Establishing Criteria for Chemical Neurotoxicity

At one time or another, we have all been guilty of making sweeping statements implying that certain groups of substances (e.g., metals, solvents, pesticides) have inherent neurotoxic properties. Such statements are not only inaccurate but also patently misleading. We need to develop much more critical criteria for neurotoxicity. The problem is illustrated immediately when a chemical is labeled a neurotoxin without any consideration of dose. Obviously, anticholinesterase chemicals have the potential for inducing dramatic neurobehavioral changes but, as Russell points out, the very same pharmacological properties have been exploited (without success, it is widely acknowledged) as symptomatic therapy for Alzheimer's disease. Thus, anticholinesterase chemicals are not neurotoxins but, at certain doses, exert neurotoxic effects. An even better example of the importance of dose is provided by the vitamin pyridoxine, well known to be an absolute requirement for normal metabolism. In this case, neuropathy follows both inadequate and megavitamin intake of Vitamin B₆. Obviously, pyridoxine is not a neurotoxin but, in sufficiently large dosage, the chemical does have neurotoxic potential. Similarly, lead, mercury, and acrylamide are not neurotoxins per se, but they are able to cause neurobehavioral changes at certain levels of exposure.

Once this point is understood, we are in a position to proceed to the question of connecting a specific chemical to one or more neurobehavioral effects. Our lack of precision here is equally as misleading. Consider the statement that acrylamide is a peripheral nervous system (PNS) neurotoxicant. Although it is certainly true that acrylamide has the potential to induce peripheral neuropathy in humans and animals, it is also apparent that this is a function of dose and duration of exposure. At certain doses, acrylamide is able to induce encephalopathy, with confusion, disorientation, memory disturbances, and hallucinations; at other doses, ataxia and dysarthria predominate. Thus, acrylamide is a chemical with neurotoxic potential capable of inducing a number of neurobehavioral effects that vary in nature with dosage and duration of exposure. An analogous situation exists with thallium, carbon disulfide, n-hexane, and a host of other chemicals with neurotoxic potential. Such considerations are more, much more, than semantic niceties; our use of language in describing chemical neurotoxicity is often so imprecise that it serves to mislead the public rather than to inform.

Our most dismal performance, however, is reserved for our failure to develop solid criteria that must be met before a chemical is identified as having neurotoxic potential in human subjects or is linked with a human neurological or developmental disorder. Stunning examples are provided by the diatribe over solvents and aluminum. In the case of solvents, where specific chemical substances are clearly associated etiologically with various types of neurological deficit, some have been prepared to indict and convict all chemicals labeled as solvents. Are we unaware that there are many different classes of chemicals which make up this heterogenous group of substances, and that the large majority of solvent chemicals has never been tested for neurotoxic potential? Apparently, this paucity of information has not prevented some from claiming solvent chemicals as causal of human dementia. Equally as disturbing is the rationale for continuing to link aluminum with the etiology of Alzheimer's disease. This idea began when neurofilamentous accumulations induced by aluminum were observed by light microscopy to be superficially similar to those seen in Alzheimer's disease. Somehow, the idea continued even though neurofilaments of experimental aluminosis were found by transmission electron microscopy to be identical to those induced by a host of other substances and quite distinct from the characteristic pairedhelical filaments of Alzheimer's disease. More fuel was added when dialysis dementia was linked to aluminum toxicosis, although the clinical features and neuropathology of dialysis dementia and Alzheimer's disease are distinctly different. Nevertheless, the idea of a link between Alzheimer's disease and aluminum toxicosis has become so entrenched that the public is encouraged to believe it. The (nonspecific) observation of aluminum (and other metals) in neurons and senile plaques in Alzheimer's disease and related disorders has further encouraged those faithful to the aluminum hypothesis. However, as Gerhard Winneke observes, the evidence supporting an association is (at best) circumstantial.

This discussion, of course, is intended neither to defend solvents—many of which may well prove to have neurotoxic potential once they are tested—nor to imply that aluminum has no etiologic relationship with Alzheimer's disease. Rather, it is a plea for the exercise of extreme caution in the best tradition of scientific conservatism before statements are made about cause-and-effect relationships between chemical substances and human neurobehavioral effects. More specifically, it is a call for the development of a set of guidelines that must be met before a chemical is accepted as causal of human neurobehavioral dysfunction. Identification of those chemicals that unequivocally fulfill criteria for a potential human neurotoxin is an essential first step in the development and validation of neurobehavioral test methods for chemical neurotoxicity. Such tests will require reference compounds linked to the various types of neurobehavioral deficit under study.

The Minamata tragedy, reviewed by Winneke, provides a graphic illustration of the steps that should be taken before a substance is accepted as causally responsible for human neurotoxic disease. Simply stated, the neurological illness seen in humans and cats was reproduced in the latter both by feeding methylmercury-contaminated fish and by administering authentic samples of the suspect chemical agent. Because pure methylmercury was able experimentally to induce in cats a disorder indistinguishable from that seen in the affected feline population of Minamata Bay, and the nature of the feline disorder clearly paralleled the attendant human disease, it was possible to make strong statements about cause-and-effect relationships between methylmercury and neurodegenerative disease. Thus, in the absence of human experimentation, the strongest foundation on which we can rest our case for chemical neurotoxicity is the experimental induction in animals of a disorder equivalent to that seen in human subjects exposed to the compound under scrutiny. Such substances, of which there are relatively few, constitute the list of chemicals with proven human and animal neurotoxic potential (class 1 chemicals). A much larger number of chemicals has been associated with neurobehavioral effects in exposed humans (class 4 chemicals), but these reports are often poorly documented exposures to mixtures of ill-defined substances that have not been subjected to experimental scrutiny either in isolation or as a mixture. Important exceptions are the recurring reports of consistent neurobehavioral deficits associated with certain therapeutic agents and mixtures, where the exact chemical compositions as well as the dose and duration of exposure are often well documented (class 3 chemicals). Our task, of course, is to help devise experimental methods that will detect neurotoxic effects of substances which can then be prevented from entering the marketplace or removed therefrom. Chemicals with neurotoxic properties recognized in experimental animal studies constitute a group of agents suspected to possess human neurotoxic potential (class 2). By far the largest group of substances, however, contains that multitude of untested chemicals which is not known to be linked to any human neurobehavioral disorder (class 5). Thus, in summary, at least five categories may be recognized in our efforts to link chemical substances with potential adverse effects on the human nervous system, the strength of the association diminishing as one descends through the list:

Class 1. Experimentally proven in animals, with similar effects in humans

Class 2. Experimentally proven in animals, with unknown effects in humans

Class 3. Effects observed in humans but experimental evidence unavailable

Class 4. Possible effects in humans but experimentally unproven

Class 5. Effects unknown in humans and untested in animals

Important consequences flow from this conservative classification. For example, in the case of industrial solvents, the majority of which falls into classes 4 or 5, there is no justification at the present time for stating that solvents as a class have human neurotoxic potential. On the other hand, because solvents are so widely employed in industry and so little is known of their chronic neurobehavioral effects, there is every reason to improve knowledge in this area. Another example is the so-called Spanish Toxic Oil Syndrome, an epidemic of a new and remarkable disease that affected thousands of people in Madrid and its surrounds. Although epidemiological studies strongly link the disorder to the consumption of an illicit cooking oil, until a viable model of the disease is produced in a laboratory animal fed the suspect agent, there is always room for a small degree of doubt.

Classifying Chemically Induced Human Neurotoxic Disorders

Russell argues convincingly that basic and clinical research designed to understand the neurochemical mechanisms by which exposures to toxicants affect behavioral indicators will lead us a long way toward a comprehension of the enormity of our task and the design of appropriate methods to detect, define, and even predict chemical neurotoxicity. In other words, we cannot simply devote our research time to the somewhat pedestrian task of developing test methods to fill the regulators' needs. With this in mind, therefore, it is appropriate to inquire how far we have come along the mechanistic path, where we need to concentrate research efforts to improve understanding, and how we should proceed in our attempts to identify the "atomic variety" of chemical time bomb. Our first chore, however, is to understand how to classify the adverse effects of chemical substances, first in the adult and then during brain development.

The Mature Nervous System as Chemical Target

Although a satisfactory scientific nosology of chemicals with neurotoxic properties in adult human subjects remains an elusive goal, it is possible to offer a surprisingly useful framework for the eventual development of a comprehensive classification system. Ideally, this should be able to link the target of chemical attack to alterations in neural function that explain observed neurobehavioral changes and, in the clinical setting, provide a logical basis for prevention. One such attempt is a 1984 classification of human neurotoxic response that was based on cellular and subcellular targets of chemical agents. This recognizes the following sites of functional disruption or damage: the neuron, glial cells and myelin, nervous system vasculature, and muscle. Points pertinent to the present discussion are readily made by selectively considering agents that act on the mature nerve cell.

Three broad types of neuronal change induced by chemical substances are recognized, namely, functional perturbations of the excitable membrane, interference with neurotransmitter systems, and structural breakdown of dendrite, perikaryon, or axon.

The first type involves direct chemical interference with the excitable surface membrane of neurons, the consequent changes in electrical transmission, and the associated generation of neurobehavioral alterations. By and large, these effects appear and reverse rapidly, and the extent of dysfunction reflects the distribution of the chemical within the central or peripheral nervous system. The best examples are provided by agents that interfere with the normal passage of sodium ions across the nerve cell membrane. Some substances, such as tetrodotoxin (from the puffer fish) and saxitoxin (paralytic "shellfish" poison), act as channel blockers, whereas ciguatoxin, scorpion and anemone toxins, DDT, and pyrethroid insecticides, act to increase membrane permeability to sodium while the membrane is in either the resting or the active state, or both. Although the neurobehavioral effects of these channel agents range from discomfort (circumoral and distalextremity paresthesias) to life-threatening dysfunction (respiratory paralysis), the salient point for the present discussion is that they are unlikely to have any long-lasting effects once the chemical has left the membrane receptor. Thus, although neurotoxicity may be fatal, agents of this type do not merit consideration as delayed-action chemical time bombs.

The second common locus of chemical attack upon the neuron is its neurotransmitter system, the proper regulation of which is often an absolute requirement for normal behavior. Chemicals may interfere with neurotransmitters at many levels, including their synthesis in the neuronal perikaryon, transport along the axon, packaging and release from synaptic vesicles, transport across the synaptic cleft, and reception by the target membrane, as well as the enzymatic breakdown or reuptake of excess transmitter in the nerve gap. Because these exigencies (points of vulnerability) exist for each and every neurotransmitter, the possible sites of chemical-induced perturbations are legion. Take, for example, agents active on cholinergic pathways, a subject discussed by Russell. Of course, the relationship between human behavioral changes and the selective actions of chemical agents on critical neurotransmitter systems extends far beyond the cholinergic system. Comparable examples may be drawn for chemicals active on central and peripheral catecholaminergic pathways. For example, monoamine oxidase inhibitors that increase the duration of action of synaptic catecholamines lead to mania, hyperreflexia, and involuntary movement. By contrast, the antihypertensive drugs tend to induce mental depression, weakness, and lethargy because they serve to deplete synaptic catecholamines. Other agents act on serotonergic pathways (lysergic acid diethylamide, LSD), gamma-aminobutyric acid (GABA) pathways (picrotoxin), glutamate pathways (beta-N-oxalylamino-L-alanine, BOAA), whereas some, such as the antipsychotics and opiates, influence a number of different pathways.

Most of the important adverse neurobehavioral effects of chemicals that perturb the function of neurotransmitter systems tend to appear rapidly, may be life threatening, and are generally considered to reverse without permanent sequelae. In this regard, many transmitter neurotoxins are as insignificant for our present thinking as compounds that only perturb excitable membranes. A notable exception is the persistent buccolingual-masticatory dyskinesia and choreiform movement of trunk and extremities (tardive dyskinesia) that often accompanies prolonged therapy with antipsychotic drugs. These involuntary movements are aggravated with emotional stress and concentration on motor tasks, and are intensified by the reduction or discontinuance of therapy, possibly because of the "unmasking" of supersensitive dopamine receptors. Chemically induced movement disorders also routinely accompany prolonged L-dopa (dihydroxyphenylalanine) therapy for parkinsonism. Although these two iatrogenic conditions probably account for a significant percentage of the disabling, chemical-induced neurological disease seen in developed countries, they have received little attention from neurobehavioral toxicologists concerned with environmental toxicity. However, for the neurologically and psychologically impaired patient who faithfully follows the doctor's prescription of L-DOPA or antipsychotics, the tardive appearance of these neurobehavioral effects may often produce an incapacitating or life-threatening state equal to or greater than the disease under treatment. A distinguished Spanish neurologist recently reported that fully 30 percent of his parkinsonian patients were found to have a drug-induced disorder that disappeared upon cessation of treatment. Thus, in addition to the mechanistic information available from such data, therapeutic agents also require much more rigorous neurobehavioral testing to predict and control the development of iatrogenic neurological and psychiatric states.

The different types of structural damage induced by chemical agents constitute for present purposes a very important third class of neuronal responses to chemical attack. Numerous chemicals and drugs are able to induce axonal degeneration without loss of the parent neurons; most require repetitive exposure. The neurobehavioral changes (usually peripheral neuropathy) develop rapidly or insidiously after weeks or months of exposure, and the disorders are reversible to the extent that damaged axons will regenerate and reconnect with sensory and motor targets in the peripheral nervous system. The clinical signs and symptoms develop in a distal, symmetrical, and temporally ascendant pattern in the extremities, with sensory loss and muscle weakness usually predominating in clinical significance over the commonly attendant autonomic dysfunction. Axonal neuropathy in humans, animals, or both, is known or (from the pattern of neurobehavioral dysfunction) suspected to occur with repetitive exposure to workplace chemicals (e.g., acrylamide, ethylene oxide, carbon disulfide, n-hexane, methyl n-butyl ketone, dimethylaminopropionitrile, certain organophosphates), therapeutic drugs (vincristine, chloramphenicol, thalidomide, disulfiram, isoniazid), household chemicals (thallium, arsenic, zinc pyridinethione), abused substances (ethanol), and natural toxins (buckthorn). Some are painful (thallium); others are associated with autonomic dysfunction (acrylamide), prominent pyramidal signs (leptophos, methyl bromide), or optic nerve changes (ethambutol). Although the molecular and cellular mechanisms underlying these toxic effects are unknown, the resulting pattern of neurobehavioral dysfunction is remarkably stereotyped and often readily studied in experimental species. Clinical experience demonstrates that recovery is usually slow, with sensory and motor deficits disappearing in the reverse order of their appearance. Recovery of sensation and strength is usually well advanced within months or years after cessation of exposure to the offending agent. Sometimes, when there has been extensive damage to ascending or descending spinal pathways, there may be permanent sensory loss, truncal ataxia, or pyramidal signs. The pathways involved in axonal neuropathies are also deleteriously affected with the normal advancement of age, but late-life appearance of neuropathy in previously recovered subjects is not recognized. In short, the fuse of this time bomb is short, and the limited damage associated with the explosion is largely reparable.

Perhaps the most important group of potential neurotoxins consists of those that trigger degeneration and loss of nerve cells. Examples include (1) thallium-induced lesions of the amygdala and periamygdaloid cortex, precipitating uncinate fits and peculiar affective disorders resuiting from disruption of the limbic system; (2) inorganic mercuryinduced cerebellar lesions associated with intention tremor, disordered speech, and ataxic gait; (3) lead-induced cerebral cortical damage leading to irreversible mental retardation; (4) organomercury-induced degeneration of dorsal root ganglion neurons and the calcarine and cerebellar cortex, with coincident sensory loss, tunnel vision, and ataxia; (5) MPTP-induced degeneration of nigrostriatal neurons causing parkinsonism; and (6) BOAA-induced loss of pyramidal neurons eliciting the spasticity of lathyrism. Because mature neurons are postmitotic and therefore irreplaceable, these types of neurobehavioral deficits are permanent. Moreover, because some neuronal groups at risk for toxic damage also normally undergo nerve cell attrition with advancing years, the combined effects of chemical-induced damage and agerelated loss may lead to a permanent deficit that becomes relentlessly progressive in old age. Finally, and most significantly, because these regions of the brain are commonly endowed with a substantial functional reserve, the initial loss of neurons associated with chemical damage may be clinically silent and only unmasked years or decades later as the deleterious effects of age deliver the coup de grace. Here is a chemical time bomb with a very long fuse, and once it explodes, the damage accrues relentlessly until death supervenes. Long-latency neurotoxicity of this type has been associated with organomercurialism, MPTP-induced parkinsonism, and the cycad-associated neurodegenerative disorder of the western Pacific that displays facets of amyotrophic lateral sclerosis (ALS), parkinsonism, progressive supranuclear palsy, and senile dementia. As a result of these new observations, a search has recently begun to identify exogenous chemicals with neurotoxic properties that may play a key role in triggering some of the devastating neurodegenerative diseases of later life, notably ALS, Parkinson's and Alzheimer's diseases.

Developing Nervous System as Chemical Target

This is the point at which we must turn our attention to the susceptibility of the nervous system during its formative stages, the subject of Krasnegor's incisive chapter. Because the developing nervous system is radically different in structure from its adult counterpart, a completely independent system must be deduced to classify the adverse effects of chemical substances. For example, whereas the adult neuron is postmitotic, static, and typically equipped with elaborate, branching dendrites for the receipt of electrochemically encoded information from neighboring nerve cells, the developing neuron divides, migrates, and has few cellular processes in contact with those of few other nerve cells. Similarly, during development, glial cells proliferate, their processes are mobile, and myelin formation is prominent. These and other factors, such as an absent blood-brain regulatory interface to control access of chemical substances to nervous tissue, radically alter the potential responses of the developing nervous system to chemical attack. Whereas specific or unique abnormalities are known to be caused by certain agents (thalidomide), structurally similar abnormalities may result from exposure to different chemicals. Conversely, different types of developmental anomalies may occur with a single noxious agent. Observations such as these suggest that there are factors besides the chemical nature of the agent which dictate the type of resulting damage. Genotype and species are important variables, but the overriding factor is developmental stage. Above all else, the timing of chemical attack appears to dictate the resulting effect.

It is well established that exposure to selected agents during development in utero may have dramatic consequences for behavior during postnatal maturation and young adulthood. Although behavioral abnormalities may be found in the absence of structural changes, either grossly or with a light microscope, it is a tenet of neurobiology that some (presently undetectable) alteration—for example, in synaptic organization or neurochemical anatomy—must underlie a change in neural function. That behavioral alterations induced by chemical agents have no structural basis whatsoever is untenable. What is needed is for the behavioral teratologist and the neurochemical anatomist to join forces to find out how to account for these behavioral abnormalities. As Krasnegor points out, the recent methodological breakthrough in studying the conceptus in the externalized uterus provides a remarkable new window of opportunity to research these phenomena. Substances of interest can now be studied for their neurochemical, structural, and behavioral effects at the time of gestation (when they have their putative action upon the developing brain), postnatally during neonatal development, at maturity, and even in old age.

Conclusion

My major goal in discussing the various points raised in the preceding four chapters is to propose a firm scientific foundation on which to accept substances as potential human neurotoxins and to place in perspective the relative severity of the adverse effects induced by chemical substances by analyzing their sites and mechanisms of action. I have argued strongly that a comprehensive understanding of behavioral neurotoxicology can only be achieved if we consider all types of chemical substances that attack the nervous system. Of course these compounds must be rigorously tested and regulated, but we will only understand the true magnitude and awfulness of their potential effects, and be able to devise appropriate test methods to detect such changes, if we are prepared to draw freely from the entire body of knowledge available to the science of neurotoxicology.

Reference

• Needleman, H. L., editor. , ed. 1980. Low Level Lead Exposure: The Clinical Implications of Current Research. New York: Raven Press.