Abstract
To select high bioavailability compounds, it is necessary to predict the first-pass metabolism in the intestine. However, in vitro-in vivo predictions of the intestinal metabolism have proven both challenging and less definitive. The purpose of this study was to investigate prediction of intestinal first-pass metabolism in humans using cynomolgus monkeys. First, we investigated intrinsic metabolic activities in intestinal microsomes of monkeys (MIM) and humans (HIM) (CLint, MIM and CLint, HIM, respectively) of 18 CYP3A substrates. The CLint, MIM values were found to be relatively high and showed excellent correlation with the CLint, HIM values. Subsequently, we determined the plasma concentrations of 9 CYP3A substrates (buspirone, carbamazepine, diazepam, felodipine, midazolam, nicardipine, nifedipine, saquinavir, and verapamil) in monkeys after an oral dose of 2 mg/kg with or without an oral dose of 5 mg/kg ketoconazole and calculated AUC(+vehicle)/AUC(+ketoconazole), defined as Fg, monkey(observed); we confirmed that the dose of ketoconazole inhibited only intestinal CYP3A metabolism by preliminary in vitro and in vivo experiments using ketoconazole. The Fg, monkey(observed) was lower than the Fg, human(observed) for most compounds, but moderate correlation was observed. Furthermore, using these data, we established a new methodology to estimate Fg, human(predicted) more precisely on the basis of the assumption that intestinal physiological conditions other than intrinsic metabolic activity would be the same between monkeys and humans. In conclusion, the in vivo model using cynomolgus monkeys in this study is useful for prediction of intestinal first-pass metabolism by CYP3A in humans because it was able to predict Fg, human of all nine compounds investigated.
Introduction
The successful prediction of human pharmacokinetic properties plays a crucial role in selection of candidate drugs and significantly reduces the number of potential failures in drug development. Oral administration is the most convenient route, but it is often associated with low bioavailability because of first-pass metabolism in the liver and the intestine. To select high bioavailability compounds, attempts have been made to predict the extent of first-pass effects in humans by in vitro methods using subcellular fractions and cell lines. Although in vitro-in vivo predictions of hepatic metabolism have become widely accepted, the same approaches to intestinal metabolism have proved to be a greater challenge, and the results are less definitive (Koster et al., 1985; Mistry and Houston, 1985; Mizuma, 2002; Yang et al., 2007). Modeling of intestinal first-pass metabolism is difficult, because intestinal first-pass metabolism is influenced not only by intrinsic metabolic activities but also by physiological complexities unique to the intestine; thus, there is a multifaceted interplay among passive permeability, active transport, tissue binding, relevant blood flows, and heterogeneity in expression of metabolic enzymes. Accordingly, we might be able to predict intestinal first-pass metabolism in humans by in vivo methods using animal models.
Monkeys may be appropriate animal models, because of their greater similarities with humans in amino acid sequences of drug-metabolizing enzymes (Komori et al., 1992; Shimada et al., 1997; Uno et al., 2007). Indeed, it has been reported that human pharmacokinetics after intravenous administration of an investigational drug can be predicted more accurately from results obtained in monkeys than from those in rats and dogs (Ward and Smith, 2004). However, the oral bioavailability of some drugs, especially cytochrome P450 (P450) 3A4 substrates, is markedly lower in monkeys than humans (Takahashi et al., 2009), and it is often speculated that the cause is extensive first-pass metabolism in the monkey intestine (Sakuda et al., 2006; Nishimura et al., 2007; Ogasawara et al., 2007). Therefore, we anticipated that the prediction of intestinal first-pass metabolism in humans by in vivo methods using monkeys would be improved by taking the species differences in intrinsic intestinal metabolic activities between monkeys and humans into account.
The CYP3A subfamily is the most important among all human drug-metabolizing enzymes because it is involved in the biotransformation of approximately 50% of therapeutic drugs on the market (Zuber et al., 2002). CYP3A is reported to be the most abundant P450 subfamily in both human intestine and liver, and percent contributions to total immunoquantified P450 content are 40 and 82%, respectively (Paine et al., 2006). It is now well known that many drugs metabolized by CYP3A undergo first-pass metabolism in the intestine.
To estimate intestinal first-pass metabolism, plasma concentration time data after intravenous and oral administration are often used (Kato et al., 2003). However, this approach cannot distinguish between the intestinal availability and the fraction absorbed. Ketoconazole is an antifungal agent that is one of the most potent CYP3A inhibitors both in vitro and in vivo (Olkkola et al., 1994; Tsunoda et al., 1999; Sai et al., 2000). Because the plasma protein binding ratio of ketoconazole is high (99%), it is an appropriate compound to selectively inhibit intestinal metabolism of CYP3A substrates before entering the blood flow (Heel et al., 1982). Ogasawara et al. (2007, 2009) have already reported inhibitory effects on CYP3A in cynomolgus monkeys using midazolam and simvastatin, and ketoconazole interaction studies focusing only on intestinal metabolism may be a particularly useful tool for studies of CYP3A4 substrates.
The purpose of the present study was to investigate prediction of intestinal first-pass metabolism in humans using cynomolgus monkeys. We estimated plasma concentrations of nine CYP3A substrates in monkeys with or without an oral dose of ketoconazole inhibiting only intestinal metabolism and then calculated AUC(+vehicle)/AUC(+ketoconazole) defined as Fg, monkey(observed). Next Fg, human(predicted) was calculated using Fg, monkey(observed) based on the assumption that intestinal physiological conditions other than intrinsic metabolic activities are the same between monkeys and humans.
Materials and Methods
Chemicals.
Astemizole, carbamazepine, cisapride monohydrate, (±)-cis-diltiazem hydrochloride, nicardipine hydrochloride, nimodipine, nitrendipine, terfenadine, (±)-verapamil hydrochloride, and flecainide acetate were purchased from Sigma-Aldrich (St. Louis, MO), and diazepam, midazolam, nifedipine, quinidine, and triazolam were from Wako Pure Chemicals (Osaka, Japan). Buspirone hydrochloride, felodipine, and ketoconazole were purchased from LKT Laboratories (St. Paul, MN), cyclosporine, lovastatin, and simvastatin were from Toronto Research Chemicals (North York, ON, Canada), saquinavir mesylate was from the U.S. Pharmacopeia (Rockville, MD), and β-NADPH was from Oriental Yeast Co., Ltd. (Tokyo, Japan). All of the other reagents and solvents were of analytical grade and commercially available. Pooled intestinal (MLM) and liver microsomes from cynomolgus monkeys (MLM) and pooled intestinal microsomes from humans (HIM) were purchased from XenoTech, LLC (Lenexa, KS).
Animals.
Male cynomolgus monkeys, 3.4 to 5.2 kg, were supplied by Guangxi Xiongsen Experimental Primate Animals Breeding and Developing Limited Company (Guangxi, China) and housed in a temperature- and humidity-controlled room with a 12-h light/dark cycle. They were fed a commercial monkey diet (PS type; Oriental Yeast Co., Ltd.) and fasted overnight before drug administration, with ad libitum access to water. Whenever overnight fasting was used, food was provided immediately after the 8-h blood sample was obtained. All procedures were approved by the Dainippon Sumitomo Pharmaceutical Committee on Animal Research.
In Vitro Metabolic Stability in Intestinal Microsomes.
CYP3A substrates were, respectively, incubated at 37°C in 100 μl of a reaction mixture consisting of 50 mM phosphate buffer (pH 7.4), MIM or HIM, and 3 mM NADPH. Linearity of metabolic activity for the microsomal concentration (0.01–0.2 mg of protein per ml), the substrate concentration (25–200 nM), and the incubation time (15 or 30 min) was confirmed, and optimal reaction conditions were set for each compound. The final concentration of organic solvent in the incubation mixture was 0.5% (v/v). After preincubation at 37°C for 5 min, reactions were initiated by the addition of NADPH solution and stopped by the addition of 200 μl of ice-cold methanol. Control samples were incubated using the same method in the absence of NADPH and with NADPH after addition of ice-cold methanol.
The reaction mixtures were spiked with 200 μl of methanol containing the internal standard, 200 nM flecainide, and centrifuged at 4500 rpm for 10 min to remove precipitated protein. Then, the supernatants were filtered through 0.45-μm 96-well filter plates (Varian, Inc., Palo Alto, CA) and diluted 2-fold with distilled water. The 10-μl portion was injected for high-performance liquid chromatography with tandem mass spectrometry (LC-MS/MS) system.
Fraction Unbound in Microsomal Incubations.
The fraction unbound in microsomal incubations (fumic) for all compounds was determined using the high-throughput dialysis method. Dialysis membranes had a 10-kDa molecular mass cutoff and were purchased from Harvard Apparatus Inc. (Holliston, MA). Compounds (1 μM, final concentration) with 0.2 mg of protein/ml HIM in 50 mM phosphate buffer (pH 7.4) were added to the acceptor chambers and 50 mM phosphate buffer (pH 7.4) was added to the donor chambers. The dialysis plate was placed in an incubator at 37°C for 22 h on a plate rotator. After equilibrium had been reached, 30 μl of samples in the acceptor chamber were mixed with 30 μl of 50 mM phosphate buffer (pH 7.4), and 30 μl of samples in the donor chamber were mixed with 30 μl of 0.2 mg of protein/ml microsomes in 50 mM phosphate buffer (pH 7.4). These samples were then mixed with 240 μl of methanol containing the internal standard, 200 nM flecainide, and centrifuged at 4500 rpm for 10 min to remove precipitated protein. Next, the supernatants were filtered using 0.45-μm 96-well filter plates (Varian, Inc.) and diluted 2-fold with distilled water for the LC/MS/MS system.
Inhibitory Effects of Ketoconazole on Metabolic Activity toward CYP3A Substrates in MIM and MLM.
Buspirone, felodipine, midazolam, nicardipine, nifedipine, saquinavir, or verapamil (200 nM, final concentration) was each incubated at 37°C for 15 min in 100 μl of a reaction mixture consisting of 50 mM phosphate buffer (pH 7.4), MIM (0.4 mg of protein/ml for buspirone, midazolam, and verapamil, 0.2 mg of protein/ml for nifedipine, 0.1 mg of protein/ml for felodipine, and 0.04 mg of protein/ml for nicardipine and saquinavir) or MLM (0.1 mg of protein/ml for buspirone, midazolam, nifedipine, and verapamil, 0.04 mg of protein/ml for felodipine, and 0.01 mg of protein/ml for nicardipine and saquinavir), ketoconazole (0, 0.005, 0.01, 0.05, 0.1, 0.5, 1, and 5 μM), and 3 mM NADPH to assess the change in percentage of substrate consumed. Subsequently, the assay was performed as described under In Vitro Metabolic Stability in Intestinal Microsomes.
Pharmacokinetic Study in Cynomolgus Monkeys.
The intravenous and oral dosing solutions were prepared in saline containing 10% (v/v) 0.1 N hydrochloric acid and 0.5% (w/v) methylcellulose aqueous vehicle, respectively. The same four male cynomolgus monkeys were used in all studies. Drug administration was performed with a washout period of at least 6 days. Animals were fasted for approximately 17 h before dosing. Immediately after oral administration of ketoconazole (5 or 100 mg/5 ml/kg), midazolam (0.2 mg/2 ml/kg) was administered intravenously to the monkeys. In addition, buspirone, carbamazepine, diazepam, felodipine, midazolam, nicardipine, nifedipine, saquinavir, or verapamil (2 mg/2 ml/kg) was administered orally to the monkeys immediately after oral administration of ketoconazole (5 mg/3 ml/kg). To obtain control values for the pharmacokinetic parameters, the vehicle for ketoconazole was administered orally immediately before oral or intravenous administration of these CYP3A substrates. Blood samples were collected from the antebrachial vein and then were centrifuged at 4500 rpm for 15 min at 4°C. The plasma samples were kept at −20°C until analysis.
Measurement of CYP3A Substrate Plasma Concentrations in Cynomolgus Monkeys.
A 50-μl aliquot of plasma, 100 μl of methanol, and 100 μl of internal standard solution (200 nM flecainide in methanol) were mixed well and kept at −80°C for 1 h and then centrifuged at 4500 rpm for 10 min to remove precipitated protein. The supernatants were filtered using 0.45-μm 96-well filter plates (Varian, Inc.) and diluted 2-fold with distilled water for LC-MS/MS.
Analytical Procedure.
Concentrations of CYP3A substrates in samples were measured by the LC-MS/MS method consisting of a TSQ 7000 (Thermo Fisher Scientific, Waltham, MA) with the Shimadzu 10A series (Shimadzu, Kyoto, Japan) or a TSQ Quantum Ultra (Thermo Fisher Scientific) with the Shimadzu 20A series (Shimadzu) or an API 4000 (Applied Biosystems, Foster City, CA) with the Shimadzu 10A series. Chromatography was performed using Inertsil ODS-3 columns (3-μm particle size, 2.1 × 50 mm; GL Science, Tokyo, Japan) warmed to 40°C. The mobile phase consisted of 0.1% formic acid (A) and methanol (B). The flow rate was 0.2 ml/min, and the gradient conditions for elution were as follows: gradient = 0 min at 10% B to 1 min at 90% B to 4 min at 90% B to 4.1 min at 10% B to 7 min at 10% B for cyclosporine, lovastatin, and simvastatin or gradient = 0 min at 10% B to 1 min at 90% B to 3 min at 90% B to 3.1 min at 10% B to 6 min at 10% B for the others. Mass spectrometry detection was performed by positive ionization electrospray. The selective reaction monitoring mode was used as follows to monitor ions (m/z precursor ion → product ion): buspirone (386.0 → 122.0), carbamazepine (237.0 → 194.0), cisapride (467.1 → 183.9), cyclosporine (1203 → 425.2), diazepam (285.0 → 154.0), diltiazem (415.3 → 177.6), felodipine (384.0 → 337.9), ketoconazole (531.0 → 489.0), lovastatin (427.0 → 325.0), midazolam (326.0 → 291.0), nicardipine (480.2 → 315.1), nifedipine (347.0 → 254.0), nimodipine (419.2 → 343.1), nitrendipine (361.0 → 315.0), quinidine (325.0 → 307.2), saquinavir (671.4 → 570.1), simvastatin (441.3 → 325.2), triazolam (344.0 → 309.2), verapamil (455.2 → 165.0), and flecainide (415.1 → 398.1).
Data Analysis.
The peak area ratios of test compounds to internal standards were used for calculation in all experiments. The mean value of duplicate determinations was plotted versus incubation time on a semilogarithmic scale, and the slope was determined by linear-regression analysis as the elimination rate constant [kel (minutes−1)]. The CLint values in MIM (CLint, MIM) and HIM (CLint, HIM) were calculated with eq. 1. When the remaining amount after incubation for 30 min with 0.2 mg of protein/ml was >90%, CLint values were not calculated [CLint <0.018 ml/(min · mg protein)] and plotted as 0.01 in Figs. 2 and 3. The fumic for each microsomal concentration was calculated by fitting the fumic for 0.2 mg of protein/ml (fumic, 0.2 mg protein/ml) in the Langmuir equation. It was assumed that the fumic in intestinal microsomes of monkeys is equivalent to that in humans. The fumic, 0.2 mg protein/ml in incubation mixtures of HIM was calculated using eq. 2, and the mean of duplicate determinations were calculated:
The percent activity of seven CYP3A substrates metabolism remaining was plotted against the range of ketoconazole concentrations on a semilog scale. The IC50 values were determined by nonlinear regression analysis using SAS (version 9.1.3; SAS Institute Inc., Cary, NC).
Pharmacokinetic parameters were calculated for individual animals by noncompartmental analysis using WinNonlin Professional (version 5.2; Pharsight, Mountain View, CA). The maximum plasma concentration (Cmax) and the time to reach Cmax (Tmax) were determined from the highest observed value in the individual plasma concentration-time profiles. The terminal elimination half-life (t1/2) was calculated as ln2/k, where k is the terminal rate constant determined by logarithmic regression analysis. The area under the plasma concentration-time curve extrapolated to infinity (AUCinf) was calculated according to eq. 3: AUCt is the area under the curve from time 0 to the time of the last measurable concentration, and Ct is the plasma concentration at the corresponding time, calculated with use of the regression equation for estimation of the elimination rate constant.
The fractions of the dose not metabolized by intestinal metabolic enzymes in monkeys (Fg, monkey) of buspirone, carbamazepine, diazepam, felodipine, midazolam, nicardipine, nifedipine, saquinavir, and verapamil were calculated from ketoconazole interaction studies, using eq. 4 and assuming that 5 mg/kg p.o. ketoconazole causes complete intestinal CYP3A inhibition: where AUC(+vehicle) and AUC(+ketoconazole) represent the AUCinf of the investigated drug after the oral dose of 2 mg/kg in the absence and presence of 5 mg/kg p.o. ketoconazole, respectively, Fa is fraction absorbed, Fh is hepatic availability, Fg is intestinal availability, and F is bioavailability. We assumed that concomitant administration of ketoconazole has no effect on the fraction of the investigated drugs absorbed or on hepatic CYP3A activity.
Fg is the fraction of dose that escapes intestinal first-pass metabolism in the enterocytes and penetrates into the blood flow of the portal vein (Fig. 1) and can be represented as in eq. 5: where CLperm is permeability clearance and CLmet is metabolic clearance.
Assuming that the intestinal physiological environment except for intestinal intrinsic metabolic activities is the same for both monkeys and humans, Fg, human is represented by eq. 6 as CLmet values of monkeys corrected by the ratio between in vitro metabolic activities of monkeys and humans: Furthermore, eq. 7 is derived from eq. 5:
Equations 8 and 9 are derived from eq. 7: where α represents coefficient value.
Equation 10 is thus derived by substituting eqs. 8 and 9 into eq. 6. Because CLint values of carbamazepine and diazepam could not be calculated [CLint <0.018 ml/(min · mg protein)], we calculated Fg, human for these two compounds assuming that CLint, MIM was 2.3-fold higher than CLint, HIM from the results in Fig. 3.
Grapefruit juice (GFJ) interaction studies represent a useful tool to estimate intestinal availability of CYP3A4 substrates in humans because GFJ inhibits only intestinal CYP3A (Gertz et al., 2008). Therefore, we calculated Fg, human(observed) with eq. 11 for 18 CYP3A substrates other than carbamazepine and diazepam. It is reported that the bioavailability of carbamazepine and diazepam is 1, so Fg, human(observed) of these two compounds was calculated as 1 (Greenblatt et al., 1980; Gérardin et al., 1990; Friedman et al., 1992): We assumed that concomitant administration of GFJ has no effect on the fraction of the investigated drugs absorbed or on hepatic CYP3A activity.
Statistical Analysis.
Statistical differences in the pharmacokinetic parameters were assessed by a two-tailed paired Student's t test. In all cases, a probability level of p < 0.05 was considered significant.
Results
Relationship between CLint, HIM and Fg, human(observed) for CYP3A Substrates.
CLint, HIM and Fg, human(observed) for CYP3A substrates are shown in Table 1 and Fig. 2. Most of the compounds that showed high Fg, human(observed) among the 18 CYP3A substrates exhibited low intestinal CLint values. However, the relationship between CLint, HIM and Fg, human(observed) was not consistent. In particular, results for buspirone, nicardipine, and saquinavir were extremely anomalous with respective values of 0.13 versus 0.11 ml/(min · mg protein), 2.04 versus 0.64 ml/(min · mg protein), and 1.80 versus 0.54 ml/(min · mg protein).
Correlations of Intestinal CLint Values between Humans and Monkeys.
Relationship of intestinal CLint values between humans and monkeys is shown in Table 1 and Fig. 3. Intestinal CLint values for the 18 CYP3A substrates in monkeys were relatively high but showed a good correlation with those in humans. The fitting line of the correlation was 2.3-fold higher in monkeys than in humans.
Inhibitory Effects of Ketoconazole on Intestinal and Hepatic Metabolic Activities of CYP3A Substrates in Monkeys.
Metabolic activity toward seven CYP3A substrates with monkey intestinal and hepatic microsomes was inhibited by ketoconazole in a concentration-dependent manner. However, the inhibitory effects in liver were lower than those in intestine with respective IC50 values of 0.016 to 0.233 and 0.007 to 0.056 μM (Table 2).
Effects of Ketoconazole on the Pharmacokinetics of Midazolam after Intravenous Administration in Monkeys.
To evaluate the dose of ketoconazole inhibiting only intestinal CYP3A of cynomolgus monkeys, we investigated the effects of ketoconazole on the pharmacokinetics of midazolam as a representative CYP3A substrate. The pharmacokinetic parameters and the plasma concentration-time profiles of midazolam and ketoconazole after 0.2 mg/kg intravenous administration of midazolam with a concomitant oral dose of vehicle or ketoconazole (5 or 100 mg/kg) in cynomolgus monkeys are shown in Table 3 and Fig. 4. The pharmacokinetic parameters of midazolam after intravenous dosing were not significantly affected by the 5 mg/kg dose, whereas the t1/2 was prolonged significantly by 100 mg/kg ketoconazole (Fig. 4A; Table 3). The plasma concentrations of ketoconazole reached a Cmax of 0.23 and 2.9 μg/ml after oral dosing of 5 and 100 mg/kg ketoconazole, respectively (Fig. 4B; Table 3).
Effects of Ketoconazole on the Pharmacokinetics of Nine CYP3A Substrates after Oral Administration in Monkeys.
To evaluate the effects of ketoconazole on the pharmacokinetics of buspirone, carbamazepine, diazepam, felodipine, midazolam, nicardipine, nifedipine, saquinavir, and verapamil, we determined concentrations of these compounds in plasma after oral administration to monkeys with concomitant oral doses of vehicle or 5 mg/kg ketoconazole. Figure 5 shows the plasma concentration-time profiles. The effects of ketoconazole on the pharmacokinetic parameters of these compounds are summarized in Table 4. The t1/2 and Tmax of the nine CYP3A substrates after oral dosing was not significantly affected by a concomitant oral dose of 5 mg/kg ketoconazole.
Relationship between Fg, monkey(observed) and Fg, human(observed) for Nine CYP3A Substrates.
Fg, monkey(observed) values for nine CYP3A substrates calculated by eq. 4 are shown in Table 5. The Fg, monkey(observed) was lower than the Fg, human(observed) for most compounds, but a moderate correlation was observed between the two (Fig. 6).
Prediction of Fg, human Values for Nine CYP3A Substrates Using Fg, monkey(observed).
Fg, human(predicted) values for the nine CYP3A substrates calculated by eq. 10 are shown in Table 5. Fg, human(predicted) of nine CYP3A substrates including buspirone, nicardipine, and saquinavir showed correspondence with Fg, human(observed) (Fig. 7).
Discussion
To select high bioavailability compounds, it is necessary to predict first-pass metabolism in the intestine. However, modeling is difficult because of the physiological complexities unique to the intestine. In the present study, we estimated plasma concentrations of nine CYP3A substrates in monkeys with or without an oral dose of ketoconazole inhibiting only intestinal CYP3A metabolism and calculated AUC(+vehicle)/AUC(+ketoconazole) defined as Fg, monkey(observed). Furthermore, Fg, human(predicted) was calculated using Fg, monkey(observed) based on the assumption that the intestinal physiological environment other than CLint values is the same in both monkeys and humans.
Kato et al. (2003) reported FaFg to be markedly reduced when the hepatic clearance was more than 100 ml/(min · kg), and this in vivo intrinsic clearance corresponds to an in vitro intrinsic clearance of 0.024 ml/(min · mg) human intestinal microsomal protein (von Richter et al., 2004). Most compounds that showed higher Fg, human values among the 18 CYP3A substrates tended to show lower intestinal CLint values (Table 1; Fig. 2), in line with the report by Kato et al. (2003). However, the relationship between CLint, HIM and Fg, human was not consistent (Table 1; Fig. 2), especially with buspirone, nicardipine, and saquinavir. The dataset in the report by Kato et al. did not include these compounds. Therefore, Fg, human cannot be predicted using only CLint, HIM values. These results showed that Fg, human were influenced not only by intrinsic metabolic activities but also by other factors (e.g., intestinal permeability), and we should take these factors into account to predict Fg, human. However, modeling of intestinal first-pass metabolism is difficult because of physiological complexities unique to the intestine. Several reports have suggested that animals and humans have much in common with regard to intestinal physiological conditions (Cao et al., 2006; Hurst et al., 2007; Nishimura et al., 2007; Mitschke et al., 2008). Accordingly, we tried to predict intestinal first-pass metabolism in humans by our model using monkeys.
The t1/2 of midazolam after intravenous dosing was not significantly affected by a concomitant oral dose of 5 mg/kg ketoconazole (Fig. 4A; Table 3). Ogasawara et al. (2007) also reported the same results, although values for midazolam and simvastatin after oral dosing were prolonged with 20 mg/kg. The plasma concentration of ketoconazole reached a Cmax of 0.23 μg/ml after oral dosing of 5 mg/kg (Fig. 4B; Table 3). Considering the 99% plasma protein binding ratio of ketoconazole (Heel et al., 1982), the free fraction of Cmax is 0.004 μM and the unbound maximum inflow concentration into the liver is calculated to be 0.006 μM (Ito et al., 1998). The inhibitory effects of ketoconazole on metabolic activities in monkey liver microsomes were lower than those in intestinal microsomes, with IC50 values of 0.016 to 0.233 μM (Table 2). Therefore, the free fraction of Cmax and the unbound maximum inflow concentration into the liver were much lower than the IC50 values. Furthermore, ketoconazole showed the strongest inhibitory effect on metabolic activity of midazolam in monkey liver (IC50 = 0.016 μM) among seven CYP3A substrates. Because the t1/2 of midazolam after intravenous dosing was not significantly affected by a concomitant oral dose of 5 mg/kg ketoconazole, we therefore conclude that this dose also did not inhibit metabolism of the other CYP3A substrates. In fact, the t1/2 values of nine CYP3A substrates after oral dosing were not significantly affected (Fig. 5; Table 4). Considering these data, we selected this dose to achieve the condition of inhibition of intestinal metabolism only.
The IC50 values of ketoconazole for metabolic activities in monkey intestinal microsomes were similar among the seven CYP3A substrates, at 0.007 to 0.056 μM (Table 2). Considering that the solubility of ketoconazole is 11.3 μM (Glomme et al., 2005), the concentration of ketoconazole in the intestinal lumen after oral dosing of 5 mg/kg greatly exceeded the IC50 values. Furthermore, considering that the monkey gut volume is 230 ml, the concentration of ketoconazole in the intestinal lumen is at least 0.32 μM (Davies and Morris, 1993). Therefore, we assumed that the increased systemic CYP3A substrate exposure caused by a concomitant oral dose of 5 mg/kg ketoconazole in monkeys would be useful in estimating Fg, monkey. Even if the oral dose of 5 mg/kg ketoconazole cannot completely inhibit monkey intestinal CYP3A, we thought that Fg, human would be predicted adequately using the increases in systemic CYP3A substrate exposure in monkeys because of the identical inhibitory effects of ketoconazole on intestinal CYP3A toward different substrates (IC50 = 0.007–0.056 μM). Thus, the oral dose of 5 mg/kg ketoconazole might be able to completely inhibit monkey intestinal CYP3A because Fg, human was able to be predicted well using eq. 10.
Intestinal CLint values in monkeys were high but showed an excellent correlation with those in humans (Table 1; Fig. 3). The excellent correlation was consistent with previous reports that showed similarities in amino acid sequences of drug-metabolizing enzymes between monkeys and humans, CYP3A8 and CYP3A5 in monkeys having especially high homology (>90%) to CYP3A4 and CYP3A5, respectively, in humans (Komori et al., 1992; Shimada et al., 1997; Uno et al., 2007). The fitting line of the relationship was 2.3-fold higher in monkeys than in humans (Table 1; Fig. 3). Fg, human values of carbamazepine and diazepam could be predicted well, assuming an CLint, HIM/CLint, MIM ratio of 2.3 (Fig. 7). In addition, Fg, human values for the other seven compounds were also reasonably predicted using the ratio of 2.3 (data not shown). Therefore, we might be able to predict Fg, human of CYP3A substrates roughly when CLint, MIM and CLint, HIM are not investigated. These results confirmed that monkeys are appropriate animals for models to predict intestinal first-pass metabolism in humans.
Fg, human(predicted) values for nine CYP3A substrates including buspirone, nicardipine, and saquinavir showed correspondence with Fg, human(observed) (Table 5; Fig. 7). Yang et al. (2007) earlier reported that the “Qgut” model for in vitro-in vivo prediction of intestinal metabolism, taking into account the interplay between permeability and metabolism, improved prediction compared with that for the “well stirred” model. However, the model using monkeys in the present study improved the predictions further. For example, the Qgut model largely underestimated the Fg, human for saquinavir (with a value of nearly 0 compared with 0.4 with our model).
Considering that interindividual variability of pharmacokinetics in monkeys was relatively large (Table 4), we recommend validation using some of these CYP3A substrates before investigating new chemical entities and using the same cynomolgus monkeys in all studies. In addition, we used the GFJ method to calculate the Fg,human(observed), emphasizing that it was not necessary to consider the “Fa” factor. However, it is necessary to discuss further which method is suitable because the GFJ methods is based on some assumptions, as indicated under Materials and Methods (Gertz et al., 2008).
To our knowledge, the present study is the first to predict intestinal first-pass metabolism in humans using monkeys on the basis of the assumption that the intestinal physiological environment except for intrinsic metabolic activity is same values for both species. We conclude from our findings that the in vivo model using cynomolgus monkeys is more useful than the in vitro-in vivo method applied previously because it was able to predict Fg, human of all compounds investigated.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.110.034561.
-
ABBREVIATIONS:
- P450
- cytochrome P450
- MIM
- monkey intestinal microsomes
- MLM
- monkey liver microsomes
- HIM
- human intestinal microsomes
- LC
- high-performance liquid chromatography
- MS/MS
- tandem mass spectrometry
- AUC
- area under the plasma concentration-time curve
- GFJ
- grapefruit juice.
- Received May 17, 2010.
- Accepted August 11, 2010.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics