ToxSci Advance Access originally published online on August 13, 2007
Toxicological Sciences 2007 100(1):146-155; doi:10.1093/toxsci/kfm203
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press 2007.
Quantitative Comparisons of the Acute Neurotoxicity of Toluene in Rats and Humans
* Human Studies Division
Neurotoxicology Division
Experimental Toxicology Division, National Health and Environmental Effects Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711
1 To whom correspondence should be addressed to Human Studies Division, Mail Code B105-06, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711. Fax: (919) 541-4849. E-mail: benignus.vernon{at}epa.gov.
Received May 9, 2007; accepted July 30, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
The behavioral and neurophysiological effects of acute exposure to toluene are the most thoroughly explored of all the hydrocarbon solvents. Behavioral effects have been experimentally studied in humans and other species, for example, rats. The existence of both rat and human dosimetric data offers the opportunity to quantitatively compare the relative sensitivity to acute toluene exposure. The purpose of this study was to fit dose-effect curves to existing data and to estimate the dose-equivalence equation (DEE) between rats and humans. The DEE gives the doses that produce the same magnitude of effect in the two species. Doses were brain concentrations of toluene estimated from physiologically based pharmacokinetic models. Human experiments measuring toluene effects on choice reaction time (CRT) were meta-analyzed. Rat studies employed various dependent variables: amplitude of visual-evoked potentials (VEPs), signal detection (SIGDET) accuracy (ACCU) and reaction time (RT), and escape-avoidance (ES-AV) behaviors. Comparison of dose-effect functions showed that human and rat sensitivity was practically the same for those two task regimens that exerted the least control over the behaviors being measured (VEP in rats and CRT in humans) and the sensitivity was progressively lower for SIGDET RT, SIGDET ACCU, and ES-AV behaviors in rats. These results suggested that the sensitivity to impairment by toluene depends on the strength of control over the measured behavior rather than on the species being tested. This interpretation suggests that (1) sensitivity to toluene would be equivalent in humans and rats if both species performed behaviors that were controlled to the same extent, (2) the most sensitive tests of neurobehavioral effects would be those in which least control is exerted on the behavior being measured, and (3) effects of toluene in humans may be estimated using the DEEs from rat studies despite differences in the amount of control exerted by the experimental regimen or differences in the behaviors under investigation.
Key Words: behavior; neurotoxicology; dose-response; risk assessment; volatile organic compounds; agents.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
Effects of toxic chemicals are usually studied in experimental animals for the purpose of understanding and predicting their effects in humans. For this purpose, it is often assumed as proposed by Lehman and Fitzhugh (1954)
that rats are less sensitive to the adverse effects of toxicants than are humans. There is, however, little direct evidence for this assumption because sufficient data for making this cross-species comparison are not available for most toxicants. Cross-species comparisons have been complicated for two reasons. First, because external exposure concentration was typically used as a dose metric and such comparisons ignore the possible species differences with respect to pharmacokinetics. Second, effects were not usually measured in the same scale for both species thus obviating direct comparisons. In the present work, effects of internal dose are compared using outcome measures that have been transformed into common scales. Here internal dose is defined as concentration of toluene in the brain. This approach provides a simpler, quantitative and defensible approach to cross-species comparisons of toxicant sensitivity. The work is restricted to acute exposures and will probably not generalize specifically to longer or repeated exposures.
Cross-species comparisons are possible for the effects of toluene, a neurotoxic chemical whose acute effects have been well documented. Experimental exposures in humans began with the qualitative descriptions of Von Oettingen et al. (1942) and progressed through the quantitative behavioral and electrophysiological studies of Cherry et al. (1983), Dick et al. (1984), Echeverria et al. (1989), Gamberale and Hultengren (1972), Iregren et al. (1986), Olson et al. (1985), and Rahill et al. (1996). Similar studies in rats have also been extensively reported (Battig and Grandjean, 1964; Boyes et al., 2007; Bushnell et al., 1994, 2007; Colotla et al., 1979; Geller et al., 1979; Harabuchi et al., 1993; Ikeda and Miyake, 1978; Kishi et al., 1988; Mullin and Krivanek, 1982; Shigeta et al., 1978, 1980; Wada et al., 1989). Toluene has also been shown to impair the function of ion channels in vitro (Bale et al., 2005; Cruz et al., 1998; Shafer et al., 2005). These studies provide a database sufficient for a quantitative comparison of the sensitivities of rats and humans to acute effects of toluene. Meta-analyses of these data have been previously presented (Benignus et al., 1998), and the present work is an extension of these analyses.
For regulatory purposes, it is necessary to quantify effects as a function of the magnitude of a subject's external exposure to a toxicant, for example, toluene concentration in inhaled air. Many acute effects depend upon the duration of exposure because of increasing uptake of the chemical with time. Concentration and duration of exposure combine nonlinearly to produce a time-varying concentration of toluene in the brain (and other organs). A number of experiments have shown that the magnitude of acute behavioral and neurophysiological effects relate more parsimoniously and closely to the concentration of toluene in the brain (internal dose) at the time of assessment (Benignus et al., 1998, 2005; Boyes et al., 2007; Bruckner and Peterson, 1981; Bushnell, 2007) than other combinations of external exposure concentration and duration. The same has been demonstrated for trichlorethylene (Boyes et al., 2005; Bushnell, 1997).
In risk assessment, the potency of a toxicant is frequently quantified by measuring the exposure or internal dose needed to produce an effect that lies on some threshold of statistical detectability, for example, a lowest observable adverse effect level (LOAEL) or a benchmark dose. Alternatively, potency can be expressed as the dose or exposure needed to obtain a certain percentage of effect, such as the dose that produces 50% of the maximum effect (ED50).
A full dose-effect curve with confidence limits provides a more complete and informative (but numerically more elaborate) description of the effect of a toxicant because it gives the statistically best estimate of the magnitude of an effect at any dose included in the experiment. A dose-effect curve of the appropriate form also acts as both an interpolation algorithm for estimating the magnitude of effect for any dose within the observed range and also as an extrapolation algorithm for doses that lie beyond the observed range.
Comparison of human and rat sensitivity to toxicants can be expressed as ratios or differences of, for example, LOAEL or ED50 values; however, just as dose-effect curves are more informative than point estimates, the relationship between dose-effect curves is more informative than ratios or differences of point estimates. A more complete description has been proposed by Benignus (2001) in which a dose-equivalence equation (DEE) was derived from the dose-effect curves of the two species. Here "dose equivalence" was defined as a dose in humans that produced the same magnitude of effect as any given dose in rats. The DEE is the function that relates equieffective doses across their known range. When rat and human dose-effect curves are the same, the DEE is linear with zero intercept and unity slope.
The purpose of this study was to estimate the DEEs between rats and humans from the dose-effect curves generated for a number of different effects that have been reported in the literature. The resulting functions enable quantitative cross-species comparisons of sensitivity across the range of tested doses and the extrapolation to doses beyond that range. Here behavioral and neurophysiological effects of acute toluene exposure were expressed as dose-effect curves by either meta-analyzing or reanalyzing previously published data. In the absence of brain concentration data, which is infrequently measured in behavioral or neurophysiological experiments, internal doses (momentary concentrations) were estimated by previously evaluated physiologically based pharmacokinetic (PBPK) models for rats (Kenyon et al., 2008) and for humans (Benignus et al., 2006).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
Data for Fitting Dose-Effect Curves
Some of the data for the following analyses came from the literature in the form of group means. These data were pooled (after appropriate transformation to assure homogenous measurement scales) and dose-effect curves were then fitted. Other data consisted of individual-subject measurements from previously published experiments. These data were simply reanalyzed to estimate dose-effect curves.
Data from different experiments using a particular dependent variable are rarely expressed in the same numerical scales. To pool data of this sort into the same database either for meta-analysis or for numerical comparison, it was necessary to transform all dependent variables into a single scale with common properties (Benignus et al., 1998, 2005). It was convenient to transform dependent variables into an "effect magnitude" scale with intuitively interpretable properties in which E (the effect magnitude) ranges from 0.0 (no effect) to 1.0 (maximum possible effect). Publications lacking sufficient detail to conduct such transformations could not be used.
The independent variable for a dose-effect curve was the concentration of brain toluene (D) in millimoles (mM) of toluene per liter. Previous analyses have used the concentration of toluene in arterial blood (Benignus et al., 1998), venous blood (Benignus et al., 2005), and brain (Boyes et al., 2007; Bushnell et al., 2007). The brain concentration was selected here to better represent the momentary dose of toluene in the target organ. In all cases, the brain concentrations were estimated from the exposure conditions via PBPK models. For rats, the PBPK model was written in ACSL Xtreme (AEGis Technologies Group, Inc., Huntsville, AL) and empirically parameterized for rats performing behavioral tasks (Kenyon et al., 2008). For humans, simulations were conducted using a general physiological and toxicokinetic (GPAT) model (Benignus et al., 2006), which permitted physiologically realistic simulations of toluene exposure and physical activities that vary as a function of time. This model was needed because some human experiments employed complex scenarios involving intermittent exercise and gradual introduction of toluene vapor.
Dose-Effect and DEEs
For meta-analyses, a form of the logistic equation (Benignus et al., 1998; Corso, 1967) was employed as the dose-effect model. The particular form was
 | (1) |
), ln is the natural logarithm, and D is the brain concentration (dose) of toluene in millimoles. This function was selected because of its flexible ogive-like shape. The parameter ß1 determines the location of the curve on the horizontal (dose) axis and ß2 governs the "slope." The curve starts at zero effect for zero D and rises toward 1.0 as D increases. A value of E = 1.0 means that the end point has been maximally affected, that is, the behavioral performance or the physiological response has entirely ceased.
The relative potency for a toxicant in rats and humans can be qualitatively compared by plotting each of the dose-effect curves for the two species on the same graph. This comparison is not directly numerical, but more importantly, the confidence limits for the comparisons are not available. In contrast, the DEE gives the x, y pairs of values that produce the same effect magnitude in the two species and their confidence intervals. This DEE can be obtained analytically if the dose-effect equations are known for the two species. The procedure is to set the two dose-effect curves equal to each other and solve for the dose in one species in terms of the dose in the other. Thus, for relating potency in humans to potency in rat
 |
 |
 | (2) |
 | (3) |
corresponds to the x-axis location and 
to the "slope."
Statistics
Experimental designs in the behavioral literature almost always involved repeated measures in each subject in various exposure concentrations and/or durations. Typically, the number of repeated measures was different for each experiment. Even for individual experiments that were reanalyzed, occasional missing data yielded a variable number of measures per subject. Curve fitting for combined data sets from these experiments was done with PROC NLMIXED (SAS, Cary, NC), which accounts for the mixed-design difficulties. For meta-analyses of data in which experimenters reported means from different-sized groups, it was desirable to weight each mean by the square root of its group size, thus balancing the emphasis of results across studies (Kleinbaum et al., 1988, pp. 219–220).
Previously, confidence limits for a DEE were not easily estimated by analytic methods (Benignus, 2001); therefore, Monte Carlo methods were employed. In the present work, the ß1 and ß2 parameters for Equation 2 were simultaneously estimated in PROC NLMIXED for the human and the rat dose-effect curves and confidence limits were analytically computed using an imbedded delta method.
Sources of Data for Meta-analyses and Reanalyses
Sources of data for the meta-analyses were reported in appendix 3 of Benignus et al. (1998) along with reasons for exclusion and inclusion of publications. The dimensions of the database for each analysis in the present work are summarized in Table 1.
|
Human laboratory studies of choice reaction time (meta-analysis).
Some form of choice reaction time (CRT) was measured in six studies (Dick et al., 1984
; Echeverria et al., 1989; Gamberale and Hultengren, 1972, Iregren et al., 1986; Olson et al., 1985; Rahill et al., 1996) among many other dependent variables and was selected for meta-analysis because it was measured most commonly. In none of the human studies was any explicit reward given for accurate or rapid responses. No individual-subject data were given. Means were therefore meta-analyzed.
Transformation of an observed CRT to effect magnitude presented problems because the maximum possible value of CRT is an indefinitely large number and could not, therefore, be used to normalize data. For this reason, the CRT data were converted to reaction speed (RS) where RS = 1.0/CRT and thus
 | (4) |
 | (5) |
Rat visual-evoked potentials (reanalysis).
Visual-evoked potential (VEP) data were taken from Boyes et al. (2007) in which groups of rats were exposed to one of five concentrations of toluene (0, 1000, 2000, 3000, and 4000 ppm). Each rat was exposed to only one concentration of toluene in one test session, but VEPs were measured one to five times per session. Each VEP waveform was an averaged response from 20 trials of a 4.55-Hz sinusoidal temporal modulation of the visual stimulus. Average VEPs were subjected to spectrum analysis, and the amplitude of the frequency-double (F2) spectrum peak was measured. Individual-subject data were used to fit the dose-effect curve, and the means of each brain concentration group were plotted with their univariate SEs. VEP amplitudes were transformed to effect magnitude by
 | (6) |
Rat SIGDET (reanalysis).
Rat SIGDET data were taken from Bushnell et al. (2007). Reaction time (RT) and accuracy (ACCU) of SIGDET were collected from the same subjects in the same test sessions. Each measurement was the mean RT or ACCU for the particular time period of performance for a particular subject in a particular time during exposure. Dose-effect curves were generated for each of the two dependent variables. These curves were plotted along with overall mean and univariate standard errors for each time period in the study. RT was transformed according to Equation 5.
ACCU, defined in the SIGDET task, has a value of 0.5 when performance is at chance level. Thus, ACCU = 0.5 is taken to be the maximum effect. The transform to compute the ACCU effect magnitude, E(ACCU), is
 | (7) |
Rat escape-avoidance behaviors (meta-analysis).
Escape-avoidance (ES-AV) behaviors data were collected from six studies (Harabuchi et al., 1993; Kishi et al., 1988; Mullin and Krivanek, 1982; Shigeta et al., 1978, 1980; Wada et al., 1989). Subjects were measured repeatedly in all four studies, but more extensively in some studies than others. Only means of correct response rate were reported, and therefore, meta-analysis was used. Correct response rate was transformed to effect magnitude using an equation of the form of Equation 6.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
Dose-Effect Curves
The dose-effect curves (solid lines) and their 95% confidence limits (broken lines) for four groups of experiments are depicted in Figures 1a–e. The axes of the figures have different scales to better depict results for the differing toluene concentrations studied. For each panel, the experimental exposure concentration and duration parameters were used to estimate brain toluene concentration at the time of testing using the appropriate PBPK model. In Figure 1a, data from human CRT experiments are presented in which the RT latency values were converted to an effect magnitude scale according to Equations 4 and 5. Figure 1b presents the amplitude of VEPs recorded from rats exposed to toluene, where effect magnitude was calculated according to Equation 6. Figures 1c and 1d present data from rats exposed to toluene while performing a SIGDET task, and effect magnitude was calculated according to Equation 5 for RT values (Panel 1c) and Equation 7 for accuracy values (Panel 1d). Figure 1e presents data from rats performing a shock ES-AV task during or after exposure to toluene, and data on the rate of correct responses were converted into effect magnitude according to Equation 6. Figure 1f gives a comparison of the dose-effect curves, all plotted on the same scales. Confidence limits and data points have been omitted for clarity. The parameters and statistics for the curve fits are given for each study in Table 2.
|
|
Dose-Equivalence Curves
Figures 2a–d are plots of DEEs and their 95% confidence limits, each calculated according to Equation 3. In each panel, the horizontal axis is the estimated brain toluene concentration in rats and the vertical axis is the estimated equivalent brain toluene concentration in humans. In each panel, one of the dependent variables from rats in Figure 1 is related to CRT, the dependent variable from human studies. A plot of the equation relating rat VEP amplitude to human CRT is presented in Figure 2a. The relationships between the RT and accuracy of rat SIGDET performance and human CRT are presented in Figures 2b and 2c, respectively. Similarly, the relationship between rat ES-AV and human CRT is presented in Figure 2d. Figure 2e is a plot of all four DEEs on the same axes to facilitate comparison.
|
The DEEs for Figure 2 are of the form of Equation 3, and the parameter values are given in Table 3. These parameters were calculated based on the parameters of the two dose-effect curves involved.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
These analyses have provided comparable quantitative dose-effect relationships for three neurotoxicity outcomes in rats and one in humans following acute exposure to toluene. The dose in these equations was expressed as estimated concentration of toluene in the brain at the time of neurotoxicity assessment. The five outcomes differed in their sensitivity to toluene exposure in that the "slopes" and location along the internal-dose dimension differed. The four rat curves were quantitatively related to the human CRT results via the empirically derived DEE.
Comparative "Sensitivity" to Toluene Exposure
In Figure 1f, it is clear that over the limited range of the human data, the rat VEP curve is nearly the same as the curve for human CRT. Each of the other dose-effect functions from the rat studies lies to the right of the function derived from the human CRT data. The similarity of rat VEP and human CRT and their difference from the curves for rat SIGDET (RT and ACCU) and for rat ES-AV suggest that the species is not the critical factor in determining the sensitivity of the response.
Figure 1f also shows that the relative sensitivity of rats to acute toluene exposure depends on some aspect of the various experiments in which effects were measured. To produce ES-AV effects of a particular magnitude in rats requires much higher brain toluene concentrations than for the equivalent magnitude of effect when other end points are measured. SIGDET RT in rats required slightly more brain toluene to produce effect magnitudes equivalent to those from rat VEP. It was also observed on two behavioral measures in the same subjects performing the same task that ACCU was less affected by a particular toluene brain concentration than was RT.
The source of the differences among the rat dose-response curves is unlikely to be pharmacokinetic because all exposures were converted to internal dose (concentration of toluene in the brain). It is possible that some or all of the differences in the task sensitivities could be due to differences in brain regional doses that could affect behaviors mediated by a particular region more than some other behavior, mediated by a more or less exposed area. Such regional brain concentration differences would have to be fairly large to explain, for example, the difference between VEP and ES-AV experiments. In any event, the idea is not testable by present PBPK models even if the neurobehavioral functions of the various brain areas were fully understood and if such functions were uniquely and fully controlled by particular areas.
Because the brain concentration contemporaneous with the behavior is a more appropriate dose metric than external exposures (Benignus et al., 1998; Boyes et al., 2005, 2007; Bushnell et al., 2007), the source of the difference in these functions could be due to differences in response to brain toluene. Thus, it could be argued that the cognitive difficulty of the task could affect the sensitivity of the performance to disruption by toluene exposure. Among the above three behaviors, however, the SIGDET task is the most cognitively demanding (as determined by the amount of training required to condition this behavior) and yet the two end points from this task fall in intermediate positions of sensitivity and also differ from each other.
Rats develop a tolerance to toluene after repeated, high-concentration exposures during performance testing (Oshiro et al., 2007). Tolerance may have developed in some of the experiments analyzed here and could account for the differences in sensitivity in the three groups of rat studies. The VEP rats were tested only once. The SIGDET rats were tested five times at varying concentrations in counterbalanced orders. For all experiments but one, ES-AV rats were tested only once. In the remaining experiment, they were tested twice at two-week intervals and different concentrations. Thus, insignificant tolerance would have developed for the VEP and ES-AV experiments and an unknown (probably small) amount in the SIGDET study. If tolerance were to account for the differences in sensitivity observed here, then the least sensitive experiment would be the SIGDET study and the VEP and ES-AV would be the most sensitive. It is possible that the SIGDET behavior would have been more sensitive if tolerance were not to have occurred, but tolerance cannot account for the order of sensitivity observed here.
Another possible reason for the differing sensitivities to brain toluene concentrations involves the extent to which the measured neurobehavioral responses are controlled. In neurobehavioral experiments, the subjects are usually placed into situations in which the measured responses are controlled (or produced) by such events as rewards and punishments that are presented in various regimens when the appropriate behavior is emitted. The purpose of such regimens is to condition and regularize the behaviors, thus reducing variance. Nevin (1974) presented extensive data to demonstrate that behavioral responses in pigeons became more difficult to disrupt as the behavior was more stringently controlled by the schedule and magnitude of rewards. In his case, the disruptive events were the discontinuation of rewards, which eventually produced cessation of responding (extinction). In general, the more controlling the environment was, the more resistant to extinction the responses became. He appealed to the construct of "response strength" as produced by the various reward schedules to describe the phenomenon. Dews (1955) and Laties and Evans (1980) demonstrated the ability to manipulate the behavioral effect in pigeons of pentobarbital and methylmercury, respectively, by adjusting the reward schedules and other cues in the task. Thompson (1978) reviewed the literature to demonstrate that the magnitude and type of effects of various drugs on behavior could be manipulated by environmental controls upon the behavior in a wide variety of experiments and species such that the more strictly controlled behaviors were less sensitive to drug-induced disruption.
A wide variety of control was exerted in the various experiments that were analyzed in the present work. The relative positions of the dose-effect curves in Figure 1 can be related to the extent of behavioral control. In all of the human studies, subjects were merely instructed to respond as quickly and accurately as possible, but neither was any reward provided for doing so nor was any adverse consequence applied for failing to do so. Thus, no measure of success was explicitly tied to performance and no cost was incurred by poor performance. Similarly, the rat VEP is not influenced by any overt control because it is a neural (sensory) response with no consequences for the subject. In contrast, rats performing the SIGDET task were food deprived and were given food pellets only for correct responding. In addition, slow responding involved only a short delay in food pellet delivery and so it may be argued that RT was less strongly controlled than accuracy of responding. Finally, it is reasonable to assume that ES-AV performance involves a stronger control than food presentation because the cost of an error (electrical shock) is more aversive than failure to receive a food pellet. Thus, the ordinal strength of the neurobehavioral control agrees inversely with the sensitivity of the end points to disruption by internal toluene dose. The systematic relationship between apparent control of behavior and the sensitivity to disruption by toluene is consistent with the literature on this topic, thus lending credence to this interpretation.
The argument about the importance of the extent to which neurobehavioral responses are controlled does not imply that other factors could not be important. In this experiment, neither were other factors measured or manipulated to test their importance nor was the strength of control of behavior varied as a factor in the study. This was an a posteriori hypothesis that should be explicitly tested.
Dose-Equivalence Equations
For VEPs in rats, the dose-equivalence curve for human CRT was nearly linear (Figs. 2a and 2d and Table 3), indicating that the value of (the slope of the DEE) was nearly one. In these two types of experiments, no explicit neurobehavioral controlling methods were used. However, if the extent of task control differed for humans and rats, then the DEE became nonlinear. In these cases, the relative sensitivity of rats and humans are not constant ratios or constant differences.
These data suggest that when predicting human behavioral sensitivity to toluene from rat data, the extent of control over the measured response must be considered. For example, it would be predicted that if humans were in a situation involving avoidance of a noxious event or escape from harm, the best predictor of toluene effects from the rat data would be the ES-AV behavior curve. Similarly, for appetitively rewarded behavior (depending on the reward magnitude and schedule), the best predictor would be from the rat SIGDET curves. Finally, for nonrewarded behavior in humans, the best predictor would be the rat VEP. Thus, according to this hypothesis, when human and rat behavior are controlled to the same extent, the linear DEE with unity slope and zero intercept would best estimate human dose equivalence. In other words, given similar test conditions and defining doses as toluene concentration in the brain, humans and rats are hypothesized to be equally sensitive to toluene.
Wide Range of Toluene Effects
Acute toluene exposure in rats has a wide range of effects on the CNS and behavior, impairing visual function, SIGDET, and ES-AV behaviors. This corresponds to the observation that a number of ion channels have been identified as potential targets of toluene exposure in vitro, and their function is affected in the same range of tissue concentrations as seen in the tests analyzed here (Bale et al., 2005; Bushnell et al., 2005; Cruz et al., 1998; Shafer et al., 2005). It is reasonable to expect that the same range of effects and sensitivities would also be found in humans acutely exposed to toluene. Indeed, ethanol, which shares many pharmacological properties with toluene and other solvents (Evans and Balster, 1991), is well known to produce a wide variety of effects in humans (e.g., Kennedy et al., 1992).
Figure 2 demonstrates that it is possible to empirically relate various effects in rats to human CRT in terms of equivalent doses. At first blush, this might seem to be a stretch of logic because not all of the rat dependent variables resemble human CRT in terms of behavioral topology and hypothetical cognitive mechanisms. However, because of the findings of a wide variety of behavioral effects in rats, one could argue that many CNS functions will be affected by toluene to some degree. Thus, it defensible to compare the effects of acute toluene exposure on various CNS functions both within and across species. In this view, the various behavioral tests are all different measures of a common generalized CNS effect. It may be argued that the DEEs from the various rat experiments are each predictors of human effects weighted by the rat sensitivity in the particular kind of experiment.
The above discussion has a particular importance to regulatory endeavors. According to the above, it may be hypothesized that the neurobehavioral effect that is most sensitive to disruption is the one in which least control is exercised over the measured behavior. This is complicated, however, by the likelihood that dependent variables measured in such situations will also have higher variability, thus reducing the sensitivity of a significance test.
In summary, given that the above hypotheses are valid and findings are replicable, to estimate human behavioral effects when human and rat behavioral controls are equal, a linear DEE with unity slope and zero intercept should be used. When estimating human behavior controlled under behavioral stringency of control that differ from those controlling behavior in the test animals (e.g., no explicit control over performance in humans estimated from rat behavior under rewarded or punished performance), the appropriate, nonlinear DEE should be used. This discussion has ignored the important problem of quantitatively or even qualitatively specifying the strength of the controls for behavior. This topic is beyond the scope of this paper.
Extrapolation
Because dose-effect data in both rats and humans are known for toluene, extrapolation across species is not necessary here. However, Benignus (2001) and Benignus et al. (2005) suggested that if the cross-species DEEs were known for several toxicants in a class, then empirical extrapolation from nonhuman species to untestable human exposures would be possible under certain conditions. In particular, cross-species extrapolation would be quantifiable if the relative potency in humans with respect to rats was always the same for any member of a class of toxicants, even though the absolute potency of the various toxicants might differ. In this case, the rat-human DEEs for the members of the toxicant class that have been characterized would have parameters that did not statistically differ from each other. If dose-equivalence parameters were known for n toxicants, an n-parameter DEE could be constructed with an n-variate sample distribution. When a case occurs in which a member of the toxicant class has been studied in rats but not humans, the extrapolation to humans can be made by finding the multivariate mean of the empirically known DEE parameters and substituting those numerical values into the general DEE. The uncertainty of the extrapolation would then be computed from the variance/covariance matrix of the sample distribution.
In this context, the dose-equivalence curve for toluene for any rat behavioral test can then be viewed as one member of a distribution of dose-equivalence curves for different solvents for that particular behavioral test. The application of this empirical approach to extrapolating from rats to humans depends on (1) finding sufficient data to construct DEEs for other solvents and (2) verification of the assumption that the relative human-rat potencies of the various solvents remain constant within that class of toxicants. Details of this method are given in Benignus (2001).
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
- Dose-effect equations were fitted to data on the behavioral and neurophysiological effects of acute toluene exposure in rats and humans. PBPK models were used to estimate internal doses, and outcome measures were transformed to a common scale of measurement.
- The sensitivity of a task to impairment by toluene exposure appears to be related to the strength of the control of performance in rat experiments. This implies that the acute effects of toluene may be overcome, at least partially, for situations where consequences for poor performance exist. This hypothesis requires explicit testing. The most sensitive neurobehavioral test appears to be one in which least control is exercised over the dependent variable by experimental procedures involving rewards or punishments in various schedules.
- Empirical DEEs were derived for describing the relative sensitivity of rats and humans, including confidence limits, to inhaled toluene exposure. The VEP measurement in rats was nearly as sensitive as the CRT task in humans.
- Relating human CRT to performance of other tasks in rats was defended on biological grounds by noting that toluene (in similar concentration ranges) has been shown to alter the functioning of several ion channels, suggesting a mechanism of action for the behavioral effects. Given that other VOCs and ethanol have similar effects on those ion channels and also produce similar suites of behavioral effects, it is reasonable to expect that many behaviors would be affected.
- If (1) DEEs were known for several solvents and (2) the relative rat and human sensitivity were constant across solvents, then data from a solvent for which only rat data existed could be used to extrapolate to human results. This would be done by estimating the parameters of the population DEE from the solvents for which data existed in both species.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
U. S. Environmental Protection Agency.
|
NOTES |
|---|
Disclaimer: This manuscript has been reviewed by the National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency and approved for publication. Approval does not signify that the contents necessarily reflect the policies of the agency, and the mention of trade names or commercial products do not constitute endorsement or recommendation for use.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUMMARY AND CONCLUSIONSFUNDINGREFERENCES |
|---|
Bale AS, Meacham CA, Benignus VA, Bushnell PJ, Shafer TJ. Volatile organic compounds inhibit human and rat neuronal nicotinic acetylcholine receptors expressed in Xenopus oocytes. Toxicol. Appl. Pharmacol. (2005) 205:77–88.[CrossRef][Web of Science][Medline]
Battig K, Grandjean E. Industrial solvents and avoidance conditioning in rats. A comparison of the effects of acetone, ethyl alcohol, carbon disulfide, carbon tetrachloride, toluene, and xylene on acquisition and extinction. Arch. Environ. Health (1964) 9:745–749.[Web of Science][Medline]
Benignus VA. Quantitative cross-species extrapolation in noncancer risk assessment. Regul. Toxicol. Pharmacol. (2001) 34:62–68.[CrossRef][Web of Science][Medline]
Benignus VA, Boyes WK, Bushnell PJ. A dosimetric analysis of behavioral effects of acute toluene exposure in rats and humans. Toxicol. Sci. (1998) 43:186–195.
Benignus VA, Bushnell PJ, Boyes WK. Toward cost-benefit analysis of acute behavioral effects of toluene in humans. Risk Anal. (2005) 25:447–456.[Web of Science][Medline]
Benignus VA, Coleman T, Eklund C, Kenyon E. A general physiological and toxicokinetic (GPAT) model for simulating complex toluene exposure scenarios in humans. Toxicol. Mech. Methods (2006) 16:27–36.[CrossRef]
Boyes WK, Bercegeay M, Krantz T, Evans M, Benignus VA, Simmons JE. Momentary brain concentration of trichloroethylene predicts the effects on rat visual function. Toxicol. Sci. (2005) 87:187–196.
Boyes WK, Bercegeay M, Krantz T, Kenyon E, Bale A, Shafer T, Bushnell PJ, Benignus VA. Acute toluene exposure alters rat visual function in proportion to momentary brain concentration. Toxicol. Sci. (2007) (in press).
Bruckner JV, Peterson RG. Evaluation of toluene and acetone inhalant abuse: II Model development and toxicology. Toxicol. Appl. Pharmacol. (1981) 61:302–312.[CrossRef][Web of Science][Medline]
Bushnell PJ. Concentration-time relationships for the effects of inhaled trichloroethylene on signal detection behavior in rats. Fundam Appl. Toxicol. (1997) 36:30–38.[CrossRef][Web of Science][Medline]
Bushnell PJ, Kelly KL, Crofton KM. Effects of toluene inhalation on detection of auditory signals in rats. Neurotoxicol. Teratol. (1994) 16:149–160.[CrossRef][Web of Science][Medline]
Bushnell PJ, Oshiro WM, Samsam TE, Benignus VA, Kenyon EM. A dosimetric analysis of the acute behavioral effects of inhaled toluene in rats. Toxicol. Sci. (2007) (in press).
Bushnell PJ, Shafer TJ, Bale AS, Boyes WK. Development and application of an exposure-dose-response model for organic solvents. Environ. Toxicol. Pharmacol. (2005) 19:607–614.[CrossRef]
Cherry N, Johnston JD, Venables H, Waldron HA, Buck L, MacKay CJ. The effect of toluene and alcohol on psychomotor performance. Ergonomics (1983) 26:1081–1087.[Medline]
Colotla VA, Bautista S, Lorenzana-Jimenes M, Rodrigues R. Effects of solvents on schedule-controlled behavior. Neurobehav. Toxicol. (1979) 1(Suppl. 1):113–118.[Medline]
Corso JF. The Experimental Psychology of Sensory Behavior (1967) New York: Holt, Rinehart and Winston.
Cruz SL, Mirshahi T, Thomas B, Balster RL, Woodward JJ. Effects of the abused solvent toluene on recombinant N-methyl-D-aspartate and non-N-methyl-D-aspartate receptors expressed in Xenopus oocytes. J. Pharmacol. Exp. Ther. (1998) 286:334–340.
Dews PB. Studies on behavior: I. Differential sensitivity to pentobarbital of pecking performance in pigeons depending on the schedule of reward. J. Pharmacol. Exp. Ther. (1955) 113:393–401.
Dick RB, Setzer JV, Wait R, Hayden MB, Taylor BJ, Tolos B, Putz-Anderson V. Effects of acute exposure of toluene and methyl ethyl ketone on psychomotor performance. Int. Arch. Occup. Environ. Health (1984) 54:91–109.[CrossRef][Web of Science][Medline]
Echeverria D, Fine L, Langolf G, Schork A, Sampaio C. Acute neurobehavioural effects of toluene. Br. J. Ind. Med. (1989) 46:483–495.[Web of Science][Medline]
Evans EB, Balster RL. CNS depressant effects of volatile organic solvents. Neurosci. Biobehav. Rev. (1991) 15:233–241.[CrossRef][Web of Science][Medline]
Gamberale F, Hultengren M. Toluene exposure II. Psychophysiological functions. Work-Environ.-Health. (1972) 9:131–139.
Geller I, Harman RJ, Tangle SR, Gauze EM. Effects of acetone and toluene vapors on multiple schedule performance of rats. Pharmacol. Biochem. Behav. (1979) 11:395–399.[CrossRef][Web of Science][Medline]
Harabuchi I, Kishi R, Ikeda T, Kiyosawa H, Miyake H. Circadian variations of acute toxicity and blood and brain concentrations of inhaled toluene in rats. Br. J. Ind. Med. (1993) 50:280–286.[Web of Science][Medline]
Ikeda T, Miyake H. Decreased learning in rats following repeated exposure to toluene: Preliminary report. Toxicol. Lett. (1978) 1:235–239.[CrossRef][Web of Science]
Iregren A, Akerstedt T, Anshelm Olsen B, Gamberale F. Experimental exposure to toluene in combination with ethanol intake. Scand. J. Work Environ. Health (1986) 12:128–136.[Web of Science][Medline]
Kennedy RS, Turnage JJ, Dunlap WP. The use of dose-equivalency as a risk assessment index in behavioral neurotoxicology. Neurotoxicol. Teratol. (1992) 14:167–175.[CrossRef][Web of Science][Medline]
Kenyon EM, Benignus VA, Eklund C, Bushnell PA. Modeling the toxicokinetics of inhaled toluene in rats: Influence of physical activity and feeding schedule. J. Toxicol. Environ. Health (2008) (in press).
Kishi R, Harabuchi I, Ikeda T, Yokota H, Miyake H. Neurobehavioural effects and pharmacokinetics in rats and their relevance to man. Br. J. Ind. Med. (1988) 45:396–408.[Web of Science][Medline]
Kleinbaum DG, Kupper LL, Muller KM. Applied Regression Analysis and Other Multivariable Methods (1988) 2nd ed. Boston: PWS-Kent.
Laties VG, Evans HL. Methylmercury-induced changes in operant discrimination by the pigeon. J. Pharmacol. Exp. Ther. (1980) 214:620–628.
Lehman AJ, Fitzhugh OG. 100-fold margin of safety. Assoc. Food Drug Off. U.S. Quarterly Bull. (1954) 18:33–35.
Mullin LS, Krivanek ND. Comparison of unconditioned reflex and conditioned avoidance tests in rats exposed by inhalation to carbon monoxide, 1,1,1-trichloromethane, toluene or ethanol. Neurotoxicology (1982) 3:126–137.[Web of Science][Medline]
Nevin JA. Response strength in multiple schedules. J. Exp. Anal. Behav. (1974) 21:389–408.[CrossRef][Web of Science][Medline]
Olson BA, Gamberale F, Iregren A. Coexposure to toluene and p-xylene in man: Central nervous functions. Br. J. Ind. Med. (1985) 42:117–122.[Web of Science][Medline]
Oshiro WM, Krantz QT, Bushnell PJ. Repeated inhalation of toluene by rats performing a signal detection task leads to behavioral tolerance on some performance measures. Neurotoxicol. Teratol. (2007) 29:247–254.[CrossRef][Web of Science][Medline]
Rahill AA, Weiss B, Morrow PE, Frampton MW, Cox C, Gibb R, Gelein R, Speers D, Utell MJ. Human performance during exposure to toluene. Aviat. Space Environ. Med. (1996) 67:640–647.[Medline]
Shafer TJ, Bushnell PJ, Benignus VA, Woodward JJ. Perturbation of voltage-sensitive Ca2+ channel function by volatile organic solvents. J. Pharmacol. Exp. Ther. (2005) 315:1109–1118.
Shigeta S, Aikawa H, Misawa T, Kondo A. Effect of single exposure to toluene on Sidman avoidance response in rats. J. Toxicol. Sci. (1978) 3:305–312.[Medline]
Shigeta S, Misawa T, Aikawa H. Effects of concentration and duration of toluene exposure on Sidman avoidance in rats. Neurobehav. Toxicol. (1980) 2:85–88.[Web of Science][Medline]
Thompson DM. Stimulus control and drug effects. In: Contemporary Research in Behavioral Pharmacology—Blackman DE, Sanger DJ, eds. (1978) New York: Plenum. 159–207.
Von Oettingen W, Neal P, Donahue D. The toxicity and potential dangers of toluene. JAMA (1942) 118:579–584.
Wada H, Hosokawa T, Saito K. Single toluene exposure and changes of response latency in shock avoidance performance. Neurotoxicol. Teratol. (1989) 11:265–272.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  V. A. Benignus, P. J. Bushnell, W. K. Boyes, C. Eklund, and E. M. Kenyon Neurobehavioral Effects of Acute Exposure to Four Solvents: Meta-analyses Toxicol. Sci., June 1, 2009; 109(2): 296 - 305. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. K. Boyes, M. Bercegeay, W. M. Oshiro, Q. T. Krantz, E. M. Kenyon, P. J. Bushnell, and V. A. Benignus Acute Perchloroethylene Exposure Alters Rat Visual-Evoked Potentials in Relation to Brain Concentrations Toxicol. Sci., March 1, 2009; 108(1): 159 - 172. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||