitory responses due to stress) that moderate the rate of learning under normal circumstances. A discussion of learning procedures and examples of chemicals that can affect learning and memory have appeared in recent reviews (Heise, 1984; WHO, 1986; Tilson, 1987; Peele, 1989).
One simple end-point procedure to measure in assessing learning and memory is habituation, which is defined as a gradual decrease in the magnitude or frequency of a response after repeated presentations of a stimulus. A toxicant can affect habituation by increasing or decreasing the number of stimulus presentations. An example of chemical-induced effects on habituation can be found in a study by Overstreet (1977), who reported that diisopropyl fluorophosphate (DFP), a choline acetyltransferase inhibitor, had no effect on the response to a novel stimulus; with repeated presentations of the stimulus, however, DFP-treated rats habituated slower than controls. Habituation is a very simple form of learning and would be perturbed by a number of chemical effects not related to learning.
A general approach to studying the effects of a chemical on learning and memory involves the pairing of a novel stimulus with a second stimulus that produces a known, observable, and quantifiable response (i.e., classical ''Pavlovian'' conditioning). The novel stimulus is known as the conditioned stimulus, and the second, eliciting stimulus the unconditioned stimulus. With repeated pairings of the two stimuli, the conditioned stimulus comes to elicit a response similar to the response to the unconditioned stimulus. The procedure has been used in behavioral pharmacology and, to a lesser extent, in neurotoxicology. Neurotoxicants that interfere with learning and memory would alter the number of presentations of the pair of stimuli required to produce conditioning or learning. Memory would be tested by determining how long after the last presentation of the two stimuli the conditioned stimulus would still elicit a response. For example, Yokel (1983) dosed rabbits repeatedly with aluminum and found that exposed rabbits learned the conditioned eyeblink response more slowly than controls; such effects on establishing the conditioned relationship between the two stimuli were seen in the absence of aluminum-induced alterations in sensitivity to the unconditioned stimulus or ability to respond. Other classically conditioned responses known to be affected by psychoactive or neurotoxic agents are the conditioned taste aversion (Riley and Tuck, 1985) and conditioned suppression (Chiba and Ando, 1976).
Other procedures use instrumental learning, which involves the pairing of a response with a stimulus that increases the probability of future response through reinforcement. Response rate can be increased by using positive reinforcement or removing negative reinforcement. Learning is usually assessed by determining the number of presentations or trials needed to produce a defined frequency of response. Memory can be defined specifically as the maintenance of a stated frequency of response after initial training. Neurotoxicants may adversely affect learning by increasing or decreasing the number of presentations required to achieve the designated criterion. Decrements in memory may be indicated by a decrease in the probability or frequency of a response at some time after initial training. Toxicant-induced changes in learning and memory should be interpreted within the context of possible toxicant-induced changes in sensory, motor, and motivational factors. An example of a test based on instrumental learning is the repeated-acquisition procedure. It requires that an animal learn how to solve a series of problems that vary from session to session. The results of the repeated-acquisition procedure are affected by carbon monoxide and microwave exposures (Schrot et al., 1980, 1984). Other examples of instrumental learning procedures used in neurotoxicology are passive and active avoidance, Y-image avoidance, spatial mazes (radial-arm maze), and delayed matching to sample (see Heise, 1984; WHO, 1986; Tilson, 1987).
One problem with almost all measures of learning and memory in animals is that it is sometimes difficult to extrapolate the procedures to human testing or to predict analogous effects in humans. However, some tasks, such as the conditioned eyeblink test and delayed matching to sample, can be easily adapted for use with humans.
Neurophysiologic Procedures
Several clinical neurophysiologic tests have been applied to exposed animals for diagnostic or prognostic evaluation of the nervous system. The tests are noninvasive or only minimally invasive, readily adapted to longitudinal studies, and capable of detecting and measuring neurologic manifestations of neurotoxicity, regardless of the initiating mechanism. When properly chosen, they can probe the functional status of particularly affected portions of neuronal networks (such as reflexes and evoked responses). Their results are reproducible within the constraints of biologic variation and the skill of the experimenter. Although of considerable value in risk assessment, they are nonetheless post hoc studies that yield results that reflect varied and often unknown exposures. The use and limitations of the tests have recently been reviewed by LeQuesne (1987).
Under conditions of the same locus and mechanism of neurotoxic action (such as demyelination), neurophysiologic testing in experimental animals appears to produce reliable indicators of what might be expected after similar exposure of other species, including humans. That can be illustrated by a brief comparison of data from clinical and experimental studies of acrylamide neurotoxicity. Clinical features of the neuropathy included ataxia, diminished or lost tendon reflexes, and sensory complaints (LeQuesne, 1980). Nerve-conduction velocities in affected patients were normal or (in a very few patients) just below normal (Takahashi et al., 1971). Studies in baboons revealed a similar picture of weakness and ataxia, also with minimal decrements in conduction velocities (Hopkins and Gilliatt, 1971). Cats with acrylamide intoxication had normal sensory and motor nerve-conduction velocities (Lowndes and Baker, 1976; Lowndes et al., 1978a), but were markedly ataxic and hyporeflexic. The close temporal association between onset and severity of signs of intoxication and impairment of static and dynamic properties of muscle spindles (Lowndes et al., 1978b) revealed that proprioceptive defects form the basis for the loss of tendon reflexes and almost certainly contribute to the ataxia.
Nerve-Conduction Studies
A number of end points can be recorded, but the critical variables are velocity of nerve conduction (e.g., in meters per second), response amplitude, and excitability, which is usually a measure of the time required before a nerve can generate another impulse (refractory period). Any long-term change in peripheral nerve function is likely to be accompanied by lesions observable with a light microscope. Although chemical-induced changes in the velocity of nerve conduction are generally rare, they are most likely to occur with decreases in body temperature and are associated with severe behavioral dysfunction. Changes in amplitude are often observed after exposure to neurotoxicants, and can be associated with sensory or motor deficits. Changes in excitability are likely to cause altered sensory thresholds, altered behavioral reaction times, and altered susceptibility to seizures. Nerve-conduction responses depend on temperature, and studies should control for this variable.
The utility of maximal motor or sensory nerve-conduction velocities depends on the neuropathologic expression of neurotoxicity. Exposure resulting in demyelination markedly reduces maximal conduction velocity. In carbon disulfide neurotoxicity, sensory conduction velocities declined earlier than motor conduction velocities (Vasilescu, 1976). Similarly, neuropathies resulting from γ-diketone precursors (n-hexane and methyl-n-butylketone) and lead produce paranodal or segmental demyelination that is manifest in diminished conduction velocities (Spencer et al., 1975; Korobkin et al., 1975; Buchthal and Behse, 1979). Toxic neuropathies with axonal degeneration as hallmarks exhibit mild or no decrease in maximal conduction velocities; neuropathies resulting from γ-diketone precursors (LeQuesne and McLeod, 1977), organophosphates (Hierons and Johnson, 1978; Shiraishi et al., 1983), or acrylamide (Fullerton, 1969; Takahashi et al., 1971) are in this category, as are mild neuropathies caused by methylbutylketone (Allen et al., 1975).
Electrodiagnostic tests that rely on amplitudes of compound action potentials (muscle or sensory nerve action potentials) provide sensitive indexes of peripheral nerve abnormality. The tests depend on the number and synchrony of conducted impulses, which are reduced when there is a loss of contributing fibers or a dispersion due to conduction slowing. Sensory nerve action potentials are the more widely studied and are reportedly decreased in intoxication with acrylamide (Takahashi et al., 1971), Dipterex (Shiraishi et al., 1983), or methylmercury (Murai et al., 1982).
Sensory Evoked Potentials
Evoked-potential methods assess the neurophysiologic response of a particular sensory system to a particular stimulus (e.g., light or sound) and can help to detect and characterize neurotoxicity (Mattsson et al., 1989). The most commonly measured end points are peak latency (time from the onset of the stimulus to the peak response) and peak amplitude. Changes in latency imply changes in conduction velocity or in synaptic transmission. Acute exposures are likely to change latencies by changing body temperature. Long-term changes in latency might actually reflect alterations in conduction, which could be produced by alterations in myelination. When such changes are interpreted to reflect central, rather than peripheral, nervous system involvement, they might indicate an irreversible effect.
Changes in amplitude of peaks can imply changes in the number of nerve cells. Several peaks are averaged across numerous trials (stimulus presentations), and a change in average peak magnitude might be associated with an increase in variability of measurement. The importance of an amplitude change depends on the particular peak and sensory system under investigation. Changes in the evoked potential are likely to be associated with agent-induced sensory changes, which can be confirmed at the behavioral level.
Sensory evoked-potential techniques are used extensively in human neurotoxicity studies and are adapted easily to laboratory animals. The biologic basis of the electro-physiologic response is generally well understood and can be collected relatively quickly (Mattsson et al., 1989). The utility of sensory evoked potentials in neurotoxicity evaluation has been demonstrated, for example, in the case of toluene-induced auditory impairment (Rebert et al., 1983; Mattsson and Albee, 1988) and hexachlorophene-induced somatosensory dysfunction (Mattsson et al., 1989). However, the time and expertise required to implant electrodes and interpret the data and the recording equipment required reduce the likelihood that this technique will be used routinely for identification of chemical hazards.
Electroencephalography
Electroencephalographic (EEG) analysis is used widely in clinical settings for the diagnosis of neurologic disorders and less often for the detection of subtle toxicity-induced dysfunction. However, it is well known that dissociation between the EEG pattern and behavior can occur, particularly in chemical-treated animals. Changes in the pattern of the EEG can be elicited by stimuli that produce arousal (e.g., light and sound), by normal sleep, and by anesthetic drugs. In studies with toxicants, changes in EEG pattern can precede alterations in other objective signs of neurotoxicity. Experiments with the EEG must be done under highly controlled conditions, and the data must be considered case by case. EEG abnormalities have been noted in patients exposed to carbon disulfide (Seppalainen and Haltia, 1980) and alkylmetals (Cossa et al., 1959; Fortemps et al., 1978).
Neurochemical Procedures
Functions within the nervous system depend on synthesis of, release of, and receptor activation by specific neurotransmitters among specific groups of neurons. Many neurochemical end points could be measured in neurotoxicologic studies, including effects on neurotransmitters (e.g., changes in synthesis, transport, storage, release, re-uptake, or degradation of serotonin, norepinephrine, acetylcholine, or amino acids) and their receptors; effects on lipids, glycolipids, glycoproteins, or other constituents of neural membranes; effects on ion channels or membrane-bound enzymes that regulate neuronal activity; and effects on metabolic processes necessary to maintain neural activity (e.g., changes in glycogen and creatine phosphate concentrations, glucose availability, and mitochondrial structure and function). The large number of possible neurochemical effects is not amenable to the development of a battery of neurochemical tests to screen large numbers of chemicals. Results of in vivo tests at the behavioral or neurophysiologic level can suggest mechanistic hypotheses to test with biochemical and neurochemical end points (Mailman, 1987).
Neurochemical effects of neurotoxicants have been investigated in many laboratories (WHO, 1986; Mailman, 1987). Those of lead, for example, have been extensively studied in animals (Silbergeld and Hruska, 1980; Winder and Kitchen, 1984). Attempts have been made to use the results of these studies to further the understanding of lead-induced neurotoxicity in humans. For example, Silbergeld and Chisolm (1976) studied monoamine metabolites in urine of children exposed to lead and found a correlation between blood lead content and 24-hour urinary excretion of the dopamine metabolite homovanillic acid (HVA). HVA was measured before initiation of chelation therapy and within a week after the children had been removed from lead-contaminated environments. Over the long term, urinary HVA content was reduced, as was blood lead content, although both blood lead and urinary HVA remained higher in treated lead-exposed children than in age-matched controls. In contrast with dopamine, some neurotransmitters reported to be altered by lead in animal models—such as GABA and enkephalin—are less amenable to clinical measurement, because they require an invasive procedure to collect cerebrospinal fluid.
Neurochemical procedures appear to have great potential in development of biologic markers of neurotoxicant-induced exposure. The organophosphorus (OP) compounds have stimulated development of such biologic markers; an erythrocyte acetylcholinesterase assay has been used to monitor human exposure to OP compounds that cause cholinergic poisoning, and assays of lymphocyte neurotoxic esterase (NTE) activity have been used to detect exposure to OP chemicals that cause delayed neurotoxicity.
Recently, hemoglobin adducts (such as the lysyl amino groups of hemoglobin modified by γ-diketones) have been evaluated as possible biologic markers of exposure to some neurotoxicants. Because of the long lifetime of the erythrocyte (120 days in humans) and because erythrocytes become more dense as they age, density-gradient centrifugation of blood samples allows differentiation between recent and earlier exposure to the neurotoxicant. The same sample of blood can be used to detect protein cross-linking, because spectrin dimers eluted from membranes of lysed erythrocytes will be stable if cross-linking occurred in vivo (St. Clair et al., 1988).
A variety of neurochemical probes of neuronal and glial integrity could be used to evaluate the possible neurotoxic effects of a candidate chemical in experimental animals (O'Callaghan, 1988). As discussed with regard to in vitro systems, these neurochemical assessments could deal with very general indicators, such as total protein or DNA, in a given region of the nervous system. Alternatively, more specific indicators—such as transmitter-related enzymes, various receptors, or the state of phosphorylation of axonal proteins—could be compared in control and experimental groups of animals. One much-used assay detects glial fibrillary acidic protein (GFAP). The immunoassay for GFAP revealed areas of neuronal damage detectable with silver stain but not with routine neuropathologic studies. Agents and conditions that increase brain GFAP include trimethyltin, triethyltin, methylmercury, cadmium, MPTP, stab wounds, and aging. The use of GFAP as a sensitive and specific markers of central neuronal damage is simple and provides quantitative data that can be useful in risk assessment. The choice of indicators should be made in the light of any structure-activity relationship or other information that suggests an optimal test strategy for a given agent.
Neuroendocrine Interactions
Neuroendocrine toxicology is an emerging field. Despite the relative paucity of formal studies, suspicion that toxicants can affect neuroendocrine processes has been voiced since the earliest case studies of poisoning. For example, reports of lead poisoning in historical times comment on the short stature of children exposed to lead (Schwartz et al., 1986). Similar observations have been made in cases of infantile methylmercury intoxication (Matsumoto et al., 1965; Takeuchi, 1977). Anecdotal reports suggesting neuroendocrine toxicity have appeared in conjunction with pesticides, drugs of abuse, medicinal agents, and industrial residues. However, there has been no systematic review of neuroendocrine toxicity in conjunction with any of those classes of chemicals.
Neuroendocrine toxicology should receive greater attention than it does now. Evidence is growing that endocrine organs are important targets for many toxicants and that deleterious effects have not been recognized, simply because no one has looked for them. For instance, prenatal exposure to TCDD at very low concentrations alters sexual behaviors in male rats by an endocrine-dependent mechanism related to early hormone imprinting of the CNS (Moore et al., 1991). It is important to note that, because damage to the neuroendocrine system might be expressed as dysfunction or injury at the target or end organ, the neural contribution to the effect could be misinterpreted. For example, some chemicals described as reproductive toxicants exert their primary effect on the pituitary-reproductive axis (NRC, 1989b). Similarly, growth effects of teratogens can result from injury to the hypothalamic neurons that control release of growth hormone from the pituitary, rather than from direct effects of the agents on somatic tissues (Rodier et al., 1991).
Many areas of the brain responsible for regulating hormone production or specific hormonal response are very small and specialized. Groups of only several thousand or even hundreds of cells can be responsible for major endocrine functions. Although the size of those specialized regions in the brain makes them difficult to identify during routine histologic assessments, their dysfunction can have major consequences. The regulation of the endocrine system has been well characterized, so routine techniques are available to identify endocrine perturbations and to pinpoint their sites of origin; these should be more widely used.
The ability to detect disturbances in the neuroendocrine axis by provocative challenge, by radioimmunoassays of plasma from peripheral blood, or by end-organ evaluation has implications for risk assessment. The techniques are somewhat less invasive than autopsy or brain biopsy, are more quantitative than many behavioral procedures, and (under optimal circumstances) can pinpoint sites of neurotoxic injury with a high degree of specificity. The neuroendocrine system is an interactive axis involving many structures, and a toxicant can act on end organ, brain, or both.
Neuropathology
Neurologic lesions can be classified according to their characteristics or site of action. Lesions can be classified as neuronopathies (changes in the neuronal body), axonopathies (changes in the axons), myelinopathies (changes in the myelin sheaths), dendropathies (changes in the dendrites), and peripheral neuropathies (changes in the peripheral nerves). For axonopathies, a more precise location of the changes should be described (i.e., proximal or distal; central or peripheral).
In general, chemical effects lead to two general types of primary alterations: the accumulation, proliferation, or rearrangement of structural elements (e.g., intermediate filaments and microtubules) or organelles (e.g., mitochondria); and the breakdown of cells, in whole or in part. Partial cellular breakdown can be associated with regenerative processes that can follow chemical exposure.
Most neurotoxic damage is evident at the microscopic level, but gross changes in structure can be reflected in significant changes in the weight of the brain. Weight changes, discoloration, discrete or massive hemorrhage, and obvious lesions are clear indicators of adverse effects.
Careful histologic analyses form the cornerstone of our knowledge of toxic neuropathies that alter nervous tissue structurally, and the changes can be unambiguously correlated with clinical outcomes and with aging. Many structural changes have been identified neurohistologically. Methylmercury produces selective granule cell loss in the cerebellum (Hunter and Russell, 1954) and axonal degeneration in sensory ganglion cells (Cavanaugh and Chen, 1971). Degeneration of the long sensory and motor axons in both the central and peripheral nervous systems is produced by OP compounds, acrylamide, and 2,5-hexanedione (Cavanaugh, 1964; Spencer and Schaumburg, 1977a,b). Myelin loss has been seen as a result of diphtheria with no axon loss (Cavanaugh and Jacobs, 1964) and as a result of exposure to lead with only slight loss (Fullerton, 1966); myelin vacuolization in the CNS and PNS has been caused by hexachlorophene (Towfighi et al., 1974) and in central neurons by triethyltin (Aleu et al., 1963).
A careful examination of properly preserved and prepared tissues (which are generally available only from experimental animals) establishes or rules out structural damage, identifies the most vulnerable sites within the nervous system, and traces the temporal evolution of pathologic changes. The nervous system can be systematically assessed by sampling its various areas or particular tissues, as recommended by different expert groups (Tables 4-7 and 4-8). Table 4-7 reflects a recommendation by the World Health Organization (1986), and Table 4-8 contains a somewhat expanded sample of tissues believed adequate to identify most known disease states and vulnerable areas of the nervous system with light microscopy (Spencer et al., 1980). As the number of anatomic regions assessed increases, so do the labor and resource requirements of the assays. Sampling all regions listed in Tables 4-7 and 4-8 could be quite costly. Fuller characterization of tissue damage and identification of mechanisms of effect at the structural level often require electron microscopy guided by light microscopy (WHO, 1986).
TABLE 4-7
Areas of the Nervous System to be Used in Neuropathologic Evaluation.
TABLE 4-8
Tissues of the Nervous System to be Used in Neuropathologic Evaluation.
Imaging Procedures
Available neuroimaging procedures to assess structural changes in the brain can be divided into three major categories: computed axial tomography (CAT), magnetic resonance imaging (MRI) and nuclear magnetic resonance spectroscopy (MRS), and positron-emission tomography (PET).
The first two categories of neuroimaging procedures can be carried out with current clinical techniques. The CAT scan combines computerized imaging techniques with radiologic tomography to develop a detailed image of the brain. Because the technique is noninvasive, it has gained widespread clinical acceptance in the last 15 years. Contrast agents can be used to enhance visualization of some abnormalities, such as tumors or vascular malformations.
The newer MRI technique is similar to CAT, but, rather than using ionizing radiation, recreates anatomic information with nuclear magnetic moments. It requires a high-intensity magnetic field, and it monitors changes in nuclear magnetic moment relaxation times after the magnetic field has been applied. MRI is unique in that it uses three kinds of data to characterize anatomic information: nuclear-specific data, i.e., data specific to the nuclear magnetic moment; imaging-specific data, i.e., data specific to the way MRI is being performed; and tissue-specific data, i.e., data specific to characteristics of the tissues that are being imaged. The three kinds of data provide some of the unusual power of MRI to define precisely the pathophysiology of CNS disease. The procedure also provides a detailed image of the brain and might be even safer than CAT, in that there is no exposure to radiation.
The use of nuclear MRS can provide a more detailed evaluation of the biochemical status of the CNS. The kinds of biochemical information obtained from nuclear MRS include rates of energy generation and use, substrate metabolism, blood flow, and tissue integrity. Taken together, the uniqueness of the anatomic and biochemical data provided by MRI and MRS suggests that the combination of the two techniques will become a powerful adjunct to neurotoxicology both in humans and in experimental animals.
PET is potentially much more sensitive. It represents a major advance in the ability to monitor some types of neurotoxicity by labeling compounds with positron-emitting elements (e.g., manganese and fluorine) so that it is possible to observe alterations in neurochemistry, as well as in receptors in the brain of living subjects. Before the introduction of PET scanning, such examination was possible only in postmortem material.
The potential value of PET was recently recognized in the study of a group of heroin addicts exposed to MPTP. A number of addicts mistakenly took synthetic heroin contaminated with the dopaminergic neurotoxicant MPTP; some became severely parkinsonian, but many did not. By using positron-emitting labeled 6-fluorodopa, striatal dopamine deficiency was shown in those who were exposed to MPTP, but were not symptomatic (Calne et al., 1985]. That finding has already led to the hope that similar techniques will make it possible to predict idiopathic Parkinson's disease.
Another PET scanning technique involves the evaluation of cerebral glucose use. Scanning revealed decreased glucose metabolism in the brains of people poisoned by ingesting mussels contaminated with domoic acid (Teitelbaum et al., 1990). In patients with Alzheimer's disease, some areas of the cerebral cortex use less glucose (Friedland et al., 1989). Toxicants, whether their effects are selective or diffuse, are likely to decrease oxygen consumption and glucose use at their target sites. That might provide a way of monitoring the physiologic consequences of exposure to a wide variety of neurotoxicants.
The potential applications of PET procedures are enormous. If a neurotoxicant damages one or more areas of the brain, causing a change in neurotransmitter concentrations, it might be possible to measure the effects of the toxicant early, while they are still subclinical. The techniques can be used in experimental animals for longitudinal studies, because it is not necessary to kill them at different times to follow the evolution of neurotoxic effects. Ligands have been developed to image various receptor populations throughout the brain. Thus, it could eventually be possible to examine the effects of neurotoxicants on a wide variety of neurotransmitter systems and their corresponding receptor populations in living subjects.
Both CAT and MRI techniques can be used to demonstrate gross structural alterations in the brain. They might be useful in defining any neurotoxic process that causes substantial atrophy in the CNS. A specific example is the cerebellar and cortical atrophy that can occur after exposure to organic solvents—glue sniffer's encephalopathy. The limitation of the procedures is that they require major, if not massive, loss of neuronal populations before atrophy is visible. Other clinical measures would probably demonstrate abnormalities much earlier in the course of neurotoxic exposure.
Imaging procedures are not useful for assessing spinal-cord, nerve, and muscle damage. They are also expensive, and there is only a small data base available on their use in neurotoxicology.
Modifying Variables
Not all people exposed to a given neurotoxicant respond similarly. The determinants of susceptibility are not always well understood. Some variables, however, are known to modify responses to neurotoxic agents and must be taken into account in screening and any testing at higher tiers. The importance of age at exposure and at testing has already been mentioned in Chapter 2. This section indicates how age, sex, and genetic background of subjects should be considered in neurotoxicity testing.
Age at Exposure
Organisms can differ in sensitivity to neurotoxicants at different ages. Developing organs, for example, resist damage from low oxygen concentrations (a condition that is normal in utero). But, because of their high mitotic activity, they are more sensitive to antimitotic agents, such as x rays. Some agents (such as diazepam) cause temporary effects in the mature CNS, but permanent effects in the developing CNS (Kellogg, 1985); others (such as methylmercury) are toxic at every age, but toxic at lower doses in developing organisms (Spyker, 1975). Some differential effects can be explained on the basis of pharmacokinetics, but that is not the case for any of the agents discussed above. Thus, it is never safe to assume that effects observed in adults can be simply extrapolated to developing subjects. Developing organisms are not always more sensitive to toxic agents than adults; they are sometimes qualitatively different in their responses, because they are biologically different.
The difference between developing animals' and mature animals' responses to toxicity in other body systems is already recognized in regulations that require developmental exposure as a separate part of toxicity testing. The teratology guidelines focus on malformations and provide no information on the integrity of the nervous system when injuries are less obvious than anencephaly or exencephaly; greater attention to neurologic effects after developmental exposures is needed. Guidelines being written for the Environmental Protection Agency are based on recommendations from several groups and on experience with various test batteries.
Butcher and Vorhees (1979) proposed a neurobehavioral test battery for developing animals. Tests were selected to reflect the guidelines for assessing reproductive toxicity then in force in Japan, Britain, and France. The battery was validated with vitamin A, a classic behavioral teratogen (Butcher and Vorhees, 1979). The Butcher and Vorhees battery was later expanded by Vorhees and others to include tests on pivoting, olfactory orientation, auditory startle, neonatal T maze, figure-8 activity, M-shaped water maze, open-field grip strength, midair righting, rotorod, crossing parallel rods, jumping down to home cage, elevated maze learning, and E-shaped water maze (Adams, 1986).
The National Center for Toxicological Research (NCTR), which serves as a research arm of the Food and Drug Administration (FDA), conducted a major project known as the NCTR Collaborative Study. It was designed to demonstrate the reliability of classical behavioral-teratology tests (Table 4-9) when administered in five laboratories. The tests proved to be highly reliable between and within laboratories, and the effects of methylmercuric chloride on auditory startle were identified in all laboratories (Buelke-Sam et al., 1985).
TABLE 4-9
Tests Used in NCTR Collaborative Study.
Developmental neurotoxicology is a relatively new and important field. Because functional development of offspring can be influenced by maternal toxicity or chemical-induced growth deficits, care must be taken in the interpretation of data from developmental-neurotoxicity studies. Whether developmental neurotoxicologic assessments should be done at the hazard-identification or hazard-detection phase is being debated. Certainly concerns raised about the ability of a chemical to affect reproduction and development would apply to the postnatal growth and function of organisms exposed during development.
Age at Testing
Age at testing is an important modifying variable, for several reasons. When injury occurs early in life, it might damage systems that are not fully mature and thus not be fully expressed until later in life. The classic demonstrations of that phenomenon are in monkeys with lesions, in which some deficits take as long as 2 years to express their full effect (Goldman, 1971). It is not a characteristic of the damage itself, but of the organism on which it is imposed. Thus, just as antimitotic injury from radiation to the hippocampus during gestation will not lead to hyperactivity until after puberty (Rodier et al., 1975), a mechanical lesion of the hippocampus will have a delayed effect (Isaacson et al., 1968). The decrease in norepinephrine content and turnover in the hypothalamus that follows fetal exposure to diazepam is minor at 30 days in the rat, when the norepinephrine system is immature, but greater than 50% at 90 days, when the system is mature (Kellogg, 1985).
A separate phenomenon is the increasing effect of lesions with aging—initially suggested to describe performance after neonatal x irradiation, which differed from performance in controls in preweaning animals, was near normal in young adults, and then differed again as animals aged (Wallace et al., 1972). Brain weights of treated subjects were always low, but the difference from control subjects became more pronounced with advancing age. The same pattern is suspected to account for the late onset of parkinsonian dementia after early exposure to cycad products. It is possible that a natural decline in function uncovers a longstanding deficit.
It is not easy to introduce aging as a variable in screening tests, because lifetime studies in animals are expensive. More basic research is needed to determine how often a toxic insult interacts with aging. In the meantime, it is reasonable to hold test animals until maturity, rather than doing short-term studies. Although short-term studies might be appropriate for teratology experiments aimed at gross structural changes, they preclude evaluation of normal adult function of the nervous system.
Sex
Toxicity to sexually dimorphic structures can be sex-specific, and the brain is in some respects a sexually dimorphic structure. A few examples of how the effects of some toxic agents on the nervous system might depend on sex are known, although not necessarily understood. Developmental effects of methylmercury on motor tests were more pronounced in boys than in girls in studies of a fish-eating population (McKeown-Eyssen et al., 1983), and neonatal male mice show more mitotic arrest and later cerebellar loss in response to methyl-mercury than do females, even though they do not have higher concentrations in the brain after being treated (Sager et al., 1984). Some sex differences might be due to pharmacokinetic differences, rather than differences in the target tissue, but that does not weaken the point that it is reasonable to test both sexes for effects.
Genetic Differences
Specific genetic differences in response to toxic agents have not been commonly demonstrated, and without some knowledge of mechanisms it is unlikely that susceptible populations can be identified in the screening of chemicals. However, when such populations are known, they should be evaluated. The classic human example is the case of the recessive gene for metabolism of the amino acid phenylalanine. People who are homozygous for the normal allele can clear excess phenylalanine; those who are homozygous for the mutant are severely brain-damaged by postnatal exposure to the amounts of phenylalanine found in a normal diet. Heterozygotes fall between the extremes in their metabolic capacity and should be sensitive to high concentrations of phenylalanine, but not harmed by the amount in a normal diet. Some toxicologists have been concerned about exposure of this group to the phenylalanine sweetener aspartame. Because most heterozygotes are unaware of their status with regard to this gene, they cannot avoid products that might be hazardous to them.
Current Regulatory Approaches
The Organization for Economic Cooperation and Development (OECD) (Brydon et al., 1990) and other organizations (Tilson, 1990b) have recommended simple and rapid observations of neurobehavioral function of experimental animals in their home cages or in large open fields in protocols for many subchronic and chronic tests. Such observations include those of naturally occurring behaviors (e.g., rearing, stereotypy, and urination) and procedurally easy and quick assessments (e.g., of click response, landing foot splay, and righting reflex) (Moser, 1989). Other home-cage testing protocols involve more elaborate 24-hour recording of naturally occurring behaviors (Evans, 1989). It is not clear how effective such relatively general and nonbinding instructions have been in fostering the detection of neurotoxicity in the context of more general toxicity testing. Moreover, such approaches do not provide for the detection of alterations in complex behaviors, such as learning and motor performance, inasmuch as their testing requires special instruments and, depending on the test, training of the animals.
No U.S. federal laws are directed specifically at protecting humans from systemic toxicity, much less from neurotoxic hazards specifically, but regulatory agencies have considerable latitude to emphasize specific health effects (Fisher, 1980; Reiter, 1987; Sette, 1987; OTA, 1990). With the exception of a test for delayed neuropathy in hens required by EPA for OP esters (Federal Register, 1978), there are no specific regulatory requirements to evaluate neurotoxicity in any current federal guidelines or regulations.
The testing requirements for neurotoxicity vary among the regulatory agencies. Pre-market testing is required for pesticides by EPA under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and for drugs and food additives by FDA, but not specifically testing for neurotoxicity. The Consumer Product Safety Commission and EPA may require testing of consumer products and industrial chemicals, respectively, but only if they can justify the need for such testing. In practice, neurotoxicity testing is rarely required, but is usually left to the discretion of the manufacturer.
Until passage of the Toxic Substances Control Act (TSCA) in 1976, there was no legal mechanism in the United States for prospective evaluation of the neurotoxicity of new industrial chemicals (OTA, 1990). Under TSCA (Section 5), there must be some indication that a ''new'' chemical might be neurotoxic before EPA can require specific tests in the premanufacturing notice (PMN) program, but then EPA can prohibit the chemical from entering into commerce until the required data are available. Generally, determination of whether additional testing is needed is based on a structure-activity analysis, because about 50% of PMNs are submitted with no toxicity data and the toxicity data that are submitted are rarely more than the results of acute lethality or irritation tests. Of the 6,120 PMNs submitted in 1984–1987, 1,200 underwent additional review (about half these were found to be associated with unreasonable risk); of those identified for additional review, 220 were suspected of being potentially neurotoxic (OTA, 1990). The SAR-based approach is now regarded as generally insensitive and problematic. Moreover, many thousands of potentially neurotoxic compounds developed before passage of TSCA remain untested. Except in unusual cases, the chemicals were simply registered with EPA and added to the TSCA inventory. For "old" chemicals (TSCA, Section 4), EPA can issue a test rule and require testing only if neurotoxicity is suspected, but little progress has been made in establishing final requirements for testing such compounds or completing evaluation of what new data have been generated (GAO, 1990a,b). Proposed test rules or consent decrees for 19 chemicals or chemical classes have included requirements for neurotoxicity testing (OTA, 1990). FIFRA also has a provision for the retroactive reassessment of the toxicity of active ingredients in existing pesticides, but this program is progressing slowly (OTA, 1990).
For assessing suspected neurotoxicants, EPA's Office of Toxic Substances has issued a series of test guidelines (Federal Register, 1985). They include guidelines for testing motor activity and a functional observational battery. Schedule-controlled operant behavior, neuropathology, peripheral nerve function, and organophosphate-induced delayed neuropathy are also addressed. A testing guideline for developmental neurotoxicity is pending. Those guidelines are generally intended to be generic guides as to how to make particular measurements, rather than lists of specific techniques. Data generated with the guidelines are used in conjunction with other toxicity data to establish potential risks and to establish reference doses.
FDA follows a pattern similar to that of EPA for assessing potential neurotoxicity. Therapeutic agents are assessed in preclinical (animal) and clinical (human) studies requiring careful, but unspecified, observations (Fisher, 1980). Specific neurotoxicity tests are required only for drugs that are intended to be neuroactive; for other drugs, a review for neurotoxicity is undertaken only if such effects are suspected (OTA, 1990). FDA requires neurotoxicity testing of food additives only if neurotoxicity is suspected for a particular chemical on the basis of evidence developed in traditional toxicity testing. However, FDA requires testing when a stated "level of concern" is reached on the basis of anticipated exposure, potential toxicity, or structure-activity relationships. A general histopathologic assessment of several sections of the brain, spinal cord, and peripheral nerve is required, with notation of any observed behavioral abnormalities in the test animals. Positive evidence that is revealed by the screening tests might, at the agency's option, lead to further assessments, whose nature is decided case by case. In fact, only instances of neuropathy or clinical signs of neurotoxicity—such as paralysis, tremor, or conclusions—have an impact on FDA decisions (Sobotka, 1986).
Rodent reproductive studies, which might permit the observation of developmental problems, are ordinarily conducted for drugs that will be administered to women of child-bearing age (OTA, 1990). On the whole, however, current tests for developmental toxicity (particularly at FDA) are designed primarily to detect structural malformations or fetal death, which are appraised after sacrifice before delivery. Even in the short-term Chernoff-Kavlock protocol (1982) and in the standard protocols for reproductive and developmental effects, the pups are sacrificed before they acquire much behavior, so opportunities for observation of behavioral teratogenicity and other developmental deficits are not provided.
Other countries have explicitly prescribed behavioral tests in animals to screen for developmental neurotoxic effects of new drugs that possibly will be used by pregnant women. To allow a new drug on the market, Japan requires a fertility study, a teratology study with female exposures, and a perinatal study with exposures during the last third of gestation and through lactation. In the latter two studies, some offspring are examined for "development (including behavioral development)." Great Britain also provides general specifications, requiring tests for "auditory, visual, and behavioral impairment" for premarket screening of new drugs. The European Economic Community guidelines require assessment of "auditory, visual and behavioral impairment'' (Vorhees, 1986).
The testing protocol of the National Toxicology Program (NTP) provides for fairly extensive neuropathologic assessment, but relatively little functional observation. NTP tests have detected the neurotoxic effects of sodium azide, for instance, because it caused neuronal destruction in chronically treated rodents. However, it is generally accepted that finding mild neuronal degeneration, at the light microscopic level, in the absence of functional or biochemical information, is unlikely to be a sensitive means of detecting neurotoxicity.
Strategies for Improved Neurotoxicity Testing
Screening
Several current proposals for generalized approaches to neurotoxicity testing of new and existing chemicals recommend a three-tiered scheme (Moser, 1989; Tilson, 1990a; OTA, 1990). The first tier (the screen) consists of tests for the existence of a neurotoxic hazard. If no evidence of such a hazard is found within the first tier, toxicity testing ceases. Positive findings in the first tier would terminate the development of many chemicals being investigated for commercial applications, raising the dilemma of false positives discussed later. For other apparently neurotoxic chemicals (those already in use and those whose commercial development is nonetheless deemed promising), neurotoxicity testing would continue to the second tier.
The experiments undertaken in the second tier would characterize the nature of a substance's neurotoxicity, e.g., its specific target tissue and dose-response and time-response relationships. The selection of specific experiments in the second tier would be directed by the results of earlier testing. Preliminary second-tier dose-response assays would establish the potency of the substance for the adverse neurotoxic end points identified in the first tier. Comparison with the chemical's potency for other types of toxicity and with estimated magnitudes of human exposure is likely to determine whether extensive additional effort is put into characterizing its neurotoxicity or whether it progresses to the third tier for investigation of its mechanisms of action in detail. If other effects would predominate or no neurologic consequences were likely to ensue from plausible exposures, research resources would be better directed toward other questions.
Testing in the third tier would aim to specify the mechanism by which the chemical produces its neurotoxic effects. This type of detailed investigation would be appropriate for substances with broad exposure that produce serious effects (e.g., lead), substances with low or rare exposure that produce a particular noteworthy effect (e.g., MPTP), or series of substances that produce lesions of concern (e.g., demyelination). The information obtained in the investigations would be of use in attempting to control the most serious existing exposure problems and in establishing SARs for the more efficient prediction of neurotoxicity of chemicals considered in the future.
Despite an apparent simplicity in the design of such a tiered strategy, implementation is not easy; many questions remain. Determination of the component tests of the first tier is the most crucial aspect of the three-tiered approach, because truly positive substances will not continue to later tiers unless detected at this point, and the course of substances that are studied further will tend to be tailored to their performance in earlier tests. The first tier is the heart of the screening aspect of neurotoxicity testing; the later tiers might produce data of great value in advancing understanding of neurotoxic processes and how the nervous system operates, but the first tier is the first line in preventing neurotoxic disease.
How are chemicals ranked for entry into the first tier? Does the first tier represent a single battery of tests to which all entering chemicals will be subjected, or is the flow of testing in the first tier itself directed by the profile of results? One way to proceed is to have an initial tier composed of a number of tests that are applied to all candidate substances. Substances that do not show signs of neurotoxicity are not tested further. Confidence in such a battery is a function of the number of tests in it, the quality of the tests, the size of the experiments (number of subjects, doses, durations of exposure, etc.), and the degree to which the tests examine a wide range of potential neurotoxic outcomes (learning, memory, developmental effects, etc.) or other biologic markers that may serve as surrogate measures for them. The sensitivity of a neurotoxicologic screen might be increased by including additional tests. Many of the tests were mentioned earlier. Inclusion of some of them might increase confidence in screening procedures, but at the price of consuming more resources. That suggests a tradeoff between testing an individual substance thoroughly and processing more substances. Tests that might be considered for addition to the screen in the future would have to be validated and their incremental value demonstrated.
The neurotoxicologic literature has many references to two-tiered and multitiered testing. In many instances, what is being discussed is the multistage nature of complete neurotoxic testing, i.e., the validation of the screening decision and the progression from screening (hazard identification) to experimental neurotoxic tests designed to establish quantitative dose-response relationships, mechanisms of action, etc. However, other authors appear to be discussing "multistaged screening" in the strict sense of testing for the purpose of hazard identification (tier one). If all steps of screening are required for every substance, then the plan is no different from a proposal for a greatly expanded comprehensive screening battery. However, if multistaged testing is required of only some substances, then it is important to specify the criteria for deciding which substances should pass beyond the initial level. Requiring additional levels or types of testing for substances in some chemical classes is, in effect, merely adding more tests to the first tier.
Some type of functional observational battery (FOB) has most often been proposed for the screening phase of neurotoxicologic assessment. It appears to be the simplest, most comprehensive option available and it, at a minimum, should be included in the first tier of neurotoxicity testing. Depending on the particular test, behavioral end points are often used in the detection phase of testing, because they may be global indicators of many of the sensory, motor, and integrative processes of the central and peripheral nervous systems. Observational batteries have been used in rodents by many investigators to assess chemicals for neurotoxicity and have been recommended by several expert panels (see Tilson, 1990b). The EPA Office of Toxic Substances prepared guidelines (EPA, 1985) for testing potential neurotoxicants that included an FOB consisting of several measures that are relatively simple, noninvasive, and quick to perform. A description of the FOB and a summary of responses produced by representative neurotoxicants were published recently by Moser and colleagues (Moser et al., 1988; Moser, 1989). Provision should be made for acute and repeated dosing and for observation of the animals over an extended period. It would also be desirable to measure motor activity and perform limited neuropathologic tests on the treated animals, as recommended by EPA (1985). Suggestive screening findings or structural alerts derived from knowledge of SARs should invoke a second level of tests, in which impaired acquisition of complex behaviors after exposure during development is used as a general indication of neurodevelopmental toxicity. In addition to tests covering the various aspects of neurotoxic effects of concern for every chemical, a screening battery might contain specific procedures used only for particular classes of chemicals, as is the case for testing OP pesticides for possible delayed distal neuropathy.
Validation
The procedures just described are the only general neurotoxicity methods whose validation approaches what would be necessary for a routinely applied screen. Before they could be adopted, additional or other tests would have to be thoroughly appraised for reproducibility and for relevance to human health. Their success in distinguishing the properties of an extensive battery of nonneurotoxicants and substances that act by a full spectrum of mechanisms to produce a given end point would have to be demonstrated. Establishing a control battery of appropriate chemicals is in itself a challenge, given the fairly limited extent to which the universe of chemicals has been tested for neurotoxicity. Ever-increasing experience might ultimately permit the substitution of a battery of short-term in vitro assays for longer procedures in whole animals. In addition to its screening capacity, such an in vitro battery would be expected to yield considerably more information about mechanisms of action than the whole-animal approach to screening with an FOB and thereby allow much more directed testing in the later tiers.
Priority-Setting and Implementation
The proposed implementation of large-scale screening for neurotoxicity has generated controversy about the appropriate strategy for approaching the first tier of testing, hazard identification. It is generally agreed that substances already present in the environment for which there is any indication of neurotoxicity in the form of structural alerts or preliminary observations should be evaluated thoroughly. Similarly, in the case of chemicals being considered for future use, such signals of possible neurotoxicity should trigger rigorous testing, if commercial development is to be pursued. When there are few or no data on which to base a judgment as to whether a substance (new or in use) presents a neurotoxic hazard, the estimated volume of production and expected pattern of human exposure (concentrations, frequencies, and numbers of people) are the most reasonable bases on which to set priorities and begin screening.
As reviewed by the Office of Technology Assessment (1990), federal regulations provide the regulatory agencies with the authority to demand that toxicity testing, including tests for neurotoxicity, be performed on chemicals as a requirement for registration or to establish standards for continued use. Federal agencies have not exercised that authority often, nor followed through completely in analyzing the data that have been generated (GAO, 1990a,b). Although it might seem obvious that the sponsors of an innovative product should be responsible for testing it adequately, the question remains of who should be responsible for conducting and financing the necessary testing on substances already widely present in the environment. It would be desirable if the burden of neurotoxicity screening not only could be divided and coordinated among U.S. industry, government, and academy, but also could involve an international effort, as is being encouraged by OECD (Brydon et al., 1990) for general toxicity testing.
Research Needs
An important question is how in vitro systems can contribute importantly to the neurotoxicologic evaluation of a broad range of agents. Research is needed to determine whether culture systems can reliably identify known neurotoxic agents and active related compounds and distinguish them from related, but inactive (or less active) compounds. Research is also needed to develop in vitro techniques that can identify neurotoxicants that require metabolic activation to be effective and those which do not, that affect either adult or developing organisms, and that are injurious after either acute or chronic exposures.
Empirical results of the many in vitro systems available need to be correlated with studies in whole animals (indeed, human populations) to establish the neurotoxicologic implications of findings from in vitro preparations. Substantial efforts will be needed to provide funding and organizational resources for that type of work; but the work is essential if reliable, efficient test systems are to be developed reasonably soon. In the long term, mechanistic studies are needed to increase understanding of the fundamental molecular targets of neurotoxic chemicals. Various schemes can be used to outline a more general approach (Table 4-10). Accumulated knowledge should eventually allow relatively efficient evaluation of agents that are likely to fall in one or another general category of neurotoxicant. A general model has emerged from developmental neurobiology that might similarly allow focusing of neurotoxicologic research with in vitro systems that capture one or another step in the progression from developing embryo to degenerating adult nervous system (Figure 4-1).
TABLE 4-10
Proposed Components for Evaluating In Vitro Neurotoxicity Screening Tests.
![Figure 4-1. Biologic markers in the stages between formation and degeneration of neural circuits.](/books/NBK234247/bin/img00008.gif)
Figure 4-1
Biologic markers in the stages between formation and degeneration of neural circuits. Selected markers are listed under the stages that are named in boxes.
Routinized screening and mechanistic studies will produce a growing body of detailed neurotoxicologic data on diverse chemicals that should be intensively analyzed. Ultimately, study of the data should reveal SARs that are more generally useful than the few that are now considered well established and that will be much better substantiated and more reliable than most current conjectures.
Much of the controversy over proper testing procedures arises from two intrinsically conflicting objectives: minimizing the incidence of false positives (substances are incorrectly identified as hazardous) and minimizing the incidence of false negatives (substances are incorrectly identified as nonhazardous). Both goals are desirable, but they cannot be maximized simultaneously. Too high an incidence of false positives will waste our resources, worsen the already severe logistic problem of testing thousands of chemicals, and cause at least some potentially valuable chemicals to be discarded. In the screening situation, however, a high incidence of false negatives is probably more undesirable. Failure to detect a truly neurotoxic substance will expose our society to health hazards, with the potential for tragic consequences like those associated with one false negative, thalidomide. Testing for characterization should expose false positives; the abandonment of an occasional new chemical on the basis of what are actually false-positive screening results is the likely cost of this process.
The expense and effort of neurotoxicologic screening would be very great if it were undertaken in isolation. Considerable practical advantages would accrue if neurotoxicologic screening were integrated with a multidisciplinary program of general toxicologic and teratologic screening.
- Testing for Neurotoxicity - Environmental NeurotoxicologyTesting for Neurotoxicity - Environmental Neurotoxicology
Your browsing activity is empty.
Activity recording is turned off.
See more...