Abstract
It is generally presumed that uptake transport mechanisms are of limited significance in hepatic clearance for lipophilic or high passive-permeability drugs. In this study, we evaluated the mechanistic role of the hepato-selective organic anion–transporting polypeptides (OATPs) 1B1/1B3 in the pharmacokinetics of compounds representing large lipophilic acid space. Intravenous pharmacokinetics of 16 compounds with molecular mass ∼400–730 Da, logP ∼3.5–8, and acid pKa <6 were obtained in cynomolgus monkey after dosing without and with a single-dose rifampicin–OATP1B1/1B3 probe inhibitor. Rifampicin (30 mg/kg oral) significantly (P < 0.05) reduced monkey clearance and/or steady-state volume of distribution (VDss) for 15 of 16 acids evaluated. Additionally, clearance of danoprevir was reduced by about 35%, although statistical significance was not reached. A significant linear relationship was noted between the clearance ratio (i.e., ratio of control to treatment groups) and VDss ratio, suggesting hepatic uptake contributes to the systemic clearance and distribution simultaneously. In vitro transport studies using primary monkey and human hepatocytes showed uptake inhibition by rifampicin (100 µM) for compounds with logP ≤6.5 but not for the very lipophilic acids (logP > 6.5), which generally showed high nonspecific binding in hepatocyte incubations. In vitro uptake clearance and fraction transported by OATP1B1/1B3 (ft,OATP1B) were found to be similar in monkey and human hepatocytes. Finally, for compounds with logP ≤6.5, good agreement was noted between in vitro ft,OATP1B and clearance ratio (as well as VDss ratio) in cynomolgus monkey. In conclusion, this study provides mechanistic evidence for the pivotal role of OATP1B-mediated hepatic uptake in the pharmacokinetics across a wide, large lipophilic acid space.
SIGNIFICANCE STATEMENT This study provides mechanistic insight into the pharmacokinetics of a broad range of large lipophilic acids. Organic anion–transporting polypeptides 1B1/1B3-mediated hepatic uptake is of key importance in the pharmacokinetics and drug-drug interactions of almost all drugs and new molecular entities in this space. Diligent in vitro and in vivo transport characterization is needed to avoid the false negatives often noted because of general limitations in the in vitro assays while handling compounds with such physicochemical attributes.
Introduction
Characterizing hepatic clearance mechanisms is of prime importance in drug discovery and development to predict systemic/target exposure and assess pharmacokinetic variability due to intrinsic and extrinsic factors, such as drug-drug interactions (DDIs), pharmacogenomics, disease state, etc. (Di et al., 2013). Preclinical and clinical data support the significance of organic anion–transporting polypeptide (OATP) 1B1 and OATP1B3 in the hepatic uptake of several high–molecular mass (MM) acidic/zwitterionic drugs (e.g., statins, sartans, and certain glinides). In addition to active uptake, hepatic metabolism and/or biliary efflux may also contribute to the human hepatic clearance of acidic/zwitterionic compounds (Alluri et al., 2020; Steyn and Varma, 2020). According to the extended clearance classification system (ECCS) framework, hepatic clearance of high MM (>400 Da) acids/zwitterions involves OATP1B1/1B3-mediated uptake irrespective of their passive membrane permeability (Varma et al., 2015; El-Kattan and Varma, 2018). The role of OATPs in the hepatic clearance of large hydrophilic acids—that is, with limited passive permeability (e.g., rosuvastatin, pravastatin, valsartan, etc.) —is well established. However, the relevance of uptake transporters, including OATP1B1/1B3, in determining the pharmacokinetics (PK) for highly permeable or lipophilic compounds has been consistently underappreciated. Indeed, recent US Food and Drug Administration DDI guidance recommends evaluation for OATP1B1/1B3 substrate activity in vitro and emphasizes the need for such characterization for investigational drugs with low passive membrane permeability (https://wwwfdagov/regulatory-information/search-fda-guidance-documents/vitro-drug-interaction-studies-cytochrome-p450-enzyme-and-transporter-mediated-drug-interactions). It is generally presumed that lipophilic compounds and/or compounds with high passive permeability can readily translocate across the hepatocyte basolateral membrane, and thus active transport contribution to hepatic clearance is limited.
According to ECCS, OATP1B1/1B3 likely play a key role in the hepatic clearance of not only large hydrophilic acids but also for most large lipophilic acids (Varma et al., 2015 2017a). However, there are limited clinical data to substantiate this thesis. We previously evaluated the clinical DDIs of victim drugs with the “first-choice” clinical probe inhibitor (rifampicin) recommended to investigate OATP1B1/1B3-mediated hepatic uptake activity, and clearly rifampicin caused moderate [area under the plasma concentration-time curve (AUC) ratio 2–5] and high (AUC ratio >5) interactions for class 1B and 3B drugs, whereas only no/low (AUC ratio <2) interactions for other ECCS classes [i.e., class 1A/3A—low MM (<400 Da) acids/zwitterions, and class 2/4—bases and neutral] (Varma et al., 2017a). With a limited clinical data set of only four drugs in this category (class 1B—atorvastatin, pitavastatin, glyburide, and grazoprevir), it is challenging to extend these findings to diverse chemicals in this space.
Here, we evaluated the role of OATP1B1/1B3-mediated hepatic uptake in the PK of large lipophilic acids using cynomolgus monkey [nonhuman primate (NHP)] as a preclinical model. Intravenous PK of a set of diverse acid drugs (n = 16) was studied in NHP in the presence and absence of a single dose rifampicin, an OATP1B1/1B3 inhibitor. In vitro uptake transport was characterized for these compounds using cultured monkey and human hepatocytes. Relationships between in vitro fraction transported by OATP1B (ft,OATP1B) and the rifampicin-induced change in the intravenous clearance and steady-state volume of distribution (VDss) were assessed. Criteria for selection of compounds included molecular mass (>400 Da), lipophilicity (log P > 3.5), and acidic pKa <7.4 (i.e., anionic at physiologic pH) (Table 1). All compounds are categorized as ECCS class 1B based on their high permeability and/or extent of metabolism in human (Table 1). Compounds represent approved drugs (n = 6) and new molecular entities (NMEs) reaching clinical development (n = 10) and comprise anionic groups, including carboxylic acids, acyl sulfonamides, N-hetaryl sulfonamides, and tetrazoles. Chemical structures of the 16 compounds are shown in Supplemental Fig. 1. These anionic groups are commonly seen in the drugs and NMEs in therapeutic areas, such as cardiovascular, diabetes, immunology/inflammation, hepatitis C, nonalcoholic fatty liver disease, etc.
Summary of physicochemical properties, in vitro data, and observed hepatic clearance in cynomolgus monkey and human of the 16 compounds
Materials and Methods
Materials.
The compounds presented in this manuscript have all been reported previously. Compounds were sourced commercially when possible or prepared according to literature routes. Compounds synthesized were solubilized in DMSO, analyzed by UPLC for purity, and used for in vitro and in vivo studies when the purity was >95% (Tess et al., 2020). One of the Pfizer compounds, PF-05089771, is commercially available from MilliporeSigma (Burlington, MA). InVitroGro-HT, CP, and HI hepatocyte media were purchased from Celsis IVT (Baltmore, MD). Collagen I–coated 24-well plates were obtained from BD Biosciences (Franklin Lakes, NJ). Cryopreserved monkey hepatocytes (10106012, female) were purchased from In Vitro ADMET Laboratories LLC (Columbia, MD), and cryopreserved human hepatocytes Hu8246 (female, Caucasian, 37 years old) were purchased from Thermo Fisher Scientific (Waltham, MA). BCA protein assay kit was purchased from PIERCE (Rockford, IL). NP-40 protein lysis buffer was purchased from Thermo Fisher (Franklin, MA). Human embryonic kidney (HEK) 293 cells stably transfected with OATP1B1 were generated at Pfizer Inc. (Groton, CT). Dulbecco’s modified Eagle’s medium (high glucose), fetal bovine serum, nonessential amino acids, GlutaMAX, sodium pyruvate, and gentamicin were obtained from Thermo Fisher Scientific (Waltham, MA). BioCoat 96-well poly-D-lysine–treated plates were obtained from Corning Inc. (Corning, NY). Hanks’ balanced salt solution (HBSS) and HEPES were obtained from Lonza (Basel, Switzerland).
In Vivo PK Studies in Cynomolgus Monkey (NHP).
All procedures performed on NHPs were conducted in accordance with regulations and established guidelines and were reviewed and approved by an Institutional Animal Care and Use Committee through an ethical review process. NHP studies were conducted at Pfizer. Male cynomolgus monkeys were purchased from Covance (Princeton, NJ), Charles River Laboratories, Inc. (Wilmington, MA) or Envigo Global Services (Indianapolis, IN); subjects 3–8 years of age were used in PK studies. NHP studies used a crossover design, wherein subjects first received test compounds intravenously (i.e., control), and this was followed by a minimum 7-day washout period; then the same subjects received rifampicin dosed with test compounds (i.e., treatment). This study procedure is similar to that reported previously (Varma et al., 2017b; Kosa et al., 2018; Ufuk et al., 2018). For each study (n = 3 or 4), a cassette of two or three compounds was dosed intravenously, typically <1 mg/kg to minimize interaction between the compounds based on preliminary in vitro uptake studies using monkey hepatocytes. Each cassette was administered via intravenous infusion for 30 minutes or a bolus injection via the cephalic vein. For the treatment portion of the study, rifampicin was administered by oral gavage at a dose of 30 mg/kg (4 ml/kg) as a suspension in 0.5% aqueous methylcellulose. One hour after the rifampicin dose, test compounds were dosed intravenously. Subjects were monitored for pain or distress throughout the study, which was followed by at least daily monitoring while off study. Intravenous dosing vehicle was optimized for each cassette such that the compounds were in solution and stable for at least 24 hours. Serial blood samples were collected via the femoral vein before dosing and at predefined time points postdose. Blood samples were collected into K2EDTA-treated collection tubes and were stored on wet ice prior to being centrifuged to obtain plasma. Each plasma sample was added to an equal volume of 0.1 M sodium acetate buffer (pH 4) and kept cold during collection, after which the plasma samples were stored frozen at −70 or −20°C. Urine samples were collected on wet ice and were added to an equal volume of 0.1 M sodium acetate buffer (pH 4.0) prior to being stored frozen at −70 or −20°C.
Liquid Chromatography Tandem Mass Spectrometry Analysis of Plasma and Urine Samples and Pharmacokinetics Analysis.
Quantitation of drug substance in plasma and urine samples was done using liquid chromatography tandem mass spectrometry (LC/MS). Standard and quality-control samples prepared in blank plasma or urine with an equal volume of 0.1 M sodium acetate buffer (pH 4) were extracted in the same manner as the in-life samples. Briefly, a Waters ACQUITY ultra performance liquid chromatography system (Waters, Milford, MA) coupled to an API 4000, 5500 or 6500 triple quadrupole quadrupole linear ion trap hybrid mass spectrometer equipped with either an atmospheric pressure chemical ionization or electrospray ionization source (depending upon analyte) (Applied Biosystems, MDS Sciex, Foster City, CA) was used. Chromatographic separation was accomplished using either a Waters ACQUITY UPLC BEH C18 column (1.7 µm, 2.1 × 50 mm), a Waters XSELECT CSH XP C18 column (2.5 µm, 2.1 × 50 mM), or a Waters XSelect HSS T3 column (2.5 µm, 2.0 × 30 mm) maintained at either room temperature or 45°C. The mobile phase (two solvents gradient) was optimized to achieve good separation between the analytes. Typically, solvent A was constituted of 0.025% formic acid and 1 mM NH4OAC in water/acetonitrile (95:5 v/v), and solvent B included 0.025% formic acid and 1 mM NH4OAC in water/acetonitrile (5:95 v/v). The gradient generally began at 3%–30% B until about 1.2 minutes, and this was followed by an increase to 50%–65% B to 1.6 minutes and then a decrease to 10%–30% B until ∼1.7–1.9 minutes. Analyst 1.4.2 or 1.6.2 software (SCIEX, Framingham, MA) was used for peak integration and standard curve regression. Data were imported into Watson LIMS version 7.5 (Thermo Fisher Scientific Inc) for standard curve regression and noncompartmental pharmacokinetics parameter calculations—AUC, intravenous clearance (CLtotal,NHP), volume of distribution (VDSS) and effective half-life (t1/2). The fraction of unchanged parent excreted into urine (fe) was calculated as follows: urine concentration × total urine volume/dose. NHP hepatic clearance (CLh,NHP) was estimated for the subsequent analyses:
(1)where Dose IV is the intravenous dose. VDSS and t1/2 were calculated using the following expressions:
(2)
(3)wherein AUMC is the total area under the first moment of the drug concentration-time curve from time 0 to infinity.
Changes in clearance and VDss on rifampicin treatment were presented as clearance (CL) ratio and VDss ratio. Note that the fe was less than 0.02 for all compounds in this data set (unpublished data), therefore, CL ratio can be considered as a representation of change in total clearance or hepatic clearance:
(4)The in vivo ft,OATP1B was calculated from intravenous clearance in the absence and presence of rifampicin:
(5)Here, subscripts “control” and “+Rifampicin” represent PK parameters in control and treatment groups, respectively. Our previous studies showed complete inhibition of OATP1B-mediated hepatic clearance of probe drugs (pitavastatin and rosuvastatin) in monkey after rifampicin 30-mg/kg oral dose (Ufuk et al., 2018).
In Vitro Uptake Studies Using Plateable Monkey and Human Hepatocytes.
The cell culture and uptake study conditions with monkey and human hepatocytes were identical. The hepatic uptake assay was performed using short-term culture format as described previously with some modifications (Bi et al., 2017, 2019). Briefly, cryopreserved monkey hepatocytes (10106012, IVAL) and human hepatocytes (Hu8246) were thawed in the InVitro-HT media at 37°C, centrifuged at 50g for 3 minutes, and resuspended in In VitroGro-CP medium. The cells were seeded into 24-well collagen I–coated plates with 0.35 × 106 cells/well in a volume of 0.5 ml/well. The cells were cultured in the InVitro-CP media overnight (∼18 hours). Plated cells were preincubated for 10 minutes at 37°C with HBSS with or without 100 µM rifampicin. The preincubation buffer was aspirated, and the uptake and inhibition reaction was initiated by addition of prewarmed buffer containing substrates (0.5 µM) with or without rifampicin (100 µM). The reactions were terminated at designated time points (0.5, 1, 2, and 5 minutes) by adding ice-cold HBSS immediately after removal of the incubation buffer. The cells were washed three times with ice-cold HBSS and lysed with 100% methanol containing internal standard, and the samples were analyzed by LC/MS (Supplemental Methods). Uptake clearance (PSinf) was estimated from the initial time course (0.5–2 minutes) by linear regression (Bi et al., 2017, 2019). The in vitro ft,OATP1B was calculated from PSinf measured in the absence and presence of 100 µM rifampicin. At this concentration, rifampicin completely inhibits OATP1B1 and OATP1B3 in vitro (Bi et al., 2017, 2019):
(6)
Measurement of Fraction Unbound in Human Hepatocytes Using Equilibrium Dialysis.
The 96-well equilibrium dialysis (HTD 96) device and cellulose membranes with molecular mass cutoff of 12–14 K were from HTDialysis, LLC (Gales Ferry, CT). Velocity V11 peelable seals were obtained from BD Falcon (Bedford, MA). Deep 96-well plates of 1.2-ml blocks were from Axygen Scientific Inc (Union City, CA). Orbital shaker and vortex mixer were from VWR (Radnor, PA). The details of the experiments measuring fraction unbound to hepatocytes (fu,heps) have been described previously (Riccardi et al., 2018).
Cryopreserved human hepatocytes were thawed and added into HBSS to make a final concentration of 0.7 × 106 cells/ml concentration matrix (similar cell density as in hepatocyte uptake studies). The dialysis membranes were prepared a day before the experiment. The cellulose membranes were immersed in deionized water for 15 minutes followed by 15 minutes in 30% ethanol/deionized water, and then they were left overnight in DPBS. A 150-μl aliquot of hepatocyte matrix spiked with 1 μM compound was added to one side of the chamber (donor), and 150 μl of DPBS was added to the other side of the dialysis membrane (receiver). The equilibrium dialysis device was sealed with a breathable easy gas permeable membrane and was placed on an orbital shaker at 200 rpm at 37°C in quadruplicate for 6 hours. After incubation, aliquots of 45 μl of dialyzed DPBS and 15 μl hepatocyte matrix were matrix-matched and quenched with cold acetonitrile containing internal standard (carbamazepine or tolbutamide) to precipitate the proteins. The plates were sealed and mixed with a vortex mixer for 1 minute and then centrifuged at 2000 × g at room temperature for 5 minutes. The supernatant was transferred to a new deep well block, sealed, and subsequently analyzed using LC/MS (Supplemental Methods). fu,heps was than calculated as ratio of receiver to donor area signal (Riccardi et al., 2018).
In Vitro–In Vivo Extrapolation of Hepatic Clearance.
Assuming hepatic uptake as the rate-determining step in the overall hepatic clearance, IVIVE was evaluated using the in vitro apparent total PSinf obtained after monkey or human hepatocyte incubations. IVIVE was assessed without and with correction for fu,heps. The fu,heps values were obtained using human hepatocytes and were assumed to be same for monkey hepatocytes. Standard physiologic scalars were used to scale up in vitro data: 1 × 106 hepatocytes/mg-measured protein, 120 × 106 hepatocytes/g-liver, 21 g-liver/kg-body weight (Davies and Morris, 1993; Hosea et al., 2009).
In vivo (observed) hepatic intrinsic clearance (CLint,h) was calculated using the well stirred clearance model (eq. 7):
(7)wherein CLh is hepatic clearance (total intravenous clearance minus renal clearance), Rbp is blood-to-plasma ratio, fu,p is fraction unbound in plasma, and Qh is hepatic blood flow (human, 20; NHP 44 ml/min per kilogram).
Results
Effect of Rifampicin on Intravenous PK of Large Lipophilic Acids in Cynomolgus Monkey.
A single dose rifampicin (30 mg/kg orally administered) significantly altered plasma exposure of almost all the large lipophilic acids dosed intravenously in cynomolgus monkey (Fig. 1). Indeed, 14 of 16 compounds evaluated showed reduction in the intravenous clearance by more than 2-fold (Table 2). CL ratio (i.e., ratio of hepatic clearance in control to rifampicin treatment groups, eq. 4) ranged from unity (epristeride) to about 27.7 ± 9.0 (MK-3577), with a; geometric mean of 4.0 across the 16 compounds. VDss also decreased considerably after rifampicin dosing compared with control, and a significant linear relationship was observed between CL ratio and VDss ratio (i.e., ratio of VDss in control to rifampicin treatment groups) (Fig. 2A). Interestingly, epristeride, which showed no change in clearance, had a marked increase in VDss (∼4×). On the other hand, glyburide showed a small change in VDss (∼1.5×) in relation to the large change in clearance (∼13×). A significant linear relationship was also observed for the VDsscontrol (volume in control group) and VDss ratio, implying that the change in VDss is higher for acids with high VDsscontrol (Fig. 2D). In the control group, a trend for an increase in VDss with an increase in clearance was apparent, although a linear relationship was not statistically significant (P > 0.05) (Fig. 2B). Also, a clear correlation between CLcontrol and CL ratio was not seen (Fig. 2C). On the other hand, effective t1/2 increased (by >2-fold) when dosed with rifampicin for only five compounds (montelukast, gliquidone, verlukast, glyburide, and MK-3577) (Table 2). Moreover, based on the plasma concentration-time profiles (Fig. 1), it is apparent that the terminal t1/2 of the lipophilic acids remained similar or decreased after rifampicin treatment.
Intravenous pharmacokinetics of 16 large lipophilic acids in cynomolgus monkey when dosed without and with single-dose oral rifampicin. Data represent plasma concentration-time profiles in individual animals in the control (dashed curves) and rifampicin treatment (solid curves) groups (n = 3 to 4). Same subjects were dosed in the control and treatment groups with at least 7-day washout period.
Pharmacokinetic parameters of large lipophilic acids in cynomolgus monkey when dosed in the absence and presence of single-dose rifampicin (30 mg/kg)
Inter-relationship between cynomolgus monkey pharmacokinetic parameters. (A) Correlation between change in clearance and VDss of large lipophilic acids on single-dose rifampicin treatment in cynomolgus monkey. (B) Correlation between clearance and VDss in the control group. (C) Correlation between clearance in the control group and change in CL ratio. (D) Correlation between VDss in the control group and change in VDss on rifampicin treatment (VDss ratio). Data points depict mean ± S.D. (n = 3 to 4). When presented, curves represent linear regression fit to the observed data and 95% confidence interval. R2, linear regression coefficient. Statistical significance of the linear relationship was tested at P value 0.05. Data points are labeled with compound code (Table 1).
Comparison of In Vitro Uptake Transport in Monkey and Human Hepatocytes.
Using short-time cultured (6 hours) monkey and human hepatocytes, uptake transport of the 16 compounds was measured in the absence and presence of 100 µM rifampicin (Fig. 3; Table 3). Firstly, total PSinf is comparable between human and monkey hepatocytes, with most of the compounds falling within 3-fold error. LY2393910 showed low PSinf (∼6.6 ml/min per kilogram), whereas pitavastatin and irbesartan showed higher uptake clearance (∼300 ml/min per kilogram) in monkey hepatocytes. Rifampicin did not inhibit uptake of a few compounds (gliquidone, GSK269984A, LY2409021, MK-0893, and montelukast) in both human and monkey hepatocytes. Interestingly, these compounds are relatively more lipophilic (logP >6.5). LY2393910, also with logP >6.5, showed about 25% reduced uptake in the presence of rifampicin. However, all others (logP <6.5) with the exception of epristeride showed significant active uptake in both human and monkey hepatocytes. The in vitro ft,OATP1B, which was estimated assuming 100 µM rifampicin inhibits OATP1B1/1B3 completely, was comparable between human and monkey hepatocytes (Fig. 3B).
Comparison of in vitro uptake in human and monkey cultured hepatocytes. (A) Comparison of total uptake clearance in human vs. monkey hepatocytes. (B) Comparison of percent active uptake in human vs. monkey hepatocytes. Data points represent mean values for compounds with log P ≤6.5 (closed points) and log P >6.5 (open points). Variability in uptake parameters for each compound is presented in Table 3. Diagonal lines represent unity and ±3-fold error or ±20% error in case of percent active uptake. Data points are labeled with compound code (see Table 1).
In vitro transport activity of 16 large lipophilic acids in the OATP1B1-transfected HEK293 cells and monkey and human primary hepatocytes
In vitro substrate studies using HEK293 cells demonstrated OATP1B1-mediated transport for all compounds with logP <6.5 except epristeride (Table 3). However, consistent with the hepatocyte uptake data, acids with logP >6.5 could not be recognized as OATP1B1 substrates under the current experimental conditions. We additionally studied OATP1B1 inhibition in transfected HEK293 cells, and notably, all 16 compounds inhibited OATP1B1-mediated rosuvastatin transport at 10 µM concentration, suggesting affinity to the transporter protein (Table 3).
In Vitro–In Vivo Extrapolation of Uptake Clearance to Predict Monkey and Human Hepatic Clearance.
In vitro apparent PSinf measured in monkey and human hepatocytes was used to predict in vivo hepatic intrinsic clearance in NHP (CLh,int,NHP) and human (CLh,int,Human), respectively, assuming uptake as the rate-determining process to the overall clearance (Fig. 4). Direct extrapolation of monkey hepatocyte data (i.e., PSinf) yielded considerable underprediction; particularly, in vitro PSinf was several-fold lower than the observed CLh,int,NHP for all acids with logP >6.5 (Fig. 4A). Corresponding predictions of human hepatic clearance from in vitro PSinf showed similar underprediction, although the bias was relatively small.
In vitro–in vivo extrapolation to predict monkey and human hepatic intrinsic clearance based on hepatocyte uptake data. (A) Observed vs. predicted hepatic intrinsic clearance in NHP. (B) Observed vs. predicted hepatic intrinsic clearance in human. Predictions are based on direct scaling of uptake clearance measured in cultured monkey (A) or human (B) hepatocytes in the control condition (open points) and correcting for fraction unbound in the hepatocyte incubations (closed points). (C) Observed vs. predicted human hepatic intrinsic clearance, wherein predictions are based on scaling of uptake clearance measured in cultured human hepatocytes (Human PSinf/fu,heps) corrected for compound-specific scaling factor (ratio of observed to predicted monkey clearance) obtained from IVIVE of monkey uptake clearance. (D) Observed human hepatic intrinsic clearance vs. observed monkey hepatic intrinsic clearance. Diagonal lines represent unity and ±3-fold error. Data points are labeled with compound code (see Table 1).
Given the lipophilic and sticky nature of the compounds in this space, we measured fu,heps via equilibrium dialysis methodology using a human hepatocyte homogenate prepared with similar cell density as employed in the plateable human hepatocyte/plateable monkey hepatocyte assay. Four of the six compounds with logP >6.5 showed fu,heps <0.004, suggesting high nonspecific binding to hepatocytes in the in vitro experimental conditions (Table 3). Other compounds showed fu,heps generally in the range of 0.20–0.70. Previous studies from our laboratory suggested that the measured fraction unbound values in human and monkey liver tissues are largely similar (Ryu et al., 2020). Therefore, the fu,heps values obtained using human hepatocytes were assumed to be same for monkey hepatocytes. Correction for the nonspecific binding in the incubations (i.e., use of PSinf/fu,heps) generally improved the IVIVE (Fig. 4, A and B). Nevertheless, systemic underprediction of hepatic intrinsic clearance was still apparent in the case of NHP scaling, in which predicted clearance of 10 of 16 compounds were <3-fold of observed values. However, in case of human clearance, about 57% of the predictions (8 of 14) were within 3-fold error when corrected for fu,heps. Underprediction in human clearance after correcting for hepatocyte binding was noted only for dexloxiglumide, glyburide, and PF-05241328, whereas three compounds showed overprediction (>3-fold).
A cross-species empirical scaling approach, as suggested previously (De Bruyn et al., 2018), was assessed to predict human clearance. In this case, the compound-specific scaling factors in NHP (SF, ratio of CLh,int,NHP to monkey PSinf/fu,heps) were used to scale human PSinf/fu,heps to predict CLh,int,Human. Cross-species scaling generally overpredicted clearance (Fig. 4C). Given that the in vitro PSinf were similar in human and monkey hepatocytes (Fig. 3A), the predicted CLh,int,Human values obtained by cross-species scaling were very consistent with the observed CLh,int,NHP (Fig. 4D).
Comparison of Fraction Transported by OATP1B In Vitro and In Vivo.
The in vitro fraction of compound transported by OATP1B1/1B3 (ft,OATP1B) in monkey and human hepatocytes was estimated by measuring uptake in the absence and presence of rifampicin (100 µM). Our previous studies have shown that rifampicin at this concentration abates OATP1B1/1B3 activity (Bi et al., 2017, 2019). Cellular accumulation of the six compounds with logP >6.5 was not affected by rifampicin in vitro, although all of these compounds had CL ratio between 2 and 6 in vivo (Fig. 5A). Note that these compounds also showed high nonspecific binding to the hepatocytes (fu,heps <0.004). For acids with logP <6.5, a good correlation was observed between the ft,OATP1B obtained using monkey hepatocytes and the in vivo CL ratio with two exceptions (glyburide and verlukast). The correlation can generally be described by the expression, CL ratio = 1/(1–ft,OATP1B), which relies on the uptake-determined clearance of the substrate drugs and the complete inhibition of OATP1B activity by rifampicin administered in vivo. Indeed, the rifampicin unbound plasma concentrations in cynomolgus monkey after 30-mg/kg oral dose well exceeded the concentration above 90% inhibition of OATP1B1 over the 12 hours postdose (Supplemental Fig. 2). Interestingly, there was an exceptional correlation between ft,OATP1B and NHP VDss ratio (Fig. 5B). Finally, a comparison of ft,OATP1B measured using human hepatocytes and CL ratio or VDss ratio in NHP also showed good concordance, although there was a relatively higher scatter (Supplemental Fig. 3).
Correlation between fraction transported by OATP1B in plated monkey hepatocytes and change in clearance (A) and change in VDss (B) in NHP after rifampicin treatment. Curves depict theoretical correlation (ratio = 1/(1−ft,OATP1B)), assuming in vitro ft,OATP1B is equal to in vivo ft,OATP1B, and rifampicin treatment completely abated OATP1B-mediated uptake in vivo. (C) In vitro vs. in vivo ft,OATP1B. Solid and dashed lines represent unity ±30% error. Data points represent compounds with log P ≤6.5 (closed points) and log P >6.5 (open points). Data points show mean, and error bars when presented depict S.D. Data points are labeled with compound code (see Table 1).
Discussion
Toward the primary objective, this study demonstrates the pivotal role of OATP1B1/1B3-mediated hepatic uptake in determining the PK of large lipophilic carboxylic acids and the related bioisosteres, and the findings can be generalized across a broad physicochemistry (MM ∼400–730 Da, logP ∼3.5–8, and acid pKa < 6.0; typically ECCS class 1B compounds). We evaluated 16 compounds (six marketed drugs and 10 NMEs that reached clinical development) in vivo in the cynomolgus monkey and in vitro using primary monkey and human hepatocytes. Single-dose rifampicin (prototypical OATP1B1/1B3 inhibitor) significantly altered the plasma concentrations of 15 of the 16 large lipophilic acids dosed intravenously in cynomolgus monkey. Additionally, there was a good correlation between the magnitude of change in clearance (CL ratio) versus the change in VDss (VDss ratio). These findings clearly imply that the rifampicin-induced change in PK of large lipophilic acids is associated with the inhibition of OATP1B1/1B3-mediated hepatic uptake. In the majority of cases, the in vivo findings are well substantiated by the in vitro data, wherein a good agreement was observed between in vitro and in vivo ft,OATP1B in NHP. These findings are of clinical relevance in this broader physicochemical space, especially since many of these compounds were not tested for the relevance of OATP1B1/1B3 in clinical studies. To our knowledge, clinical DDI data with OATP1B probe inhibitors (e.g., rifampicin and cyclosporine) are available only for danoprevir (Brennan et al., 2015), glyburide (Zheng et al., 2009), and pitavastatin (Prueksaritanont et al., 2014), wherein the human PK of these substrate drugs was modulated by an inhibitor. For the other compounds, preclinical data presented in this study may provide basis for further clinical investigation and/or improved mechanistic rationalization of their human PK variability because of intrinsic and extrinsic factors (e.g., age, disease state, DDIs, etc.).
VDss is generally thought to be determined by the physicochemical properties (e.g., logP and ionization) of compounds (Rodgers and Rowland, 2007; Smith et al., 2015). Carboxylic acids and related bioisosteres (acid pKa <6.0) are anionic at physiologic pH and have limited lipid partitioning that restricts the compound to circulating blood, and they are therefore expected to have low VDss (<0.2 l/kg). NHP VDss of the compounds evaluated in this study ranged from 0.13 l/kg (glyburide) to about 5 l/kg (irbesartan, MK-0893, MK-3577), which was clearly reduced on rifampicin treatment. Notably, compounds with high VDss showed relatively higher VDss ratio (Fig. 2D). A linear correlation between the CL ratio and VDss ratio further signifies the role of hepatic uptake in the PK of these compounds. In contrast, inhibition of hepatic metabolism or biliary secretion alone typically does not alter VDss (Grover and Benet, 2009). Overall, the data suggest that large lipophilic acids may show larger-than-expected VDss primarily driven by their transport rates across the basolateral membrane of hepatocytes and that the liver acts as a major distribution compartment for such compounds. Accounting for such transporter-mediated distribution should further help predict plasma concentration-time profiles and dose optimization in drug discovery.
We employed cynomolgus monkey, which has been shown as a useful animal model for the assessment of OATP1B-mediated disposition and drug-drug interactions due to the high degree of gene and amino acid sequence identity between NHP and human orthologs (>93% for OATP1B uptake transporters) (Ebeling et al., 2011; Shen et al., 2013). Recent studies have also drawn several parallels at the level of functional activity, including the following: 1) The mechanisms of transporter-mediated drug-drug interactions in humans and NHPs appear to be comparable (Shen et al., 2013; Takahashi et al., 2013; Shen et al., 2015; De Bruyn et al., 2018; Kosa et al., 2018; Ufuk et al., 2018; Liao et al., 2019), 2) in vivo studies showed similarity in biliary excretion (Kimoto et al., 2017) and handling of some endogenous OATP1B biomarkers (Chu et al., 2015; Shen et al., 2016; Thakare et al., 2017), 3) members of monkey cytochrome P450 exhibit >90% homology in amino acid sequences with corresponding human CYPs and similar hepatic metabolic activity in vitro and in vivo (Iwasaki and Uno, 2009), and 4) human hepatic clearance was shown to be well predicted based on the species scaling of monkey hepatic clearance for a large set of OATP substrates (Tess et al., 2020). The current in vitro data (PSinf and % active uptake) also corroborate similarities in human and NHP transport mechanisms and thus suggest the latter as a valuable preclinical model. On the other hand, rodents do not express key human hepato-selective OATPs (OATP1B1/1B3) but instead express orthologs with only ∼70% sequence similarity in liver. Such differences can present uncertainty when projecting disposition mechanisms in human (Chu et al., 2013; Tess et al., 2020). Alternatively, humanized rodent models (Oatp1a/1b knockout and OATP1B1/1B3 knock in) are emerging as useful tools for studying the role of these transporters in drug disposition (Higgins et al., 2014; Salphati et al., 2014). However, some reports showed limited in vivo function for probe substrates (e.g., pitavastatin and rosuvastatin) in humanized OATP1B1 mice, suggesting challenges in the translation of the findings from humanized mice to humans (Salphati et al., 2014). Further understanding and refinement are therefore required to employ these models for characterizing human hepatic uptake.
Our secondary objective was to provide in vitro data (e.g., studies using transfected cell and monkey and human hepatocytes) in support of the in vivo findings in NHP. Generally, a good agreement was observed between in vitro and in vivo ft,OATP1B in NHP for compounds with logP ≤6.5 (Fig. 5). In this study, some very lipophilic acids (logP >6.5; e.g., montelukast, LY2409021, MK-0893) did not produce expected results in the in vitro assays (hepatocyte uptake, OATP1B1 substrate assay, etc.). Also, the presence of rifampicin in the incubations did not reduce cellular accumulation for these compounds, although a clear in vivo signal was seen in cynomolgus monkey with CL ratio reaching as high as 6.3 (MK-0893). It is noteworthy that these compounds show high binding to hepatocytes (fu,heps < 0.004), and under such conditions the in vitro assays may not have the fidelity to distinguish active uptake from nonspecific cell binding. It is worth noting that highly lipophilic compounds can be generally problematic to study absorption, distribution, metabolism, and excretion properties in vitro, including permeability, plasma protein binding, metabolism, and transport kinetics (Press and Di Grandi, 2008). This may be attributed to one or several experimental reasons, including nonspecific binding to the cell surface and assay apparatus, etc. (associated with high lipophilicity). Therefore, additional diligence is necessary in characterizing the uptake mechanism in this space to avoid unexpected clinical readouts. In particular, for compounds in the current physicochemical space (MM ∼400–730 Da and logP ∼3.5–8, acid pKa <6.0), if in vitro studies do not provide appropriate guidance, preclinical in vivo studies may support the qualitative assessment of the contribution of hepatic uptake to overall clearance, which in turn may provide impetus for early clinical assessment of inhibitor (e.g., rifampicin) and genotype impact on pharmacokinetics.
Considerable underprediction was noted in the IVIVE of NHP and human hepatic clearance when using in vitro PSinf directly and scaled assuming hepatic uptake as the rate-determining process to the overall clearance in both species. Correction for the binding to hepatocytes (fu,heps) markedly improved the predictions, although the overall performance may still be suboptimal for prospective predictions (Fig. 4). Potential sources for the IVIVE inconsistencies in scaling uptake clearance are discussed in several previous reports (Watanabe et al., 2009; Jones et al., 2012; (Ménochet et al., 2012); Varma et al., 2014; Kimoto et al., 2017). Additionally, recent studies have suggested “albumin-mediated uptake,” wherein presence of albumin/plasma in the in vitro incubations could improve IVIVE for OATP1B substrates (Bowman and Benet, 2018; Miyauchi et al., 2018; Bteich et al., 2019; Kim et al., 2019; Riccardi et al., 2019; Liang et al., 2020; Bi et al., 2021). Further investigation is warranted in this area. One of the limitations of our study is the use of a single hepatocyte lot. Nonetheless, transport data, particularly in plateable human hepatocytes, are consistent with our earlier reports across multiple hepatocyte lots for several OATP1B substrates evaluated here (pitavastatin, glyburide, montelukast, PF-05241328) (Bi et al., 2021). Finally, it is noteworthy that the measurement of nonspecific binding to cell systems at conditions used in the in vitro assays can be important in the translation of in vitro data. This is a common practice in the translation of in vitro metabolic clearance data obtained using liver microsomes or hepatocytes (Obach, 1997; Hallifax and Houston, 2012) and should be routinely considered in scaling transport clearance.
Earlier studies with extensive data sets suggested considerable underperformance in the translation of in vitro metabolic clearance measured using human liver microsomes or hepatocytes to predict human hepatic clearance for drugs thought to be cleared by metabolism alone (i.e., do not involve transporters) (Hallifax and Houston, 2012; Wood et al., 2017). However, it is well demonstrated that the fraction metabolism by a specific enzyme measured using these human reagents provides reliable quantitative predictions for the victim DDIs, particularly involving CYPs and UDP glucuronosyltransferases (Bjornsson et al., 2003; Di et al., 2013; https://wwwfdagov/regulatory-information/search-fda-guidance-documents/vitro-drug-interaction-studies-cytochrome-p450-enzyme-and-transporter-mediated-drug-interactions). Analogous to this, although we note considerable disconnect in the IVIVE of hepatic intrinsic clearance using in vitro PSinf in both monkey and human, in vitro ft,OATP1B values obtained using hepatocytes are in good agreement with the in vivo change in clearance and VDss in the cynomolgus monkey after rifampicin administration (Fig. 5). Moreover, there is good concordance in the in vitro ft,OATP1B between monkey and human hepatocytes, reflecting on the likely translation of observed rifampicin-induced PK changes in cynomolgus monkey to human.
Conclusions
In conclusion, this study provides robust evidence for the important role of OATP1B1/1B3-mediated hepatic uptake in the clearance and VDss of large lipophilic acids (typically highly permeable and ECCS class 1B drugs/compounds). Although in vitro transporter data are helpful in many cases to quantify the uptake mechanism, certain limitations in the cell-based studies may result in considerable false negative readout. Consequently, further diligence using in vivo animal models, such as the cynomolgus monkey, becomes important to characterize hepatic uptake in this broader physicochemical space.
Acknowledgments
The authors wish to thank Brendan Tierney for supporting pharmacokinetic studies in cynomolgus monkeys. Keith Riccardi is acknowledged for conducting plasma protein binding measurements. John Curto is acknowledged for sourcing several of the study compounds. Robert Mongillo is acknowledged for providing analytical chemistry support. The authors also wish to thank David Edmonds, David Rodrigues, Theunis Goosen, Li Di, Dennis Scott, and Tristan Maurer for valuable input during this study.
Authorship Contributions
Participated in research design: Eng, Bi, Tess, Griffith, Litchfield, Kalgutkar, Varma.
Conducted experiments: Eng, Bi, West, Ryu, Yamaguchi, Kosa.
Performed data analysis: Eng, Bi, Tess, Varma.
Wrote or contributed to the writing of the manuscript: Eng, Bi, West, Ryu, Yamaguchi, Kosa, Tess, Griffith, Litchfield, Kalgutkar, Varma.
Footnotes
- Received December 7, 2020.
- Accepted January 25, 2021.
All authors are full-time employees of Pfizer Inc. The authors have no conflicts of interest that are directly relevant to this study. No funding was received for the work reported here.
↵
This article has supplemental material available at jpet.aspetjournals.org.
Abbreviations
- AUC
- area under the plasma concentration-time curve
- CL
- clearance
- CLh,int,Human
- hepatic intrinsic clearance in human
- CLh,int,NHP
- hepatic intrinsic clearance in NHP
- DDI
- drug-drug interaction
- DPBS
- Dulbecco’s PBS
- ECCS
- extended clearance classification system
- ft,OATP1B
- fraction transported by OATP1B
- fu,heps
- fraction unbound to hepatocytes
- fu,p
- fraction unbound in plasma
- HBSS
- Hanks’ balanced salt solution
- HEK
- human embryonic kidney
- IVIVE
- in vitro–in vivo extrapolation
- LC/MS
- liquid chromatography tandem mass spectrometry
- MM
- molecular mass
- NHP
- nonhuman primate
- NME
- new molecular entity
- OATP
- organic anion–transporting polypeptide
- PK
- pharmacokinetics
- PSinf
- uptake clearance
- Rbp
- blood-to-plasma ratio
- t1/2
- half-life
- VDss
- steady-state volume of distribution
- Copyright © 2021 by The American Society for Pharmacology and Experimental Therapeutics
References
In this issue
OATP1B and PK of Large Lipophilic Acids
