FINAL REPORT: Validation of Test Methods for Assessing Neurodevelopment in Children1
Testing for human neurodevelopment effects of exposure to neurotoxicants almost always involves use of an extensive battery of conventional tests typically used in clinical neurodevelopmental assessment (see Amler, et al., 1996, Anger, et al., 1994, and Krasnegnor, et al., 1994 for comprehensive reviews). Psychologists involved in this field have argued that testing should be accomplished by assessing developmental domains (discrete areas of development, such as executive brain functions, sensation, language, social behavior, and perceptual-motor functions). In this approach, the choice of specific domains to be assessed should be determined by biological hypotheses and domains should be evaluated in a hierarchical model, e.g., hypothesis testing should proceed from the most to the least likely domain to be affected (Davidson, et al., 1995; White, et al., 1993). Typically, numerous time-consuming tests are needed to accomplish this type of assessment, especially when mild or subtle effects may be suspected, such as might result from low-dose exposures. A clinical processes approach might be more useful to organize and simplify the task (Bellinger, 1995).
Many of these questions lie outside the bounds of conventional tests. They are posed as questions about performance rather than as questions about diagnosis. Furthermore, they are elicited in a context of risk—the potential for adverse effects—rather than a context in which scores are designed to provide a comparison with some reference sample. These are among the considerations that led to the development and application of test instruments aimed at more specific neurotoxicological assays.
The developmental domains of concern in assessing subtle effects of low dose exposures include overall cognition, including memory, auditory and visual information processing, somatosensory functions, fine motor control, complex perceptual motor functions, and language. In this study, we focused on those domains for which limited valid performance measures were available for use with children. We largely excluded direct tests of language function since we were interested in language-free measures that would be applicable across cultures.
Overall Cognitive. The predominant measures found in developmental neurotoxicology consist of those assessing overall cognitive function. Fiedler (1996), in reviewing the array of tests that have been used to assess neurotoxic effects in adults, highlights psychometric tests of verbal and spatial abilities, the sub-domains typically defined by clinical psychologists as comprising overall cognitive ability. There are two types of tests available to determine these functions, and they may be used for assessing children. The first group consists of traditional IQ tests, such as the Wechsler Intelligence Scale for Children. IQ scores and associated indices have been widely used in human behavioral neurotoxicology research because of their long history and their familiarity to child development specialists. Studies of exposures to lead, PCBs, illicit drugs, methylmercury, and other agents largely have been framed by those indices. Another, complimentary approach is to employ neuropsychological assessment batteries, such as the Halstead-Reitan battery (Heaton, Grant & Matthews, 1991), or the Cambridge Neuropsychological Test Automated Battery (CANTAB). The CANTAB was derived from laboratory research and has proved useful in defining the nature of neurobehavioral deficits in a variety of adult clinical disorders (Fray, Robbins & Sahakian, 1996). Cognitive functions also play a role in multi-tasking performance, which calls upon simultaneous use of attributes such as monitoring, vigilance, attention, and memory, which no single available psychometric or neuropsychological test is designed to measure.
Information Processing. Exposure to many neurotoxicants can affect multiple sensory systems. We selected the auditory, visual and somatosensory systems because of their contribution to successful execution of daily living skills and complex perceptual and neurocognitive functions.
Auditory .Many neurotoxicants in high doses affect auditory functions. Lead is a documented auditory system toxicant (e.g., Otto and Fox, 1993) and its literature has demonstrated the utility of sensory assessment as an index of adverse effects. Methylmercury studies in monkeys indicate difficulties with auditory discrimination tasks (Rice, 1989). Also, human hearing deficits were seen with high level exposure to MeHg (Al-Damluji et al, 1976; Harada, 1977; Ino and Mizukoshi, 1977; Amin-Zaki et al, 1974, 1979 and Brenner and Snyder, 1980). PCBs, in animals, have now also been shown to induce low-frequency hearing loss (Crofton et al, 1999). Solvents such as trichloroethylene (TCE) and toluene can induce auditory deficits in both animals (cf., Crofton et al, 1994; Fechter et al, 1998) and humans (Murata et al, 1997; 2003). Recent data provide evidence for a cochlear origin (Fechter et al, 1998).
Tsubaki and Irukayama (1977) have indicated that cochlear lesions may be responsible for some of the hearing loss in Minamata disease. In addition, changes in the Organ of Corti have been seen in guinea pigs and squirrel monkeys with mercury exposure (Wu et al, 1985 and Tsubaki and Takahashi, 1986). In the Faroe Islands, at exposure dosages presumed to be many times lower than in Minamata, increased latencies of waves III and V of the Auditory Brainstem Response (ABR) at 40 Hz were associated with higher levels of mercury exposure (Grandjean et al., 1997). Specifically, the absolute latencies of waves III and V were significantly prolonged in those children with the highest prenatal mercury exposures when stimulated at either 20 or 40 clicks/sec. However, the inter-wave latencies showed no association with exposure to mercury. In Madeira, Murata and colleagues (1997) reported that the I-V (20/sec) and the I-III (40/sec) inter-wave latencies were significantly correlated to the methylmercury levels in mothers. Recently, Murata and colleagues (2003) reported that when Faeroese children were examined, the I-III interpeak interval was associated with prenatal MeHg exposure at 7 years, but the association was weaker when they were reexamined at 14 years. A new effect on III-V the interpeak interval was also reported (Murata, et al., 2004).
Visual . The visual system has been known as a prime target of methylmercury for over fifty years, and motivated the design of our procedure, based on primate research, for studying form discrimination under scotopic illumination. Lead, too, is a visual system neurotoxicant. Like methylmercury, but by a different mechanism, it impairs scotopic vision (Fox et al, 1997), and also reduces acuity in adult rats, but effects in children have yet to be explored in detail. PCBs and related compounds have largely been ignored in this literature, but a recent publication by Kremer et al (1999) indicates that, in rats exposed prenatally, coplanar PCB congeners produce reductions in the scotopic electroretinogram. Similar results were reported for children exposed to PCBs in the Faeroe Islands (Murata, et al., 2004)
Common solvents, such as those found at Superfund sites, have been extensively documented as visual system neurotoxicants. Deficits in color discrimination in workers are the clearest indication of excessive exposure, but other markers of visual dysfunction also need to be explored. Visual evoked potentials to sinusoidal gratings or as checkerboard reversals are one potentially useful technique to join with psychophysical measures.
The organophosphorous (OP) insecticides, at higher levels, produce complaints of blurred vision. Evidence of persistent visual deficits in children exposed to low levels is ambiguous. Japanese data showing such an association (Ishikawa and Miyata, 1980; Dementi, 1994) has prompted the US Environmental Protection Agency to undertake an evaluation of the evidence (e.g., Boyes et al, 1994; Geller et al, 1998).
Somatosensory. At least in adults, paresthesias seem to be the earliest symptoms of methylmercury toxicity, and data from monkeys (Rice & Gilbert, 1995) reveal deficits in vibration sensitivity. Similar deficits arise from exposure to acrylamide (Maurissen, et al., 1983). OP insecticides are documented axonopathic agents and some have been assessed directly for impaired vibration sensitivity (McConnell et al, 1994; Beach et al, 1996). Organic solvents can also induce deficits in tactile sensitivity, as shown by a study in microelectronics workers (Broadwell et al, 1995). The consequences of exposure to organic solvents on changes in developmental parameters have not been studied to date.
Fine Motor. Motor dysfunction is a common element in neurotoxicity and appears in a variety of ways. Excessive limb tremor accompanies many neurological disorders, is a product of many drug treatments, and can result from exposure to many different classes of toxicants (Newland & Weiss, 1991). Tremor is associated with exposure to metals such as mercury and manganese; to insecticides such as chlordecone, dieldrin, and the organophosphates; and to solvents such as carbon disulfide. In a recent survey, Anger (1990) noted tremor as a response to 177 chemicals or chemical classes. Because of its pervasiveness as a marker of nervous system dysfunction, many techniques have been developed to measure tremor. The availability of relatively inexpensive digital computers and appropriate programs makes the task of recording and analyzing tremor much simpler now than in the past. Coupling tremor measures with measures of response time, movement time, and positioning accuracy, as described later, provides a comprehensive assay of fine motor control.
Complex Perceptual-Motor. Test batteries commonly strive to assess different functions with individual measures. For example, reaction time will be measured with one test, attention with another test, motor coordination with a third test and cognitive function with additional tests. Such an approach is consistent with traditional analyses of function, but, simultaneously, overlooks the performance properties that are demanded outside the testing situation. For example, some investigators have reported that ethanol may degrade automobile driving performance at doses that have no influence on reaction time. In one of our own investigations of the effects of toluene on performance (Rahill et al, 1996), a task requiring the subjects to perform four concurrent tasks proved sensitive to toluene concentration at the official threshold limit value. Multi-task performance evaluates the capacity of subjects to respond to the kinds of simultaneous demands that press the limits of functional capacity and that may prove to offer a useful measure of attention deficit disorders.