Abstract
The pharmacokinetics, metabolism, and excretion of torcetrapib, a selective inhibitor of human cholesteryl ester transfer protein, were investigated in healthy human male volunteers after oral administration of [14C]torcetrapib (120-mg dose). The total mean recovery of radiolabeled dose after 21 days was 75.7%, and most of the dose (63%) was excreted in the urine. The total circulating radioactivity and unchanged torcetrapib plasma concentrations increased over the first 6 h and then declined slowly with mean terminal elimination half-lives of 373 and 211 h. Metabolism of torcetrapib was extensive in humans. Only 5.2% of the total dose constituted unchanged torcetrapib in the feces, whereas no parent was excreted unchanged in the urine. Similarly, pharmacokinetic analysis of total radioactivity and unchanged torcetrapib revealed that the area under the concentration versus time curve from zero to infinity of torcetrapib accounted for ∼7.0% of the circulating radioactivity. Torcetrapib was metabolized to numerous metabolites via oxidation. The primary metabolic pathway involved initial oxidative decarbamoylation followed by extensive further oxidation, resulting in the formation of bistrifluoromethylbenzoic acid (M1) and quinaldic acid (M4) metabolites. A mean 40% of the total dose was excreted in the urine as M4 (and its glucuronide and urea conjugates), whereas 7.0% of the total dose was excreted as M1. In vitro studies using human subcellular fractions suggested that the initial metabolism of torcetrapib proceeds via CYP3A-mediated decarbamoylation. Subsequent oxidations lead to the major circulating and excretory metabolites M1 and M4.
Torcetrapib {(-)-[2R,4S] 4-[(3,5-bis-trifluoromethylbenzyl)-methoxycarbonyl-amino]-2-ethyl-6-trifluoromethyl-3,4-dihydro-2H-quinoline-1-carboxylic acid ethyl ester} was being developed to treat hypercholesterolemia (elevated cholesterol levels) and prevent cardiovascular disease. It was withdrawn from development in 2006 when phase III studies showed excessive mortality in the treatment group receiving a combination of atorvastatin and the study drug (Howes and Kostner, 2007). Torcetrapib acts by inhibiting human cholesteryl ester transfer protein, which normally transfers cholesterol from high-density lipoprotein (HDL) cholesterol to very-low-density lipoprotein or low-density lipoprotein (LDL). Inhibition of this process results in higher HDL levels (the “good” cholesterol-containing particle) and reduces LDL levels (the “bad” cholesterol) (Clark et al., 2006). Administration of torcetrapib alone or in combination with atorvastatin led to an increase in HDL and a decrease in LDL concentrations (Brousseau et al., 2004; Clark et al., 2004; McKenney et al., 2006).
The pharmacokinetics of torcetrapib has been evaluated in the rat and monkey. The results from these studies have indicated that the drug was moderately absorbed with a bioavailability of 33 to 45% and readily distributed throughout the body (volume of distribution is 1.1-2.5 l/kg). The clearance of torcetrapib in both the species was low, ranging from 6 to 12 ml/min/kg. Absorption, distribution, metabolism, and excretion studies with [14C]torcetrapib in the mouse, rat, and monkey resulted in good recovery of the radiolabeled dose, ranging from 74% in the monkey to 94% in the rat. Feces was the major route of excretion in all species. The percentages of dose excreted as unchanged torcetrapib in the mouse, rat, and monkey were 34, 37, and 40%, respectively, suggesting that torcetrapib was extensively metabolized in all three species, and metabolism was the primary route of clearance. The primary metabolic pathway was the oxidative cleavage of the drug to 6-trifluoromethylquinoline-2-carboxylic acid (M4) and 3,5-bistrifluoromethylbenzoic acid (M1) (structures shown in Table 1). In vitro studies in human, rat, and monkey liver microsomes resulted in very low turnover of torcetrapib. The primary metabolite identified in the microsomal incubation is the decarbamoylated metabolite (M2).
The pharmacokinetics of torcetrapib has also been evaluated in several phase I and phase II clinical trials after a single dose and multiple doses. Administration of 10 to 1000 mg of torcetrapib as a Miglyol 812 emulsion to groups of fasting, healthy males and healthy females of non-child-bearing potential resulted in a nonlinear increase in maximum plasma torcetrapib concentrations and areas under the concentration versus time curves (AUCs) with an increase in dose. The mean terminal half-life of torcetrapib was estimated to be 25 h in this study. Torcetrapib also showed greater exposure in volunteers who were given a low- or high-fat meal compared with that in the fasted state. The objective of the current study was to investigate the metabolism and excretion of torcetrapib in humans after a single oral administration of [14C]torcetrapib (Scheme 1). An attempt was also made to gain some mechanistic insight into the formation of the novel metabolite M4 using in vitro systems.
Materials and Methods
Reference Compounds, Radiolabeled Torcetrapib, and Chemicals. All synthetic standards for the metabolites were synthesized at Pfizer Global Research and Development (Groton, CT) using standard procedures. [14C]Torcetrapib was synthesized by the radiochemistry group at Pfizer Global Research in Groton under good manufacturing practice conditions, and the label was located in the C-4 position of the tetrahydroquinoline ring (Scheme 1). The purity of the radiolabeled material was >99%. All reagents and solvents were of highest grade available and were obtained from commercial sources. Ecolite(+) scintillation cocktail was obtained from MP Biomedicals (Irvine, CA). Carbosorb and Permafluor E+ scintillation cocktails were purchased from PerkinElmer Life and Analytical Sciences (Waltham, MA).
Study Design, Dosing, and Sample Collection. This was an open-label, single dose inpatient study conducted with six nonsmoking healthy male human volunteers aged between 18 and 55 years. Before the study started, an institutional review committee approved the protocol and the informed consent document. All study participants gave written informed consent before initiation of the study.
All volunteers were given 120 mg of [14C]torcetrapib (92 μCi) in an emulsion of Miglyol 812 and 0.1% aqueous Polysorbate 80. The specific activity of the dose was 0.77 μCi/mg. Urine was collected into containers surrounded by dry ice at predose (-8 to 0), 0 to 12, and 12 to 24 h and at 24-h intervals during the study through 504 h postdose (21 days). Feces were collected before dosing and then over 24-h intervals up to 504 h postdose (21 days). The total weight of the urine and feces was recorded after each collection. Blood samples (sufficient to provide 6 ml of plasma) were collected for pharmacokinetic evaluation of total radioactivity, torcetrapib, and metabolite M1 at times 0 (just before dosing), 0.5, 1, 2, 3, 4, 6, 8, 12, 24, 36, and 48 h and at 24-h intervals through 504 h postdose. Additional blood samples sufficient to provide a minimum of 20 ml of plasma were collected at 1, 4, 8, and 12 h postdose for metabolite identification. Samples were stored frozen until the day of analysis.
Quantitation of Radioactivity. Radioactivity in the plasma, urine, and feces was determined by liquid scintillation counting. Aliquots of plasma (500 μl) and urine (100-1000 μl) were counted in triplicate by mixing with Ecolite(+) scintillation cocktail (6 ml with 100 μl and 12 ml with 1000 μl) and counted in a Wallac 1409 liquid scintillation counter (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). For determination of radioactivity in feces, the weight of each fecal sample was determined, and the samples were homogenized in 2 parts of deionized water using a Stomacher blender 400. After homogenization, triplicate aliquots (250-500 μl) of each sample were transferred into tared cones and pads, weighed, and combusted in an automatic sample PerkinElmer 308 oxidizer (PerkinElmer Life and Analytical Sciences). The resulting 14CO2 was trapped in Carbo-Sorb (PerkinElmer Life and Analytical Sciences) and mixed with Perma-Fluor E+ (PerkinElmer Life and Analytical Sciences) scintillation fluid, and the radioactivity was quantified by liquid scintillation counting. The combustion efficiency was determined daily, before combustion of the study samples, using a 14C standard. To prevent carryover during combustion, one blank cone and pad were placed between each set of samples. The measured radioactivity content in combusted samples was adjusted using the combustion efficiency values. Samples were analyzed for radioactivity for 2 min (5 min for plasma). Scintillation counter data were automatically corrected for counting efficiency using an external standardization technique and an instrument-stored quench curve generated from a series of sealed quench standards. Radioactivity less than twice the background value was considered to be below the limit of determination. Samples collected before dosing were used as controls and counted to obtain a background count rate.
The radioactivity in the plasma was expressed as nanogram-equivalents of torcetrapib per milliliter. The compound equivalents were determined by dividing the microcuries per gram sample by the specific activity of the compound (0.77 μCi/mg). Samples containing radioactivity (disintegrations per minute) less than or equal to twice the background were considered to be zero in the calculation of the means. Radioactivity in the urine and feces was expressed as a percentage of the administered dose per time interval.
Quantitation of Torcetrapib and Metabolite M1 in Human Plasma. The plasma concentrations of torcetrapib and M1 were determined by GC-MS/MS method. The analytical method consisted of protein precipitation followed by liquid-liquid extraction and chemical derivatization. In brief, a 250-μl aliquot of each plasma sample was treated with a 50-μl solution of an internal standard that was structurally similar to torcetrapib in which the ethyl carbamate was replaced with an isopropyl carbamate and the 2-ethyl substituent was replaced with a cyclopropyl group (0.500 μg/ml, synthesized at Pfizer) and 2,4-bistrifluoromethylbenzoic acid as an internal standard for M1 (1 μg/ml; Sigma-Aldrich, St. Louis, MO), vortexed, and quenched with acetonitrile (1 ml). The mixture was then centrifuged at 3500 rpm for approximately 10 min, and the supernatant was poured in a tube containing 1.0 M potassium phosphate buffer (1 ml, pH 11) and 20 μl of 2,3,4,5,6-pentafluorobenzyl bromide solution in isopropyl alcohol and toluene (10:90). The mixture was heated for 30 min at 85°C in a heat block, and water (2 ml) was added to it. The aqueous mixture was then extracted with methyl tert-butyl ether. The samples were then centrifuged at 3500 rpm for approximately 5 min, and 100 μl of the methyl tert-butyl ether layer was transferred to an amber-colored gas chromatography autosampler vial and capped. Aliquots of ∼0.5 to 2 ml were injected onto the gas chromatograph (model 3400; Varian Chromatography Systems, Walnut Creek, CA)/MAT TSQ 7000 mass spectrometer (Thermo Fisher Scientific, Waltham, MA). The analytes were separated chromatographically using a ZB-5 column (0.25 × 15 mm × 0.1 μm film thickness; Phenomenex, Torrance, CA). The GC-MS/MS conditions were as follows: carrier gas, hydrogen; head pressure, 20 psi; injector temperature, 300°C; transfer line temperature, 250°C; initial column temperature, 80°C; initial time, 0.75 min; temperature ramp rate, 25°C/min; final temperature, 250°C; and final time, 0.1 min. The compounds were detected using negative ion chemical ionization, and the transitions were m/z 373 > 269 and 399 > 324 for torcetrapib and for the internal standard for torcetrapib and m/z 257 > 213 for M1 and its internal standard. The MS conditions were as follows: ionization gas, ammonia; source temperature, 200°C; manifold temperature, 75°C; ionizing energy, 200 V; and emission current, 300 μA. Data collection and integration were performed using Finnigan ICIS software (Thermo Fisher Scientific). Quantitation was based on quadratic regression analysis of calibration curves weighted 1/x2 with the area ratio versus concentration using Watson DMLIMS (version 5.4.1.02; Pharsight, Mountain View, CA). The dynamic range of the assay was 0.500 to 100 ng/ml for torcetrapib and 12.5 to 2500 ng/ml for M1.
Determination of the Pharmacokinetic Parameters. Pharmacokinetic parameters were determined by noncompartmental methods using WinNonlin (version 2.1; Pharsight). The maximum plasma concentration (Cmax) and the time at which this concentration was achieved (Tmax) were taken directly from the concentration data. The area under the plasma concentration versus time curve (AUC0-Tlast) was calculated from 0 to the last quantifiable time point (Tlast), using linear trapezoidal approximation. The plasma terminal elimination rate constant (kel) was estimated by linear regression analysis of the terminal slope of log plasma concentration versus time curve. The area from Tlast to infinity (∞) was estimated by CTlast/kel, where CTlast represents the estimated plasma concentration at Tlast, based on the aforementioned regression analysis. The area under the plasma concentration versus time curve from zero to ∞ (AUC0-∞) was estimated as the sum of AUC0-Tlast and AUCT--∞. The terminal elimination half-life (t½) was estimated as ln2/kel. For estimation of the means and pharmacokinetic parameters, concentrations at 0 h and those <0.500 and <12.5 ng/ml, for torcetrapib and M1, respectively, were assumed to be 0 ng/ml. The means were calculated only if more than 50% of the data were less than the lower limits of quantitation.
Quantitation of Metabolite M1 in Human Urine. Analysis of M1 in urine samples was performed with a nonvalidated LC-MS/MS assay but with inclusion of quality control samples prepared in human urine and a standard curve. A 100-μl aliquot of each urine sample containing M1 was treated with a 20-μl aliquot of the solution of internal standard (60 μg/ml 3,5-dinitrobenzoic acid). Samples above the upper limit of quantitation were diluted 10- or 20-fold by adding 10 or 5 μl of the sample to predosed urine. The urine samples were centrifuged at 1400 rpm for 10 min, and 20 μl of the supernatant was injected onto a Luna C-18(2) column (50 × 2.00 mm, 5 μm; Phenomenex). The mobile phase consisting of 5 mM ammonium formate, pH 7.0 (solvent A) and acetonitrile (solvent B) was used to separate the internal standard and M1. All HPLC analyses were performed at a 0.2 ml/min flow rate at 35°C. The analytes were separated using a gradient system as follows: 0 to 3 min, 5% B; 3 to 7 min, 70% B; and 7 to 7.01 min, 5% B. The column was reequilibrated at 5% B for the next 3 min before next injection. The compounds were detected in negative ion electrospray tandem mass spectrometry on a Micromass Quattro Ultima mass spectrometer (Waters, Milford, MA). The retention times of M1 and the internal standard were 7 and 6 min, respectively. The ions monitored were m/z 257.1 and 211.1 for M1 and the internal standard, respectively. The dynamic range of the assay was 20 to 1500 ng/ml. Data were collected and integrated using MassLynx software (version 3.5; Waters). The ratio of peak area responses of drug relative to the internal standard was used to construct a standard curve using a quadratic regression with a 1/x weighting.
The percentage of dose excreted over 0 to 504 h was then calculated as follows. The number of moles of M1 in the urine was first determined from the mass of M1 (nanograms per milliliter of M1 × volume of urine per sampling time), which was then normalized by the moles of torcetrapib dosed as shown:
Metabolic Profiling.Urine. Urine was pooled from 0 to 96 h so that >90% of the excreted dose was accounted for. The pooling was proportional to the volume of urine collected at each time point. Pooled urine samples were quenched with acetonitrile (5:1) and then centrifuged (3000 rpm for 10 min). The supernatants were evaporated in a TurboVap concentration evaporator (Caliper Life Sciences, Hopkinton, MA) at 35°C under nitrogen, and the residue obtained was reconstituted in 500 μl of 50% acetonitrile solution. An aliquot was injected onto the column (100 μl).
Plasma. Plasma samples obtained from 0 to 12 h were pooled according to the method reported by Hamilton et al. (1981) for profiling of circulating metabolites. The pooled samples were treated with acetonitrile at a ratio of 5 volumes of acetonitrile to 1 volume of plasma. The mixture was then centrifuged, and the supernatant was transferred to another tube. The pellets were washed once more to ensure that most of the radioactivity was recovered. The supernatants were mixed and evaporated to dryness in a TurboVap at 35°C under nitrogen. The residues were reconstituted in 500 μl of 50% acetonitrile solution, and an aliquot (100 μl) was injected onto the column. Aliquots (30 μl) of the reconstituted samples were also counted on the liquid scintillation counter to determine the radioactivity extraction recovery.
Feces. Fecal homogenates from all of the healthy human volunteers were pooled on a weight basis to account for 90% or greater radioactivity. Each pooled fecal sample was diluted with 30 ml of acetonitrile and vortexed. The sample was then centrifuged, and the supernatant was separated. The process was repeated several times until >90% of the radioactivity was extracted. All supernatants were mixed and evaporated to approximately 1 ml in a TurboVap at 35°C under nitrogen. The concentrated residue was extracted with 10 ml of ethyl acetate until all radioactivity was recovered, and the organic layer was evaporated to dryness in a TurboVap at 35°C under nitrogen. The residue obtained was reconstituted in ∼300 μl of 50% acetonitrile solution, an aliquot was injected onto the column (100 μl), and a small sample was analyzed by liquid scintillation counting for radioactivity extraction recovery.
Separation, Quantification, and Identification of Metabolites. Metabolic profiling was performed using the HPLC system that consisted of an HP-1100 membrane degasser, HP-1100 autoinjector, and HP-1100 binary gradient pump (Agilent Technologies, Palo Alto, CA). Chromatography was performed on a Zorbax C18 column (5 μm, 4.5 × 150 mm) by injecting 100 μl of the reconstituted sample. The mobile phase was initially composed of acetonitrile (solvent B) and 10 mM ammonium formate, pH 2.0 (solvent A). The flow rate was 1.0 ml/min, and separation was achieved at ambient temperature. The 60-min gradient was as follows: 0 to 10 min, 20% B; 10 to 35 min, 24% B; 35 to 37 min, 50% B; 37 to 53 min, 95% B; and 53 to 55 min, 20% B. The column was reequilibrated at 20% B for the next 5 min before the next injection. The fecal metabolites were detected using a β-RAM detector fitted with a liquid scintillation cell (IN/US, Riviera Beach, FL). The β-RAM was operated in the homogeneous liquid scintillation counting mode with the addition of 3 ml/min of Tru-Count scintillation cocktail (IN/US) to the effluent. The β-RAM response was recorded as a real time analog signal by the MS data collection system. The metabolites were quantified by measuring radioactivity in the individually separated peaks in the radiochromatogram using WinFLOW software (IN/US). The β-RAM provided an integrated printout in counts per minute and percentage of the radiolabeled material. The urinary and circulating metabolites were detected by a liquid chromatography-accurate radioisotope counting system (LC-ARC; AIM Research Company, Wilmington, DE). The LC-ARC was operated in the homogeneous liquid scintillation counting mode with the addition of 2.5 ml/min of Tru-Count scintillation cocktail to the effluent. The urinary and circulating metabolites were quantified by LC-ARC software (AIM Research Company) by measuring the radioactivity in the chromatographically separated peaks.
The metabolites were identified using a PE Sciex API 2000 (PerkinElmer-Sciex Instruments, Boston, MA) or a Finnigan LCQ Ion Trap mass spectrometer (Thermo Fisher Scientific) equipped with an electrospray ion source. All mass spectrometers were operated in the positive ion mode. The instrument settings and potentials (e.g., collision energy) were adjusted to provide optimal data after the instrument was tuned with a solution of torcetrapib. The PE Sciex API 2000 was operated at an ionspray voltage of 4000 V and orifice voltage of 36 V. The collision-induced dissociation (CID) studies (precursor ion scan or product ion scan) were performed using nitrogen gas at a collision energy of 30 V and collision gas thickness of 4 × 1014 molecules/cm2. The MS data were analyzed by MultiView 1.4 software (PerkinElmerSciex Instruments).
The operating parameters for the ion trap were as follows: capillary temperature, 270°C; spray voltage, 4.0 kV; capillary voltage, 4.0 V; sheath gas flow rate, 90; and auxiliary gas flow rate, 30. The mass spectrometer was operated in the positive ion mode with data-dependent scanning. The ions were monitored over a full mass range of m/z 125 to 1000. For a full scan, the automatic gain control was set at 5.0 × 108, the maximum ion time was 100 ms, and the number of microscans was set at 3. For MSn scanning, the automatic gain control was 1.0 × 108, maximum ion time was 400 ms, and the number of microscans was set at 2. For data-dependent scanning, the default charge-state was 1, default isolation width was 3.0, and normalized collision energy was 40 V. The MS data were analyzed using Xcalibur software (version 1.2; Thermo Fisher Scientific).
The metabolites were identified using a Q1 (full scan), neutral loss, precursor ion scan. The structures of metabolites were identified using a product ion scan of the molecular ions that were identified in the above scanning modes and multiple reaction monitoring scanning.
Investigation of the in Vitro Metabolism of Torcetrapib to M2. An initial incubation study to determine reaction velocity linearity was carried out at 37°C with pooled human liver microsomes (3 mg/ml) or the S9 fraction (3.7 mg/ml) and torcetrapib (20 μM) in the presence and absence of NADPH (1 mM) in a total volume of 1 ml of 0.1 M potassium phosphate buffer, pH 7.4. At 0, 10, 30, and 45 min after initiation of the reaction, 0.1 ml of sample was removed and added to 0.2 ml of ice-cold acetonitrile containing 200 ng of internal standard (used previously in the quantitation of torcetrapib; see above). The precipitate was removed by centrifugation at 3000 rpm for 10 min, and the supernatant was analyzed by HPLC-MS as described below. In subsequent studies an incubation time of 5 min and a protein concentration of 0.1 mg/ml were used. For the saturation experiment the torcetrapib concentrations ranged from 0.1 to 20 μM. For chemical inhibition tests, torcetrapib (2 μM) was incubated in the presence and absence of furafylline (20 μM), sulfaphenazole (10 μM), quinidine (1 μM), ticlopidine (10 μM), and ketoconazole (0.01-1 μM). For recombinant P450 enzymes, torcetrapib (2 μM) was incubated with recombinant P-450 (1 mg/ml; 10 pmol/ml of P-450) for 5 min and for the determination of enzyme kinetic parameters with recombinant CYP3A4 and CYP3A5 the range of torcetrapib concentrations was 0.1 to 20 μM. Analysis of samples for M2 was done by selected reaction monitoring on a PE Sciex API 3000 LC-MS/MS system using an Advantage ARMOR C8 column (4.6 × 50 mm, 5 μm particle size) and a mobile phase of 80% acetonitrile and 20% 10 mM ammonium acetate at a flow rate of 0.5 ml/min. Torcetrapib (m/z 618.1 → 300.1), M2 (m/z 529.2 → 228.0), and the internal standard (m/z 644.1 → 325.9) had retention times of 2.0, 1.7, and 2.3 min, respectively. M2 was quantitated using a standard curve range of 4.00 to 500 ng/ml. Enzyme kinetic data were fit to the function using SigmaPlot (version 8; Systat Software Inc., San Jose, CA):
Elucidation of the Pathway to Metabolite M4. Four potential substrates: M24, 2-ethyl-6-trifluoromethylquinoline (1), M3, and M17, were incubated at 10 μM with pooled human liver microsomes (2 mg/ml), MgCl2 (3.3 mM), and NADPH (1.3 mM) in 2 ml of potassium phosphate buffer (100 mM; pH 7.4). Incubations were begun with the addition of NADPH and incubated at 37°C for 60 min open to the air. The incubations were terminated by addition of 5 volumes of acetonitrile, the precipitate was removed by centrifugation, and the supernatant was evaporated under N2. The residue was reconstituted in 0.2 ml of 0.1% formic acid containing 20% acetonitrile and injected onto a Polaris C18 column (4.6 × 250 mm; 5 μm) equilibrated in 0.1% formic acid in 25% acetonitrile at a flow rate of 0.8 ml/min. The mobile phase was held at this composition for 5 min followed by a linear gradient to 60% acetonitrile over the next 45 min. The effluent was monitored at λ = 305 nm, split ∼9:1, and introduced into an electrospray ion source of a Finnigan LTQ ion trap mass spectrometer operated in the positive ion mode. The source parameters of operation of the LTQ ion trap mass spectrometer were similar to those of the ion trap described previously. A similarly analyzed incubation was done for M17 as substrate using pooled human liver S9 (0.5 mg/ml) with addition of NAD+ instead of NADPH.
Results
Excretion. Urine and feces were collected over 504 h (21 days) from six healthy male volunteers after oral administration of 120 mg of [14C]torcetrapib. The percentage of radioactivity (expressed as percentage of dose) excreted in the urine and feces of six healthy male volunteers over 504 h ranged from 65.4 to 85.6% (mean 75.7 ± 8.7%) (Fig. 1). Table 2 shows the percent recovery of the radiolabel in the urine and feces. The percent recovery of dose in urine ranged from 52.4 to 70.5% (mean 63.0 ± 7.0%), whereas the percentage of dose in feces ranged from 6.90 to 15.5% (mean 12.7 ± 3.1%) over the collection period (Table 2). The majority of recovered radioactivity (>90%) was excreted within the first 5 days (120 h) after administration of the dose in all subjects (Fig. 1).
Pharmacokinetics of Total Radioactivity, Torcetrapib, and Metabolite M1 in Healthy Male Volunteers. The mean plasma concentration versus time profiles of total radioactivity and torcetrapib after a single oral administration of [14C]torcetrapib in healthy male volunteers at a dose of 120 mg are presented in Fig. 2. The individual and mean pharmacokinetic parameters of total radioactivity and torcetrapib determined by noncompartmental analysis are summarized in Table 3. Because metabolite M1 was a major metabolite in the preclinical species and was devoid of the radiolabel, its concentrations in the plasma were also monitored and the pharmacokinetic parameters were determined (Fig. 2; Table 3).
The mean Cmax values for total circulating radioactivity, torcetrapib, and M1 were 2240 ng-Eq/ml, 576 ng/ml, and 2920 ng/ml and peaked at 6.3, 6.3, and 24 h, respectively (Table 3). The mean AUC(0-∞) values of total circulating radioactivity, torcetrapib, and M1 were 108 × 103 ng-Eq-h/ml, 7.57 × 103 ng-h/ml and 452 × 103 ng-h/ml, respectively. The mean terminal elimination half-lives of the total circulating radioactivity, torcetrapib, and M1 were 373, 211, and 117 h, respectively. The exposure of total radioactivity was substantially higher than that of unchanged torcetrapib. The AUC(0-∞) of torcetrapib accounted for only 7.0% of that of the total circulating radioactivity. Similarly, the exposure of M1 was 139-fold higher than that of the unchanged torcetrapib.
Metabolism of Torcetrapib. The metabolism of torcetrapib in humans was investigated by profiling urine, plasma, and feces collected from six healthy male volunteers. As a first step the biological matrices were pooled to account for >95% of the recovered radioactivity.
Urine. The representative HPLC radiochromatogram of pooled urine is shown in Fig. 3. The individual and mean percentages of urinary metabolites of torcetrapib are shown in Table 4. A total of eight radioactive peaks were observed in the urinary profile. The peaks eluting at ∼35 (M4) and 40 (M22) min accounted for majority of the dose excreted in the urine and represented means of 28.7 and 8.2% of the dose, respectively. The peaks eluting at 16, 19, 21, and 22 min (M11A-M11D) accounted for 18% of the total dose whereas the peaks at ∼14 and 18 min (M5A and M5B) accounted for only 3.5% of the dose. No unchanged torcetrapib was detected in the urine. Metabolite M1 was also detected in the pooled urine sample. The percentage of M1 excreted in the urine over 0 to 504 h was determined by quantifying the metabolite using a nonvalidated assay and normalizing the mass of M1 in urine to the percentage of dose as described under Materials and Methods. On the basis of these calculations, M1 accounted for ∼7.0% of the total dose in the urine (Table 4).
Feces. A representative HPLC radiochromatogram of the extract of pooled human feces is shown in Fig. 4. The individual and mean percentages of metabolites of torcetrapib in human fecal extracts are shown in Table 4. A total of seven radioactive peaks including unmodified torcetrapib were observed in the fecal extracts. The major peak in the radiochromatogram eluting at ∼51 min had a retention time that was similar to that of unchanged torcetrapib and accounted for 5.2% of the dose. Other peaks (M2, M6A and M6B, M7, M12, M13, M24, and M25) represented <2% of the dose. Metabolite M1 was not detected in the fecal extracts.
Circulating metabolites. A representative HPLC radiochromatogram of circulating metabolites in pooled plasma (0-12 h) is shown in Fig. 5. A mean of 97% of the total circulating radioactivity could be identified. The plasma was pooled as described by Hamilton et al. (1981) to get a good assessment of the exposure of each metabolite over 0 to 12 h. Table 5 represents the individual and mean percentages of each circulating metabolite. Four radioactive peaks were observed in plasma. The peak eluting at 51 min corresponded to the retention time of unchanged torcetrapib and accounted for 21.5% of the total radioactivity. The peaks eluting at 18.2 (M3), 31 (M4), and 50 (M2) min constituted 7.2, 63.2, and 3.1% of the circulating radioactivity, respectively.
Identification of Metabolites. The structures of metabolites were elucidated by ionspray LC-MS/MS using a combination of full, precursor ion, and neutral loss scanning techniques. All metabolites were further characterized using the product ion scans of the identified masses. Torcetrapib was detected as an ammonium adduct during its mass spectral analysis and gave a signal at m/z 618 [M + NH4]+ in the positive ion mode. The product ion mass spectrum of m/z 618 gave characteristic major fragment ions at m/z 601 [M + H]+ and 300 (loss of the methyl N-(bis-trifluoromethylbenzyl)-carbamate moiety) and minor fragment ions at m/z 254 and 228, as shown in Table 1. The fragment ions at m/z 272 and 254 resulted from the loss of the hydroxyethyl and ethoxycarbamoyl moieties from m/z 300. Wherever possible, the structures of the major metabolites were confirmed either by nuclear magnetic resonance or by comparing their retention times and mass spectra to synthetic standards. The structures/proposed structures of metabolites of torcetrapib are shown in Scheme 2.
Metabolite M1. Metabolite M1 was not detected in the radiolabeled chromatogram because of the lack of radiolabel in that portion of the molecule. Because the synthetic standard of this metabolite was readily available, the presence of this metabolite was confirmed by comparison of its retention time and mass spectrum, and its exposure was determined in plasma and urine using a GC-MS/MS or a LC-MS/MS assay (see Materials and Methods).
Metabolite M2. Metabolite M2 gave a signal at m/z 529, suggesting that it was the decarbamoylated derivative of torcetrapib. A CID spectrum of M2 at m/z 529 gave a major fragment ion at m/z 228 (Table 1). The fragment ion of m/z 228 suggested a loss of the methyl N-(bis-trifluoromethylbenzyl)carbamate moiety from the molecule, whereas the fragment ion at m/z 200 suggested a loss of an ethyl group from m/z 228. Metabolite M2 was further confirmed by comparing its retention time and its mass spectrum with that of the authentic standard.
Metabolite M3. Metabolite M3 gave a molecular ion at m/z 212. The mass spectrum of M3 at m/z 212 gave one fragment ion at m/z 196 (Table 1). The presence of the radiolabel in the peak indicated that it was torcetrapib-related. Accurate mass analysis of the protonated ion was determined to be 212.0701 (Δ 1.0 ppm, theoretical), corresponding to an empirical formula of C11H9NF3, which suggested that the metabolite was 2-methyl-6-trifluoromethylquinoline. Comparison of the retention time and its mass spectrum with that of the synthetic standard further confirmed this proposal.
Metabolite M4. Metabolite M4 gave a molecular ion at m/z 242. The presence of the radiolabel in the peak suggested that this was a torcetrapib-related product. The molecular ion of M4 was 30 amu greater than that of M3, which suggested that the metabolite was a carboxylic acid. The CID spectrum of M4 at m/z 242 gave fragment ions at m/z 224 and 196, which resulted from the loss of 18 and 46 amu, respectively, from m/z 242 (Table 1). Comparison of the retention time and mass spectrum of M4 with those of the authentic standard of 6-trifluoromethylquinoline-2-carboxylic acid further confirmed its structure.
Metabolite 5. Metabolites M5A and 5B gave a molecular ion at m/z 418, which was 176 amu greater than M4, suggesting that both metabolites were glucuronide conjugates. The mass spectrum of m/z 418 resulted in fragment ions at m/z 400, 242, 224, and 196 (Table 1). Treatment of M5A and 5B with 1 N sodium hydroxide resulted in the hydrolysis of both metabolites to M4, which further confirmed its structure.
Metabolite M6. Metabolites M6A and M6B gave signals at m/z 545, which was 16 amu greater than that of M2 (m/z 529), suggesting that both the metabolites were hydroxylated products of M2. Both metabolites had a similar fragmentation pattern with ions at m/z 527, 244, and 200 (Table 1). The fragment ion of m/z 527 suggested a loss of a water molecule from m/z 545, whereas the fragment ions at m/z 244 and 200 suggested a loss of the methyl N-bis-trifluoromethylbenzyl carbamate moiety followed by loss of the acetaldehyde moiety. The exact location of the hydroxyl group in both of the metabolites could not be determined from the spectrum.
Metabolite M7. Metabolite M7 gave a signal at m/z 559, which was 30 amu greater than that of M2 (m/z 529), suggesting that the metabolite was a carboxylic acid derivative. The CID spectrum of M7 at m/z 559 gave major fragment ions at m/z 258 and 198 (Table 1), which probably resulted from the loss of the methyl N-bis-trifluoromethylbenzyl carbamate moiety and a subsequent loss of acetic acid from the molecule. The structure of M7 was further confirmed by comparing its retention time and its mass spectrum with that of the authentic standard.
Metabolite M11. Metabolites M11A, M11B, M11C, and M11D gave a molecular ion at m/z 418. The CID spectrum at m/z 418 showed one fragment ion at m/z 242, indicating a loss of 176 amu and indicating that all metabolites were glucuronide conjugates (Table 1). Treatment with 1 N sodium hydroxide did not hydrolyze the glucuronide conjugates, suggesting that these were glucuronide conjugates of hydroxylated 2-ethyl-6-trifluoromethylquinoline. The positions of the glucuronide attachment could not be confirmed from the mass spectra.
Metabolite M12. Metabolite M12 gave a signal at m/z 617, which was 16 amu greater than that of the protonated molecular ion of torcetrapib (m/z 601), suggesting that M12 was a hydroxylated derivative of torcetrapib. The mass spectrum of m/z 617 showed fragment ions at m/z 316 and 254 (Table 1). The fragment ion at m/z 316 resulted from the loss of the methyl N-bis-trifluoromethylbenzyl carbamate moiety. The fragment ion at m/z 254 was similar to that observed in the mass spectrum of torcetrapib, which suggested that the ethoxycarbamoyl moiety was the site of hydroxylation (Table 1). Although the site of the hydroxy group could not be determined from the spectrum, comparison of the retention time and mass spectrum of M12 with those of the authentic standard suggested that the 2-position of the ethoxycarbamoyl position was the site of hydroxylation (Table 1).
Metabolite M13. Metabolite M13 gave a molecular ion at m/z 631, which was 30 amu greater than that for the protonated molecular ion of torcetrapib (m/z 601). This suggested that M13 was a carboxylic acid derivative of torcetrapib. The CID spectrum of M13 at m/z 631 showed fragment ions at m/z 330 and 254 (Table 1). The fragment ions at m/z 330 resulted from the loss of the methyl N-bis-trifluoromethylbenzyl carbamate moiety. A loss of 76 amu (hydroxyacetic acid) from m/z 330 yielded m/z 254, a fragment ion that was similar to one observed in the mass spectrum of torcetrapib. This suggested that the ethoxycarbamoyl moiety of torcetrapib was modified.
Metabolite M22. Metabolite M22 gave a signal at m/z 284. The presence of the radiolabel in the peak suggested that it was torcetrapib-related. The molecular ion showed an addition of 42 amu to the molecular ion of M4 (m/z 242) and suggested that it was a derivative of M4. The CID spectrum of m/z 284 showed fragment ions at m/z 267, 224, and 196 (Table 1). The presence of m/z 224 and 196 in the mass spectrum further suggested that the modification was on the carboxy group. The fragment ion of 267 suggested a loss of ammonia from the molecule, which is characteristic of an amide or an urea. The accurate mass of MH+ (284.0652) supports the empirical formula of C12H9N3O2F3 (Δ -0.2 ppm, theoretical), which corresponded to a urea conjugate of M4.
Metabolite M24. Metabolite M24 gave a signal at m/z 471 and suggested a loss of 58 amu from metabolite M2 (m/z 529). A CID spectrum of M24 at m/z 471 gave one fragment ion at m/z 228 (Table 1), which was similar to the one observed in the mass spectrum of M2. This result suggested that M24 was a decarbamoylated derivative of M2, which was confirmed by comparison of the retention time and mass spectrum of M24 with those of its authentic standard.
Metabolite M25. Metabolite M25 gave a signal at m/z 487, suggesting addition of 16 amu to metabolite M24. A CID spectrum of m/z 487 showed one fragment ion of m/z 244, indicating the loss of the methyl bis-trifluoromethylcarbamate moiety (Table 1). It also indicated an addition of 16 amu to the fragment ion at m/z 228 observed in the mass spectrum of M24, suggesting that the tetrahydroquinoline portion of the molecule was modified. The exact position of the hydroxy group could not be determined from the spectrum.
In Vitro Metabolism of Torcetrapib to Metabolite M2. Metabolite M2 was the predominant metabolite identified after incubation of torcetrapib with NADPH-supplemented pooled human liver microsomes and S9 fractions (data not shown). Control experiments lacking NADPH in the incubation mixture did not form M2. This reaction was subsequently investigated with liver microsomes. Enzyme kinetics of the reaction yielded Km and Vmax values of 2.9 μM and 1.3 nmol/min/mg of protein (Table 6), and the shape of the v versus [S] curve supported possible substrate inhibition at high substrate concentrations. Incubation in the presence of P450 selective inhibitors (furafylline, sulfaphenazole, ticlopidine, quinidine, and ketoconazole) demonstrated that only ketoconazole had an inhibitory effect with an IC50 value of 0.1 μM consistent with its potency for inhibition of CYP3A enzymes. Incubation of torcetrapib with several recombinant heterologously expressed human P450 enzymes (CYP1A1, 1A2, 1B1, 2A6, 2B6, 2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 3A4, 3A5, and 4A11) showed that only CYP3A4 and CYP3A5 catalyzed conversion to M2. The Michaelis constants for the recombinant CYP3A enzymes were similar to those measured in liver microsomes.
Investigation of the Metabolic Pathway to Metabolite M4. A summary of metabolites observed in in vitro incubations performed to investigate the pathway to M4 is listed in Table 7. Incubation of M24 with human liver microsomes under conditions supporting P450 activity yielded M4 and intermediate metabolites M9 and M3 (Fig. 6). Interestingly, incubation of the analog lacking the 3,5-trifluorobenzyl substituent did not yield metabolites, indicating that this compound is not an intermediate metabolite in the pathway to M4 (data not shown). Under oxidative conditions, incubation of 2-ethyl-6-trifluoromethylquinoline (1) yielded metabolites that were one carbon atom less on the 2-position (M3, M17, and M4) (Fig. 6). As expected, 2-methyl-6-trifluoromethylquinoline (M3) was sequentially oxidized to the benzyl alcohol metabolite (M17) and quinaldic acid (M4), with the latter step being observed under conditions that support P450 activity as well as alcohol dehydrogenase activity (Table 7).
Discussion
In the present study, the metabolism and disposition of torcetrapib were investigated in six healthy male volunteers after a single oral administration of [14C]torcetrapib at a dose of 120 mg. Unlike preclinical species (Prakash et al., 2008), the mass balance study suggested that urine (63%) was the major route of excretion in humans in contrast with the preclinical species in which most of the radioactivity was excreted via the feces. Only 76% of the dose was recovered in the humans over a period of 504 h (21 days). Assessment of the ratio of AUC0-Tlast to AUC0-∞ (0.70) for total radioactivity in plasma suggested that complete recovery could probably be obtained with continued collection of excreta. The low recovery of dose in humans was probably due to the possible adipose tissue uptake of total radioactivity followed by slow elimination. This speculation was based on the results from whole-body autoradioluminography studies in Long-Evans rats, which demonstrated sequestration of radioactivity in adipose tissue (D. K. Dalvie, W. Chen, C. Zhang, A. Vaz, and M. J. Potchoiba, unpublished data). Although a proper assessment of absorption could not be made in this study because of the low total radioactivity recovery, a 63% recovery of dose in the urine and the presence of only 5.2% unchanged torcetrapib in the feces suggested absorption of most of the dose in humans.
Pharmacokinetic analysis of total radioactivity and unchanged torcetrapib suggested that the absorption was slow as plasma concentrations for both peaked at 6.3 h. The elimination of total radioactivity and torcetrapib was also slow as indicated by long terminal elimination half-lives for both (373 h for total radioactivity and 211 h for torcetrapib). This result was also speculated to be due to the adipose tissue uptake of the drug-related material.
As observed in the preclinical species, torcetrapib was extensively metabolized in humans. Only 5.2% of unchanged torcetrapib was detected in the feces, and no parent was detected in the urine. In addition, the majority of the total circulating radioactivity comprised metabolites as AUC0-∞ and Cmax of unchanged torcetrapib accounted for only 7.0 and 25.7% of AUC0-∞ and Cmax of total circulating radioactivity in plasma. Oxidation was the major route of metabolism of torcetrapib in humans. The proposed metabolic scheme of torcetrapib in healthy male volunteers is shown in Scheme 2. Metabolites M1 and M4 constituted major metabolites in plasma and urine. Approximately 40% of the dose was excreted as M4 and its glucuronide and urea conjugates in the urine, whereas circulating concentrations of M4 were ∼3-fold greater than those of unchanged torcetrapib.
The exposures to M1 in human plasma as determined by AUC0-∞ and Cmax were 139- and 12-fold greater than the exposure of unchanged torcetrapib, whereas M1 accounted for approximately 7% of the dose excreted in the urine. The plasma concentrations of M1 peaked at 24 h, which was approximately 18 h later than those of torcetrapib and total radioactivity suggesting that M1 formation could not be attributed only to the parent drug and that other metabolites also contributed to its formation. Neither metabolite M1 nor M4 inhibited cholesteryl ester transfer protein. This finding was not surprising, given the structural and lipophilicity differences of these two metabolites and torcetrapib. M1 and M4 were not detected in the fecal extracts in contrast with urine. Instead, the metabolic profile of the fecal extracts showed decarbamoylation of torcetrapib as the primary pathway (M2, M6, M7, M24, and M25) and oxidative metabolites of torcetrapib (M12 and M13). All metabolites (except M24 and M25) detected in the plasma, urine, and feces of humans were present in the toxicology species. Although unique in humans, both M24 and M25 constituted only ∼1% of the total excreted dose. In addition, neither metabolite was present in human plasma. Thus, no additional studies were performed to assess their safety in preclinical species.
In the in vitro studies undertaken to assess the formation of the major metabolites M1 and M4, torcetrapib was incubated with human liver microsomal and S9 fractions. These studies revealed that torcetrapib primarily underwent decarbamoylation, resulting in M2 in both the fractions. Although, oxidative decarbamoylation as well as hydrolysis of carbamates by intestinal and hepatic esterases has been well documented in the literature (Yumibe et al., 1996; Khanna et al., 2000; Quinney et al., 2005), formation of M2 was NADPH-dependent and led to the conclusion that this was a P450-mediated process rather than hydrolysis. Further phenotyping studies using recombinant P450s and inhibitors indicated that this pathway was CYP3A4/3A5 catalyzed. Thus, a pathway similar to oxidative decarbamoylation of loratadine (Yumibe et al., 1996) could be proposed for the formation of M2 (Scheme 3). This pathway could involve initial hydrogen atom abstraction from the carbon atom of the carbamoyloxy group followed by oxygen rebound and subsequent cleavage of the corresponding α-hydroxycarbamate via concomitant decarboxylation to yield M2. Metabolites M1 and M4 were not detected in these experiments, suggesting that these were downstream products of M2.
Studies to gain further mechanistic insight into the formation of the novel metabolite M4 were conducted by incubating the benzylamine M24 and other synthetic metabolites 1, M3, M9, and M17 with pooled liver microsomes and cytosolic fractions. Detection of M4 in NADPH-supplemented microsomal incubations of M24 indicated that one possible pathway leading to M4 was via the route depicted in Scheme 2. Briefly, the tetrahydroquinoline metabolite (M24) is first oxidized at the α-position to yield the imine (2), which will be in equilibrium with the corresponding enamine (3). The latter can undergo aromatization via elimination of the 3,5-bistrifluoromethylbenzylamine moiety to yield the quinoline intermediate 1. Although the exact mechanism for elimination of the substituted benzylamine was not explored using unsubstituted analogs, one possible trigger for this could be the protonation of the amino group, which could make 3,5-disubstituted benzylamine a good leaving group in addition to the drive for aromatization. Analogous P450-mediated aromatization of alicyclic amines (tetrahydroquinolines and 4-substituted piperidines) to quinolines or pyridines has been reported previously (Aventa et al., 1997; Fang et al., 2001; Gu et al., 2006). The intermediate (1) can subsequently undergo modification of the ethyl moiety via sequential oxidation (to M9 and 4) and decarboxylation of 4 to yield M3 (Scheme 2). The proposed C-C bond cleavage, resulting in the conversion of intermediate 4 to M3, has precedence in organic chemistry (Borsche and Manteuffel, 1936; Doering and Pasternak, 1950). Furthermore, the presence of an electron-withdrawing substituent such as the trifluoromethyl group on the quinoline ring could further facilitate such a cleavage. Sequential P450-mediated oxidation of the benzylic methyl group of M3 through M17 to M4 is a common pathway in drug metabolism. The intermediate products of this mechanism such as M9, M3, and M17 have also been identified as metabolites in the preclinical species. Metabolite M3 has also been detected as a circulating metabolite in humans (Fig. 5; Table 5). Incubations of M17 with NAD+-supplemented human S9 fractions also resulted in M4, indicating that this reaction could be catalyzed by alcohol dehydrogenase in addition to P450.
Although several pathways could be envisioned for M1 formation, one direct pathway could involve oxidative N-dealkylation of torcetrapib or its metabolites M2, M6, M7, M12, M13, M24, or M25 (Scheme 4). This pathway is consistent with the previous reports that have demonstrated N-dealkylation of amides or carbamates in the past (Morino et al., 1985; Hall and Hanzlik, 1991; Labroo et al., 1995; Hey and Tolando, 2000). Alternatively, M24 can also undergo dealkylation, which, on oxidation, would result in M1.
In conclusion, this study demonstrated the metabolism of torcetrapib in humans. Administration of torcetrapib to humans resulted in extensive metabolism of the drug to multiple oxidative metabolites. The initial metabolism of torcetrapib proceeds via CYP 3A4-mediated decarbamoylation. Subsequent oxidations lead to the major circulating and excretory metabolites M1 and M4.
Acknowledgments
We thank Drs. Klaas Schildknegt, Roger Ruggeri, and Gregory Dolnikowski for providing radiolabeled torcetrapib and synthetic metabolite standards.
Footnotes
-
doi:10.1124/dmd.108.023176.
-
ABBREVIATIONS: HDL, high-density lipoprotein; LDL, low-density lipoprotein; AUC, area under the concentration versus time curve; GC, gas chromatography; MS/MS, tandem mass spectrometry; MS, mass spectrometry; HPLC, high-performance liquid chromatography; CID, collision-induced dissociation; amu, atomic mass units.
-
↵1 Current affiliation: Arena Pharmaceuticals, San Diego, California.
-
↵2 Current affiliation: Genentech, San Francisco, California.
- Received June 29, 2008.
- Accepted August 8, 2008.
- The American Society for Pharmacology and Experimental Therapeutics