In vitro OATP1B1 and OATP1B3 inhibition studies

Transporter‐transfected HEK293 cells were used to measure the OATP1B1 and OATP1B3 inhibition (Table 1). A 30‐min pre‐incubation with test compound or control inhibitor at 37°C, 5% CO₂, and 90% relative humidity was done prior to the assay incubation. Rosuvastatin was used as the probe substrate, at a concentration (0.3 µM) below its OATP1B1 Km (3.8 µM) and OATP1B3 Km (28.3 µM). Detailed procedure is provided in Supplementary Methods.

Whole‐body PBPK modeling and simulations were performed using population‐based absorption, distribution, metabolism, and excretion (ADME) simulator, Simcyp (version 19.0; Certara, Sheffield, UK). Unless mentioned otherwise, the virtual populations of healthy subjects had a body weight of about 80 kg, with age ranging from 18 to 65 years, and included both sexes. Dosage regimen was matched to the clinical study design. Model for substrate compound, rosuvastatin, was built by adopting a similar approach described elsewhere.12, 13, 14, 15, 16, 17 Briefly, an initial model was developed using the physicochemical properties and the data from in vitro studies (Table 2). Advanced dissolution absorption and metabolism (ADAM) model was used to predict the oral absorption based on Caco‐2 permeability data. BCRP Km was obtained using membrane vesicle studies. Concentration‐dependent Caco‐2 studies did not yield adequate saturation kinetics likely due to competing OATP2B1 and BCRP kinetics (data not shown). Consequently, BCRP maximum flux (J_max) and OATP2B1 uptake clearance (CL_int) were optimized to recover ~ 40% Fa. The Roger and Rowland method (Method 2)16, 18 was used to capture the distribution of rosuvastatin assuming rapid equilibrium between blood and all organs, except the liver and kidneys.

Model input parameters for the full‐PBPK model of rosuvastatin, a probe substrate drug

Abbreviations: ADAM, advanced dissolution absorption and metabolism gut model; CL_int,T, intrinsic transport clearance; CL_PD, passive transport clearance; fu, fraction unbound; fu_IW/ fu_EW, fraction unbound in the intracellular and extracellular water; RAF, relative activity factor; SCHH, sandwich culture human hepatocytes.

Permeability‐limited hepatic disposition of rosuvastatin was considered, for which, sinusoidal active and passive uptake and canalicular active efflux obtained from our sandwich culture human hepatocyte (SCHH) transport studies were incorporated. Recent SCHH studies suggest intrinsic biliary and basolateral (MRP4) efflux are similar, so basolateral efflux was accounted with CL_int assuming to be similar to BCRP‐mediated biliary clearance obtained from SCHH studies.19 The geometric mean relative activity factor (RAF) for OATP1B (RAF_OATP) of 10.6, previously established utilizing SCHH (3 hepatocyte lots) from 10 different OATP1B substrates, was initially applied.20 RAF_OATP was further refined to a value of about 20 to better recover both the plasma concentration‐time profiles and urinary excretion data (~ 25% dose) from the intravenous study. Contribution from specific uptake transporters (OATP1B1, OATP1B3, OATP2B1, and NTCP) to the overall hepatic uptake were measured using “SLC‐phenotyping” methodology recently reported,21 which showed ~ 95% of active uptake is driven by OATP1B1, OATP1B3, and OATP2B1. Rosuvastatin metabolic clearance was assumed to be negligible based on its general enzymatic stability.

Transporter‐mediated renal secretion was built into the PBPK model. Uptake clearance via OAT3 was obtained from the rate studies using HEK293‐OATs cells, as previously described.22 RAF_OAT3 of ~ 4.1, which we previously established based on the in vitro–in vivo extrapolation of OAT3‐mediated transport using static model,22 was initially applied. This value was further refined (final RAF_OAT3 = 2.7) to recover the observed renal clearance. We assumed an apical efflux transporter on the proximal tubule cells mediate secretion from cells to urine compartment with similar transporter kinetics and RAF. Passive transport clearance was estimated from uptake rates in HEK293 wild type cells.

Model input parameters for 12 inhibitor drugs used in the PBPK evaluation, and their simulated plasma concentration‐time profiles are presented in Supplementary Material (Tables S2–S11, Figures S6–S11). Models were additionally verified for Fa, or F when possible. Inhibitors models for capmatinib, darolutamide, fostamatinib/metabolite‐R406, fenebrutinib, elbasvir, and grazoprevir were developed and verified as part of the current work. Cyclosporine PBPK model, previously developed and verified with several known OATP probe drugs, was applied without any modification.12, 13 Rifampicin, gemfibrozil/glucuronide parent‐metabolite pair, and itraconazole simulator default models are adopted with minor changes. Fenebrutinib and velpatasvir Simcyp models were previously described, and used here after minor modifications.23, 24 Inhibition potency against BCRP and OATP1B1/1B3 generated in the current study were used for all models. Rosuvastatin concentration used to measure BCRP IC₅₀ and OATP1B1/1B3 IC₅₀ is well below its corresponding Km. Therefore, IC₅₀ and Ki values are assumed same for BCRP. However, Ki was assumed to be IC₅₀/2 for OATP1B inhibition as a conservative measure. It is noteworthy that majority of inhibitor drugs (except rifampicin, cyclosporine, and gemfibrozil) did not inhibit OATPs at the clinically relevant concentrations.

Finally, PBPK model performance was evaluated by the ratio of predicted and observed area under the concentration‐time curve ratio (AUCR) values, and presented as R _pred/obs. An R _pred/obs value between 0.8 and 1.25 was assumed as acceptable model performance.25