(W0930-05-26) Physiologically Based Pharmacokinetic (PBPK) Modeling of Rifampicin and Its Application for Drug-Drug Interaction with Midazolam in Adults

Purpose: Rifampicin (RIF) is an essential part of the tuberculosis therapy and the pharmacokinetics (PK) of RIF has been of interest due to its non-linear and auto-induction behavior. RIF acts as a perpetrator, causing clinically relevant drug-drug interactions (DDIs) via induction and inhibition of multiple metabolic enzymes and transporters. The purpose was to develop a mechanistic PBPK model for RIF which accounts for all the relevant mechanisms after IV and PO administration in healthy and tuberculosis (TB) subjects . This model was validated by simulating the effect of RIF on midazolam (MDZ) PK.

Methods: The PBPKPlus^TM module in GastroPlus^TM v.9.8.2 was used to model the PK of RIF. The Advanced Compartmental Absorption and Transit (ACAT^TM) model was used to describe the intestinal absorption and dissolution of RIF after PO administration. Human physiologies were generated by the program’s internal Population Estimates for Age-Related (PEAR^TM) physiology^TM module. Tissue/plasma partition coefficients (Kps) for all the compounds were calculated using the Lukacova algorithm based on tissue composition and in vitro and in silico physiochemical properties. For the permeability limited liver, a specific permeability*surface area product (sPStc) was optimized to a value of 1 mL/s/mL tissue resulting in liver PStc = 842 mL/s for a 74 kg male subject. The biopharmaceutical parameters for RIF were obtained from literature or predicted by ADMET Predictor® v10.2. Steady-state volume of distribution (Vdss) was adjusted to match IV noncompartmental Vdss by changing the log P value to 1.5 to calculate the Kp values and then returned to 1.3 for simulations. In addition, the in vitro drug permeability from a parallel artificial membrane permeability assay (PAMPA) was converted to human jejunal P<sub>eff and then scaled up by 6.2-fold in order to fit the absorption phase [1]. The metabolism of RIF mediated by CYP3A4 was modelled using in vitro enzyme kinetic parameters and the GastroPlus built-in expression levels of 3A4 in gut and liver. The K_m of RIF for 3A4 was derived from in vitro studies and was kept constant across all studies (16.2 mM) [1] and the V_max was fitted. ADMET Predictor suggested the involvement of UGT1A3 for the formation of a reported water-soluble metabolite [2]. UGT enzyme kinetics in vitro were not found in the literature. To account for the dose related non-linearity of the observed C_P‑time profiles, a K_m value (ten times lower than the 3A4 K_m was assumed for UGT1A3), and the V_max was fitted. To account for the deacetylation of RIF linear systemic clearance of 20 L/h was added to liver. The hepatic uptake of RIF mediated by OATP1B1 was modeled using in vitro K_m and V_max obtained from literature and a relative activity factor (RAF) of 0.28 [3]. RIF is also a substrate for the hepatic and renal apical efflux transporter MRP2 [4]. In the absence of in vitro values, K_m was assumed to be equal to the MRP2 inhibitor constant (K_i) value and the V_max value was fitted. The DDI module in GastroPlus was used to predict the autoinduction and autoinhibition of RIF and the effect on midazolam PK for varying RIF and MDZ doses and administration times. Table 1 presents the induction and inhibition parameters of enzymes and transporters for RIF.

Results: Table 2 summarizes the study design of clinical studies used as part of model development and DDIs between RIF and MDZ [11-19]. MDZ C_P‑time profiles before and during co-administration with RIF are reasonably well predicted over the full range of administered doses and protocols. Correlation of predicted to observed DDI AUC and C_max ratios of interactions studies, illustrating the performance of the DDI network, is presented in Figure 1.

Conclusion: This work aimed to develop the RIF PBPK model and demonstrate the use of GastroPlus PBPK approach to predict the potential DDI interactions between RIF and MDZ. The overall results presented in Figure 1 show that the model accurately predicts the interaction effect for both IV and PO dosage forms. In conclusion, the GastroPlus PBPK approach, integrating all relevant physicochemical processes, perpetrator mechanisms, and physiological details is a highly reliable utility to estimate the potential of DDIs.

References: 1. Baneyx G. et al., Eur. J. Pharm. Sci. (2014) 56:1
2. Nakagawa K. et al., Tuberculosis. (1981) 56(12): 577
3. Tirona RG. Et al., J. Pharmacol. Exp. Therap., (2003) 304(1):223
4. Yoshikado T. et al., CPT Pharmacomet. Sys. Pharmacol. (2018) 7(11):739
5. Asaumi, CPTPharmacometSystPharmacol 2018;7:186
6. Moscovitz, JPET 2018;365(2):262
7. Kajosaari, Basic ClinPharmacolToxicol 2005;97:249
8. Takashima, JNuclMed 2012;53:741
9. Dong, Br J Pharmacol 2018;175;3486
10. Pedersen JM, Eur J Pharm Sci 2017;103(70)
11. Nitti, Chemotherapy 1977;23(1):1
12. Peloquin, Antimicrob Agents Chemother 1997;41(12)
13. Loos, KlinWochenschr 1985;63(23):1205
14. Kharasch, ClinPharmacolTher 2004;76(5):452
15. Backman, Clin Pharmacol Ther 1996;59(1)
16. Link B, Br J Clin Pharmacol 2008;66(4):473
17. Gorski, Clin Pharmacol Ther 2003;74(3):275
18. Kriby, DMD 2011;39(6):1070
19. Chung, Clin Pharmacol Ther 2006;79(4):350


Table 1: Induction and Inhibition Parameters of Enzymes and Transporters of Rifampicin


Table 2: Clinical Studies used for the Model Development and DDI verification for the PBPK Model


Figure 1: Predicted versus Observed Plasma Concentration Cmax and AUC