In-vivo kinetics of the interaction between midazolam and erythromycin in rats, taking account of metabolic intermediate complex formation.

To predict, quantitatively, the extent of drug interaction during repeated administration of a metabolic inhibitor, we analysed the effects of erythromycin treatment under several regimens on the area under the concentration curve (AUC) of midazolam in rats. Midazolam was administered into the portal vein 12 h after erythromycin treatment for 1, 2 or 3 days, or 12, 24, 36, 48, 72 and 96 h after erythromycin treatment for 4 days, and the plasma-concentration profiles of midazolam were analysed to assess the AUC. Moreover, the contents of total cytochrome P450 and inactive metabolic intermediate (MI) complex were simultaneously quantitated. While the AUC value of midazolam was not affected by the administration of erythromycin for 1 day, repeated administration of erythromycin evoked an increase in AUC ratio (AUC in erythromycin-treated rats/AUC in vehicle-treated rats), which reached a maximum value of 1.99 at 12 h after 4 days' treatment with erythromycin. The total content of cytochrome P450 in liver microsomes was unaffected by erythromycin treatment. Although the MI complex was undetectable after 1 day's treatment with erythromycin, its content increased with duration of erythromycin treatment, and the complex disappeared after the end of erythromycin treatment with a half-life of 12.3 h. In conclusion, the interaction between erythromycin and midazolam could be well predicted when the formation of MI complex in the liver was taken into account.

Prediction of midazolam-CYP3A inhibitors interaction in the human liver from in vivo/in vitro absorption, distribution, and metabolism data.

The extent of decreases in apparent hepatic clearance and intrinsic hepatic clearance of midazolam (MDZ) after intravenous administration of MDZ with concomitant oral administration of cimetidine (CIM), itraconazole (ITZ), or erythromycin (EM) was predicted using plasma unbound concentrations and liver unbound concentrations of inhibitors. When MDZ was concomitantly administered with CIM, the observed increase in MDZ concentration was successfully predicted using inhibition constants assessed by human liver microsome and liver-to-plasma unbound concentration ratios in rats. However, the extent of interaction with ITZ or EM was still underestimated even taking into account the concentrative uptake of inhibitors into liver. We could predict the degree of "mechanism-based" inhibition by EM on the hepatic metabolism of MDZ, after repeated administration of EM, by a physiological model incorporating the amount of active enzyme as well as the concentration of inhibitor. The maximum inactivation rate constant and the apparent inactivation constant of EM on MDZ metabolism were 0.0665 min(-1) and 81.8 microM, respectively. These kinetic parameters for the inactivation of the enzyme were applied to the physiological model with pharmacokinetic parameters of EM and MDZ obtained from published results. Consequently, we estimated that cytochrome P450 3A4 in the liver after repeated oral administration of EM was inactivated, resulting in 2.6-fold increase in the plasma concentration of MDZ. The estimated extent of increase in MDZ concentration in our study correlated well with the observed value based on metabolic inhibition by EM from published results.

Quantitative prediction of metabolic inhibition of midazolam by erythromycin, diltiazem, and verapamil in rats: implication of concentrative uptake of inhibitors into liver.

To evaluate the degree of drug-drug interaction concerning metabolic inhibition in the liver quantitatively, we tried to predict the plasma concentration increasing ratio (R) of midazolam (MDZ) by erythromycin (EM), diltiazem (DLZ), or verapamil (VER) in rats. MDZ was administered through the portal vein at the steady state of plasma concentration of these inhibitors. The R values in the area under the plasma concentration curve of MDZ in the presence of EM, DLZ, and VER were 2.02, 1.64, and 1.30, respectively. The liver to plasma unbound concentration ratios of EM, DLZ, and VER at the steady state after infusion were 20.8, 1.02, and 3.01, respectively, suggesting concentrative uptake of EM and VER into the liver. The predicted R value in the presence of EM calculated by use of plasma unbound concentration was 1.03, whereas the value calculated with liver unbound concentration was 1.61, which was very close to the observed value. These findings indicated the need to consider the concentrative uptake of inhibitors into the liver for the quantitative prediction of metabolic inhibition. However, the predicted values in the presence of DLZ or VER calculated by use of liver unbound concentration were still underestimated. This result may be due to the metabolic inhibition by the metabolites of both inhibitors. Therefore, when predicting the degree of metabolic inhibition quantitatively, the inhibitory effect by coadministered drugs and the disposition of these metabolites in the liver must also be considered.

Correlation between in vivo and in vitro hepatic uptake of metabolic inhibitors of cytochrome P-450 in rats.

To predict the degree of accumulation of hepatic metabolic inhibitors in the liver from the in vitro data, we investigated the relationship between cell/medium concentration ratios (C/M ratios) in isolated rat hepatocytes and liver/blood unbound concentration (K(Bf)) after i.v. administration of various metabolic inhibitors such as itraconazole, ketoconazole, verapamil, diltiazem, enoxacin, ciprofloxacin, clarithromycin, cimetidine, and nizatidine. The C/M ratios of itraconazole were approximately 6,000 and 200 at the concentrations of 0.1 and 10 microg/ml, respectively, and the uptake of ketoconazole and verapamil into the hepatocytes also showed a concentration dependence, although the degree was smaller than that of itraconazole. The uptake of diltiazem, enoxacin, ciprofloxacin, and clarithromycin into the hepatocytes showed linear profiles on concentration dependence. There was an excellent correlation between C/M ratios and K(Bf) values of all nine drugs with a slope of 1. This finding suggested the possibility of predicting drug concentrations in the liver (C(H)) from C/M ratios, the blood concentrations of drugs (C(B)) and unbound fraction in blood (f(B)), which was expressed by C(H) = (C/M). C(B). f(B). It may be possible to predict the drug concentrations in human liver from K(Bf) values estimated with isolated human hepatocytes and concentrations in the blood in a similar manner as in rats.

Quantitative prediction of the interaction of midazolam and histamine H2 receptor antagonists in rats.

To quantitatively evaluate drug-drug interactions involving metabolic processes in the liver, we attempted to predict the increasing ratio of the plasma concentration of midazolam (MDZ) in the presence of cimetidine (CIM) or nizatidine (NZD) in rats. Under steady-state conditions for the plasma concentration of CIM or NZD, MDZ was administered through the portal vein. The AUC of MDZ in the presence of CIM was 2.5-fold higher than that in the absence of CIM. There was no effect of NZD on the AUC of MDZ. The liver/plasma concentration ratios for CIM and NZD were 4.0 and 2.7, respectively. The estimated liver unbound concentration (CH,f)/plasma unbound concentration (Cp,f) ratios for CIM and NZD were 1.9 and 2.4, respectively, suggesting concentrative hepatic accumulation of both drugs. The oxidative metabolism of MDZ in rat liver microsomes was competitively inhibited by CIM or NZD, and the Ki values of CIM and NZD were 110 and 2600 microM, respectively. Based on these data obtained in vivo and in vitro, the increasing ratios for MDZ in the presence of CIM or NZD were predicted using the equations Rp = 1 + Cp,f/Ki and RH = 1 + CH,f/Ki. The observed increasing ratios in the presence of CIM were very close to RH, compared with Rp. However, Cp, f and CH,f were much less than Ki and there was no difference between Rp and RH in the presence of NZD. Consequently, Cp,f and CH, f were greater than or equal to Ki and Cp,f was not equal to CH,f, as in the presence of CIM, and it was indicated that CH,f was more suitable for quantitatively predicting the drug-drug interactions than was Cp,f.