The standard PK-Sim whole-body PBPK structural model for small molecules was utilized to build a combined bedaquiline and M2 model [21 (link)–23 (link)]. The standard PK-Sim whole-body PBPK model consists of key tissues and organs, including, the brain, heart, lungs, liver, kidneys, GI tract, etc., connected through vascular and arterial blood circulation. Each compartment is divided into four subcompartments, i.e., vascular, blood cells, interstitial, and intracellular [22 (link)]. Physicochemical parameters for bedaquiline and M2 were obtained from literature (Table 1 ) [24 (link)]. Different values have been reported in literature for bedaquiline lipophilicity and fraction unbound; therefore, model evaluation using each of the reported values was conducted to select the lipophilicity and fraction unbound values that provide the best fit to bedaquiline plasma PK data [24 (link), 25 (link)]. Bedaquiline oral absorption has previously been described as atypical with delay and double peaks [26 (link)–28 (link)]. The Weibull absorption model built within the PK-Sim software was selected due to its flexibility in describing atypical absorption profiles, and the parameters were estimated by fitting to the plasma PK data. Partition coefficients and cellular permeability parameters of bedaquiline and M2 in various tissues were calculated using the PK-Sim standard method [22 (link), 29 (link)]. In PK-Sim, the standard calculation method uses lipophilicity and plasma protein binding parameters along with lipid, protein, and water fractions in each compartment and subcompartment to calculate partition coefficients. CYP3A4 enzyme is involved in the metabolism of bedaquiline to M2 [25 (link)]. Therefore, CYP3A4-mediated metabolism conversion from bedaquiline to M2 was modeled using the Michaelis–Menten equation. Experimental data also suggest contributions of CYP2C8 and CYP2C19 enzymes in the metabolism of bedaquiline to M2 [31 (link)] and, thus, were evaluated in the model using the Michaelis–Menten equation. Expression profiles for all three enzymes based on the RNA-sequencing (RNA-seq) method were obtained from the Bgee (https://www.bgee.org/ ) database accessible within PK-Sim [30 (link)]. The parameter Michaelis–Menten constant (Km) for the enzymatic reactions was obtained from literature from in vitro experiments [31 (link)]. Residual bedaquiline liver plasma clearance was obtained from literature [24 (link)]. Next, the model was simultaneously fitted to bedaquiline and M2 PK data following 400–300–200 QD dosing in patients with pulmonary TB to estimate Weibull absorption parameters, enzymatic reaction rates (Vmax), and M2 liver plasma clearance. The combined bedaquiline–M2 plasma PK model was validated by comparing the simulations versus observed plasma PK data for bedaquiline following 200–100 mg QD, 500–400–300 mg QD, and 700–500–400 mg QD dosing regimens (clinical trial: NCT01215110). M2 PK data for this study were not available.
Parameters for the bedaquiline-M2 PBPK model with CNS distribution
| Parameter | Unit | Value | Source |
|---|---|---|---|
| Bedaquilinea | |||
| Molecular weightb | g/mol | 555.5 | PubChem Database |
| Lipophilicity | log unit | 5.14 | [24 (link)] |
| Fraction unbound in plasma | Dimensionless | 0.0003 | [24 (link)] |
| pKa (base) | Dimensionless | 9.10 | [24 (link)] |
| Weibull dissolution time (50% dissolved) | Min | 125.21 | Estimated |
| Weibull dissolution shape | Dimensionless | 1.51 | Estimated |
| Vmax CYP3A4 | umol/L/min | 407.85 | Estimated |
| Km CYP3A4 | umol/L | 8.5 | [31 (link)] |
| Vmax CYP2C8 | umol/L/min | 163.73 | Estimated |
| Km CYP2C8 | umol/L | 13.1 | [31 (link)] |
| Additional hepatic clearance | L/h/kg | 0.03 | [24 (link)] |
| Permeability across BBB and BCSFB (assumed half of the calculated permeability from plasma-to-interstitial due to lipid bilayer in BBB and BCSFB) | dm/min | 0.00217 | PK-Sim Calculated |
| Cellular permeability from plasma to interstitial | dm/min | 0.013 | |
| Brain interstitial water partition coefficient | Dimensionless | 0.0013 | |
| Brain intracellular water partition coefficient | Dimensionless | 6.2 × 10−5 | |
| Plasma-to-CSF partition coefficient | Dimensionless | 0.0082 | Calculated (Eq. 1) |
| Molecular weightc | g/mol | 541 | PubChem Database |
| Lipophilicity | Log unit | 6.5 | [24 (link)] |
| Fraction unbound in plasma | Dimensionless | 0.0005 | [24 (link)] |
| Hepatic clearance | L/h/kg | 0.14 | Estimated |
| Permeability across BBB and BCSFB (assumed half of the calculated permeability from plasma-to-interstitial due to lipid bilayer in BBB and BCSFB) | dm/min | 0.185 | PK-Sim Calculated |
| Cellular permeability from plasma to interstitial | dm/min | 0.36 | |
| Brain interstitial water partition coefficient | Dimensionless | 0.0013 | |
| Brain intracellular water partition coefficient | Dimensionless | 2.8 × 10−6 | |
| Plasma-to-CSF partition coefficient | Dimensionless | 0.0084 | Eq. 1 |
aWater solubility was assumed 0.01 mg/mL because both bedaquiline and M2 are poorly soluble in water
bBedaquiline number of halogens Cl is 2, thus, effective molecular weight is 511.5 g/mol
cM2 number of halogens Cl is 2, thus, effective molecular weight is 497 g/mol
