Neurotoxicological and Statistical Analyses of a Mixture of Five Organophosphorus Pesticides Using a Ray Design
http://www.100md.com
《毒物学科学杂志》
Neurotoxicology Division, NHEERL/ORD, US EPA, RTP, North Carolina
Department of Biostatistics, Virginia Commonwealth University, Richmond, Virginia
Experimental Toxicology Division, NHEERL/ORD, US EPA, RTP, North Carolina
ABSTRACT
Environmental exposures generally involve chemical mixtures instead of single chemicals. Statistical models such as the fixed-ratio ray design, wherein the mixing ratio (proportions) of the chemicals is fixed across increasing mixture doses, allows for the detection and characterization of interactions among the chemicals. In this study, we tested for interaction(s) in a mixture of five organophosphorus (OP) pesticides (chlorpyrifos, diazinon, dimethoate, acephate, and malathion). The ratio of the five pesticides (full ray) reflected the relative dietary exposure estimates of the general population as projected by the US EPA Dietary Exposure Evaluation Model (DEEM). A second mixture was tested using the same dose levels of all pesticides, but excluding malathion (reduced ray). The experimental approach first required characterization of dose-response curves for the individual OPs to build a dose-additivity model. A series of behavioral measures were evaluated in adult male Long-Evans rats at the time of peak effect following a single oral dose, and then tissues were collected for measurement of cholinesterase (ChE) activity. Neurochemical (blood and brain cholinesterase [ChE] activity) and behavioral (motor activity, gait score, tail-pinch response score) endpoints were evaluated statistically for evidence of additivity. The additivity model constructed from the single chemical data was used to predict the effects of the pesticide mixture along the full ray (10–450 mg/kg) and the reduced ray (1.75–78.8 mg/kg). The experimental mixture data were also modeled and statistically compared to the additivity models. Analysis of the 5-OP mixture (the full ray) revealed significant deviation from additivity for all endpoints except tail-pinch response. Greater-than-additive responses (synergism) were observed at the lower doses of the 5-OP mixture, which contained non-effective dose levels of each of the components. The predicted effective doses (ED20, ED50) were about half that predicted by additivity, and for brain ChE and motor activity, there was a threshold shift in the dose-response curves. For the brain ChE and motor activity, there was no difference between the full (5-OP mixture) and reduced (4-OP mixture) rays, indicating that malathion did not influence the non-additivity. While the reduced ray for blood ChE showed greater deviation from additivity without malathion in the mixture, the non-additivity observed for the gait score was reversed when malathion was removed. Thus, greater-than-additive interactions were detected for both the full and reduced ray mixtures, and the role of malathion in the interactions varied depending on the endpoint. In all cases, the deviations from additivity occurred at the lower end of the dose-response curves.
Key Words: organophosphate mixtures; cumulative risk; neurotoxicity; chlorpyrifos; acephate; malathion; diazinon; dimethoate.
INTRODUCTION
The evaluation of potential health risks of exposures to chemical mixtures is a growing area of environmental and toxicological research. Research needs in this area were described recently in a consensus paper published by the Society of Toxicology Expert Panel on Mixtures (Teuschler et al., 2002). This Expert Panel advocated a scientifically based approach to the risk assessment of mixtures, which would involve biologically based hypotheses and experimental approaches in the design of laboratory studies. Because of the uncertainty about the health risks of exposures to multiple chemicals, research concerning chemical mixtures will be one of the key environmental health issues for several decades to come.
Recent advances in the statistical modeling of mixture data have moved the field from the analysis of binary combinations to the analysis of multiple chemicals using various designs. The classical statistical approach to the analysis of binary chemical mixtures is to construct full dose-response curves for each compound in the presence of a range of doses of the second compound (isobolographic analyses; Gessner, 1974). For interaction studies using more than two chemicals, response-surface methodology frequently requires factorial-type designs (e.g., Carter, 1995; Carter et al., 1985; Solana et al., 1990a,b). A more economical (e.g., fewer treatment groups, fewer test subjects, more straightforward analyses) approach is to compare an additivity model to selected mixtures of the compounds that are experimentally tested (Gennings, 1995; Gennings et al., 2002). An extension of this approach is to test combinations of chemicals in fixed ratios or proportions, i.e., a ray design (e.g., Brunden and Vidmar, 1989; Finney, 1964). Characteristics of the fitted dose-response curve along the ray using total dose for the dependent variable can be compared to that predicted under additivity. This methodology shows great promise and has recently been modified for use with neurotoxicity data using a relevant pesticide mixture (Casey et al., 2004, 2005, in press; Gennings et al., 2004).
Given that pesticide application patterns generally result in exposure to mixtures instead of single chemicals, the detection and characterization of pesticide interactions is critical for regulatory decisions. Studies of mixtures toxicity can be simplified by grouping chemicals with a known or presumed common mode of action. This approach was codified in the 1996 Food Quality Protection Act, which directed the U.S. Environmental Protection Agency (US EPA) to base regulatory decisions on mixtures of pesticides which act through a common mode of action, rather than on individual pesticides. Organophosphorus (OP) pesticides which inhibit acetylcholinesterase (AChE), a widely used class of pesticides, were the first class to undergo a cumulative (i.e., mixtures) risk assessment (US EPA, 2002). Despite waning household uses, OPs are still used widely on a variety of foods and commercial crops, and there is potential for multiple OP exposures. A default assumption in these assessments was dose-additivity.
Previous research on OP pesticide interactions has involved primarily lethality studies of binary chemical combinations in adult laboratory animals (e.g., Casida et al., 1963; Cohen, 1984; DuBois, 1961; reviewed in NRC, 1989). For example, DuBois (1961) recorded lethality in rats administered pairs of 13 OP pesticides, and reported that the results showed additivity in 21 of 43 pairs; four pairs showed marked synergy. Potentiation of malathion toxicity by other OPs is often cited as an example of interactions due to inhibition of detoxification pathways (Casida et al., 1963; Cohen and Murphy, 1971a,b; DuBois, 1969; Frawley et al., 1957; Murphy and DuBois, 1957; Murphy et al., 1959; Su et al., 1971). More recently, some have begun to investigate sequential dosing of binary combinations (Karanth et al., 2001, 2004) and, using lethality and symptomatology as endpoints, have demonstrated that particular dose sequences can produce marked potentiation. Some other interaction studies involve OPs (mostly nerve agents) given with other chemical classes, i.e., antidotes and pre-treatments for exposure (e.g., Solana et al., 1990a,b). These studies do not address the larger issue of cumulative mixture to numerous pesticides at lower dose levels that do not produce such overt toxicity.
OP pesticides produce specific neurobehavioral effects, many of which are due to their AChE-inhibiting properties. Exposure to high doses produces cholinergic signs of toxicity, including salivation, lacrimation, muscle fasciculations, polyuria, diarrhea, progressing to tremors, respiratory depression, and death (Ecobichon, 1996). Behavioral effects that occur at lower doses include lowered activity, antinociception, and gait abnormalities. These functional changes often correlate well with depression of brain and/or blood AChE (Moser, 1995; Nostrandt et al., 1997). A combination of both AChE measurements and functional outcomes provides a more detailed evaluation of the neurotoxicity following OP exposure.
In this study, we tested the hypothesis of dose-additivity for a mixture of five OP pesticides (acephate, diazinon, chlorpyrifos, dimethoate, and malathion). Since non-additivity was observed for the 5-OP mixture, we hypothesized this may be due to malathion; therefore, a second mixture excluding malathion was tested to examine the influence of malathion on the observed interaction. Neurobehavioral measures and cholinesterase inhibition in blood and brain were evaluated at a single time point (the time of peak effect of the component pesticides) following acute exposure to the mixture. The mixing ratio of the pesticides was determined to reflect potential dietary exposure for the general population.
MATERIALS AND METHODS
Chemicals.
Chlorpyrifos (O,O-diethyl-O-[3,5,6-trichloro-2-pyridyl]phosphorothioate; purity, 99.2–99.9%), acephate (O,S-dimethyl acetylphosphoroamidothioate; purity, 98.7%), diazinon (O,O-diethyl-O-[2-isopropyl-4-methyl-6-pyridimyl]phosphorothioate; purity, 99.2–99.4%), and malathion (S-[1.2-dicarbethoxyethyl]-O,O-dimethyl dithiophosphate; purity, 98.2–99.5%) were obtained from Chem Service (West Chester, PA), and dimethoate (O,O-dimethyl-S-[N-methylcarbomoylmethyl]phosphorodithioate; purity, 98.7%) was a gracious gift from Cheminova (Lemvig, Denmark). Acephate was dissolved in deionized water. The other pesticides were dissolved in a vehicle of corn oil that contained 5% ethanol, as this was required to dissolve the dimethoate.
For the radiometric cholinesterase (ChE) assay, [3H]acetylcholine iodide (specific activity, 82.0 mCi/mmol) was obtained from DuPont/NEN Life Science Products (Boston, MA) and unlabeled acetylcholine iodide from Sigma (St. Louis, MO). All other reagents were obtained from commercial sources and were of the highest available grade.
Animals.
Adult (66–70 days of age at testing) Long-Evans male rats were obtained from Charles River Laboratories (Raleigh, NC) and individually housed on heat-treated pine shavings with feed (Purina Rodent Chow 5001) and water (filtered tap) freely available. The animal facility was accredited fully by the Association for Assessment and Accreditation of Laboratory Animal Care International and maintained at 70 ± 2°F, 50 ± 10% humidity, with a 12-h light/dark cycle.
Behavioral testing.
Neurobehavioral function was evaluated using several endpoints shown previously to be sensitive to OP exposure (Moser, 1995). Upon removing the rat from the cage, the observer noted the presence of miosis, mouth smacking, salivation, or lacrimation. The rat was then placed in an open field for observation, and tremors, gait abnormality, and the response to a tail pinch were scored. Motor activity was then measured during a 30-min session, using an automated chamber shaped like a figure eight (Reiter, 1983). The same observer conducted all studies, and was unaware of the dose level of each rat.
Immediately after the motor activity assessment, rats were decapitated quickly under CO2-induced anesthesia. The brain was rapidly removed, and trunk blood was collected in heparinized tubes and diluted. All tissues were stored at –80°C until the time of assay.
Cholinesterase assay.
A radiometric assay was used to determine brain and blood ChE activity (Johnson and Russell, 1975). Both tissues were diluted in two volumes of 0.1 M sodium phosphate buffer (pH 8.0) with 1% Triton X-100, followed by homogenization (brain only) for 30 s (Polytron homogenizer, Kinematica Model PT3100, Littau, Switzerland). The final acetylcholine iodide concentration was 1.2 mM.
Chemical treatments.
Single-chemical dose-response data were collected at the time of peak effect, previously determined by pilot and/or published studies (Moser, 1995; Moser and Padilla, 1998; Poet et al., 2004; Timchalk et al., 2002; Wu et al., 1996). Where published data were not available, the time to peak effect was evaluated by monitoring behavioral changes (lacrimation, salivation, tremors, uncoordinated gait, ataxia, decreased arousal, and fasciculations) at hourly intervals. The time(s) of maximum effect on each measure was noted, and summed across doses. Calculated this way, the sum of peak effects over three doses at 2, 3, 4, and 5 h were: dimethoate, 10, 15, 4, and 2 and acephate, 11, 13, 11, and 10. Malathion had no observable effects at the dose levels tested. The time course was similar for the four pesticides for which effects were observed, with maximum effects protracted over 3 to 4 h. We therefore chose the testing time to encompass the peak times of each chemical. Neurobehavioral testing began 3 h after dosing, and rats were euthanized between 4 to 4 h after dosing.
All pesticides were administered by oral gavage (water solutions, 1 ml/kg; corn oil suspensions, 2 ml/kg). Because of the two different vehicles, the mixture study required that the rats be gavaged twice. The water dosing vehicle (acephate) was given first, followed immediately by the corn oil/ethanol dosing vehicle (containing the other pesticides). Although the single chemical data were collected using a single gavage dose, confirmatory studies were conducted to compare the effects of the two-gavage dosing procedure to the single gavage, and no differences were detected.
The single-chemical dose-response curves included at least five dose levels plus control, with n = 8/dose group. For chlorpyrifos and diazinon, a second study was conducted to obtain an adequate range of responses. The data for ChE and motor activity in the two studies were compared and found not to differ significantly; the data were then combined for all endpoints. The dose-response curve for malathion included only two dose levels, since both were non-effective on most endpoints and the highest dose tested, 500 mg/kg, was considered to be adequately high for these studies.
Dietary exposure was estimated using the Dietary Exposure Evaluation Model (DEEM-FCID) software (US EPA, 2001). This is a probabilistic analysis conducted by combining representative data on concentrations of OP pesticides on foods, market basket monitoring studies, residue decline and degradation studies, and consumer practices such as washing, cooking, etc. with distributions of anticipated consumption of these foods (USDA's Continuing Survey of Food Intake by Individuals) by different segments of the U.S. population. The daily anticipated exposure value designated for 95% of the general population was obtained for each pesticide, and the ratio of these values was used for the mixture. The proportions of the chemicals in the mixture were: chlorpyrifos 0.031; acephate 0.040; diazinon 0.002; dimethoate 0.102; and malathion 0.825.
The mixture based on this composition was considered the "full ray." The mixture dose levels ranged from 10 to 450 mg/kg, and the pesticide proportions (mixing ratio) remained the same across dose levels. Table 1 presents the actual doses used in each mixture dose. The second mixture ("reduced ray") did not contain malathion, but did have the remaining four OPs at the same dose levels, and therefore the same relative ratios, as in the full ray. The reduced ray mixture dose levels ranged from 1.75 to 78.8 mg/kg, and dimethoate was the most prevalent pesticide, comprising 58.5% of the 4-OP mixture.
The mixture studies were conducted using six dose levels of each mixture plus control. In each mixture study, a single dose group of each of the individual chemicals was included to assure that the chemical response was the same as that predicted from the earlier data ("positive controls"). Furthermore, a single dose of the 5-OP mixture (100 mg/kg) was repeated to replicate the effects of that dose in the first study. These repeated-dose data were not used in the statistical models.
Power analyses based on the motor activity variable indicated that a sample size of 12/mixture dose group was needed to detect a 25% decrease in the slope of the dose-response curve compared to that predicted under additivity with 70% power (Casey et al., in press). Thus, the conduct of the mixture study included six mixture dose groups (n = 12/dose group), control (n = 8), and a single dose level of each chemical (n = 8/chemical).
Statistical methods.
Data for eight behavioral endpoints were collected, but only three (motor activity, gait abnormality, and tail pinch response), along with the ChE measurements, were analyzed using the additivity models. Gait abnormality and the tail pinch response were ranked, but for the analyses the data were converted to a binary response (i.e., presence or absence of gait abnormality, normal or decreased tail pinch response). To constrain the probability of a response to be between 0 and 1, a logit link function was used. Total activity counts during the motor activity session, as well as blood and brain ChE activity, were continuous variables. Due to differences in the control brain ChE activity in one study, the data for this measure were analyzed as a proportion of the respective control value.
The additivity model is based on the single chemical dose-response data and a definition of dose additivity (e.g., Berenbaum, 1985). If the observed response along the fixed-ratio mixture ray is more extreme than that predicted under additivity, then it is reasonable to claim a greater-than-additive response (i.e., synergism); if the response is less extreme than that predicted under additivity, a less-than-additive response (i.e., antagonism) can be claimed; otherwise, the curves are coincident and departure from additivity cannot be claimed for the mixture ray(s) considered.
The development of the methods and preliminary use of these data are presented in several manuscripts (Casey et al., 2004, 2005, in press; Gennings et al., 2004). A threshold additivity model (Gennings et al., 1997) was used for motor activity and brain ChE, whereas for blood ChE the data fit a threshold outside the experimental range, and a generalized linear model was used. For each endpoint, the methodology uses the single chemical dose-response data to develop a model that predicts the response of the mixture ray under additivity. This predicted additivity model for decreasing curves, which relates the doses of the chemicals under study to the mean through a link function, g(μ;), and possibly a set of nonlinear parameters, , is written as:
where g(μadd;) is the link function (McCullagh and Nelder, 1989) of the response of interest, xi is the dose of the ith chemical, 0 is the unknown parameter associated with the intercept (common intercept across all curves), i is the unknown slope parameter for the ith chemical, is the unknown parameter associated with the threshold, and is the dose threshold (in mg/kg units) parameter for the ith chemical. If none of the dose threshold parameters were within the experimental region, the threshold additivity model was replaced by the corresponding generalized linear model:
The single chemical data were used to estimate the additivity model. Let [a1, ..., ac] represent the proportions of the c (here, c = 5) chemicals in the fixed-ratio mixture. The additivity model was used to predict the response of the mixture along the fixed-ratio ray in terms of total dose, t. Noting that xi = ait, the slope of the additivity model along the mixture ray is given by
For comparison, the mixture data along the fixed-ratio ray were fit to a similarly parameterized model:
When the threshold is estimated outside of the experimental region, the parameter is removed and the following model was used:
Before testing for departure from additivity, a goodness-of-fit test was performed on the model to ensure that it adequately fit the mixture data. In the event of significant lack-of- fit, higher-order terms were added until an appropriate model was found for the mixture data. The test of additivity is a test of coincidence of these two models along the specified fixed-ratio mixture. For example, if the threshold models are fit to the single chemical and mixture data, then the hypothesis of additivity is given by:
Following Gennings et al. (2002) and Casey et al. (2004), likelihood ratio tests were used to test this hypothesis of additivity.
The full and reduced rays were compared using the method described by Casey et al. (2004), which required correcting the curve from the full ray so that both rays fall in the same dose range. That is, the effect of removing malathion from the mixture was tested by comparing the dose-response curves for the two rays while noting that treduced = tfull(1 – a5) for a5 = 0.825, the proportion of malathion in the full mixture. Thus, the hypothesis of no malathion effect on the mixture is given by
Following Casey et al. (2004), likelihood ratio tests were used to test this hypothesis.
The estimation of these models was done in SAS version 8.2 using the maximum quasi-likelihood criterion in a Nelder-Mead direct-search algorithm embedded in the nonlinear programming (NLP) procedure in SAS and using the maximum quasi-likelihood criterion using a Fisher scoring algorithm in PROC GENMOD in SAS.
RESULTS
No lethality or unexpected toxicity occurred during any of the chemical testing. At the highest doses tested, mild signs of cholinergic stimulation were evident, including uncoordinated gait, lowered activity, lacrimation, miosis, and mild fasciculations or tremors. Malathion produced no observable effects. The control groups for each of the individual dose-response curves, as well as for the mixture study, were similar in almost all responses. The only exception was brain ChE, for which the control data from one study was higher than all the others, even though the inhibition relative to control was the same as a previous study. The tissues were analyzed a second time with the same result; thus, brain ChE data were analyzed as a proportion of the concurrent control mean value. The models for each endpoint were then fit simultaneously in an overall model with a common intercept parameter.
Each mixture study included one dose level of each individual chemical, as a "positive control" group (open circles on Figures 1, 3, 6, and 8). For all endpoints, the "positive control" data were not significantly different from the dose-response curves fit for each chemical.
Brain ChE
The individual dose-response data for brain ChE are presented in Figure 1. The curves were plotted for the individual dose-response curves, and the open symbols indicate the "positive control" data taken as part of the mixture studies. It is evident that the responses to the single doses were generally close to the curve fitted to the dose-response data, and they were not statistically different from the fitted curve. Hatch marks on the graphs indicate the location of the dose levels used in the mixture studies; these doses ineffective for some and in the effective dose-range for other pesticides. For example, the dose levels of diazinon used in the mixtures were far below the effective dose range for diazinon alone, whereas most of the dimethoate dose levels in the mixtures produced slight to moderate effects. The largest proportion of the mixture was malathion, which had no effect on this endpoint (or any other, except blood ChE). Being the second most prevalent OP in the mixture, dimethoate could be considered to be driving the mixture in terms of biological activity. For brain ChE, the higher mixture dose levels included active dose levels of both acephate and dimethoate.
The threshold parameter for the additivity models for the full and reduced rays was significant (p < 0.001) indicating that at least one of the curves showed a threshold within the experimental dose range. Specifically, the thresholds for chlorpyrifos and diazinon were 3.2 mg/kg and 17.7 mg/kg, respectively (Fig. 1). The slope parameter for malathion was not significant (p = 0.646), and therefore it was removed from the additivity models.
The additivity models predicted by the individual dose-response curves for the full and reduced rays are shown in Figure 2, along with the model fit to the experimental mixture data. The likelihood ratio test of additivity was rejected (p < 0.001) for both curves, with the experimental mixture model showing greater inhibition than the predicted additivity model, indicating greater-than-additive effects (synergism) in the lower dose range. A single dose level of the 5-OP mixture (100 mg/kg) was repeated several weeks later (open circle in full ray graphs, Figures 2, 4, 5, 7, and 9), and the data generally replicated the previous effect at that dose.
The predicted additivity models along both rays showed threshold values (full ray, 13.4 mg/kg; reduced ray, 2.3 mg/kg), whereas the models fitted to the mixture data revealed thresholds less than the lowest doses tested. This reveals a significant threshold shift, indicating an interaction even at the low end of the dose-response curves. As described by Casey et al. (2004), the full and reduced rays can be compared directly. For brain ChE inhibition, the two models were not significantly different (p = 0.421), so there was no evidence that malathion was responsible for the observed interactions.
The magnitude of change is important for understanding the biological significance of non-additivity. For purposes of illustration, ratios of point estimates have been calculated from the equations using the predicted additivity model and the actual mixture model; these are presented in Table 2. The purpose of these estimates is to show the ratio of the effective doses, or magnitude of change. Doses were calculated that (1) produced a 20% (ED20) or 50% (ED50) change from control (ChE activity or activity counts), or (2) produced an abnormal response in 20% (ED20) or 50% (ED50) of the treatment group (gait score, tail-pinch response). For brain ChE inhibition, the point estimates for a 20% effect on brain ChE illustrate that the shift in effective doses was about two-fold, and somewhat less at the ED50.
Evaluation of the thresholds provided additional information with which to compare the curves. The thresholds for the predicted additivity curves were 13.4 mg/kg for the full ray, and 2.3 mg/kg for the reduced ray. The experimental mixture model revealed much lower thresholds (2.3 mg/kg for the full ray, 0.12 mg/kg for the reduced ray) which represent 6-fold and 19-fold decreases from the predicted full and reduced rays, respectively.
Blood ChE
The individual dose-response data for blood ChE are presented in Figure 3. The threshold parameter was estimated outside of the experimental region for all individual dose-response curves, and was therefore removed from the model. The slope parameters for all pesticides were negative and significant. This was true even for malathion, even though the highest dose tested produced only about 50% inhibition. As with the brain ChE, the "positive control" data fell generally close to, and were not statistically different from, the curve fitted to the dose-response data.
The additivity models predicted by the individual dose-response curves are shown in Figure 4, along with the models fit to the experimental mixture data. For both the full and reduced rays, the likelihood ratio test of additivity was rejected (p < 0.001), and there is evidence of synergism in the low dose range. Comparison of the two mixture rays showed that they were significantly different from each other (p < 0.001), indicating that, for this measure, malathion interacted with the other four pesticides in the mixture. Indeed, the reduced ray showed a greater deviation from the predicted model. Table 2 presents the ED20 and ED50 for blood ChE inhibition for the predicted and the actual mixture data. There was only a small shift in doses producing 20% inhibition, whereas for the reduced ray, the shift in the ED50 was almost two-fold.
Tail-Pinch Response
Tail-pinch response was ranked on an ordinal scale, but for these analyses data were converted to binary responses. OP pesticide exposure only decreased the tail-pinch response, but the data were not particularly consistent, especially the dose-response curves for chlorpyrifos and diazinon (data not shown). For each of these, the dose-response data were collected twice, and there were differences between them. Furthermore, the "positive controls" sometimes appeared different from the fitted curve. Therefore, adequacy of the model fit may be questionable. The mixture data are shown in Figure 5, and the likelihood ratio test of additivity was not rejected (p = 0.499). Since the mixture rays showed only additivity, point estimates were not determined.
Motor Activity
A preliminary analysis of the motor activity data is presented in Casey et al. (2004); however, more specific data are presented here. The individual dose-response curves shown in Figure 6 identified significant thresholds within the dose range tested for all pesticides except malathion, which produced no effect on this measure up to 500 mg/kg. The individual thresholds were: acephate, 5.7 mg/kg; chlorpyrifos, 4.7 mg/kg; diazinon, 25 mg/kg; and dimethoate, 7.5 mg/kg (Casey et al., 2004).
There was sufficient evidence to reject the hypothesis of additivity for both the full and reduced rays (p < 0.001), shown in Figure 7. Comparison of the full and reduced rays showed no difference (p = 0.378), indicating that malathion did not influence the interaction. The point estimates in Table 2 illustrate an approximately two-fold shift in the ED20 dose, and somewhat less difference in the ED50 values. The thresholds for the predicted additivity curves were significant (36.6 mg/kg for the full ray, 6.4 mg/kg for the reduced ray); however, the fitted mixture models had no significant threshold parameter. Quantification is therefore problematic, but this indicates at least a three-fold shift in the thresholds since they must be below the lowest doses tested (<10 mg/kg for the full ray, <1.75 mg/kg for the reduced ray).
Gait Score
The gait score data (incidence of abnormal gait) were analyzed and presented in Gennings et al. (2004). Individual dose-response curves are presented in Figure 8, and the mixture data are presented in Figure 9. The full mixture data was close to significance for a deviation from additivity (p = 0.053), but the reduced ray was not different (p = 0.314) from the additivity model. This indicates that the deviation of the full mixture from additivity was due to the malathion, and the comparison of the full and reduced rays showed that they were different (p = 0.014). The point estimates in Table 2 illustrate that the full ray data produced point estimates about 1.7-fold less than predicted under additivity.
Other Endpoints
The other endpoints evaluated in this study were not subject to statistical analyses, but some patterns were clear in qualitatively assessing the data. When comparing the signs observed in the mixture studies with those produced by the individual pesticides, several patterns become obvious. Since a large proportion of the mixture was malathion and dimethoate, it was of interest to explore the contribution of those specific pesticides. Since malathion alone produced no behavioral changes, we focused on the effects of dimethoate. Mixture data for several endpoints are presented in Figure 10, plotted as a function of the dimethoate dose contained in each corresponding mixture dose level. The dimethoate single chemical dose-response is also plotted, to compare the effects of dimethoate alone and in the mixture. The increasing incidence of lacrimation is shown in Figure 10. The only pesticide which produced lacrimation at doses that were used in the mixture was dimethoate, albeit only a small proportion (1 of 8). In contrast, the higher doses of the mixture produced lacrimation in at least 20% of the animals, and except for the highest dose, there was not much difference between the full and reduced rays. This again suggests a non-additive response, and that malathion was not a factor at the lower doses; however, there is a suggestion of an influence at the highest dose. The incidence of mouth-smacking showed a similar pattern (data not shown). In contrast to this pattern, the incidence of miosis suggested more of an additive effect (Fig. 10). All doses of the mixture caused miosis in at least some rats, and the full and reduced rays look very similar. This effect could be due mostly to the dimethoate in the mixture, as that dose-response was close to the ray data. The other pesticides were not effective on this measure at the doses used in the mixture, even at the highest dose. The incidence of mild tremors mirrored that of miosis (data not shown).
DISCUSSION
This dose-additive model revealed greater-than-additive (synergistic) responses, using both neurochemical and behavioral endpoints, when five OPs were administered concurrently. Malathion could be held responsible for the non-additivity observed in the gait score measure, but not for the ChE and motor activity measures. Although the synergism was of relatively small magnitude (1.5–2-fold) when comparing point estimates (ED20s and ED50s), it was consistent across four of the five endpoints tested. The threshold shift observed at the low end of the dose-response curves was at least 3-fold for motor activity, and 6- to 19-fold for brain ChE. However, since these values fall outside the experimental region (i.e., doses tested), this analysis should be interpreted with caution.
For all endpoints, the deviations from additivity were observed in the lower end of the dose range (i.e., doses producing less than 20% effect). The high end of the dose range did not show appreciably greater inhibition. This finding appears contrary to the general assumption that deviations from additivity will become evident only at higher doses where metabolism becomes rate-limiting (e.g., Carpy et al., 2000; NRC, 1989). Of course, "high" and "low" doses are relative terms, but in this study, "low" was defined by effects on brain ChE. The doses of the individual chemicals used for all but the higher mixture doses were ineffective in inhibiting brain ChE, with the exception of acephate and dimethoate. The threshold shift was detected below the lowest mixture dose, which was comprised of ineffective doses of almost all chemicals. While the doses used in this mixture are higher than environmental levels to which the population may be exposed chronically, these acute doses are still much lower than those producing lethality, convulsions, or extreme cholinergic signs.
There could be both toxicokinetic and/or dynamic mechanisms for non-additivity with the OPs included in this mixture. Following the demonstration by Frawley and coworkers (1957) of potentiation between malathion and EPN, subsequent studies reported such non-additivity between other OPs as well, including TOTP, TOCP, parathion, paraoxon, soman, sarin, and others (e.g., Casida et al., 1963; Clement, 1984; Cohen and Murphy, 1971a,b; DuBois, 1961; Fleisher et al., 1963; Lauwerys and Murphy, 1969; Murphy et al., 1959, 1976; Su et al., 1971). These OPs compete for binding sites and/or block hydrolysis by carboxylesterases (CaEs), which are non-target esterases that play a major role in their detoxification (Maxwell, 1992). It was suggested that inhibition of CaE could be predictive of the potential for non-additive interactions with other OPs; that is, if the pesticide was more potent in inhibiting CaE compared to ChE, it could be expected to potentiate malathion at doses lower than the no-effect levels for ChE inhibition (Cohen and Murphy, 1974; DuBois, 1969).
Of the OPs tested in the present mixture, chlorpyrifos and dimethoate have been reported to inhibit CaE at doses less than or equal to those that inhibit ChE (Chambers and Carr, 1993; Chambers et al., 1994; Chanda et al., 1997; Su et al., 1971), but less is known about diazinon and acephate. CaEs hydrolyze malathion but are inhibited by other OPs; thus, interactions could be a function of blocking malathion detoxification or competition for binding to the limited number of CaE molecules. The latter explanation is suggested by the finding of synergism even without malathion in the mixture. Considering stochiometric binding of the enzyme and inhibitor, however, it would be expected that this kinetic factor would be dose-dependent, and show greatest effects at the highest doses. The only endpoint which showed greater effect at higher doses was blood ChE inhibition, but in that case the larger deviation from additivity was observed without malathion in the mixture. Direct measurement of CaE activity in liver and blood could aid in determining the role of CaE detoxification in these interactions.
While the interaction involving the CaE pathway is considered a primary mechanism of potentiation for some OPs, the degree of CaE inhibition is not sufficient in some mixtures to predict the potentiation observed (Cohen and Murphy, 1971a; Murphy et al., 1976; Seume et al., 1960). Other esterases, e.g., A-esterases, also hydrolyze chlorpyrifos and diazinon (Maxwell, 1992; Pond et al., 1995). Since these enzymatic reactions are non-saturable, it is less likely that this plays a role in the interactions. In addition, some phosphorothionates, including diazinon, induce hepatic microsomal enzymes which could alter activation of the pesticides which are not active in the parent form (DuBois, 1969; Uchiyama et al., 1975). Chlorpyrifos, malathion, dimethoate, and diazinon must be metabolized to their active oxons, and acephate is metabolized to a more active form (methamidophos) (Hussain et al., 1985; Pope, 1999). A study with direct administration of the active metabolites could remove the potential interaction at the activation step, and therefore reveal its influence; however, this would also greatly alter the time to onset of effects. In vitro measurement of the activation/detoxification produced by microsomal enzymes may provide support for some of these hypotheses.
Microsomal enzymes also serve to detoxify some of the OPs, which would have an influence in the opposite direction of activation. It has been suggested, however, that increasing the rate of microsomal detoxification was more important than the increased activation (DuBois, 1969). Due to the fairly complex metabolic pathways of these pesticides, and several different potential sites for interactions, physiologically based pharmacokinetic models may best be employed to determine the final outcome. Such approaches have recently been described for other pesticide mixtures (e.g., El-Masri et al., 2004; Timchalk et al., in press), and a similar approach could be used with this mixture.
Another factor which could influence the mixture response is the time of testing for the acute effects. The individual pesticides tested here showed the same time of peak effect, that is, about 3.5–4 h after dosing. These pesticides were administered together, although others have reported that varying the time of administration can markedly change the interaction (DuBois, 1969; Karanth et al., 2001, 2004). While this time-course has been published for chlorpyrifos (Moser, 1995; Moser and Padilla, 1998; Timchalk et al., 2002) and diazinon (Moser, 1995; Poet et al., 2004; Wu et al., 1996), pilot studies (described in the Methods) established the time-course data for the other pesticides. A potential mechanism for the interaction would be a change in the time-course, perhaps via altered absorption or metabolic activation. Since we did not conduct a time-course study of the mixture per se, we cannot make definitive statements regarding this possibility; however, follow-up studies could address this issue. Casual observation of the animals during the study, however, did not suggest that the effects were accelerated. Since all rats were euthanized for tissues by 4.5 h after dosing, we did not evaluate any measure of recovery.
We measured ChE inhibition as the accepted biomarker of exposure, and perhaps of biological effect, of these pesticides. There are, however, other dynamic actions of these pesticides which may serve to modulate the level of inhibition as well as the functional outcome of the ChE inhibition. Recent papers have suggested a peripheral binding site on the acetylcholinesterase enzyme which may be allosterically modulated by some OPs; this raises the possibility of interactions at that site that could alter the actual inhibition of the catalytic site (Kardos and Sultatos, 2000; Kousba et al., 2004). In addition, several of the pesticides in this mixture have been reported to have direct actions on cholinergic receptors, for example, chlorpyrifos oxon and malaoxon bind muscarinic receptors in vitro and inhibit cAMP formation (Ward and Mundy, 1996), and chlorpyrifos alters presynaptic muscarinic receptor-mediated functions in vivo (Chaudhuri et al., 1993; Pope, 1999). These pesticide actions may interact at the receptor level to produce the observed deviations from additivity. On the other hand, while these other actions could contribute to the behavioral effects we measured, it is doubtful that they could directly change ChE activity.
The only endpoint which did not deviate from additivity was the tail pinch response. One possibility for this exception could be the specific neuronal function being measured with this test. Stimulation of muscarinic, but not nicotinic, spinal cord receptors produces antinociception (Naguib et al., 1997; Pedigo et al., 1975). The endpoints that showed the non-additive interaction, i.e., gait score and motor activity, are more apical and are not specific for one receptor system or the other. Thus, the synergism observed with those endpoints could be due to interactions at sites other than muscarinic receptors. On the other hand, the tail pinch response was the most variable endpoint in this study. The lack of significant non-additivity could simply be due to lack of power caused by the high variability in the data.
The chemical proportions studied in the mixture can be a critical factor for the interpretation of the data. For the OP cumulative risk assessment, the joint dose-response was predicted using relative potency factors for the single chemical data (US EPA, 2002). The toxic equivalency factor (TEF) and relative potency factor (RPF) approaches have been used for pesticides and other chemicals (Chen et al., 2001; Safe, 1998). This implies, however, that the population exposure will reflect these relative potencies. In line with the Society of Toxicology Expert Panel recommendations for testing "real world" exposures (Teuschler et al., 2002), we chose chemicals and their ratios which would reflect potential human dietary exposures. We chose to test a mixture of five pesticides, since co-occurrence of five or fewer pesticides occurs in >98% of products tested (US FDA, 2000). These pesticides were chosen for study based on usage patterns and food residue data (i.e., pesticides used on the same or similar crops; USDA, 2000; US FDA, 2000) and market share (i.e., highest volume pesticides; US EPA, 1999). The ratio of pesticides in the test mixture was determined by comparing the relative dietary exposure to humans as projected by the U.S. EPA probabilistic model of exposure. Thus, while the outcome of this study should be extrapolated with caution to other OP mixtures, we felt that this mixture represents one appropriate for potential environmental exposure. Additional studies could directly address this factor by using other pesticide ratios.
In summary, these data demonstrate greater than additive responses (synergism) in several behavioral effects as well as ChE inhibition produced by a mixture of five OP pesticides. The non-additive interaction is not due solely to the malathion in the mixture, since synergism was also detected in the 4-OP mixture (excluding malathion). The proportion of each pesticide in the mixture reflected anticipated human exposure as projected by the EPA dietary exposure model. While the magnitude of the interactions was not large, at least a two-fold shift in effective doses was noted. Because a threshold shift was also observed but was more difficult to quantify, the magnitude of the difference could actually be much greater. These effects at the lower end of the dose-response curve contradict common assumptions that interactions will be additive in the low-dose region. Other ratios or testing conditions could show greater or lesser deviations from additivity. These findings suggest several testable hypotheses with which to examine the mechanisms for these interactions, which would in turn allow better prediction of the neurotoxicity of OP mixtures.
NOTES
This research has been funded in part by the U.S. Environmental Protection Agency through a cooperative agreement (NHEERL-RTP #CR-82811401-0). This manuscript has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents necessarily reflect the views of the Agency nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
Portions of this research were presented at the 43rd annual meeting of the Society of Toxicology, March 2004, Baltimore, MD.
ACKNOWLEDGMENTS
The authors acknowledge invaluable assistance from A. Lowit, D. Miller, and D. Hrdy in the US EPA Office of Pesticide Programs (Office of Prevention, Pesticides and Toxic Substances) in the selection and determination of DEEM values for the pesticides in this mixture. We also gratefully acknowledge the excellent technical assistance of Ms. P. Phillips and K. McDaniel in the meticulous conduct of this study.
REFERENCES
Berenbaum, M. C. (1985). The expected effect of a combination of agents: The general solution. J. Theor. Biol. 114, 413–431.
Brunden, M., and Vidmar, T. (1989). Optimal 3 x 3 factorial and 3-ray designs for drug interaction experiments with dichotomous responses. Commun. Statistics 18, 2835–2859.
Carpy, S. A., Kobel, W., and Doe, J. (2000). Health risk of low-dose pesticides mixtures: A review of the 1985–1998 literature on combination toxicology and health risk assessment. J. Toxicol. Environ. Health, Part B 3, 1–25.
Carter, W. H., Jr. (1995). Relating isobolograms to response surfaces. Toxicology 105, 181–188.
Carter, W. H., Jr., Jones, D. E., and Carchman, R. A. (1985). Application of response surface methods for evaluating the interactions of soman, atropine, and pralidioxime chloride. Fundam. Appl. Toxicol. 5, S232–S241.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (2004). Detecting interaction(s) and assessing the impact of component subsets in a chemical mixture using fixed-ratio mixture ray designs. J. Agr. Biol. Environ. Statistics 9, 339–361.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (2005). Ds-optimal designs for studying combinations of chemicals using multiple fixed-ratio ray experiments. Environmetrics 16, 129–147.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (in press). Power and sample size calculations for linear hypotheses associated with mixtures of many components using fixed-ratio ray designs. J. Environ. Ecol. Statistics.
Casida, J. E., Baron, R. L., Eto, M., and Engel, J. L. (1963). Potentiation and neurotoxicity induced by certain organophosphates. Biochem. Pharmacol. 12, 73–78.
Chambers, J. E., and Carr, R. L. (1993). Inhibition patterns of brain acetylcholinesterase and hepatic and plasma aliesterases following exposures to three phosphorothionate insecticides and their oxons in rats. Fund. Appl. Toxicol. 21, 111–119.
Chambers, J. E., Ma, T., Boone, J. S., and Chambers, H. W. (1994). Role of detoxication pathways in acute toxicity levels of phosphorothionate insecticides in the rat. Life Sci. 54, 1357–1364.
Chanda, S. M., Mortensen, S. R., Moser, V. C., and Padilla, S. (1997). Tissue-specific effects of chlorpyrifos on carboxylesterase and cholinesterase activity in adult rats: An in vitro and in vivo comparison. Fundam. Appl. Toxicol. 38, 148–157.
Chaudhuri, J., Chakraborti, T. K., Chanda, S., and Pope, C. N. (1993). Differential modulation of organophosphate-sensitive muscarinic receptors in rat brain by parathion and chlorpyrifos. J. Biochem. Toxicol. 8, 207–216.
Chen, J. J., Che, Y.-J., Rice, G., Teuschler, L. K., Hamernik, K., Protzel, A., and Kodell, R. L. (2001). Using dose addition to estimate cumulative risks from exposure to multiple chemicals. Reg. Toxicol. Pharmacol. 34, 35–41.
Clement, J. G. (1984). Role of aliesterase in organophosphate poisoning. Fundam. Appl. Toxicol. 4, S96–S105.
Cohen, S. D. (1984). Mechanisms of toxicological interactions involving organophosphate insecticides. Fundam. Appl. Toxicol. 4, 315–324.
Cohen, S. D., and Murphy, S. D. (1971a). Carboxylesterase inhibition as an indicator of malathion potentiation in mice. J. Pharmacol. Exp. Ther. 176, 733–742.
Cohen, S. D., and Murphy, S. D. (1971b). Malathion potentiation and inhibition of hydrolysis of various carboxylic esters by triorthotolyl phosphate (TOTP) in mice. Biochem. Pharmacol. 20, 575–587.
Cohen, S. D., and Murphy, S. D. (1974). A simplified bioassay for organophosphate detoxification and interactions. Toxicol. Appl. Pharmacol. 27, 537–550.
DuBois, K. P. (1961). Potentiation of the toxicity of organophosphorus compounds. Adv. Pest Control Res. 4, 117–151.
DuBois, K. P. (1969). Combined effects of pesticides. Can. Med. Assoc. J. 100, 173–179.
Ecobichon, D. J. (1996). Toxic effects of pesticides. In Casarett and Doull's Toxicology (C. D. Klaassen, M. O. Amdur, and J. Doull, Eds.), pp. 643–689. McGraw-Hill, New York.
El-Masri, H. A., Mumtaz, M. M., and Yushak, M. L. (2004). Application of physiologically-based pharmacokinetic modeling to investigate the toxicological interaction between chlorpyrifos and parathion in the rat. Environ. Toxicol. Pharmacol. 16, 57–71.
Finney, D. J. (1964). Probit Analysis. Cambridge University Press, London.
Fleisher, J. H., Harris, L. W., Prudhomme, C., and Bursel, J. (1963). Effects of ethyl p-nitrophenyl thionobenzene phosphonate (EPN) on the toxicity of isopropyl methyl phosphonofluoridate (GB). J. Pharmacol. Exp. Ther. 139, 390–396.
Frawley, J. P., Fuyat, H. N., Hagan E. C., Blake, J. R., and Fitzhugh, O. G. (1957). Marked potentiation in mammalian toxicity from simultaneous administration of two anticholinesterase compounds. J. Pharmacol. Exp. Ther. 121, 96–106.
Gennings, C. (1995). An efficient experimental design for detecting departure from additivity in mixtures of many chemicals. Toxicology 105, 189–197.
Gennings, C., and Carter, W. H., Jr. (1995). Utilizing concentration-response data from individual components to detect statistically significant departures from additivity in chemical mixtures. Biometrics 51, 1264–1277.
Gennings, C., Carter, W. H., Jr., Campain, J. A., Bae, D., and Yang, R. S. H. (2002). Statistical analysis of interactive cytotoxicity in human epidermal keratinocytes following exposure to a mixture of four metals. J. Agric. Biol. Environ. Stat. 7, 58–73.
Gennings, C., Carter, W. H., Jr., Casey, M., Moser, V., Carchman, R., and Simmons, J. E. (2004). Analysis of functional effects of a mixture of five pesticides using a ray design. Environ. Toxicol. Pharmacol. 18, 115–125.
Gennings, C., Schwartz, P., Carter, W. H., Jr., and Simmons, J. E. (1997). Detection of departures from additivity in mixtures of many chemicals with a threshold model. J. Agric. Biol. Environ. Stat. 2, 198–211.
Gessner, P. K. (1974). The isobolographic method applied to drug interactions. In Drug Interactions (P. L. Morselli, S. Garattini, and S. N. Cohen, Eds.), pp. 349–362. Raven Press, New York.
Hussain, M. A., Mohamad, R. B., and Oloffs, P. C. (1985). Studies on the toxicity, metabolism, and anticholinesterase properties of acephate and methamidophos. J. Environ. Sci. Health B20, 129–147.
Johnson, C. D., and Russell, R. L. (1975). A rapid, simple radiometric assay for cholinesterase, suitable for multiple determinations. Anal. Biochem. 64, 229–238.
Karanth, S., Liu, J., Olivier, K., Jr., and Pope, C. (2004). Interactive toxicity of the organophosphorus insecticides chlorpyrifos and methyl parathion in adult rats. Toxicol. Appl. Pharmacol. 196, 183–190.
Karanth, S., Olivier, K., Jr., Liu, J., and Pope, C. (2001). In vivo interaction between chlorpyrifos and parathion in adult rats: Sequence of administration can markedly influence toxic outcome. Toxicol. Appl. Pharmacol. 177, 247–255.
Kardos, S. A., and Sultatos, L. G. (2000). Interactions of the organophosphates paraoxon and methyl paraoxon with mouse brain acetylcholinesterase. Toxicol. Sci. 58, 118–126.
Kousba, A. A., Sultatos, L. G., Poet, T. S., and Timchalk, C. (2004). Comparison of chlorpyrifos-oxon and paraoxon acetylcholinesterase inhibition dynamics: Potential role of a peripheral binding site. Toxicol. Sci. 80, 239–248.
Lauwerys, R. R., and Murphy, S. D. (1969). Interaction between paraoxon and tri-o-tolyl phosphate in rats. Toxicol. Appl. Pharmacol. 14, 348–357.
Maxwell, D. M. (1992). Detoxication of organophosphorus compounds by carboxylesterase. In Organophosphates: Chemistry, Fate, and Effects (J. E. Chambers and P. E. Levi, Eds.), pp. 183–199. Academic Press, New York.
McCullagh, P., and Nelder, J. A. (1989). Generalized Linear Models, 2nd ed. Chapman and Hall, New York.
Moser, V. C. (1995). Comparisons of the acute effects of cholinesterase inhibitors using a neurobehavioral screening battery in rats. Neurotoxicol. Teratol. 17, 617–625.
Moser, V. C., and Padilla, S. (1998). Age- and gender-related differences in the time course of behavioral and biochemical effects produced by oral chlorpyrifos in rats. Toxicol. Appl. Pharmacol. 149, 107–119.
Murphy, S. D., Anderson, R. L., and DuBois, K. P. (1959). Potentiation of toxicity of malathion by triorthotolyl phosphate. Proc. Soc. Exp. Biol. Med. 100, 483–387.
Murphy, S. D., Cheever, K. L., Chow, A. Y. K., and Brewster, M. (1976). Organophosphate insecticide potentiation by carboxylesterase inhibitors. Proc. Eur. Soc. Tox. XVII Esc. Med. Int. Cong. 376, 292–300.
Murphy, S. D., and DuBois, K. P. (1957). Quantitative measurement of inhibition of the enzymatic detoxification of malathion by EPN (ethyl p-nitrophenyl thionobenzene phosphate). Proc. Soc. Exp. Biol. Med. 96, 813–818.
Naguib, M., Farsci, F., and Yaksh, T. L. (1997). Characterization of muscarinic receptor subtypes that mediate antinociception in the rat spinal cord. Anesth. Analg. 85, 847–853.
National Research Council. (1989). Acetylcholinesterase inhibitors: Case study of mixtures of contaminants with similar biological effects. In Drinking Water and Health. Selected Issues in Risk Assessment. Part II. Mixtures. Vol. 9, pp. 146–161. National Academy Press, Washington, DC.
Nostrandt, A. C., Padilla, S., and Moser, V. C. (1997). The relationship of oral chlorpyrifos effects on behavior, cholinesterase inhibition, and muscarinic receptor density in rat. Pharmacol. Biochem. Behav. 58, 15–23.
Pedigo, N. W., Dewey, W. L., and Harris, L. S. (1975). Determination and characterization of the antinociceptive activity of intraventricularly administered acetylcholine in mice. J. Pharmacol. Exp. Ther. 193, 845–852.
Poet, T. S., Kousba, A. A., Dennison, S. L., and Timchalk, C. (2004). Physiologically based pharmacokinetic/pharmacodynamic model for the organophosphorus pesticide diazinon. Neurotoxicol. 25, 1013–1030.
Pond, A. L., Chambers, H. W., and Chambers, J. E. (1995). Organophosphate detoxication potential of various rat tissues via A-esterase and aliesterase activities. Toxicol. Letters 78, 245–252.
Pope, C. N. (1999). Organophosphorus pesticides: Do they all have the same mechanism of toxicity J. Toxicol. Environ. Health, Part B, 2, 161–181.
Reiter, L. W. (1983). Chemical exposures and animal activity: utility of the figure-eight maze. Dev. Toxicol. Environ. Sci. 11, 73–84.
Safe, S. H. (1998). Hazard and risk assessment of chemical mixtures using the toxic equivalency factor approach. Environ. Health Perspect. 106(Suppl. 4), 1051–1058.
Seume, F. W., Casida, J. E., and O'Brien, R. D. (1960). Effects of parathion and malathion separately and jointly upon rat esterases in vivo. Agricul. Food Chem. 8, 43–47.
Solana, R. P., Gennings, C., Carter, W. H., Jr., Anderson, D., Lennox, W. J., Carchman, R. A., and Harris, L. W. (1990a). Evaluation of the efficacy of two carbamates, physostigmine and pyridostigmine, when used in conjunction for protection against organophosphate exposure. Fundam. Appl. Toxicol. 15, 814–819.
Solana, R. P., Harris, L. W., Carter, W. H., Jr., Talbot, B. G., Carchman, R. A., and Gennings, C. (1990b). Evaluation of a two-drug combination pretreatment against organophosphorus exposure. Toxicol. Appl. Pharmacol. 102, 421–429.
Su, M.-Q., Kinoshita, F. K., Frawley, J. P., and DuBois, K. P. (1971). Comparative inhibition of aliesterases and cholinesterase in rats fed eighteen organophosphorus insecticides. Toxicol. Appl. Pharmacol. 20, 241–249.
Teuschler, L., Klaunig, J., Carney, E., Chambers, J., Connolly, R., Gennings, C., Giesy, J., Hertzberg, R., Klaassen, C., Kodell, R., Paustenbach, D., and Yang, R. (2002). Support of science-based decisions concerning the evaluation of the toxicology of mixtures: A new beginning. Reg. Toxicol. 36, 34–39.
Timchalk, C., Nolan, R. J., Mendrala, A. L., Dittenber, D. A., Brzak, K. A., and Mattsson, J. L. (2002). A physiologically based pharmacokinetic and pharmacodynamic (PBPK/PD) model for the organophosphate insecticide chlorpyrifos in rats and humans. Toxicol. Sci. 66, 34–53.
Timchalk, C., Poet, T. S., Hinman, M. N., Busby, A. L., and Kousba, A. A. (in press). Pharmacokinetic and pharmacodynamic interaction for a binary mixture of chlorpyrifos and diazinon in the rat. Toxicol. Appl. Pharmacol.
Uchiyama, M., Yoshida, T., Homma, K., and Hongo, T. (1975). Inhibition of hepatic drug-metabolizing enzymes by thiophosphate insecticides and its drug toxicological implications. Biochem. Pharmacol. 24, 1221–1225.
U.S. Department of Agriculture. (2000). Pesticide Data Program: Annual Summary Calendar Year 1998, http://www.ams.usda.gov/science/pdp/.
U.S. Environmental Protection Agency. (1999). Pesticide Industry Sales and Usage, 1996 and 1997 Market Estimates, http://www.epa.gov/oppbead1/pestsales/.
U.S. Environmental Protection Agency (2001). Organophosphate Pesticides: Preliminary OP Cumulative Risk Assessment, http://www.epa.gov/oppsrrd1/cumulative/pra-op/newdesign.htm.
U.S. Environmental Protection Agency. (2002). Organophosphate Pesticides: Revised OP Cumulative Risk Assessment, http://www.epa.gov/pesticides/cumulative/rra-op/.
U.S. Food and Drug Administration. (2000). Pesticide Program: Residue Monitoring, 1999, http://www.cfsan.fda.gov/download/pes99db.html.
Ward, T. R., and Mundy, W. R. (1996). Organophosphorus compounds preferentially affect second messenger systems coupled to M2/M4 receptors in rat frontal cortex. Brain Res. Bull. 39, 49–55.
Wu, H. X., Evreux-Gros, C., and Descotes, J. (1996). Diazinon toxicokinetics, tissue distribution and anticholinesterase activity in the rat. Biomed. Environ. Sci. 9, 359–369.(V. C. Moser, M. Casey, A.)
Department of Biostatistics, Virginia Commonwealth University, Richmond, Virginia
Experimental Toxicology Division, NHEERL/ORD, US EPA, RTP, North Carolina
ABSTRACT
Environmental exposures generally involve chemical mixtures instead of single chemicals. Statistical models such as the fixed-ratio ray design, wherein the mixing ratio (proportions) of the chemicals is fixed across increasing mixture doses, allows for the detection and characterization of interactions among the chemicals. In this study, we tested for interaction(s) in a mixture of five organophosphorus (OP) pesticides (chlorpyrifos, diazinon, dimethoate, acephate, and malathion). The ratio of the five pesticides (full ray) reflected the relative dietary exposure estimates of the general population as projected by the US EPA Dietary Exposure Evaluation Model (DEEM). A second mixture was tested using the same dose levels of all pesticides, but excluding malathion (reduced ray). The experimental approach first required characterization of dose-response curves for the individual OPs to build a dose-additivity model. A series of behavioral measures were evaluated in adult male Long-Evans rats at the time of peak effect following a single oral dose, and then tissues were collected for measurement of cholinesterase (ChE) activity. Neurochemical (blood and brain cholinesterase [ChE] activity) and behavioral (motor activity, gait score, tail-pinch response score) endpoints were evaluated statistically for evidence of additivity. The additivity model constructed from the single chemical data was used to predict the effects of the pesticide mixture along the full ray (10–450 mg/kg) and the reduced ray (1.75–78.8 mg/kg). The experimental mixture data were also modeled and statistically compared to the additivity models. Analysis of the 5-OP mixture (the full ray) revealed significant deviation from additivity for all endpoints except tail-pinch response. Greater-than-additive responses (synergism) were observed at the lower doses of the 5-OP mixture, which contained non-effective dose levels of each of the components. The predicted effective doses (ED20, ED50) were about half that predicted by additivity, and for brain ChE and motor activity, there was a threshold shift in the dose-response curves. For the brain ChE and motor activity, there was no difference between the full (5-OP mixture) and reduced (4-OP mixture) rays, indicating that malathion did not influence the non-additivity. While the reduced ray for blood ChE showed greater deviation from additivity without malathion in the mixture, the non-additivity observed for the gait score was reversed when malathion was removed. Thus, greater-than-additive interactions were detected for both the full and reduced ray mixtures, and the role of malathion in the interactions varied depending on the endpoint. In all cases, the deviations from additivity occurred at the lower end of the dose-response curves.
Key Words: organophosphate mixtures; cumulative risk; neurotoxicity; chlorpyrifos; acephate; malathion; diazinon; dimethoate.
INTRODUCTION
The evaluation of potential health risks of exposures to chemical mixtures is a growing area of environmental and toxicological research. Research needs in this area were described recently in a consensus paper published by the Society of Toxicology Expert Panel on Mixtures (Teuschler et al., 2002). This Expert Panel advocated a scientifically based approach to the risk assessment of mixtures, which would involve biologically based hypotheses and experimental approaches in the design of laboratory studies. Because of the uncertainty about the health risks of exposures to multiple chemicals, research concerning chemical mixtures will be one of the key environmental health issues for several decades to come.
Recent advances in the statistical modeling of mixture data have moved the field from the analysis of binary combinations to the analysis of multiple chemicals using various designs. The classical statistical approach to the analysis of binary chemical mixtures is to construct full dose-response curves for each compound in the presence of a range of doses of the second compound (isobolographic analyses; Gessner, 1974). For interaction studies using more than two chemicals, response-surface methodology frequently requires factorial-type designs (e.g., Carter, 1995; Carter et al., 1985; Solana et al., 1990a,b). A more economical (e.g., fewer treatment groups, fewer test subjects, more straightforward analyses) approach is to compare an additivity model to selected mixtures of the compounds that are experimentally tested (Gennings, 1995; Gennings et al., 2002). An extension of this approach is to test combinations of chemicals in fixed ratios or proportions, i.e., a ray design (e.g., Brunden and Vidmar, 1989; Finney, 1964). Characteristics of the fitted dose-response curve along the ray using total dose for the dependent variable can be compared to that predicted under additivity. This methodology shows great promise and has recently been modified for use with neurotoxicity data using a relevant pesticide mixture (Casey et al., 2004, 2005, in press; Gennings et al., 2004).
Given that pesticide application patterns generally result in exposure to mixtures instead of single chemicals, the detection and characterization of pesticide interactions is critical for regulatory decisions. Studies of mixtures toxicity can be simplified by grouping chemicals with a known or presumed common mode of action. This approach was codified in the 1996 Food Quality Protection Act, which directed the U.S. Environmental Protection Agency (US EPA) to base regulatory decisions on mixtures of pesticides which act through a common mode of action, rather than on individual pesticides. Organophosphorus (OP) pesticides which inhibit acetylcholinesterase (AChE), a widely used class of pesticides, were the first class to undergo a cumulative (i.e., mixtures) risk assessment (US EPA, 2002). Despite waning household uses, OPs are still used widely on a variety of foods and commercial crops, and there is potential for multiple OP exposures. A default assumption in these assessments was dose-additivity.
Previous research on OP pesticide interactions has involved primarily lethality studies of binary chemical combinations in adult laboratory animals (e.g., Casida et al., 1963; Cohen, 1984; DuBois, 1961; reviewed in NRC, 1989). For example, DuBois (1961) recorded lethality in rats administered pairs of 13 OP pesticides, and reported that the results showed additivity in 21 of 43 pairs; four pairs showed marked synergy. Potentiation of malathion toxicity by other OPs is often cited as an example of interactions due to inhibition of detoxification pathways (Casida et al., 1963; Cohen and Murphy, 1971a,b; DuBois, 1969; Frawley et al., 1957; Murphy and DuBois, 1957; Murphy et al., 1959; Su et al., 1971). More recently, some have begun to investigate sequential dosing of binary combinations (Karanth et al., 2001, 2004) and, using lethality and symptomatology as endpoints, have demonstrated that particular dose sequences can produce marked potentiation. Some other interaction studies involve OPs (mostly nerve agents) given with other chemical classes, i.e., antidotes and pre-treatments for exposure (e.g., Solana et al., 1990a,b). These studies do not address the larger issue of cumulative mixture to numerous pesticides at lower dose levels that do not produce such overt toxicity.
OP pesticides produce specific neurobehavioral effects, many of which are due to their AChE-inhibiting properties. Exposure to high doses produces cholinergic signs of toxicity, including salivation, lacrimation, muscle fasciculations, polyuria, diarrhea, progressing to tremors, respiratory depression, and death (Ecobichon, 1996). Behavioral effects that occur at lower doses include lowered activity, antinociception, and gait abnormalities. These functional changes often correlate well with depression of brain and/or blood AChE (Moser, 1995; Nostrandt et al., 1997). A combination of both AChE measurements and functional outcomes provides a more detailed evaluation of the neurotoxicity following OP exposure.
In this study, we tested the hypothesis of dose-additivity for a mixture of five OP pesticides (acephate, diazinon, chlorpyrifos, dimethoate, and malathion). Since non-additivity was observed for the 5-OP mixture, we hypothesized this may be due to malathion; therefore, a second mixture excluding malathion was tested to examine the influence of malathion on the observed interaction. Neurobehavioral measures and cholinesterase inhibition in blood and brain were evaluated at a single time point (the time of peak effect of the component pesticides) following acute exposure to the mixture. The mixing ratio of the pesticides was determined to reflect potential dietary exposure for the general population.
MATERIALS AND METHODS
Chemicals.
Chlorpyrifos (O,O-diethyl-O-[3,5,6-trichloro-2-pyridyl]phosphorothioate; purity, 99.2–99.9%), acephate (O,S-dimethyl acetylphosphoroamidothioate; purity, 98.7%), diazinon (O,O-diethyl-O-[2-isopropyl-4-methyl-6-pyridimyl]phosphorothioate; purity, 99.2–99.4%), and malathion (S-[1.2-dicarbethoxyethyl]-O,O-dimethyl dithiophosphate; purity, 98.2–99.5%) were obtained from Chem Service (West Chester, PA), and dimethoate (O,O-dimethyl-S-[N-methylcarbomoylmethyl]phosphorodithioate; purity, 98.7%) was a gracious gift from Cheminova (Lemvig, Denmark). Acephate was dissolved in deionized water. The other pesticides were dissolved in a vehicle of corn oil that contained 5% ethanol, as this was required to dissolve the dimethoate.
For the radiometric cholinesterase (ChE) assay, [3H]acetylcholine iodide (specific activity, 82.0 mCi/mmol) was obtained from DuPont/NEN Life Science Products (Boston, MA) and unlabeled acetylcholine iodide from Sigma (St. Louis, MO). All other reagents were obtained from commercial sources and were of the highest available grade.
Animals.
Adult (66–70 days of age at testing) Long-Evans male rats were obtained from Charles River Laboratories (Raleigh, NC) and individually housed on heat-treated pine shavings with feed (Purina Rodent Chow 5001) and water (filtered tap) freely available. The animal facility was accredited fully by the Association for Assessment and Accreditation of Laboratory Animal Care International and maintained at 70 ± 2°F, 50 ± 10% humidity, with a 12-h light/dark cycle.
Behavioral testing.
Neurobehavioral function was evaluated using several endpoints shown previously to be sensitive to OP exposure (Moser, 1995). Upon removing the rat from the cage, the observer noted the presence of miosis, mouth smacking, salivation, or lacrimation. The rat was then placed in an open field for observation, and tremors, gait abnormality, and the response to a tail pinch were scored. Motor activity was then measured during a 30-min session, using an automated chamber shaped like a figure eight (Reiter, 1983). The same observer conducted all studies, and was unaware of the dose level of each rat.
Immediately after the motor activity assessment, rats were decapitated quickly under CO2-induced anesthesia. The brain was rapidly removed, and trunk blood was collected in heparinized tubes and diluted. All tissues were stored at –80°C until the time of assay.
Cholinesterase assay.
A radiometric assay was used to determine brain and blood ChE activity (Johnson and Russell, 1975). Both tissues were diluted in two volumes of 0.1 M sodium phosphate buffer (pH 8.0) with 1% Triton X-100, followed by homogenization (brain only) for 30 s (Polytron homogenizer, Kinematica Model PT3100, Littau, Switzerland). The final acetylcholine iodide concentration was 1.2 mM.
Chemical treatments.
Single-chemical dose-response data were collected at the time of peak effect, previously determined by pilot and/or published studies (Moser, 1995; Moser and Padilla, 1998; Poet et al., 2004; Timchalk et al., 2002; Wu et al., 1996). Where published data were not available, the time to peak effect was evaluated by monitoring behavioral changes (lacrimation, salivation, tremors, uncoordinated gait, ataxia, decreased arousal, and fasciculations) at hourly intervals. The time(s) of maximum effect on each measure was noted, and summed across doses. Calculated this way, the sum of peak effects over three doses at 2, 3, 4, and 5 h were: dimethoate, 10, 15, 4, and 2 and acephate, 11, 13, 11, and 10. Malathion had no observable effects at the dose levels tested. The time course was similar for the four pesticides for which effects were observed, with maximum effects protracted over 3 to 4 h. We therefore chose the testing time to encompass the peak times of each chemical. Neurobehavioral testing began 3 h after dosing, and rats were euthanized between 4 to 4 h after dosing.
All pesticides were administered by oral gavage (water solutions, 1 ml/kg; corn oil suspensions, 2 ml/kg). Because of the two different vehicles, the mixture study required that the rats be gavaged twice. The water dosing vehicle (acephate) was given first, followed immediately by the corn oil/ethanol dosing vehicle (containing the other pesticides). Although the single chemical data were collected using a single gavage dose, confirmatory studies were conducted to compare the effects of the two-gavage dosing procedure to the single gavage, and no differences were detected.
The single-chemical dose-response curves included at least five dose levels plus control, with n = 8/dose group. For chlorpyrifos and diazinon, a second study was conducted to obtain an adequate range of responses. The data for ChE and motor activity in the two studies were compared and found not to differ significantly; the data were then combined for all endpoints. The dose-response curve for malathion included only two dose levels, since both were non-effective on most endpoints and the highest dose tested, 500 mg/kg, was considered to be adequately high for these studies.
Dietary exposure was estimated using the Dietary Exposure Evaluation Model (DEEM-FCID) software (US EPA, 2001). This is a probabilistic analysis conducted by combining representative data on concentrations of OP pesticides on foods, market basket monitoring studies, residue decline and degradation studies, and consumer practices such as washing, cooking, etc. with distributions of anticipated consumption of these foods (USDA's Continuing Survey of Food Intake by Individuals) by different segments of the U.S. population. The daily anticipated exposure value designated for 95% of the general population was obtained for each pesticide, and the ratio of these values was used for the mixture. The proportions of the chemicals in the mixture were: chlorpyrifos 0.031; acephate 0.040; diazinon 0.002; dimethoate 0.102; and malathion 0.825.
The mixture based on this composition was considered the "full ray." The mixture dose levels ranged from 10 to 450 mg/kg, and the pesticide proportions (mixing ratio) remained the same across dose levels. Table 1 presents the actual doses used in each mixture dose. The second mixture ("reduced ray") did not contain malathion, but did have the remaining four OPs at the same dose levels, and therefore the same relative ratios, as in the full ray. The reduced ray mixture dose levels ranged from 1.75 to 78.8 mg/kg, and dimethoate was the most prevalent pesticide, comprising 58.5% of the 4-OP mixture.
The mixture studies were conducted using six dose levels of each mixture plus control. In each mixture study, a single dose group of each of the individual chemicals was included to assure that the chemical response was the same as that predicted from the earlier data ("positive controls"). Furthermore, a single dose of the 5-OP mixture (100 mg/kg) was repeated to replicate the effects of that dose in the first study. These repeated-dose data were not used in the statistical models.
Power analyses based on the motor activity variable indicated that a sample size of 12/mixture dose group was needed to detect a 25% decrease in the slope of the dose-response curve compared to that predicted under additivity with 70% power (Casey et al., in press). Thus, the conduct of the mixture study included six mixture dose groups (n = 12/dose group), control (n = 8), and a single dose level of each chemical (n = 8/chemical).
Statistical methods.
Data for eight behavioral endpoints were collected, but only three (motor activity, gait abnormality, and tail pinch response), along with the ChE measurements, were analyzed using the additivity models. Gait abnormality and the tail pinch response were ranked, but for the analyses the data were converted to a binary response (i.e., presence or absence of gait abnormality, normal or decreased tail pinch response). To constrain the probability of a response to be between 0 and 1, a logit link function was used. Total activity counts during the motor activity session, as well as blood and brain ChE activity, were continuous variables. Due to differences in the control brain ChE activity in one study, the data for this measure were analyzed as a proportion of the respective control value.
The additivity model is based on the single chemical dose-response data and a definition of dose additivity (e.g., Berenbaum, 1985). If the observed response along the fixed-ratio mixture ray is more extreme than that predicted under additivity, then it is reasonable to claim a greater-than-additive response (i.e., synergism); if the response is less extreme than that predicted under additivity, a less-than-additive response (i.e., antagonism) can be claimed; otherwise, the curves are coincident and departure from additivity cannot be claimed for the mixture ray(s) considered.
The development of the methods and preliminary use of these data are presented in several manuscripts (Casey et al., 2004, 2005, in press; Gennings et al., 2004). A threshold additivity model (Gennings et al., 1997) was used for motor activity and brain ChE, whereas for blood ChE the data fit a threshold outside the experimental range, and a generalized linear model was used. For each endpoint, the methodology uses the single chemical dose-response data to develop a model that predicts the response of the mixture ray under additivity. This predicted additivity model for decreasing curves, which relates the doses of the chemicals under study to the mean through a link function, g(μ;), and possibly a set of nonlinear parameters, , is written as:
where g(μadd;) is the link function (McCullagh and Nelder, 1989) of the response of interest, xi is the dose of the ith chemical, 0 is the unknown parameter associated with the intercept (common intercept across all curves), i is the unknown slope parameter for the ith chemical, is the unknown parameter associated with the threshold, and is the dose threshold (in mg/kg units) parameter for the ith chemical. If none of the dose threshold parameters were within the experimental region, the threshold additivity model was replaced by the corresponding generalized linear model:
The single chemical data were used to estimate the additivity model. Let [a1, ..., ac] represent the proportions of the c (here, c = 5) chemicals in the fixed-ratio mixture. The additivity model was used to predict the response of the mixture along the fixed-ratio ray in terms of total dose, t. Noting that xi = ait, the slope of the additivity model along the mixture ray is given by
For comparison, the mixture data along the fixed-ratio ray were fit to a similarly parameterized model:
When the threshold is estimated outside of the experimental region, the parameter is removed and the following model was used:
Before testing for departure from additivity, a goodness-of-fit test was performed on the model to ensure that it adequately fit the mixture data. In the event of significant lack-of- fit, higher-order terms were added until an appropriate model was found for the mixture data. The test of additivity is a test of coincidence of these two models along the specified fixed-ratio mixture. For example, if the threshold models are fit to the single chemical and mixture data, then the hypothesis of additivity is given by:
Following Gennings et al. (2002) and Casey et al. (2004), likelihood ratio tests were used to test this hypothesis of additivity.
The full and reduced rays were compared using the method described by Casey et al. (2004), which required correcting the curve from the full ray so that both rays fall in the same dose range. That is, the effect of removing malathion from the mixture was tested by comparing the dose-response curves for the two rays while noting that treduced = tfull(1 – a5) for a5 = 0.825, the proportion of malathion in the full mixture. Thus, the hypothesis of no malathion effect on the mixture is given by
Following Casey et al. (2004), likelihood ratio tests were used to test this hypothesis.
The estimation of these models was done in SAS version 8.2 using the maximum quasi-likelihood criterion in a Nelder-Mead direct-search algorithm embedded in the nonlinear programming (NLP) procedure in SAS and using the maximum quasi-likelihood criterion using a Fisher scoring algorithm in PROC GENMOD in SAS.
RESULTS
No lethality or unexpected toxicity occurred during any of the chemical testing. At the highest doses tested, mild signs of cholinergic stimulation were evident, including uncoordinated gait, lowered activity, lacrimation, miosis, and mild fasciculations or tremors. Malathion produced no observable effects. The control groups for each of the individual dose-response curves, as well as for the mixture study, were similar in almost all responses. The only exception was brain ChE, for which the control data from one study was higher than all the others, even though the inhibition relative to control was the same as a previous study. The tissues were analyzed a second time with the same result; thus, brain ChE data were analyzed as a proportion of the concurrent control mean value. The models for each endpoint were then fit simultaneously in an overall model with a common intercept parameter.
Each mixture study included one dose level of each individual chemical, as a "positive control" group (open circles on Figures 1, 3, 6, and 8). For all endpoints, the "positive control" data were not significantly different from the dose-response curves fit for each chemical.
Brain ChE
The individual dose-response data for brain ChE are presented in Figure 1. The curves were plotted for the individual dose-response curves, and the open symbols indicate the "positive control" data taken as part of the mixture studies. It is evident that the responses to the single doses were generally close to the curve fitted to the dose-response data, and they were not statistically different from the fitted curve. Hatch marks on the graphs indicate the location of the dose levels used in the mixture studies; these doses ineffective for some and in the effective dose-range for other pesticides. For example, the dose levels of diazinon used in the mixtures were far below the effective dose range for diazinon alone, whereas most of the dimethoate dose levels in the mixtures produced slight to moderate effects. The largest proportion of the mixture was malathion, which had no effect on this endpoint (or any other, except blood ChE). Being the second most prevalent OP in the mixture, dimethoate could be considered to be driving the mixture in terms of biological activity. For brain ChE, the higher mixture dose levels included active dose levels of both acephate and dimethoate.
The threshold parameter for the additivity models for the full and reduced rays was significant (p < 0.001) indicating that at least one of the curves showed a threshold within the experimental dose range. Specifically, the thresholds for chlorpyrifos and diazinon were 3.2 mg/kg and 17.7 mg/kg, respectively (Fig. 1). The slope parameter for malathion was not significant (p = 0.646), and therefore it was removed from the additivity models.
The additivity models predicted by the individual dose-response curves for the full and reduced rays are shown in Figure 2, along with the model fit to the experimental mixture data. The likelihood ratio test of additivity was rejected (p < 0.001) for both curves, with the experimental mixture model showing greater inhibition than the predicted additivity model, indicating greater-than-additive effects (synergism) in the lower dose range. A single dose level of the 5-OP mixture (100 mg/kg) was repeated several weeks later (open circle in full ray graphs, Figures 2, 4, 5, 7, and 9), and the data generally replicated the previous effect at that dose.
The predicted additivity models along both rays showed threshold values (full ray, 13.4 mg/kg; reduced ray, 2.3 mg/kg), whereas the models fitted to the mixture data revealed thresholds less than the lowest doses tested. This reveals a significant threshold shift, indicating an interaction even at the low end of the dose-response curves. As described by Casey et al. (2004), the full and reduced rays can be compared directly. For brain ChE inhibition, the two models were not significantly different (p = 0.421), so there was no evidence that malathion was responsible for the observed interactions.
The magnitude of change is important for understanding the biological significance of non-additivity. For purposes of illustration, ratios of point estimates have been calculated from the equations using the predicted additivity model and the actual mixture model; these are presented in Table 2. The purpose of these estimates is to show the ratio of the effective doses, or magnitude of change. Doses were calculated that (1) produced a 20% (ED20) or 50% (ED50) change from control (ChE activity or activity counts), or (2) produced an abnormal response in 20% (ED20) or 50% (ED50) of the treatment group (gait score, tail-pinch response). For brain ChE inhibition, the point estimates for a 20% effect on brain ChE illustrate that the shift in effective doses was about two-fold, and somewhat less at the ED50.
Evaluation of the thresholds provided additional information with which to compare the curves. The thresholds for the predicted additivity curves were 13.4 mg/kg for the full ray, and 2.3 mg/kg for the reduced ray. The experimental mixture model revealed much lower thresholds (2.3 mg/kg for the full ray, 0.12 mg/kg for the reduced ray) which represent 6-fold and 19-fold decreases from the predicted full and reduced rays, respectively.
Blood ChE
The individual dose-response data for blood ChE are presented in Figure 3. The threshold parameter was estimated outside of the experimental region for all individual dose-response curves, and was therefore removed from the model. The slope parameters for all pesticides were negative and significant. This was true even for malathion, even though the highest dose tested produced only about 50% inhibition. As with the brain ChE, the "positive control" data fell generally close to, and were not statistically different from, the curve fitted to the dose-response data.
The additivity models predicted by the individual dose-response curves are shown in Figure 4, along with the models fit to the experimental mixture data. For both the full and reduced rays, the likelihood ratio test of additivity was rejected (p < 0.001), and there is evidence of synergism in the low dose range. Comparison of the two mixture rays showed that they were significantly different from each other (p < 0.001), indicating that, for this measure, malathion interacted with the other four pesticides in the mixture. Indeed, the reduced ray showed a greater deviation from the predicted model. Table 2 presents the ED20 and ED50 for blood ChE inhibition for the predicted and the actual mixture data. There was only a small shift in doses producing 20% inhibition, whereas for the reduced ray, the shift in the ED50 was almost two-fold.
Tail-Pinch Response
Tail-pinch response was ranked on an ordinal scale, but for these analyses data were converted to binary responses. OP pesticide exposure only decreased the tail-pinch response, but the data were not particularly consistent, especially the dose-response curves for chlorpyrifos and diazinon (data not shown). For each of these, the dose-response data were collected twice, and there were differences between them. Furthermore, the "positive controls" sometimes appeared different from the fitted curve. Therefore, adequacy of the model fit may be questionable. The mixture data are shown in Figure 5, and the likelihood ratio test of additivity was not rejected (p = 0.499). Since the mixture rays showed only additivity, point estimates were not determined.
Motor Activity
A preliminary analysis of the motor activity data is presented in Casey et al. (2004); however, more specific data are presented here. The individual dose-response curves shown in Figure 6 identified significant thresholds within the dose range tested for all pesticides except malathion, which produced no effect on this measure up to 500 mg/kg. The individual thresholds were: acephate, 5.7 mg/kg; chlorpyrifos, 4.7 mg/kg; diazinon, 25 mg/kg; and dimethoate, 7.5 mg/kg (Casey et al., 2004).
There was sufficient evidence to reject the hypothesis of additivity for both the full and reduced rays (p < 0.001), shown in Figure 7. Comparison of the full and reduced rays showed no difference (p = 0.378), indicating that malathion did not influence the interaction. The point estimates in Table 2 illustrate an approximately two-fold shift in the ED20 dose, and somewhat less difference in the ED50 values. The thresholds for the predicted additivity curves were significant (36.6 mg/kg for the full ray, 6.4 mg/kg for the reduced ray); however, the fitted mixture models had no significant threshold parameter. Quantification is therefore problematic, but this indicates at least a three-fold shift in the thresholds since they must be below the lowest doses tested (<10 mg/kg for the full ray, <1.75 mg/kg for the reduced ray).
Gait Score
The gait score data (incidence of abnormal gait) were analyzed and presented in Gennings et al. (2004). Individual dose-response curves are presented in Figure 8, and the mixture data are presented in Figure 9. The full mixture data was close to significance for a deviation from additivity (p = 0.053), but the reduced ray was not different (p = 0.314) from the additivity model. This indicates that the deviation of the full mixture from additivity was due to the malathion, and the comparison of the full and reduced rays showed that they were different (p = 0.014). The point estimates in Table 2 illustrate that the full ray data produced point estimates about 1.7-fold less than predicted under additivity.
Other Endpoints
The other endpoints evaluated in this study were not subject to statistical analyses, but some patterns were clear in qualitatively assessing the data. When comparing the signs observed in the mixture studies with those produced by the individual pesticides, several patterns become obvious. Since a large proportion of the mixture was malathion and dimethoate, it was of interest to explore the contribution of those specific pesticides. Since malathion alone produced no behavioral changes, we focused on the effects of dimethoate. Mixture data for several endpoints are presented in Figure 10, plotted as a function of the dimethoate dose contained in each corresponding mixture dose level. The dimethoate single chemical dose-response is also plotted, to compare the effects of dimethoate alone and in the mixture. The increasing incidence of lacrimation is shown in Figure 10. The only pesticide which produced lacrimation at doses that were used in the mixture was dimethoate, albeit only a small proportion (1 of 8). In contrast, the higher doses of the mixture produced lacrimation in at least 20% of the animals, and except for the highest dose, there was not much difference between the full and reduced rays. This again suggests a non-additive response, and that malathion was not a factor at the lower doses; however, there is a suggestion of an influence at the highest dose. The incidence of mouth-smacking showed a similar pattern (data not shown). In contrast to this pattern, the incidence of miosis suggested more of an additive effect (Fig. 10). All doses of the mixture caused miosis in at least some rats, and the full and reduced rays look very similar. This effect could be due mostly to the dimethoate in the mixture, as that dose-response was close to the ray data. The other pesticides were not effective on this measure at the doses used in the mixture, even at the highest dose. The incidence of mild tremors mirrored that of miosis (data not shown).
DISCUSSION
This dose-additive model revealed greater-than-additive (synergistic) responses, using both neurochemical and behavioral endpoints, when five OPs were administered concurrently. Malathion could be held responsible for the non-additivity observed in the gait score measure, but not for the ChE and motor activity measures. Although the synergism was of relatively small magnitude (1.5–2-fold) when comparing point estimates (ED20s and ED50s), it was consistent across four of the five endpoints tested. The threshold shift observed at the low end of the dose-response curves was at least 3-fold for motor activity, and 6- to 19-fold for brain ChE. However, since these values fall outside the experimental region (i.e., doses tested), this analysis should be interpreted with caution.
For all endpoints, the deviations from additivity were observed in the lower end of the dose range (i.e., doses producing less than 20% effect). The high end of the dose range did not show appreciably greater inhibition. This finding appears contrary to the general assumption that deviations from additivity will become evident only at higher doses where metabolism becomes rate-limiting (e.g., Carpy et al., 2000; NRC, 1989). Of course, "high" and "low" doses are relative terms, but in this study, "low" was defined by effects on brain ChE. The doses of the individual chemicals used for all but the higher mixture doses were ineffective in inhibiting brain ChE, with the exception of acephate and dimethoate. The threshold shift was detected below the lowest mixture dose, which was comprised of ineffective doses of almost all chemicals. While the doses used in this mixture are higher than environmental levels to which the population may be exposed chronically, these acute doses are still much lower than those producing lethality, convulsions, or extreme cholinergic signs.
There could be both toxicokinetic and/or dynamic mechanisms for non-additivity with the OPs included in this mixture. Following the demonstration by Frawley and coworkers (1957) of potentiation between malathion and EPN, subsequent studies reported such non-additivity between other OPs as well, including TOTP, TOCP, parathion, paraoxon, soman, sarin, and others (e.g., Casida et al., 1963; Clement, 1984; Cohen and Murphy, 1971a,b; DuBois, 1961; Fleisher et al., 1963; Lauwerys and Murphy, 1969; Murphy et al., 1959, 1976; Su et al., 1971). These OPs compete for binding sites and/or block hydrolysis by carboxylesterases (CaEs), which are non-target esterases that play a major role in their detoxification (Maxwell, 1992). It was suggested that inhibition of CaE could be predictive of the potential for non-additive interactions with other OPs; that is, if the pesticide was more potent in inhibiting CaE compared to ChE, it could be expected to potentiate malathion at doses lower than the no-effect levels for ChE inhibition (Cohen and Murphy, 1974; DuBois, 1969).
Of the OPs tested in the present mixture, chlorpyrifos and dimethoate have been reported to inhibit CaE at doses less than or equal to those that inhibit ChE (Chambers and Carr, 1993; Chambers et al., 1994; Chanda et al., 1997; Su et al., 1971), but less is known about diazinon and acephate. CaEs hydrolyze malathion but are inhibited by other OPs; thus, interactions could be a function of blocking malathion detoxification or competition for binding to the limited number of CaE molecules. The latter explanation is suggested by the finding of synergism even without malathion in the mixture. Considering stochiometric binding of the enzyme and inhibitor, however, it would be expected that this kinetic factor would be dose-dependent, and show greatest effects at the highest doses. The only endpoint which showed greater effect at higher doses was blood ChE inhibition, but in that case the larger deviation from additivity was observed without malathion in the mixture. Direct measurement of CaE activity in liver and blood could aid in determining the role of CaE detoxification in these interactions.
While the interaction involving the CaE pathway is considered a primary mechanism of potentiation for some OPs, the degree of CaE inhibition is not sufficient in some mixtures to predict the potentiation observed (Cohen and Murphy, 1971a; Murphy et al., 1976; Seume et al., 1960). Other esterases, e.g., A-esterases, also hydrolyze chlorpyrifos and diazinon (Maxwell, 1992; Pond et al., 1995). Since these enzymatic reactions are non-saturable, it is less likely that this plays a role in the interactions. In addition, some phosphorothionates, including diazinon, induce hepatic microsomal enzymes which could alter activation of the pesticides which are not active in the parent form (DuBois, 1969; Uchiyama et al., 1975). Chlorpyrifos, malathion, dimethoate, and diazinon must be metabolized to their active oxons, and acephate is metabolized to a more active form (methamidophos) (Hussain et al., 1985; Pope, 1999). A study with direct administration of the active metabolites could remove the potential interaction at the activation step, and therefore reveal its influence; however, this would also greatly alter the time to onset of effects. In vitro measurement of the activation/detoxification produced by microsomal enzymes may provide support for some of these hypotheses.
Microsomal enzymes also serve to detoxify some of the OPs, which would have an influence in the opposite direction of activation. It has been suggested, however, that increasing the rate of microsomal detoxification was more important than the increased activation (DuBois, 1969). Due to the fairly complex metabolic pathways of these pesticides, and several different potential sites for interactions, physiologically based pharmacokinetic models may best be employed to determine the final outcome. Such approaches have recently been described for other pesticide mixtures (e.g., El-Masri et al., 2004; Timchalk et al., in press), and a similar approach could be used with this mixture.
Another factor which could influence the mixture response is the time of testing for the acute effects. The individual pesticides tested here showed the same time of peak effect, that is, about 3.5–4 h after dosing. These pesticides were administered together, although others have reported that varying the time of administration can markedly change the interaction (DuBois, 1969; Karanth et al., 2001, 2004). While this time-course has been published for chlorpyrifos (Moser, 1995; Moser and Padilla, 1998; Timchalk et al., 2002) and diazinon (Moser, 1995; Poet et al., 2004; Wu et al., 1996), pilot studies (described in the Methods) established the time-course data for the other pesticides. A potential mechanism for the interaction would be a change in the time-course, perhaps via altered absorption or metabolic activation. Since we did not conduct a time-course study of the mixture per se, we cannot make definitive statements regarding this possibility; however, follow-up studies could address this issue. Casual observation of the animals during the study, however, did not suggest that the effects were accelerated. Since all rats were euthanized for tissues by 4.5 h after dosing, we did not evaluate any measure of recovery.
We measured ChE inhibition as the accepted biomarker of exposure, and perhaps of biological effect, of these pesticides. There are, however, other dynamic actions of these pesticides which may serve to modulate the level of inhibition as well as the functional outcome of the ChE inhibition. Recent papers have suggested a peripheral binding site on the acetylcholinesterase enzyme which may be allosterically modulated by some OPs; this raises the possibility of interactions at that site that could alter the actual inhibition of the catalytic site (Kardos and Sultatos, 2000; Kousba et al., 2004). In addition, several of the pesticides in this mixture have been reported to have direct actions on cholinergic receptors, for example, chlorpyrifos oxon and malaoxon bind muscarinic receptors in vitro and inhibit cAMP formation (Ward and Mundy, 1996), and chlorpyrifos alters presynaptic muscarinic receptor-mediated functions in vivo (Chaudhuri et al., 1993; Pope, 1999). These pesticide actions may interact at the receptor level to produce the observed deviations from additivity. On the other hand, while these other actions could contribute to the behavioral effects we measured, it is doubtful that they could directly change ChE activity.
The only endpoint which did not deviate from additivity was the tail pinch response. One possibility for this exception could be the specific neuronal function being measured with this test. Stimulation of muscarinic, but not nicotinic, spinal cord receptors produces antinociception (Naguib et al., 1997; Pedigo et al., 1975). The endpoints that showed the non-additive interaction, i.e., gait score and motor activity, are more apical and are not specific for one receptor system or the other. Thus, the synergism observed with those endpoints could be due to interactions at sites other than muscarinic receptors. On the other hand, the tail pinch response was the most variable endpoint in this study. The lack of significant non-additivity could simply be due to lack of power caused by the high variability in the data.
The chemical proportions studied in the mixture can be a critical factor for the interpretation of the data. For the OP cumulative risk assessment, the joint dose-response was predicted using relative potency factors for the single chemical data (US EPA, 2002). The toxic equivalency factor (TEF) and relative potency factor (RPF) approaches have been used for pesticides and other chemicals (Chen et al., 2001; Safe, 1998). This implies, however, that the population exposure will reflect these relative potencies. In line with the Society of Toxicology Expert Panel recommendations for testing "real world" exposures (Teuschler et al., 2002), we chose chemicals and their ratios which would reflect potential human dietary exposures. We chose to test a mixture of five pesticides, since co-occurrence of five or fewer pesticides occurs in >98% of products tested (US FDA, 2000). These pesticides were chosen for study based on usage patterns and food residue data (i.e., pesticides used on the same or similar crops; USDA, 2000; US FDA, 2000) and market share (i.e., highest volume pesticides; US EPA, 1999). The ratio of pesticides in the test mixture was determined by comparing the relative dietary exposure to humans as projected by the U.S. EPA probabilistic model of exposure. Thus, while the outcome of this study should be extrapolated with caution to other OP mixtures, we felt that this mixture represents one appropriate for potential environmental exposure. Additional studies could directly address this factor by using other pesticide ratios.
In summary, these data demonstrate greater than additive responses (synergism) in several behavioral effects as well as ChE inhibition produced by a mixture of five OP pesticides. The non-additive interaction is not due solely to the malathion in the mixture, since synergism was also detected in the 4-OP mixture (excluding malathion). The proportion of each pesticide in the mixture reflected anticipated human exposure as projected by the EPA dietary exposure model. While the magnitude of the interactions was not large, at least a two-fold shift in effective doses was noted. Because a threshold shift was also observed but was more difficult to quantify, the magnitude of the difference could actually be much greater. These effects at the lower end of the dose-response curve contradict common assumptions that interactions will be additive in the low-dose region. Other ratios or testing conditions could show greater or lesser deviations from additivity. These findings suggest several testable hypotheses with which to examine the mechanisms for these interactions, which would in turn allow better prediction of the neurotoxicity of OP mixtures.
NOTES
This research has been funded in part by the U.S. Environmental Protection Agency through a cooperative agreement (NHEERL-RTP #CR-82811401-0). This manuscript has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents necessarily reflect the views of the Agency nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
Portions of this research were presented at the 43rd annual meeting of the Society of Toxicology, March 2004, Baltimore, MD.
ACKNOWLEDGMENTS
The authors acknowledge invaluable assistance from A. Lowit, D. Miller, and D. Hrdy in the US EPA Office of Pesticide Programs (Office of Prevention, Pesticides and Toxic Substances) in the selection and determination of DEEM values for the pesticides in this mixture. We also gratefully acknowledge the excellent technical assistance of Ms. P. Phillips and K. McDaniel in the meticulous conduct of this study.
REFERENCES
Berenbaum, M. C. (1985). The expected effect of a combination of agents: The general solution. J. Theor. Biol. 114, 413–431.
Brunden, M., and Vidmar, T. (1989). Optimal 3 x 3 factorial and 3-ray designs for drug interaction experiments with dichotomous responses. Commun. Statistics 18, 2835–2859.
Carpy, S. A., Kobel, W., and Doe, J. (2000). Health risk of low-dose pesticides mixtures: A review of the 1985–1998 literature on combination toxicology and health risk assessment. J. Toxicol. Environ. Health, Part B 3, 1–25.
Carter, W. H., Jr. (1995). Relating isobolograms to response surfaces. Toxicology 105, 181–188.
Carter, W. H., Jr., Jones, D. E., and Carchman, R. A. (1985). Application of response surface methods for evaluating the interactions of soman, atropine, and pralidioxime chloride. Fundam. Appl. Toxicol. 5, S232–S241.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (2004). Detecting interaction(s) and assessing the impact of component subsets in a chemical mixture using fixed-ratio mixture ray designs. J. Agr. Biol. Environ. Statistics 9, 339–361.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (2005). Ds-optimal designs for studying combinations of chemicals using multiple fixed-ratio ray experiments. Environmetrics 16, 129–147.
Casey, M., Gennings, C., Carter, W. H., Jr., Moser, V. C., and Simmons, J. E. (in press). Power and sample size calculations for linear hypotheses associated with mixtures of many components using fixed-ratio ray designs. J. Environ. Ecol. Statistics.
Casida, J. E., Baron, R. L., Eto, M., and Engel, J. L. (1963). Potentiation and neurotoxicity induced by certain organophosphates. Biochem. Pharmacol. 12, 73–78.
Chambers, J. E., and Carr, R. L. (1993). Inhibition patterns of brain acetylcholinesterase and hepatic and plasma aliesterases following exposures to three phosphorothionate insecticides and their oxons in rats. Fund. Appl. Toxicol. 21, 111–119.
Chambers, J. E., Ma, T., Boone, J. S., and Chambers, H. W. (1994). Role of detoxication pathways in acute toxicity levels of phosphorothionate insecticides in the rat. Life Sci. 54, 1357–1364.
Chanda, S. M., Mortensen, S. R., Moser, V. C., and Padilla, S. (1997). Tissue-specific effects of chlorpyrifos on carboxylesterase and cholinesterase activity in adult rats: An in vitro and in vivo comparison. Fundam. Appl. Toxicol. 38, 148–157.
Chaudhuri, J., Chakraborti, T. K., Chanda, S., and Pope, C. N. (1993). Differential modulation of organophosphate-sensitive muscarinic receptors in rat brain by parathion and chlorpyrifos. J. Biochem. Toxicol. 8, 207–216.
Chen, J. J., Che, Y.-J., Rice, G., Teuschler, L. K., Hamernik, K., Protzel, A., and Kodell, R. L. (2001). Using dose addition to estimate cumulative risks from exposure to multiple chemicals. Reg. Toxicol. Pharmacol. 34, 35–41.
Clement, J. G. (1984). Role of aliesterase in organophosphate poisoning. Fundam. Appl. Toxicol. 4, S96–S105.
Cohen, S. D. (1984). Mechanisms of toxicological interactions involving organophosphate insecticides. Fundam. Appl. Toxicol. 4, 315–324.
Cohen, S. D., and Murphy, S. D. (1971a). Carboxylesterase inhibition as an indicator of malathion potentiation in mice. J. Pharmacol. Exp. Ther. 176, 733–742.
Cohen, S. D., and Murphy, S. D. (1971b). Malathion potentiation and inhibition of hydrolysis of various carboxylic esters by triorthotolyl phosphate (TOTP) in mice. Biochem. Pharmacol. 20, 575–587.
Cohen, S. D., and Murphy, S. D. (1974). A simplified bioassay for organophosphate detoxification and interactions. Toxicol. Appl. Pharmacol. 27, 537–550.
DuBois, K. P. (1961). Potentiation of the toxicity of organophosphorus compounds. Adv. Pest Control Res. 4, 117–151.
DuBois, K. P. (1969). Combined effects of pesticides. Can. Med. Assoc. J. 100, 173–179.
Ecobichon, D. J. (1996). Toxic effects of pesticides. In Casarett and Doull's Toxicology (C. D. Klaassen, M. O. Amdur, and J. Doull, Eds.), pp. 643–689. McGraw-Hill, New York.
El-Masri, H. A., Mumtaz, M. M., and Yushak, M. L. (2004). Application of physiologically-based pharmacokinetic modeling to investigate the toxicological interaction between chlorpyrifos and parathion in the rat. Environ. Toxicol. Pharmacol. 16, 57–71.
Finney, D. J. (1964). Probit Analysis. Cambridge University Press, London.
Fleisher, J. H., Harris, L. W., Prudhomme, C., and Bursel, J. (1963). Effects of ethyl p-nitrophenyl thionobenzene phosphonate (EPN) on the toxicity of isopropyl methyl phosphonofluoridate (GB). J. Pharmacol. Exp. Ther. 139, 390–396.
Frawley, J. P., Fuyat, H. N., Hagan E. C., Blake, J. R., and Fitzhugh, O. G. (1957). Marked potentiation in mammalian toxicity from simultaneous administration of two anticholinesterase compounds. J. Pharmacol. Exp. Ther. 121, 96–106.
Gennings, C. (1995). An efficient experimental design for detecting departure from additivity in mixtures of many chemicals. Toxicology 105, 189–197.
Gennings, C., and Carter, W. H., Jr. (1995). Utilizing concentration-response data from individual components to detect statistically significant departures from additivity in chemical mixtures. Biometrics 51, 1264–1277.
Gennings, C., Carter, W. H., Jr., Campain, J. A., Bae, D., and Yang, R. S. H. (2002). Statistical analysis of interactive cytotoxicity in human epidermal keratinocytes following exposure to a mixture of four metals. J. Agric. Biol. Environ. Stat. 7, 58–73.
Gennings, C., Carter, W. H., Jr., Casey, M., Moser, V., Carchman, R., and Simmons, J. E. (2004). Analysis of functional effects of a mixture of five pesticides using a ray design. Environ. Toxicol. Pharmacol. 18, 115–125.
Gennings, C., Schwartz, P., Carter, W. H., Jr., and Simmons, J. E. (1997). Detection of departures from additivity in mixtures of many chemicals with a threshold model. J. Agric. Biol. Environ. Stat. 2, 198–211.
Gessner, P. K. (1974). The isobolographic method applied to drug interactions. In Drug Interactions (P. L. Morselli, S. Garattini, and S. N. Cohen, Eds.), pp. 349–362. Raven Press, New York.
Hussain, M. A., Mohamad, R. B., and Oloffs, P. C. (1985). Studies on the toxicity, metabolism, and anticholinesterase properties of acephate and methamidophos. J. Environ. Sci. Health B20, 129–147.
Johnson, C. D., and Russell, R. L. (1975). A rapid, simple radiometric assay for cholinesterase, suitable for multiple determinations. Anal. Biochem. 64, 229–238.
Karanth, S., Liu, J., Olivier, K., Jr., and Pope, C. (2004). Interactive toxicity of the organophosphorus insecticides chlorpyrifos and methyl parathion in adult rats. Toxicol. Appl. Pharmacol. 196, 183–190.
Karanth, S., Olivier, K., Jr., Liu, J., and Pope, C. (2001). In vivo interaction between chlorpyrifos and parathion in adult rats: Sequence of administration can markedly influence toxic outcome. Toxicol. Appl. Pharmacol. 177, 247–255.
Kardos, S. A., and Sultatos, L. G. (2000). Interactions of the organophosphates paraoxon and methyl paraoxon with mouse brain acetylcholinesterase. Toxicol. Sci. 58, 118–126.
Kousba, A. A., Sultatos, L. G., Poet, T. S., and Timchalk, C. (2004). Comparison of chlorpyrifos-oxon and paraoxon acetylcholinesterase inhibition dynamics: Potential role of a peripheral binding site. Toxicol. Sci. 80, 239–248.
Lauwerys, R. R., and Murphy, S. D. (1969). Interaction between paraoxon and tri-o-tolyl phosphate in rats. Toxicol. Appl. Pharmacol. 14, 348–357.
Maxwell, D. M. (1992). Detoxication of organophosphorus compounds by carboxylesterase. In Organophosphates: Chemistry, Fate, and Effects (J. E. Chambers and P. E. Levi, Eds.), pp. 183–199. Academic Press, New York.
McCullagh, P., and Nelder, J. A. (1989). Generalized Linear Models, 2nd ed. Chapman and Hall, New York.
Moser, V. C. (1995). Comparisons of the acute effects of cholinesterase inhibitors using a neurobehavioral screening battery in rats. Neurotoxicol. Teratol. 17, 617–625.
Moser, V. C., and Padilla, S. (1998). Age- and gender-related differences in the time course of behavioral and biochemical effects produced by oral chlorpyrifos in rats. Toxicol. Appl. Pharmacol. 149, 107–119.
Murphy, S. D., Anderson, R. L., and DuBois, K. P. (1959). Potentiation of toxicity of malathion by triorthotolyl phosphate. Proc. Soc. Exp. Biol. Med. 100, 483–387.
Murphy, S. D., Cheever, K. L., Chow, A. Y. K., and Brewster, M. (1976). Organophosphate insecticide potentiation by carboxylesterase inhibitors. Proc. Eur. Soc. Tox. XVII Esc. Med. Int. Cong. 376, 292–300.
Murphy, S. D., and DuBois, K. P. (1957). Quantitative measurement of inhibition of the enzymatic detoxification of malathion by EPN (ethyl p-nitrophenyl thionobenzene phosphate). Proc. Soc. Exp. Biol. Med. 96, 813–818.
Naguib, M., Farsci, F., and Yaksh, T. L. (1997). Characterization of muscarinic receptor subtypes that mediate antinociception in the rat spinal cord. Anesth. Analg. 85, 847–853.
National Research Council. (1989). Acetylcholinesterase inhibitors: Case study of mixtures of contaminants with similar biological effects. In Drinking Water and Health. Selected Issues in Risk Assessment. Part II. Mixtures. Vol. 9, pp. 146–161. National Academy Press, Washington, DC.
Nostrandt, A. C., Padilla, S., and Moser, V. C. (1997). The relationship of oral chlorpyrifos effects on behavior, cholinesterase inhibition, and muscarinic receptor density in rat. Pharmacol. Biochem. Behav. 58, 15–23.
Pedigo, N. W., Dewey, W. L., and Harris, L. S. (1975). Determination and characterization of the antinociceptive activity of intraventricularly administered acetylcholine in mice. J. Pharmacol. Exp. Ther. 193, 845–852.
Poet, T. S., Kousba, A. A., Dennison, S. L., and Timchalk, C. (2004). Physiologically based pharmacokinetic/pharmacodynamic model for the organophosphorus pesticide diazinon. Neurotoxicol. 25, 1013–1030.
Pond, A. L., Chambers, H. W., and Chambers, J. E. (1995). Organophosphate detoxication potential of various rat tissues via A-esterase and aliesterase activities. Toxicol. Letters 78, 245–252.
Pope, C. N. (1999). Organophosphorus pesticides: Do they all have the same mechanism of toxicity J. Toxicol. Environ. Health, Part B, 2, 161–181.
Reiter, L. W. (1983). Chemical exposures and animal activity: utility of the figure-eight maze. Dev. Toxicol. Environ. Sci. 11, 73–84.
Safe, S. H. (1998). Hazard and risk assessment of chemical mixtures using the toxic equivalency factor approach. Environ. Health Perspect. 106(Suppl. 4), 1051–1058.
Seume, F. W., Casida, J. E., and O'Brien, R. D. (1960). Effects of parathion and malathion separately and jointly upon rat esterases in vivo. Agricul. Food Chem. 8, 43–47.
Solana, R. P., Gennings, C., Carter, W. H., Jr., Anderson, D., Lennox, W. J., Carchman, R. A., and Harris, L. W. (1990a). Evaluation of the efficacy of two carbamates, physostigmine and pyridostigmine, when used in conjunction for protection against organophosphate exposure. Fundam. Appl. Toxicol. 15, 814–819.
Solana, R. P., Harris, L. W., Carter, W. H., Jr., Talbot, B. G., Carchman, R. A., and Gennings, C. (1990b). Evaluation of a two-drug combination pretreatment against organophosphorus exposure. Toxicol. Appl. Pharmacol. 102, 421–429.
Su, M.-Q., Kinoshita, F. K., Frawley, J. P., and DuBois, K. P. (1971). Comparative inhibition of aliesterases and cholinesterase in rats fed eighteen organophosphorus insecticides. Toxicol. Appl. Pharmacol. 20, 241–249.
Teuschler, L., Klaunig, J., Carney, E., Chambers, J., Connolly, R., Gennings, C., Giesy, J., Hertzberg, R., Klaassen, C., Kodell, R., Paustenbach, D., and Yang, R. (2002). Support of science-based decisions concerning the evaluation of the toxicology of mixtures: A new beginning. Reg. Toxicol. 36, 34–39.
Timchalk, C., Nolan, R. J., Mendrala, A. L., Dittenber, D. A., Brzak, K. A., and Mattsson, J. L. (2002). A physiologically based pharmacokinetic and pharmacodynamic (PBPK/PD) model for the organophosphate insecticide chlorpyrifos in rats and humans. Toxicol. Sci. 66, 34–53.
Timchalk, C., Poet, T. S., Hinman, M. N., Busby, A. L., and Kousba, A. A. (in press). Pharmacokinetic and pharmacodynamic interaction for a binary mixture of chlorpyrifos and diazinon in the rat. Toxicol. Appl. Pharmacol.
Uchiyama, M., Yoshida, T., Homma, K., and Hongo, T. (1975). Inhibition of hepatic drug-metabolizing enzymes by thiophosphate insecticides and its drug toxicological implications. Biochem. Pharmacol. 24, 1221–1225.
U.S. Department of Agriculture. (2000). Pesticide Data Program: Annual Summary Calendar Year 1998, http://www.ams.usda.gov/science/pdp/.
U.S. Environmental Protection Agency. (1999). Pesticide Industry Sales and Usage, 1996 and 1997 Market Estimates, http://www.epa.gov/oppbead1/pestsales/.
U.S. Environmental Protection Agency (2001). Organophosphate Pesticides: Preliminary OP Cumulative Risk Assessment, http://www.epa.gov/oppsrrd1/cumulative/pra-op/newdesign.htm.
U.S. Environmental Protection Agency. (2002). Organophosphate Pesticides: Revised OP Cumulative Risk Assessment, http://www.epa.gov/pesticides/cumulative/rra-op/.
U.S. Food and Drug Administration. (2000). Pesticide Program: Residue Monitoring, 1999, http://www.cfsan.fda.gov/download/pes99db.html.
Ward, T. R., and Mundy, W. R. (1996). Organophosphorus compounds preferentially affect second messenger systems coupled to M2/M4 receptors in rat frontal cortex. Brain Res. Bull. 39, 49–55.
Wu, H. X., Evreux-Gros, C., and Descotes, J. (1996). Diazinon toxicokinetics, tissue distribution and anticholinesterase activity in the rat. Biomed. Environ. Sci. 9, 359–369.(V. C. Moser, M. Casey, A.)