Purpose: A semi-PBPK model for CIP after different routes of administration was developed and validated to predict and assess the sensitivity of plasma and sputum concentration-time profiles (cp(t), and cs(t)) as well as their respective exposure metrics to model parameters.
Methods: The final model included two catenary pulmonary disposition compartments (peripheral and central lung), representing pulmonary excretion/absorption/mucociliary clearance connected to a central body compartment (plasma, cp(t)). Systemic disposition was incorporated as a two-compartment body model with renal and hepatic elimination from the central body compartment. All transfer processes were assumed to be first-order, consistent with results from noncompartmental analyses. Sputum concentrations (cs(t)) were modeled as the amount of CIP in the central lung compartment corrected by the fraction of the surface fluid volume attributable to the central lung. To account for known differences in PK of CIP between HV and CF, a binary covariate model was used on pertinent systemic disposition (Vdcc, CLren and CLnonren) and oral absorption (Foral) parameters.
Initial estimates for pulmonary and systemic disposition as well as oral absorption parameters were based on available in-vitro and in-vivo studies. Mucociliary clearance values were based on an in-vivo study; in-vivo pulmonary deposition of INH CIP was obtained from the literature. Eighteen clinical PK studies with I.V. CIP (n=8, 50-400 mg) in HV and CF, with P.O. CIP (n=6, 100-1020 mg) in HV and CF and with INH CIP (n=4, 32.5-65 mg as DPI) in HV and CF were used as reference for parameter optimization and model validation; a total of 29 plasma and 2 sputum profiles were included in the analysis.
Point and variability estimates for the final model parameters were optimized sequentially (IV, PO and INH) to match respective model-predicted with observed (mean) cp(t) profiles, cs(t) profiles, and systemic exposure metrics (AUC). Monte Carlo simulations (MCS) were performed using normal and log-normal distributions for each parameter; model validity was assessed by visual predictive check (VPC, 5%-95%-percentiles) relative to both the reported cp(t) and cs(t) profiles and comparison of exposure metrics for all PK studies. Sensitivity analysis of cp(t) and cs(t) and associated exposure metrics after INH was performed for a 25-fold change around the point estimates of systemic disposition and oral absorption parameters, and 100-fold change around the pulmonary disposition final model parameters. R-Studio Version 1.0.136 with the deSOLVE package add-in was used for modeling.
Results: Model parameter optimization during MCS required addition of parameter variability, ranging from 20% to 75% (COV), primarily in pulmonary disposition parameters. Using the final parameter estimates, all observed cp(t) and cs(t) profiles data fall within the predicted 5%-95%-percentiles. Additionally, the predicted systemic and sputum exposure metrics (central tendencies and dispersion) were within <45% of their reported values.
The final, optimized PBPK model parameters indicate that pulmonary excretion of CIP accounts for 27% of the total (IV) administered dose, while 73% is eliminated via (renal and hepatic) CLtot from the central body compartment. Systemic distribution is limited (Vdss = 2.9 L/kg and 1.3 L/kg for HV and CF, respectively); BW-corrected total clearance comprises both renal and non-renal (CLtot = 9 mL/min/kg and 12 mL/min/kg in HV and CF, respectively). CF patients had consistently lower BW than HV (55 vs. 70 kg), resulting in similar CLtot, uncorrected for BW, but a higher Vdss in HV uncorrected for BW. Oral bioavailability (Foral) is high in both HV (73%) and in CF patients (85%). After inhalation, bioavailability (Finh) was 32% of the deposited dose - via both pulmonary and GI absorption -, while pulmonary bioavailability (Fpul) was only 8% of the deposited dose. The pulmonary absorption/excretion compartments are characterized by equilibration half-lives with plasma of 23 and 17 minutes in the peripheral and central lung, respectively. Pulmonary excretion and absorption appear to be similar from both lung compartments.
Sensitivity analysis indicated that, after INH, systemic/plasma AUC depends primarily on systemic disposition parameters (CLtot and k12). On the other hand, pulmonary/sputum AUC and cmax depends primarily on mucociliary clearance (kcm) and pulmonary absorption from the central lung (kca).
Conclusion: This semi-PBPK model for CIP with an optimized model parameter space allowed estimation of mechanistically plausible parameter values and was able to adequately predict cp(t) and cs(t) profiles from all 18 available PK studies in HV and CF. Pulmonary excretion of CIP appears to be less important than its renal and hepatic elimination. Fpul (even after correction for pulmonary deposition) is much lower than Foral, reflecting poor pulmonary absorption as a result of pulmonary excretion and efficient mucociliary clearance.
Systemic exposures were found to be sensitive to CLtot and k12, but insensitive to all systemic disposition, oral and pulmonary disposition parameters. However, with regards to pulmonary exposures, only mucociliary clearance and pulmonary absorption from the central lung appeared to be influential.
Based on the limited available data, there was no evidence of differences in pulmonary disposition of CIP between CF and HV.
Jurgen Venitz– Virginia Commonwealth University