Physiologically-Based Pharmacokinetic Models

A number of models have been developed specifically for methanol. PBPK models incorporate

Appendix II

Appendix II

Appendix II

species-specific parameters such as blood flow rates, tissue volumes and relative levels of blood perfusion, and known metabolic mechanisms. Once developed, PBPK models can then be validated using available data on the disposition of the chemical of interest in various species. Based on the validity of the model, a decision can then be made on its use for predicting human risk from chemi-cal exposure. These models are briefly described below.

Aone-compartment,“semi-physiologic”PBPKmodelwasdevelopedbyPerkinsetal.(57, 62)to describemethanoldispositioninmiceandrats.Modelpredictionsformethanoldispositioninmice during and after inhalation exposure were compared to those previously determined in the female Sprague-Dawleyrat,andthedispositioninmiceaftervariousexposurerouteswasalsoexamined.

Using published kinetic parameters determined after IV and oral administration of methanol in humans and other primates, and estimated fraction of absorbed methanol (Φ) and physiological parameters, Perkins et al. next applied the inhalation pharmacokinetic model for rodents to humans (57). Data for the IV exposure were modeled with the one-compartment model described in Perkins et al. (62), with saturable elimination that was first-order at low levels of blood methanol. Data for oral exposure were modeled similarly but with a factor for gut absorption. Maintaining the fraction of absorbed methanol (Φ) as the dependent variable, and using kinetic parameters from the oral or intravenous data, inhalation data were then fitted to the previously determined pharmacokinetic model. Background human blood methanol from both endogenous and exogenous sources was set at 1.0 mg/L for the initial time step. The authors estimated that following an 8-hour exposure to 5,000 ppm methanol vapor (6,550mg/m³), blood methanol concentrations in the mouse would be 13–18-fold higher than in humans, whereas methanol concentrations in the rat would be 5-fold higher than the value for humans.

The semi-physiologic model was further applied to methanol disposition in rodents when absorp-tion was confined to the upper respiratory tract, where the majority of methanol absorpabsorp-tion occurs (63). Their research results support the hypothesis that absorption of inhaled methanol takes place entirely in the upper respiratory tract of rodents. Methanol absorption was increased by decreased ventilation, but unaffected by increased ventilation. The semi-physiologic pharmacokinetic model developed by the study authors incorporated the body burden of methanol computed from blood methanol measurements, methanol elimination estimates, ventilation rate, and fractional absorption.

Because ventilation rate varies with blood methanol concentration, and fractional absorption varies with environmental methanol concentration and ventilation rate, additional equations were derived to modulate these values using nonlinear least-squares regression.

A two-compartment model for methanol disposition in pregnant rodents which utilized Michaelis-Menten elimination from the maternal compartment was developed (61). Pregnant Sprague-Dawley rats were given a single dose of 100 or 2,500 mg/kg/methanol by gavage or by IV. Pregnant CD-1 mice were also given a single dose of 2,500 mg/kg by gavage or IV. Methanol disposition was de-termined in non-pregnant rats, and at gd 7, 14, and 20 (to approximate three trimesters); in mice, non-pregnant animals and pregnant animals at gd 9 and 18 were examined. Blood samples were taken via jugular vein cannula. Rat concentration-time data were modeled using two-compartment models for each dose; mouse data were modeled with a one-compartment model with Michaelis-Menten elimination. Blood methanol levels after oral exposure rose more rapidly in pregnant than

Appendix II

non-pregnant rats, but the opposite was true for mice (61). Peak blood levels in rats were higher during pregnancy. V_max for elimination in near-term rats and mice was 65– 80% of that in non-pregnant animals. Mice eliminated methanol twice as quickly as rats. Qualitatively, the disposition between pregnant and non-pregnant animals was similar, with the same model, incorporating differ-ent parameter estimates, adequately fitting both conditions. In vitro studies showed that adult near-term livers have a V_max for methanol metabolism of 85% that in livers from non-pregnant rodents.

Mouse liver homogenates metabolized methanol twice as fast as rat liver homogenates. Fetal rodent livers had a V_max less than 5% that of adults.

A PBPK model was developed by Ward et al. (64) for the disposition of methanol in rat and mouse dams and the conceptus. The model was validated by exposing rats on gd 14 and 20, and mice on gd 18, via injection into the jugular cannula, and using intrauterine microdialysis to measure trans-placental methanol toxicokinetics.

The conceptal/maternal diffusion constant ratio consistently decreased with increasing dose in pregnant rats and mice, consistent with earlier observations that methanol limits its own delivery to the conceptus (65). The validated model described methanol elimination as occurring primarily in the liver by a saturable, first-order metabolic process, as has been demonstrated in other studies (45, 53). Methanol tended to partition to tissues with high water content. Peak methanol concentration

) increased slightly but non-significantly in maternal blood as gestation progressed, consis -(C_max

tent with the decrease in V_max for methanol elimination described by Ward and Pollack (61). The conceptal/maternal ratio of AUCs decreased with dose and gestation progression; at low doses con-ceptal AUC exceeded maternal, but at higher doses, maternal exceeded concon-ceptal AUC. Pregnant mice data from the Dorman et al. (66) study were also used to validate the model; at 10,000 ppm the conceptal methanol AUC exceeded maternal blood AUC by 10%, while at 15,000 ppm, the ma-ternal blood methanol AUC exceeded the conceptal AUC by 30%.

A disadvantage of the microdialysate procedure is the need to keep animals anesthetized. Urethane was used in this study for anesthesia, and it may have had some effect on pharmacokinetic param-eters. However, the parameters obtained here fit well with those obtained from other studies with non-anesthetized animals.

The Panel concluded that the PBPK studies described above represent an extensive series of care-fully conducted experiments to develop pharmacokinetic models for rodents exposed to methanol and to begin to apply the results to humans. The strengths of these studies are the use of appropri-ate techniques to measure blood methanol, good study design, and justification of the models. This work has the most utility for understanding rodent toxicity studies.

As discussed earlier in Section 2.1.3, Horton et al. (53) developed a four-compartment PBPK mod-el that does not include a fractional absorption parameter (Φ). The modmod-el utilized a double pathway for metabolism to formaldehyde in the liver: one pathway using rodent catalase K_m and V_max, and one using smaller K_m and V_max values to simulate an enzyme with higher affinity and lower capac-ity. The compartments were richly perfused tissue (adrenals, brain, gastrointestinal tract), slowly perfused tissue (muscle, fat), kidney, and liver (the major metabolizing compartment). The model was scaled up for humans using the 0.74 power of body weight.

Appendix II

Appendix II

Appendix II

Horton et al. (53) is a careful attempt to develop PBPK models for methanol in rats, monkeys, and humans. The Horton models differ from those discussed in the preceding section in that they include more compartments but do not account for fractional absorption. Another important differ-ence is that Horton et al. used a much lower range of methanol exposure conditions for the rodent studies, therefore there is one more confidence extrapolating the results to humans. The inclusion of data on primates that was developed in the same laboratory, using the same techniques, is a plus.

As discussed earlier in Section 2.1.3, the publication of Fisher et al. (56) quantitatively estimated relative respiratory uptake of methanol, demonstrated the linearity of uptake over a range of doses, and proposed that correction for uptake can be readily incorporated in PBPK models.

Environ (67) performed a comparative analysis of the Perkins et al. (57, 62) and Horton et al. (53) models on behalf of the American Forest and Paper Association (AF&PA). The analysis included the presentation of the exact algebraic forms of the models’ mathematical relationships, and the application of these relationships to the prediction of human, monkey, and rodent blood methanol levels following exposure to low (83 mg/m³) and higher (260, 1,300, and 2,600 mg/m³) levels of methanol vapor. Both models produced similar results for steady-state blood methanol levels at various exposures, with the exception of the failure of the Perkins et al. model to achieve steady state at the highest exposure concentration (2,600 mg/m³) in mice and rats. Because the Perkins et al. (57, 62) model exhibited consistently smaller initial rates of methanol uptake across species, the Horton et al. (53) model predicts higher blood methanol levels prior to achieving steady state.

This difference may be due to the fact that the Horton et al. model does not incorporate a fractional absorption parameter (Φ). The Perkins et al. (57, 62) model, however, incorporates only a single metabolic compartment, and does not consider lung or kidney elimination, resulting in its inability to reach steady state at high methanol vapor concentrations. Environ (67) concluded that both mod-els support a similar, prepredicted result. The Environ (67) analysis also provides additional insights and explanation of the models used in the above studies.