Physiologically Based Pharmacokinetic Modeling and Simulation of Thiazolidinediones in Patients with Hepatic Impairment

Abstract

Background The prevalence of liver disease in diabetic patients is very high. As the liver is the major organ responsible for glycemic control and drug elimination, pharmacokinetic (PK) evaluation of antidiabetic drugs is essential in patients with hepatic impairment. Thiazolidinedione (TZD) drugs (e.g., lobeglitazone, rosiglitazone, and pioglitazone) — agonists of peroxisome proliferator-activated receptor−γ (PPAR−γ) —increase insulin sensitivity. TZDs share many PK characteristics. Lobeglitazone, rosiglitazone, and pioglitazone have high bioavailability (80—90%) with short times to reach peak concentration (Tmax ~1h). These drugs are highly bound to serum albumin (>99%), are mainly metabolized by liver enzymes, and excreted as metabolites; the renal excretion of the parent drugs is negligible. Hepatic metabolism is the main route of elimination. Therefore, the PK parameters of TZDs can be significantly altered in patients with hepatic impairment, varying according to the severity of impairment. In principle, clinical trials should be conducted on patients with hepatic impairment to obtain the PK parameters for these drugs. However, there are many difficulties associated with such trials, especially with respect to recruiting patients and safety issues. Recently, PK modeling and simulation methods have been focused to overcome such limitations of clinical trials. The physiologically based pharmacokinetic (PBPK) modeling and simulation approach combines chemical properties, biochemical reactions, and physiological characteristics to predict the absorption, distribution, and elimination of a drug after administration. This method can be used, instead of conducting clinical trials, to predict PK parameters in patients with various diseases, children, and pregnant women. The purpose of this study was to develop PBPK models of lobeglitazone, rosiglitazone, and pioglitazone in patients with hepatic impairment of varying severity, as defined by the Child-Pugh classification system. Method The population-based software, SimCYPTM version 18.1 (Certara, St. Louis, MO, USA) was used to develop PBPK models of lobeglitazone, rosiglitazone, and pioglitazone. The input parameters related to the physicochemical properties of these drugs were obtained from the open-source database, DrugBank (go.drugbank.com). The required in vitro experimental values were obtained by searching the literature. The patient demographics of hepatic impairment were defined based on previous reports of different phenotypes of hepatic enzymes and varying degrees of pathophysiology. To develop PBPK models, a first-order kinetic absorption model was applied for lobeglitazone and rosiglitazone, whereas an advanced dissolution, absorption, and metabolism (ADAM) model was applied for pioglitazone. The full-body distribution model was applied to represent each major organ as a compartment connected by blood flow. Based on in vitro information about relevant hepatic enzymes, the enzyme kinetic model was selected for all TZDs to analyze different contributions of the intrinsic clearance of each enzyme to decreased liver function, with the severity of the hepatic impairment stratified according to the Child-Pugh classification system. The developed PBPK models were validated with the major PK parameter values (especially for the peak concentrations, Cmax, and the areas under the concentration‒ time curve, AUCs) from simulation results obtained for virtual populations and compared to those from clinical trials in age- and sex-matched healthy subjects. The acceptable range for the predicted capability of the models was from 0.5 to 2.0 for the ratios of the predicted and observed PK parameters. For lobeglitazone, an additional evaluation of the model was accomplished by comparing the simulated PK parameters to the values from the available clinical trials in Child-Pugh A and B patients to determine whether the model was applicable to patients with hepatic impairment. The validated PBPK models of the TZDs were simulated in 100 virtual, healthy, Child-Pugh A, B, or C subjects cases aged 18‒65 years, with a 50:50 male:female ratio. The major PK parameters of the simulation results were subsequently compared among all groups (healthy population, Child-Pugh A, B, and C). Results The developed PBPK models of lobeglitazone, rosiglitazone, and pioglitazone in this study were valid, with the ratio of each predicted major PK parameter value of simulations to the observed value of clinical trials within the acceptable range for healthy subjects. The prediction errors (Pes) of Cmax and AUC were 0.84 and 0.97 for lobeglitazone; 0.86 and 0.88 for rosiglitazone; and 0.94 and 0.91 for pioglitazone, respectively. There were no significant differences between the predicted PK values and the observed values for lobeglitazone in Child-Pugh A and B patients. Hence, the validity of the model was additionally confirmed: the PEs of Cmax and AUC were 0.94 and 1.49 in Child-Pugh A patients and 1.10 and 1.20 in Child-Pugh B patients. The simulation results showed that the major PK parameters of all TZDs in the Child-Pugh A patients were not significantly different from those in healthy populations; the geometric mean ratios (GMRs) were between 0.5 and 2.0 for Cmax and AUC. In Child-Pugh B patients, the GMRs for AUC and Cmax were 1.36 and 0.97 for lobeglitazone, 1.93 and 0.92 for rosiglitazone, 2.19 and 1.03for pioglitazone, respectively. The GMRs of AUC and Cmax were 1.94 and 0.92 for lobeglitazone, 2.26 and 0.81 for rosiglitazone, and 2.76 and 0.91 for pioglitazone, respectively, in Child- Pugh C patients. Additional analysis also revealed a significant alteration in the unbound fraction (fu) based on the severity of hepatic impairment. The GMRs of AUC for fu of lobeglitazone were as follows: 1.51, 2.50, and 3.62 for lobeglitazone; 1.73, 3.23, and 5.66 for rosiglitazone; 1.78, 2.82, and 4.56 for pioglitazone in Child- Pugh A, B, and C patients, respectively. Discussion The current PBPK models of TZDs, applied as full-body distribution model considering the functional liver volume, hepatic enzyme activities, blood flow, and serum proteins could be used to predict PK alterations in patients with hepatic impairment, grouped according to the Child-Pugh classification system. Compared to the previous report, the simulation results of the lobeglitazone PBPK model were inconsistent with the observed values in healthy subjects, and Child-Pugh A and B patients. Thus, the simulation results in Child-Pugh C patients could be used for reference, although the model tended to slightly overpredict the AUC in patients with hepatic impairment. The simulation of the rosiglitazone model showed increased AUCs, consistent with previous reports in patients with hepatic impairment. However, the simulation of the pioglitazone model overpredicted AUCs in patients with hepatic impairment. This could be due to the nature of PBPK modeling to overpredict AUCs in hepatic impairment, not taking the transporter activity into account in this model, or not considering CYP2C8 genetic polymorphisms in the observed data used for the validation of the model. Nonetheless, the PBPK model and simulation in this study were noteworthy because the results provided evidence supporting the results of clinical trials. The PK parameters could be compared among healthy subjects and patients with Child- Pugh A, B, and C based on the simulation results. In addition, the PBPK models enabled the evaluation of the unbound fraction in healthy subjects and patients with hepatic impairment by severity, which was not feasible due to technical limitations to measure these values in the clinical trials. Conclusion The predictive capability of PBPK modeling and simulation in this study would be appropriate and could be applied to prediction of PK parameters of TZDs in patients with hepatic impairment graded according to the Child-Pugh classification system. Thus, the PBPK modeling and simulation could be notable to give supplementary information under those limited condition to conduct of clinical trials in specific patient groups.