Physiological modeling and derivation of the rat to human toxicokinetic uncertainty factor for the carbamate pesticide aldicarb.

Aldicarb (ALD, 2-methyl-2-(methylthio)-propionalaldehyde O-(methyl-carbamoyl) oxime, Temik) is widely used as an insecticide, nematocide and acaricide and it is oxidized to aldicarb sulfoxide (ALX) and aldicarb sulfone (ALU). Neither a toxicokinetic model nor an estimate of the target tissue dose of ALD and its metabolites in exposed organisms is available. The objective of this study was: (i) to develop a physiologically based toxicokinetic (PBTK) model for ALD in the rat and humans, and (ii) to determine the interspecies toxicokinetic uncertainty factor (UF(AH-TK)) of ALD. The model consists of a series of mass balance differential equations that describe the time course behavior of ALD in blood, liver, kidney, lungs, brain, fat, and rest of the body compartments. The physiological parameters of the model (blood flow rates, cardiac output, and tissue volumes) were obtained from the literature, while the maximum velocity (mg/kg/min) and the Michaelis constant (mg/l) for ALD oxidation in rats and humans were determined by in vitro AH-TK microsomal assays. The estimation of the tissue:blood partition coefficient was accomplished within the PBTK model by representing the tissues as a composite of neutral lipids, phospholipids and water, and providing the vegetable oil:water partition coefficient as input parameter. The validity of the rat PBTK model was assessed by comparing the model simulations of ALX time course blood concentrations and the inhibition patterns of acetylcholinesterase (AChE) in erythrocytes and plasma obtained by administering rats ALD (0.1 and 0.4 mg/kg, iv). The human PBTK model was validated by comparing the simulations of AChE inhibition patterns in blood with human experimental data obtained from oral administrations of ALD. The UF(AH-TK) for ALD was determined by dividing the areas under the blood and brain concentration vs time curve (AUCCV, AUCCBR) for ALD and ALX in the rat and in human exposed to the same dose. The results indicate that with respect to parent chemical, equivalent applied doses in rats and humans result in a 9.5-fold difference in the AUC(CV) and AUC(AH-TK) respectively, in the two species, and 17-fold difference in the AUC(CV) and AUC(CBR) with respect to the metabolite. In other words, in order to have toxicokinetic equivalence in rats and humans, the former species must be exposed to a dose that is 9.5 and 17 times higher than the human with respect to the parent chemical and the metabolite respectively. Overall, the present study demonstrates the applicability of PBTK models in the quantitative evaluation of UH(AH-TK), and shows that their current default values are inaccurate, at least with respect to ALD, which has potential negative implications in the alleged protection of risk estimates derived from them.

Physiological-model-based derivation of the adult and child pharmacokinetic intraspecies uncertainty factors for volatile organic compounds.

The intraspecies uncertainty factor (UF(HH)=10x) is used in the determination of the reference dose or reference concentration and accounts for the pharmacokinetic and pharmacodynamic heterogeneity within the human population. The Food Quality Protection Act of 1996 mandated the use of an additional uncertainty factor (UF(HC)=10x) to take into account potential pre- and postnatal toxicity and lack of completeness of the data with respect to exposure and toxicity to children. There is no conclusive experimental or theoretical justification to support or refute the magnitude of the UF(HH) and UF(HC) nor any conclusive evidence to suggest that a factor of 100 is needed to account for intrahuman variability. This study presents a new chemical-specific method for estimating the pharmacokinetic (PK) component of the interspecies uncertainty factor (UF(HH-PK) and UF(HC-PK)) for volatile organic compounds (VOCs). The approach utilizes validated physiological-based pharmacokinetic (PBPK) models and simplified physiological-model-based algebraic equations to translate ambient exposure concentration to tissue dose in adults and children the ratio of which is the UF(HH-PK) and UF(HC-PK). The results suggest that: (i) the UF(HH-PK) and UF(HC-PK) are chemical specific; (ii) for the chemicals used in this study there is no significant difference between UF(HH-PK) and UF(HC-PK); (iii) the magnitude of UF(HH-PK) and UF(HC-PK) varies between 0.033 and 2.85 with respect to tissue and blood concentrations; (iv) the body weight, the rate of ventilation, the fraction of cardiac output flowing to the liver, the blood : air partition coefficient, and the hepatic extraction ratio are the only parameters that play a critical role in the variability of tissue and blood doses within species; and (v) the magnitude of the UF(HH-PK) and UF(HC-PK) obtained with the simplified steady-state equations is essentially the same with that obtained with PBPK models. Overall, this study suggests that no adult-children differences in the parent chemical concentrations of the VOCs are likely to be observed during inhalation exposures. The physiological-model-based approaches used in the present study to estimate the UF(HH-PK) and UF(HC-PK) provide a scientific basis for their magnitude. They can replace the currently used empirical default approaches to provide chemical-specific UF(HH-PK) in future risk assessments.

Evaluation of subchronic toxicity data using the benchmark dose approach.

We used the benchmark dose (BMD) methodology devised by Crump (Fundam. Appl. Toxicol. 4, 854-871, 1984) to estimate BMDs for 90-day toxicological data and several fabricated data sets. From a toxicological perspective, dose-response modeling offers certain advantages over using a point estimate, such as the currently used no-observable-adverse-effect level (NOAEL) approach. However, there are many variables associated with the BMD that could be set to produce unreasonable BMD estimates. Some of these variables and decisions are examined in this study. BMDs were calculated for discrete and continuous endpoints using a variety of different variables (e.g., maximum likelihood estimates [MLEs], lower-confidence limits [LCLs], and different risk levels). In addition, the fabricated data sets were manipulated (i.e., dose groups eliminated) and the BMDs recalculated. This process tested how the BMD estimates varied using different forms of the data. For the 90-day toxicological studies, the BMDs were typically within an order of magnitude of the NOAEL for discrete endpoints. For the discrete endpoints, the MLEs were typically greater than the NOAEL and the LCLs were typically less than the NOAEL. The BMD was insensitive to changes in the data points one to two dose groups beyond the NOAEL/LOAEL. With the continuous data, the ratios of MLEs and LCLs to the NOAEL were highly variable, and no general trend could be determined. The BMD methodology offers potential improvements in the risk assessment process since dose-response characteristics are used to calculate the BMD. Depending upon how the BMD is defined, i.e., the form of the dose-response model, and how the BMD is used in the risk assessment process, BMD estimates may produce reference doses/concentrations that are more or less conservative than the NOAEL approach. Active involvement in discussions with regulatory agencies is needed to ensure that inappropriate models and unreasonable BMDs are not used. In addition, further discussions on how BMDs should be used in the risk assessment process are needed.

Determination of the rate of aldicarb sulphoxidation in rat liver, kidney and lung microsomes.

1. The rate of sulphoxidation of aldicarb (2-methyl-2-(methylthio) propanal O-[(methylamino) carbonyl oxime], Temik) in rat hepatic, renal and pulmonary microsomes was determined by quantitating the levels of aldicarb sulphoxide and aldicarb sulphone produced during incubations. Under in vitro experimental conditions used in the present study, aldicarb sulphoxide was the only metabolite produced, and further metabolism of aldicarb sulphoxide to aldicarb sulphone was negligible. 2. The average maximal velocity (mumol/min/mg protein) for the sulphoxidation of aldicarb, based on measurements of product formation, in liver, kidney and lung microsomes was 5.41, 39.51 and 2.45 respectively. The corresponding values for the Michaelis constant (microM) were 184, 1050 and 188 respectively. 3. These results imply that under in vivo conditions (1) aldicarb sulphoxidation is not likely to be saturable even at lethal doses in the rat, and (2) aldicarb clearance in rat liver and kidney will be limited by the rate of blood flow and not metabolizing enzyme levels.

Assessing the relevance of rodent data on chemical interactions for health risk assessment purposes: a case study with dichloromethane-toluene mixture.

Several descriptive studies have reported the occurrence of infra-additive and supra-additive toxic interactions in rodents given high doses of chemicals by routes different from anticipated human exposures. In order to assess the relevance of such rodent data on chemical interactions for humans, the route, species, and dose extrapolations need to be conducted on the basis of proven/hypothetical interaction mechanisms. The present study initially developed a physiologically based model of the toxicological interaction reported in rats receiving high oral doses of dichloromethane (DCM) and toluene (TOL). This predictive model was then used to asses the relevance of DCM-TOL interaction for humans exposed to threshold limit values (TLVs) of these chemicals, following the conduct of the various, essential extrapolations (i.e., rat to human, oral to inhalation, high dose to low dose). The interaction modeling approach involved (i) obtaining validated rat and human physiologically based pharmacokinetic (PBPK) models for TOL and DCM from the literature, and (ii) linking them via the modified Michaelis-Menten equation accounting for hypothetical mechanisms of interactions (no interaction, competitive inhibition, noncompetitive inhibition, and uncompetitive inhibition). Of the various interaction mechanisms investigated, the noncompetitive and uncompetitive metabolic inhibitions were found to adequately describe the reduction of carboxyhemoglobinemia (COHB) observed in rats during combined exposures (18.8 mmol/kg TOL, +6.2 mmol/kg DCM, po; 0.005 mmol/kg TOL, ip +5000 ppm DCM, 1 hr). The simulation model, based on noncompetitive and uncompetitive inhibition mechanisms, suggests that only < 10% reduction in the area under the COHB vs time curve (AUCCOHB) is likely to occur in humans exposed to the current TLVs of DCM and TOL (compared to AUCCOHB resulting from an 8-hr exposure to TLV of DCM alone). The present modeling approach, based on hypothetical mechanisms of interaction, then indicates that rodent data on DCM-TOL interaction are not relevant for humans, particularly with respect to the COHB effect. The application of this kind of a predictive modeling approach should be useful in screening the available reports on chemical interactions for identifying those of greater concern at relevant human exposure levels (RfD, RfC, TLV).

A methodology for solving physiologically based pharmacokinetic models without the use of simulation softwares.

The objective of the present study was to develop and validate a methodology for solving physiologically based pharmacokinetic (PBPK) models without the use of simulation software. The approach involves keying the parameter values and model equations into Microsoft Excel spreadsheets, and conducting simulations by solving the model equations according to Euler's method of numerical integration. This approach was applied to simulate the pharmacokinetics of styrene in rats exposed to 80 and 600 ppm for 6 h. The simulation results were plotted along with experimental data using the regular graphic features available in Excel, and validated by comparing them with simulation results obtained using a commercially available software (Advanced Continuous Simulation Language, ACSL). The simulations obtained with ACSL and Excel, in general, differed by <1%. The methodology developed in the present study should help informed individuals understand and solve PBPK models, without having to use "black-box' kind of computer programs and simulation softwares.

Hematotoxic interactions: occurrence, mechanisms and predictability.

The available data on the binary chemical interactions involving hematotoxicants, particularly organic chemicals causing a reduction in either the number of white/red blood cells or the capacity of hemoglobin to transport oxygen, are limited. These observations are limited to investigations in rodents of the enhancement or attenuation of the hematotoxicity of benzene, dichloromethane and dimethylanilines following prior administration of inducers of CYP 2E1 or co-administration of substrates for this isoenzyme. The relevance of these data on interactions for humans exposed at low concentrations can be assessed only when the mechanism of interaction is understood at a quantitative level, and incorporated within a physiological modeling framework. The present study exemplifies the predictability of the magnitude of binary chemical interactions in humans exposed to low concentrations, by developing a physiological model of the modulation by toluene of dichloromethane-induced carboxyhemoglobinemia. Consistent with the basic biochemical principles, this modeling exercise suggests that, with competitive metabolic inhibition mechanism, the threshold for binary chemical interactions will follow a downward trend with increasing number of substrates or structurally-similar substances in a mixture. The use of this kind of mechanistic models, along with data from descriptive chemical interaction studies, will form the very basis of mechanistic risk assessment methods for complex chemical mixtures.

An approach for incorporating tissue composition data into physiologically based pharmacokinetic models.

The objective of this study was to develop an approach for incorporating tissue composition data into physiologically based pharmacokinetic (PBPK) models in order to facilitate "built-in" calculation of tissue: air partition coefficients (PCs) of volatile organic chemicals. The approach involved characterizing tissue compartments within PBPK models as a mixture of neutral lipids, phospholipids, and water (instead of using the conventional description of them as "empty" boxes). This approach enabled automated calculation of the tissue solubility of chemicals from n-octanol and water solubility data, since these data approximate those of solubility in tissue lipids and water. Tissue solubility was divided by the saturable vapor concentration at 37 degrees C to estimate the tissue: air PCs within PBPK models, according to the method of Poulin and Krishnan (1995c). The highest and lowest lipid and water levels for human muscle, liver, and adipose tissues were obtained from the literature and incorporated within the tissue composition-based PBPK model to calculate the tissue: air PCs of dichloromethane (DCM) and simulate the pharmacokinetics of DCM in humans. The PC values predicted for human tissues were comparable to those estimated using rat tissues in cases where the relative levels of lipids and water were comparable in both species. These results suggest that the default assumption of using rat tissue: air PCs in human PBPK models may be acceptable for certain tissues (liver, adipose tissues), but questionable for others (e.g., muscle). The PBPK modeling exercise indicated that the interindividual differences in tissue dose arising from variations of tissue: air PCs may not be reflected sufficiently by venous blood concentrations. Overall, the present approach of incorporating tissue composition data into PBPK models would not only enhance the biological basis of these models but also provide a means of evaluating the impact of interindividual and interspecies differences in tissue composition on the tissue dose surrogates used in PBPK-based risk assessments.

A simple index for representing the discrepancy between simulations of physiological pharmacokinetic models and experimental data.

The objective of this study was to develop an index that would provide a quantitative measure of the degree of discrepancy between simulations of physiologically based pharmacokinetic (PBPK) models and experimental data. The approach we developed involves the calculation of the root mean square of the error (representing the difference between the individual simulated and experimental values for each sampling point in a time course curve), and dividing it by the root mean square of the experimental values. The resulting numerical values of discrepancy measures for several data sets (each corresponding to an end point) obtained in a single experimental study are then combined on the basis of a weighting proportional to the number of data points contained in each data set. Such consolidated discrepancy indices obtained from several experiments (e.g., exposure scenarios, doses, routes, species) are averaged to get an overall discrepancy index, referred to as the PBPK index. This empirical index reflects the overall, weighted average percent difference between the a priori PBPK model simulations and experimental data. The proposed methodology is illustrated using previously published experimental and simulated data on dichloromethane pharmacokinetics in humans. The application of this kind of a "quantitative" method should help remove the ambiguity in communicating the degree of concordance or discrepancy between PBPK model simulations and experimental data.