INVESTIGATION OF UNDERLYING MECHANISMS

One way of relating animal findings and human hazard is to evaluate underlying structural, biochemical, and physiologic correlates of overt functional changes seen in animals. The rationale is that the determination of the target and degree of toxicity produced by developmental exposure will yield information relevant to the human situation.

Mirmiran and colleagues (1985) recently reviewed the relationships between behavioral alterations observed in humans and animals and the underlying neurochemical and electrophysiologic disturbances observed in rats that were exposed to pharmaceutical agents during development. Table 27–1 presents some of the findings.

TABLE 27–1

Sequelae of Developmental Exposure to Drugs in Humans and Animals.

The immaturity of the blood-brain barrier and greater accumulation of many of these compounds in the developing brain make the fetal brain a major target of its mother's medication. Mirmiran et al. (1985) have shown that neonatal exposure of rats to clonidine, an antihypertensive agent, and clomipramine, an antidepressant that acts on norepinephrine and serotonin neurotransmission, suppresses rapid-eye-movement (REM) sleep in the developing rats. In adulthood, the offspring rats showed hyperactivity, hyperanxiety, reduced sexual behavior, disturbed sleep patterns, and smaller cerebral cortex.

STUDY OF COMPARABLE FUNCTIONAL END

A second approach to determining the relationship of animal findings in postnatal functional studies to the human situation is to select a functional response that is comparable across species. A variety of potentially relevant end points are available, e.g., sleep patterns, neonatal vocalizations, and suckling patterns. The development of these end points and their sensitivity to toxic insult could be compared directly across species.

The startle reflex is valuable in such an effort for several reasons:

• Startle can be elicited in all mammals, including humans.
• The startle reflex is mediated via simple neuronal circuits.
• The startle reflex is modulated via several neurotransmitter systems.
• The startle reflex can be measured at early ages in many species.
• The startle reflex is quantifiable.
• Inhibitory or excitatory effects can be determined.
• Startle displays different types of plasticity.

Davis (1984) has reviewed aspects of the mammalian startle reflex. It consists of a characteristic, very rapid sequence of muscular responses elicited by a sudden, intense stimulus. Under comparable circumstances, the more intense the stimulus, the greater the response. The graded amplitude of the mammalian startle response can be detected in direct muscle recordings (e.g., electromyographic recordings from a limb or muscles involved in blinks) or in the output from transducers that measure cage movements when whole-body startle is measured. A standard feature of this reflex is its very short latency; the response occurs only milli-seconds after the onset of the eliciting stimulus.

Although the neural circuitry that mediates startle is at lower levels of the CNS, higher neural networks can modulate it. Nearly all defined neurotransmitter systems interact to modulate the startle response (Fechter, 1974; Davis and Aghajanian, 1976; Davis and Sheard, 1976; Handley and Thomas, 1979; Davis and Astrachan, 1981; Gallager et al., 1983; Holson et al., 1985). In the spinal cord and facial motor nucleus, serotonin and norepinephrine increase auditory startle and glycine tonically inhibits it; it appears that GABA can also inhibit the response at this level. Supraspinally, dopamine and perhaps GABA receptor stimulation increases startle, and serotonin activation depresses it. Startle is also modulated in several brain regions distant from the primary startle pathway itself. Therefore, the reflex can provide a sensitive indicator of function after toxicant exposure. Developmental insults that result in changes in a neurotransmitter system might be expressed as changes in the latency, amplitude, or modification of the response. The type of change observed can suggest which systems have been affected by exposure.

Auditory startle has been used often in studies of animal developmental toxicology. Recently, automated procedures have been applied in such studies, thus allowing more specific characterization of changes in this reflex. Automated procedures for stimulus presentation and data collection have yielded useful information for evaluating sensitization, habituation, prepulse inhibition, and reflex modification by prior associative learning after toxicant exposure (Hoffman, 1984). Startle thus represents a potentially powerful tool in developmental toxicology for investigating sensorimotor reactivity. The simplicity of the response and the plasticity displayed within it across animal species, including humans, suggest that specific efforts to investigate the comparability of startle alterations in animals and humans after developmental insult are warranted.

DIRECT COMPARISONS BETWEEN ANIMALS AND HUMANS

A third approach to determining the relationships among human and animal developmental toxicity is to compare observed effects when data are available for several species. Few human behavioral-teratology studies have been reported, and most experimental behavioral-teratology studies have used rodents. The comparisons outlined here reflect that situation. In addition, similarities and differences in design and conduct between experimental and clinical research must be considered in any comparison of results. The similarities between the two include the following:

• Physical growth and development are the most commonly measured end points.
• Several behavioral subsystems are assessed with a battery of functional tests.
• Experimental and control subjects are matched for maternal and environmental characteristics.
• The majority of studies are designed to provide descriptive information.

In human studies, weight and motor development usually are measured for 1–2 years after birth. In rodent studies, weight is monitored repeatedly, most often throughout the duration of the study, and assessments of preweaning reflex development are often carried out as well.

A battery of functional tests usually are used for neurobehavioral evaluation in both human and animal studies. The use of a single assessment technique that incorporates multiple evaluations is most common in human research. The Apgar test (Apgar, 1953) is used routinely 1 and 5 minutes after birth; it consists of a 10-point scale based on five components: appearance (skin color), heart rate, latency of the cry reflex, muscle tone, and respiration. The Bayley scale of infant development (Bayley, 1969) is used commonly for evaluation of older infants; it contains sensory, motor, verbal, and cognitive items, and results are summarized in motor and mental development scores. Neurobehavioral function in rodents is evaluated with a test battery that often includes assessment of reflex and sensorimotor development, activity level, and some evaluation of learning ability. Each category of function is evaluated with separate tests.

Another similarity in study designs is the use of experimental and control subjects matched for maternal and environmental characteristics. In human studies, mothers are matched as closely as possible for age, parity, and nutritional and socioeconomic status. In animal studies, maternal weight, parity, diet, and housing conditions routinely are controlled across groups.

Both clinical and experimental studies designed to evaluate neurobehavioral outcomes after prenatal drug or chemical exposures provide primarily descriptive information. The methods permit a description of functional deficits after insult, but not of underlying physiologic or neurochemical mechanisms responsible for the observed behavioral alterations. As noted above, that situation is changing in animal studies. Multidisciplinary efforts can provide information on the mechanisms involved and thus might suggest types of intervention that could be effective in alleviating or improving clinical outcomes.

Several basic differences in design and conduct between human and animal studies are common. They include differences in the relative age range, in timing of administration of tests, in attempts at standardization, and in methods of reporting results (Adams, 1986).

The relative age range studied is broader in much of the animal research than in human studies. Practically, it is very difficult to follow a prenatally exposed person for more than a year or two after birth. In the best of circumstances, clinical investigators can extend evaluations to 6 or 8 years of age. Funding, time requirements, and population mobility and attrition all contribute to the difficulty. In contrast, rodent studies often include behavioral evaluations into early adulthood. Such a longitudinal approach can be even more valuable if testing is identical across the age span studied.

In human research, several functions usually are evaluated at a single age. Neurobehavioral function in rodents most often is evaluated through separate testing at different ages. Furthermore, unlike the results of multifunctional evaluations in humans, rodent performance across tests is not integrated into a single value or score.

A greater effort is made in human than in animal studies to perform testing on infants in a comparable behavioral state, e.g., alert, drowsy, or asleep. Such control contributes to both absolute response levels and decreased variability in behavioral data collected in infants and children (Clifton and Nelson, 1976). Animal researchers at best attempt to control such factors by balancing time of day during testing across experimental groups.

Developmental delays in physical, motor, and cognitive end points are considered more important in human than in animal studies. Such delays can be assessed only during particular periods of development. Their biologic meaning in rodents after prenatal exposures is not clear, and they have been viewed as problematic (Tilson and Wright, 1985). One contributing factor might be the time disparity in postnatal developmental schedules between humans and rodents, i.e., months and years versus days.

Finally, a characteristic difference between human and animal studies involves the method of reporting results. Clinical studies typically identify the incidence of behavioral dysfunction in individual control versus individual exposed subjects. Animal data usually are presented in terms of the presence or absence of group mean differences. Thus, the incidence of affected (and nonaffected) rodent offspring in the exposed group is not available.

In the light of these differences in design, conduct, and reporting between human and animal behavioral-teratology studies, findings are discussed below if appropriate data were available for comparison. The number of reported human studies was the limiting factor in the following brief overview; once those were identified, the animal literature was evaluated. In all cases, if human sub-system dysfunction was reported, corroborative evidence was found in the animal data if a comparable end point had been evaluated. In that manner, effects observed after developmental exposures to lead, mercury, PCBs, phenytoin, ethanol, and methadone are compared. Articles reviewing the spectrum of effects observed in humans and animals are cited in the following discussion.

Table 27–2 summarizes the comparability of neuromotor effects after exposure to particular toxicants. Delayed motor development has been reported in both humans and rodents exposed to lead (Rutter, 1980; Reiter, 1982), mercury (Reuhl and Chang, 1979), alcohol (Abel, 1980), and phenytoin (Hanson et al., 1976; Vorhees, 1983). Cerebral palsy and seizure disorders have been reported in humans developmentally exposed to mercury, and motor dysfunction and increased susceptibility to seizure induction have been reported in rats and mice. Developmental exposures to PCBs have resulted in motor dysfunction in both humans (Jacobson et al., 1984) and mice (Tilson et al., 1979). A prolonged neonatal abstinence syndrome with neuromotor sequelae has been identified in human infants and rodents prenatally exposed to methadone (Hutchings, 1983). The specific motor alterations observed in humans and rodents were not always identical, but the normal behavioral repertoires of the two also are different. The data do indicate that the motor systems of humans and rodents are susceptible to disruption after developmental exposures to the agents in question.

TABLE 27–2

Examples of Motor Dysfunction After Behavioral-Teratogen Exposures.

The clinical relevance of experimental data on cognitive functions is more difficult to evaluate. Tests of human and animal cognitive abilities might evaluate very different functions; i.e., a rodent brain is not capable of the many complex functions evaluated in human assessments. Techniques used to assess cognitive function measure responses that are modified by sensory and motivational processes, as well as motor capabilities. However, performance on such tests can provide useful information concerning the postexposure integrity of underlying systems that contribute to an animal's or child's ability to learn, process, store, and retrieve relevant information. Table 27– 3 summarizes some of the cognitive deficits noted after developmental insult. Reduced general intelligence, as measured on standardized tests, has been found in some children who were exposed prenatally to lead (Needleman et al., 1979), mercury (Harada, 1976), ethanol (Streissguth et al., 1984), and phenytoin (Hanson et al., 1976). The IQs of those children are often less than 70. Mental retardation is one of the most serious results of exposure to the agents in question. Attentional deficits have been reported in some children prenatally exposed to lead, mercury, and ethanol. Experimental studies have indicated impairments in visual recognition memory in infants exposed to PCBs (Jacobson et al., 1985) and increased reaction times during a vigilance task in ethanol-exposed children.

TABLE 27–3

Examples of Cognitive Dysfunction After Behavioral-Teratogen Exposures.

The animal literature indicates impaired learning and memory abilities in rodents after developmental exposures to the agents. Performance deficits on avoidance tasks have been reported after exposures to lead (Kimmel et al., 1978), mercury (Spyker et al., 1972), PCBs (Tilson et al., 1979), ethanol (Abel, 1980), and phenytoin (Vorhees, 1983). Results of water-maze tasks have indicated impaired function in rodents exposed to mercury (Spyker et al., 1972), ethanol (Abel, 1980), and phenytoin (Vorhees, 1983). Hughes and Sparber (1979) found that prenatal mercury exposure disrupted operant performance.

It is interesting that prenatal methadone exposure does not appear to alter cognitive performance in either humans or animals (Hutchings, 1983). The fact that both humans and animals showed no effects on cognitive performance supports the utility of experimental-teratology data in assessing toxic effects.

Sensory/perceptual processes have not been carefully evaluated in most behavioral-teratology studies (Adams and Buelke-Sam, 1981; Ison, 1984). As shown in Table 27–4, clinical case reports have suggested that some persons exposed in utero to ethanol (Clarren and Smith, 1978) or to phenytoin (Hill et al., 1974) have unspecified hearing defects. Visual impairments have been reported in some children exposed in utero to phenytoin (Wilson et al., 1978). However, specific sensory functions have not been evaluated in animals that have been exposed prenatally to alcohol or phenytoin. Vorhees (1983) reported delayed development of auditory responsiveness in rats treated with phenytoin prenatally.

TABLE 27–4

Examples of Sensory/Perceptual Processing Dysfunction After Behavioral-Teratogen Exposures.

Decreases in visual acuity have been reported to occur in children (Rummo et al., 1979), rats (Fox et al., 1977, 1979; Fox and Wright, 1982), and monkeys (Bushnell et al., 1977) after exposure to inorganic lead. Altered electrophysiologic brain activity in response to visual stimulation has also been found in lead-exposed children (Otto et al., 1981, 1982, 1985; Otto and Reiter, 1983) and rats (Fox et al., 1979).

Fetal exposure to methylmercury has been reported to produce postnatal alterations in reactivity to visual and auditory stimulation in humans (Harada, 1976, 1977). Studies done on prenatally exposed rats have shown increased reactivity to auditory stimuli (Buelke-Sam et al., 1985).

Wilson et al. (1979) reported that children prenatally exposed to methadone had deficits in visual, auditory, and tactile perception, but these were interpreted as resulting from poor attentional and strategic processing abilities, rather than from specific sensory deficits. Lodge (1976) reported alterations in brain electrophysiologic responses to visual stimuli and hypersensitivity to auditory stimulation in children exposed to methadone before birth. The integrity of sensory functioning has not been specifically evaluated in studies carried out in rodents, but hypersensitivity to aversive stimulation has been reported by Hutchings (1983).

In nearly all the cases outlined above, the clinical problem was identified before the development of animal models to explore such toxicity. The literature of animal behavioral teratology has expanded greatly in recent years and suggests that a number of additional drugs and chemicals warrant clinical investigation. However, the potential value of animal data in predicting human hazard cannot be determined fully until clinical studies to look for behavioral dysfunctions identified in experimental animals are designed and conducted.

DATA INTERPRETATION

We have discussed many problems that contribute to the difficulty of evaluating the relevance of animal data in identifying potential health hazards in developing humans. The first three issues that follow bear on the interpretation of both human and animal data; the last two are related to problems in cross-species extrapolation of results.

The first problem centers on determining whether behavioral-teratology findings are a result of primary developmental toxicity or are secondary to maternal toxicity or to primary toxicity produced in other organ systems, e.g., liver or kidney. The heart of the issue is the relative susceptibility of the developing organism (whether human or animal) to toxic insult. If postnatal dysfunction only accompanies maternal toxicity, such information might be of value in alerting clinicians to moni tor such pregnancies more closely. If postnatal functional deficits are obtained in the absence of overt toxic signs, the potential for selective developmental toxicity must be considered in human risk assessment.

Genetics might play a large role in susceptibility or expression of postnatal dysfunction. Genetic makeup could predispose the parents or offspring to greater sensitivity to a toxicant. Toxicants could produce postnatal dysfunction via genetic mechanisms, or such dysfunction could be transmitted to later generations (cf. Fujii et al., 1987; Stoetzer et al., 1987). Thus, whether developmental-toxicology studies are performed with inbred strains of mice or in the highly diverse human and whether one or both parents have been exposed to a toxicant are important considerations in evaluating and interpreting postnatal functional data.

The postnatal environment can have an impact on developmental toxicity, maximizing or minimizing the expression of damage. In human studies, the role of maternal socioeconomic status is great in that regard, as well as contributing to overall prenatal and perinatal status. In animal research, controlling litter size is one means of standardizing this factor. In both types of research, the degree of environmental experience and enrichment might contribute to the manifestations of toxicity.

Once those issues are considered, two additional aspects must be dealt with in determining the relevance of animal data to the human situation. The first concerns the disparity in timing of organ-system development across species. The rat gestation period covers approximately 3 weeks, and CNS development, including cell differentiation and migration, continues into the immediate neonatal period. More CNS development occurs in humans during the 9-month gestation period, although completion of histogenesis does not occur until well after birth. Agent exposure is timed specifically in animal studies, and care is taken to standardize doses within treatment groups. Such control usually is not available in human studies, and retrospective investigations often rely on maternal reporting of exposure to drugs and when it occurred.

Postnatal development schedules also differ between humans and animals. Postnatal development through puberty in a rat requires approximately 6 weeks. It takes years to reach that stage in humans. Thus, prenatal and postnatal disparities in timing must be accounted for, as well as the variations in agent exposures during the comparative process.