Abstract
Rosiglitazone is a potent peroxisome proliferator-activated receptor gamma agonist that decreases hyperglycemia by reducing insulin resistance in patients with type 2 diabetes mellitus. The disposition of 14C-labeled rosiglitazone was determined after oral and i.v. dosing of rosiglitazone solution, and the disposition of nonradiolabeled rosiglitazone was determined after oral dosing of tablets in this open-label, three-part, semirandomized, crossover study. The absorption of rosiglitazone was rapid and essentially complete, with absolute bioavailability estimated to be ∼99% after oral tablet dosing and ∼95% after oral solution dosing, and clearance was primarily metabolic. The time to maximal concentration of radioactivity and the elimination half-life for two metabolites in plasma were significantly longer than for rosiglitazone itself (4–6 h versus 0.5–1 h, and ca. 5 days versus 3–7 h). Radioactivity was excreted primarily via the urine (∼65%) and was excreted similarly after oral and i.v. dosing. The major routes of metabolism were N-demethylation and hydroxylation with subsequent conjugation, of which neither was affected by the route of drug administration. The major metabolites, those of intermediate importance, and nearly all of the trace metabolites in humans have been identified previously in preclinical studies. Rosiglitazone was well tolerated in all formulations.
Rosiglitazone [AVANDIA, BRL-49653C, (±)-5-[[4-[2-methyl-2-(pyridinylamino)ethoxy]phenyl]methyl]methyl]-2,4-thiazolidinedione (Z)-2-butenedioate (1:1)] is a potent antihyperglycemic agent (Patel et al., 1998; Young et al., 1998) that reduces insulin resistance in patients with type 2 diabetes (Matthews et al., 1999). Rosiglitazone is a member of the thiazolidinedione class, a group of oral antidiabetic agents that exert their glucose-lowering effect by binding to peroxisome proliferator-activated receptors gamma (PPARγ) (Lehmann et al., 1995; Berger et al., 1996; Willson et al., 1996; Young et al., 1998), which preferentially bind to DNA as heterodimers and activate transcription of a wide variety of metabolic regulators (Forman et al., 1996). These regulators are associated with the differentiation of stem cells into adipocytes (Lehmann et al., 1995; Gimble et al., 1996; Tai et al., 1996; Adams et al., 1997) and increased expression of a number of genes involved in the regulation of glucose and lipid metabolism (Tontonoz et al., 1995; Pearson et al., 1996; Schoonjans et al., 1996;Lefebvre et al., 1997; Martin et al., 1997; Motojima et al., 1998). Like all drugs of this class, which includes pioglitazone, englitazone, and troglitazone (Perry and Petrie, 1998), rosiglitazone contains a thiazolidinedione core, but differs from other thiazolidinediones in the presence of an aminopyridyl side chain (Henry, 1997; Young et al., 1998). Such substitutions among side chains are believed to be responsible for differences in disposition, metabolism, and antidiabetic efficacy among thiazolidinediones (Lehmann et al., 1995;Berger et al., 1996).
The objectives of this study were 3-fold: 1) to determine the disposition and routes of elimination of rosiglitazone in humans after oral and i.v. dosing of solution; 2) to quantify and characterize the major compound-related components of rosiglitazone in human plasma and excreta; and 3) to obtain estimates of the absolute bioavailability of rosiglitazone from both the radiolabeled solution dose and the nonradiolabeled tablet dose.
Materials and Methods
Rosiglitazone (BRL-49653C).
All dose amounts and concentrations are expressed in terms of pure base. Ampoules of nonradiolabeled rosiglitazone injection solution (0.098 mg/ml) and tablets (4 mg) were supplied by Pharmaceutical Technologies, SmithKline Beecham (Upper Merion, PA) (See Scheme FS1.).
[14C]Rosiglitazone injection solution (chemical purity 99.4%, radiochemical purity 99.5%, 0.100 mg/ml, 2.56 μCi/ml) was formulated for human use by Pharmaceutical Technologies, SmithKline Beecham, as follows. Appropriate amounts of sodium chloride and citric acid were dissolved sequentially in Water for Injection (Fresenius, Runcorn, UK). [14C]Rosiglitazone [specific activity 19.2 μCi/mg (712 kBq/mg), synthesized and supplied by the Synthetic Isotope Chemistry Unit, SmithKline Beecham, Harlow, UK] was then dissolved in this solution. The weight of the final solution was made up as required with Water for Injection and filtered (0.2-μm polyvinylidene difluoride, Minisart NML; Sartorius, Epsom, UK). The filtrate was finally refiltered into washed and sterilized 5-ml ampoules, which were then sealed and stored protected from light.
For i.v. dosing, this solution was used directly. For oral administration, individual radiolabeled doses were prepared on the day of dosing by mixing [14C]rosiglitazone injection solution (28.7 ml) and nonradiolabeled rosiglitazone injection solution (53.3 ml) in prewashed and sterilized bottles (100 ml). The solutions were protected from light until administration. Immediately before dosing, each bottle of dose solution was weighed. Approximately 2 g of each solution was removed for the assessment of radiochemical purity and concentration, and each bottle was reweighed.
Study Subjects.
Four healthy male volunteers, aged between 40 and 65, with normal medical history, vital signs (blood pressure and heart rate), 12-lead ECG, clinical chemistry, hematology, and urinalysis participated in the study. Exclusion criteria included: 1) receipt of a total body radiation dose greater than 5.0 mSv or participation in a study with radiolabeled medication in the 12 months before the study start; 2) use of any prescribed medication within 7 days of the study start; 3) a family history of diabetes in first degree relatives; and 4) a history of any condition known to interfere with the absorption, distribution, metabolism, and excretion of drugs. The study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. The protocol was approved by the Independent Ethics Committee in Harlow, UK, before the start of the study, and all subjects provided their written informed consent.
Treatment Protocol.
On the first study day, subjects received [14C]rosiglitazone either as an i.v. infusion (2 mg, 20 ml, 50 μCi) over 60 min or as an oral solution (8 mg, 80 ml, 70 μCi, followed by two 50-ml water washings of the solution container). Subjects crossed over to the other regimen on the second study day, after a 50-day washout period. On the third study day, 42 days later, subjects received nonradiolabeled rosiglitazone tablets (2 × 4 mg), administered with 180 ml of water. Subjects fasted for at least 7.5 h before administration of all study drugs and for an additional 4 h after dosing.
Subjects receiving oral formulations remained ambulatory throughout the procedure; those receiving i.v. rosiglitazone solution were dosed in bed and remained there for 3 h after dosing. Adverse events, vital signs, ECG, clinical chemistry, hematology, urinalysis, and physical examinations were recorded throughout the study. Safety assessments were also made at a follow-up visit 7 to 14 days after the last study dose. Spontaneously reported adverse experiences were also recorded. Subjects remained in the Clinical Pharmacology Unit for 24 h after dosing.
Collection of Biological Specimens.
Blood, urine, and feces were collected at appropriate times after each dose from all subjects. Blood samples were collected into potassium EDTA tubes, mixed gently, and centrifuged (ca. 1500gav) at 4°C without delay. During and after i.v. dosing, blood samples were obtained from the contralateral arm. Blood samples were collected predose (baseline sample), at a dozen time points within the first 24 h, and 2, 3, 4, 7, 14, 21, and 42 days after dosing. The sampling schedule had to be extended from that initially protocoled, due to the slower than anticipated clearance of radioactivity from the plasma. Based on the results from the first period, the sampling schedule was modified so that samples were collected predose and at 14 or 15 time points up to 21 days postdose. In the third study period, blood samples were collected at 10 time points up to 48 h after dosing.
Radiodetection and Quantification.
All materials and instruments used in the radioassay of samples were obtained from Canberra-Packard (Pangbourne, UK). All samples were assayed for radioactivity by liquid scintillation counting (LSC) using Tri-Carb liquid scintillation spectrometers (2200CA, 2500TR, and 2700TR). Quench correction was achieved using an automatic external standard ratio method. Unless stated otherwise, 10 ml of Ultima Gold scintillant was added to each sample before LSC for 10 min, with automatic background subtraction. When possible, radio-HPLC data were captured on-line by homogeneous detection. Alternatively, where levels of radioactivity were too low for on-line radiodetection, HPLC-radio data were captured by off-line fraction collection into minivials followed by direct LSC of the individual fractions. Fractions collected during off-line radio-HPLC analysis were counted for 2 min (without background subtraction) after the addition of 4 ml of Ultima Gold MV scintillant. UV data for these analyses were captured by Laura v1.2 (Lablogic, Sheffield, UK).
Radiochemical Purity Determinations.
Aliquots (100 or 200 μl) of each dose were analyzed by reverse-phase HPLC with on-line radiodetection (Canberra Packard A-525AX radiodetector, with a 500-μl homogeneous flow cell). They were applied to a Kromasil C18 column (5 μm, 150 × 4.6 mm), which was operated at 40°C. The UV absorbance was monitored at 247 nm. Solvent A of the mobile phase was 50 mM sodium acetate (pH 5.0), and solvent B of the mobile phase was acetonitrile. A gradient was used, starting at 100% solvent A at 0 min, changing to 50% solvent B over 15 min, and held isocratic for an additional 25 min. The flow rates were 1.0 ml/min for the mobile phase and 3.0 ml/min for the scintillant. The HPLC column recoveries were calculated from the total dpm recovered in the eluate during the entire gradient run as a percentage of the total dpm injected.
Radioassay of Samples.
Two or three weighed samples of each dose solution (approximately 0.2 g) and dose residue were diluted by weight and radioassayed in triplicate. Plasma and urine samples were radioassayed using three and four aliquots, respectively. Each fecal homogenate was assayed using six aliquots. During sample extractions, recovery of radioactivity was determined at each procedural step using two or three aliquots.
Plasma Analysis of Rosiglitazone.
Plasma concentrations of rosiglitazone were quantified using a validated automated Automated Sequential Trace Enrichment of Dialysates/HPLC assay using fluorescence detection. The lower limit of quantification for rosiglitazone was 3.00 ng/ml using a 200-μl sample volume. Quality control samples were assayed with each batch of samples against calibration standards to assess the reproducibility of the assay.
Sample Preparation for HPLC Radiometabolite Profiling.
Urine
For each subject, a pooled sample representative of the urinary excretion over the first 8 days was prepared by mixing an equal percentage (by weight) of the sample collection of each day. Urine samples (12–24 h and day 8) were also analyzed. On-line radiodetection was used to produce radiochromatograms from unextracted urine samples (0–8 day pooled and 12–24 h) and from solid-phase extracts (day 8 urine and urine hydrolysates). All urine samples and urine extracts were centrifuged (ca. 11,000gav, 5 min) before HPLC analysis.
Enzymic hydrolysis of urine samples.
Selected urine samples were subjected to enzymic hydrolysis to aid metabolite structural identification. Three enzyme preparations were used in these analyses: 1) an enzyme solution containing β-glucuronidase (EC 3.2.1.21, Sigma type B-1, bovine liver, 1000 U/ml) in 0.2 M sodium acetate buffer pH 5.0; 2) a solution containing β-glucuronidase and the β-glucuronidase inhibitor,d-saccharic acid-1,4-lactone (Sigma, 7.5 mg/ml); and 3) a solution containing sulfatase (EC 3.1.61, Sigma Type H-2, Helix pomatia, 115 U/ml). Incubations were prepared by diluting the enzyme and enzyme plus inhibitor 1:1 with buffer/sample mixtures. Identical control incubations were prepared in each case, replacing enzyme with buffer. Samples were incubated overnight at 37°C in a shaking water bath.
Solid-phase extraction (SPE) of urine samples.
Solid-phase extracts were prepared for radiochromatography of day 8 urine and for those urine samples to be analyzed by liquid chromatography-mass spectrometry (LC-MS). SPE cartridges (Varian Mega Bond Elut C18, 60 cc) were preconditioned with methanol and distilled water. Samples of urine (approximately 10–70 ml) were run through the column. The column was washed with distilled water (approximately 25 ml) and then eluted with methanol (approximately 25 ml). The eluent was dried under N2 at room temperature before being resuspended in deionized water (0.5–7 ml) and the resulting solution was centrifuged before HPLC analysis. On average, recovery of radioactivity after SPE was 98% (range, 87–113%) and at subsequent reconstitution was 87% (range, 60–100%). Column recoveries for radioactivity were on average 103% (range, 100–106%). Therefore, no adjustments for this have been made in the calculations.
Samples of 0 to 8 day pooled and day 8 urine intended for enzymic hydrolysis were prepared as described above. These samples were re-extracted after incubation by SPE before LC-MS analysis. SPE cartridges (Varian C18, 100 mg, 1 cc) were preconditioned with methanol and distilled water. Urine enzymic hydrolysates (approximately 1–2 ml) were then run through the column. The column was washed with distilled water and then eluted with methanol. The eluent was dried under N2 at room temperature before reconstitution in deionized water (0.5–1.5 ml). On average, recovery of radioactivity at SPE elution was 106% (range, 99–120%) and at subsequent reconstitution was 78% (range, 72–82%).
Feces.
For each subject, fecal subsamples from days 0 to 7 (0–6 day, subject 4, i.v.; and 0–9 day, subject 3, i.v.) were pooled by total homogenate wet weight ratio and were analyzed by HPLC. To each fecal subsample, one-quarter volume of 0.1 M citric acid was added to improve recovery of radioactivity and the suspension was mixed by vortexing. An equal volume of acetonitrile to fecal homogenate was added, the suspension was mixed and centrifuged (ca. 1300gav, 5 min), the supernatant was removed and retained, and the pellet was extracted twice more. The three supernatants for each sample were combined and recovery of radioactivity was determined. The total volume of the extract was reduced under N2 at room temperature and the samples were frozen with ethanol/dry ice and freeze-dried. The dry extracts were suspended in approximately 15 ml of methanol. Particulates in the suspension were removed by centrifugation, and radioactivity in the supernatant was determined before reducing the preparation to near dryness under N2 at room temperature. The resulting residue was resuspended in 1 to 2 ml of methanol, and the recovery of radioactivity was determined. Before analysis by HPLC, particulates in each extract were removed by centrifugation (ca. 11,000gav, 5 min) and the recovery of radioactivity was again determined.
Radioactivity was extracted for LC radioprofiling with an average efficiency of 84% (range, 77–89%). Recovery of radioactivity after reconstitution for HPLC was on average 86% after the five extraction steps (range, 75–100%). Column recoveries for radioactivity were, on average, 91.5% (range, 90–93%); no corrections to the data have been made to account for this.
Fecal extracts were prepared for LC-MS by omitting the addition of 0.1 M citric acid in the first step and mixing the final methanol solutions with equal volumes of distilled water before centrifugation. In the absence of citric acid, the overall recovery of radioactivity was lower (44–61%), but unchanged in composition.
Plasma.
HPLC analysis was performed on pooled plasma samples from day 1 and on individual plasma samples from day 2. An equal volume of 0.1 M citric acid was added to each plasma sample (2 or 4 ml), and the suspension was mixed by vortexing. An equal volume of acetonitrile was then added and the suspension was thoroughly mixed before being centrifuged (ca. 11,000gav). The supernatant was removed and retained, and the pellet was extracted twice more. The three supernatants for each sample were combined and recovery of radioactivity was determined (99%, range 80–133%). The extract was reduced to dryness under N2 at room temperature, and the resulting residue was resuspended in distilled water. Before HPLC analysis, particulates were removed by centrifugation and recovery of radioactivity was again determined (86%, range 49–112%). Plasma extracts obtained on study day 1 were combined according to dose route and time point for radioprofiling and LC-MS. Plasma extracts obtained on study day 2 were analyzed separately for each subject and time point. An average radioactive concentration value was used for samples from the first dose administration that had been combined. Column recoveries for radioactivity were 92.5% (range, 90–95%); no corrections to the data have been made to account for this.
Quantitative HPLC Radiometabolite Profiling.
HPLC analysis of radiometabolites in plasma, urine, hydrolyzed urine, and feces was performed as follows. Authentic rosiglitazone and synthetic metabolites (SB-243914, SB-244675, and SB-237216) were obtained from Chemical Development, SmithKline Beecham (Harlow, UK) and were used as chromatographic and MS standards. A Canberra Packard A-525AX radiodetector with a 500-μl homogeneous flow cell was used. The UV absorbance was monitored at 247 nm. For analysis of all samples, a Hypersil BDS C18 column (5 μm, 150 × 4.6 mm) was used. The column was operated at 40°C. Solvent A of the mobile phase was 50 mM ammonium acetate (pH 8.0). Solvent B of the mobile phase was acetonitrile/methanol (1:1). A gradient was used, starting at 100% solvent A held for 5 min, changing to 50% solvent B at a rate of 1% per min for 50 min, finally increasing to 95% solvent B over 5 min and held for an additional 10 min. The flow rates were 1.0 ml/min for the mobile phase and 3.0 ml/min for the scintillant. All samples were coinjected with rosiglitazone, SB 243914, SB 244675, and SB 237216 to confirm retention times.
LC-MS.
The HPLC system, column, and conditions were essentially identical with those described under Quantitative HPLC Radiometabolite Profiling except that detection was by MS and the column eluent was split in an approximate ratio of 3:1, to give an approximate flow rate of 250 μl/min into the MS interface. For MS, a VG Quattro mass spectrometer with a PC running Mass Lynx v2.10 data system was used. The ionization mode for MS was electrospray, positive and negative ion, and selected ion recording (SIR) was used. For tandem mass spectrometry, the ionization mode was electrospray, positive ion, the collision energy was 25 eV, and the mass range was 50 to 300 Da (daughters of 281). All samples analyzed by MS were compared with control (predose) samples, prepared, and analyzed under identical conditions.
Data Analysis.
Excretion
The total radioactivity in each collection of excreta has been expressed as a percentage of the total administered radioactive dose. The total radioactivity in plasma has been expressed as nanograms rosiglitazone equivalents per milliliter.
HPLC.
On-line radio-HPLC data were processed and calculated using the software Radiomatic Flo-One, version 3.52 (Packard Instrument Company, Meriden, CT). The LSC data obtained after off-line radiodetection were converted into a format compatible with Labchrom version 2.10 (Lablogic, Sheffield, UK) using its LSC import function to reconstruct radio-HPLC chromatograms. Radioactive peaks were marked using an on-screen cursor. The software integrated the area under the radiochromatogram and also expressed each region of interest as a fraction of the total number of counts detected. The metabolites in urine and feces were quantified in terms of percentage of the radioactive dose administered, and those in plasma as a percentage of the total radioactivity in the sample.
Mass spectral characterization of metabolites.
Peak assignments were made by comparison of chromatographic retention times with the corresponding radiotraces. Numbers were assigned to each individual metabolite identified together with descriptive names.2 Structural assignments were made from molecular ion and retention time data by reference to the fuller mass spectral and NMR data obtained for these metabolites in the preclinical species (Bolton et al., 1996).
Pharmacokinetic analysis.
Plasma concentration-time data for each subject in each regimen were analyzed by noncompartmental methods using a validated in-house computer software program, #PROTOCOL (version 1.2; SmithKline Beecham). All calculations were based on actual sampling times. Pharmacokinetic parameter values determined included maximum observed concentration (Cmax), the time to reachCmax (Tmax), elimination phase half-life (t1/2), and the area under the plasma concentration-time curve from time zero to infinity (AUC0-inf). AUC0-inf was calculated using the linear trapezoidal rule for each incremental trapezoid and the log trapezoidal rule for each decremental trapezoid (Chiou, 1978). In addition, plasma clearance and the volume of distribution at steady state were calculated for rosiglitazone using the plasma concentration data obtained after i.v. administration of [14C]rosiglitazone. The observed absolute bioavailability was calculated as dose-normalized AUC0-inf oral/dose-normalized AUC0-inf i.v. Pharmacokinetic linearity previously had been shown over the relevant dose range and plasma concentrations (D. A. Boyle, personal communication). Thus the use of dose-normalized AUC values for the i.v. dose (2 mg) and the oral dose (8 mg) was valid in the calculation of absolute bioavailability. Descriptive statistics (mean, S.D., median, minimum, and maximum) were determined to summarize the data for each pharmacokinetic parameter.
Results and Discussion
Tolerability.
Four healthy male Caucasian subjects aged 52 ± 6 years, with a mean weight of 80 ± 16 kg and a mean height of 174 ± 8 cm, enrolled into and completed the study. Rosiglitazone was generally well tolerated during administration. A total of six adverse events were reported during the study, all of which occurred only once, and none were considered likely to be drug-related. There were no clinically important changes in ECG, laboratory parameters, blood pressure, or heart rate. One subject did not complete the study due to a falling hemoglobin concentration, which was most likely due to the intensive blood sampling schedule.
Plasma Concentration-Time Profiles.
Rosiglitazone was rapidly cleared from plasma in all subjects, being quantifiable only up to 24 h after dosing. Plasma concentration-time data obtained after both oral and i.v. administration displayed an overall monoexponential decline from peak values (Fig. 1). Pharmacokinetic parameter values are shown in Table 1. On the other hand, plasma concentrations of radioactivity changed little between Cmax (ca. 6 h) and 24 h after dosing, and then displayed an overall biexponential decline (Fig.1), being measurable for the entire 21-day sampling period. Thus, systemic exposure to rosiglitazone, after both oral and i.v. administration, was a small percentage (<5%) of the exposure to all drug-related materials (total radioactivity).
Mass Balance.
The overall excretion of radioactivity in urine and feces is summarized in Table 2, and mean cumulative excretion data are shown in Fig. 2. Elimination of radioactivity was similar after either route of administration, and was predominantly into the urine. There were two known instances of significant noncompliance with urine and fecal collections. Samples for days 9 to 17 were not collected for subject 3 who, after oral dosing, unexpectedly traveled abroad for 1 week. After i.v. administration, subject 4 mistakenly thought he had completed the study and the day 7 samples were not collected. Estimation of the losses, by comparison with the excretion patterns of the other volunteers over the relevant periods, suggested that overall recovery for subject 4 (i.v.) should have been ca. 3% higher and for subject 3 (oral) ca. 8% higher, which would have given an acceptable mass balance. There was, however, a very poor mass balance for subject 2 after both doses (10% lower than the mean for the other three subjects after i.v. dosing, 24% lower after oral administration). No errors could be detected in either the dose estimation or the radioassay results for urine and feces, thus giving no reason for the poor recoveries other than the possibility of consistent noncompliance in sample collection.
The most notable feature of the excretion of radioactivity in humans was the prolonged period over which it occurred, in comparison to the rate of excretion in the rat and dog. In the preclinical species, approximately 90% of the excreted radioactivity was recovered within 48 h of dosing (Bolton et al., 1996). In humans, however, only about 35% of the administered radioactivity was recovered in 48 h and it took over a week to recover 90% of the excreted radioactivity. This slow excretion of radioactivity mirrored the slow elimination from the systemic circulation.
This prolonged elimination of radioactivity could be a consequence of the extremely high plasma protein binding of M10 (SB-332650), the principal drug-related component in the circulation after 24 h, and the dominant component in urine on day 8 (see below). Binding in human plasma was measured using ultrafiltration and an LC-tandem mass spectrometric assay (detection limit of 0.5 ng/ml) and found to be in excess of 99.99%, compared with 99.72 ± 0.04% for rosiglitazone (A-M. Harris, S. E. Fowles, and P. B. East, unpublished data). Other mechanisms, e.g., enterohepatic recycling, may play contributory roles. In the rat, biliary secretion of the para-O-sulfates M4 and M10 was substantial, suggesting the possibility of enterohepatic recycling, but excretion of radioactivity was essentially complete within 48 h in both the intact dog (Bolton et al., 1996) and intact rat (S. M. Wheeler, A. W. Harrell, P. J. Cox, R. J. Chenery, and J. P. Keogh, unpublished data). This excretion pattern was consistent with the rapid clearance of radioactivity from the plasma of the preclinical species. Thus it cannot be assumed that biliary recycling is responsible for the prolonged elimination of rosiglitazone metabolites in humans.
Characterization of Human Urinary, Fecal, and Plasma Radiometabolites.
Molecular ion and retention time data for metabolites and authentic reference compounds (rosiglitazone, SB-243914, SB-244675, and SB-237216) are given in Table3.
Urine.
Analysis of 12–24 h, 0–8 day pooled, and day 8 urine after either the oral or i.v. dose showed that at least 15 radiometabolites were detected by radio-HPLC analysis (Fig. 3). The 0–8 day pooled urine radiochromatograms showed very little variation between dose route or subject. Table4 shows that metabolites M10 and M4, bothpara-hydroxylated sulfate conjugates, together accounted for approximately 35% of the dose excreted over 8 days. All other metabolites individually accounted for <4% of the dose (Table 4).
Plasma.
Analysis of 1, 4, 8, and 24 h as well as day 4 plasma extracts, after either the oral or i.v. dose, showed that at least four metabolites and parent were present. Rosiglitazone was the predominant plasma component at 1 h postdose. Peaks for M10 and M12 could also be observed at this time point (Table5). Plasma from i.v. dosed subjects in comparison to orally dosed subjects showed lower proportions of metabolites compared with rosiglitazone at 1 h after dosing, as expected. From 4 h onwards, oral and i.v. plasma extract radiochromatograms were very similar, with M10 as the predominant component and M4 and M12 also at significant levels. From 24 h onwards, plasma radioactivity was due almost exclusively to M4 and M10, and therefore the decline in radioactivity concentrations may be used to calculate an approximate half-life for these para-hydroxy sulfate metabolites of about 5 days. The persistence in the systemic circulation of M10, M4, and, for 24 h, of theN-desmethyl metabolite M12, suggests that these slowly cleared metabolites are likely to accumulate on repeated daily dosing of rosiglitazone to humans.
The potential contribution of some of these metabolites to the pharmacodynamic activity of rosiglitazone has been studied by comparing their abilities to activate proliferated-activated receptors gamma (PPARγ) in a cell-based gene transcription assay (Lehmann et al., 1995). The phase I metabolites, N-desmethyl-rosiglitazone (M12, SB-237216) and unconjugated para-hydroxyrosiglitazone (M13, SB-275286, not detected in plasma) were 20-fold less potent than and equipotent with rosiglitazone, respectively. However, the conjugated and highly protein-bound plasma metabolite, rosiglitazone-para-O-sulfate (M10, SB-332650), was 55-fold less potent than rosiglitazone in this assay (S.A. Smith, personal communication). Thus, although theoretically possible, it seems unlikely that the slowly cleared plasma metabolites contribute significantly to the pharmacodynamic activity of rosiglitazone.
Feces.
Analysis of extracts of feces collected during days 1 or 2 and 6 or 7, together with pooled day 0 to 7 feces extracts, after either the oral or i.v. dose, gave similar profiles and showed that at least four radiometabolites were present. Metabolites M7 and M13, bothpara-hydroxylated metabolites, were predominant (Table 4).
Total Percentage of Dose to Which Structures Were Assigned.
Approximately 62 and 65% of the administered dose were unambiguously assigned structures after oral administration of 8 mg of [14C]rosiglitazone and i.v. administration of 2 mg of [14C]rosiglitazone, respectively. Tentative assignments were made for another 6% by both dosing routes. Urine and feces samples collected from about day 8 on contained insufficient radioactivity to allow radiochromatographic analysis. These samples accounted for 11 and 14% of the dose (oral and i.v., respectively). On the assumption that the metabolic profiles in urine and feces collected from about day 8 onward were similar to the profiles obtained for day 7 or 8 postdose, the overall percentage of the dose to which structures could be assigned is 79 and 86% for oral and i.v. doses, respectively. The remaining radioactivity was present as components that were too small to be distinguished above background radioactivity.
Metabolic Scheme.
A proposed scheme for the metabolism of rosiglitazone in humans, based on the results of this study, is presented in Fig.4 and is closely similar to that proposed for rat and dog (Bolton et al., 1996). The major routes of metabolism in humans were N-demethylation, hydroxylation, and subsequent conjugation, and these were unaffected by the route of dose administration. Approximately 44% of the metabolites wereN-demethylated, and about 76% had been hydroxylated on the pyridinyl ring, with or without N-demethylation. The ratio of ortho- to para-hydroxylation was approximately 1 to 9. The hydroxylated products were extensively conjugated, accounting for about two-thirds of the total metabolites. The ratio of sulfation to glucuronidation was approximately 6 to 1. It should be noted that only unconjugated metabolites were detected in the feces. It is likely that these were first secreted into the bile as conjugates and then hydrolyzed by intestinal microflora. Thus it can be estimated that 90% of the phase I metabolites of rosiglitazone are initially conjugated. Unlike the preclinical species, cleavage of the molecule to give a phenoxyacetic acid metabolite (M1) was a minor route of elimination in humans, accounting for less than 4% of the dose.
Although there were differences between species in the persistence of the circulating metabolites of rosiglitazone (measured as total radioactivity), its principal metabolites were accurately predicted from preclinical studies.
Acknowledgments
We thank K.M. Lawrie for the synthesis of [14C]rosiglitazone; P. J. Cummings and C. L. Lowi for formulating the radiolabeled doses; W. Guiney, J.P. Keogh, and R. McCarthy for expert assistance with sample analysis; and E. Minthorn for the pharmacokinetic analysis.
Footnotes
-
Send reprint requests to: Dr. Peter J. Cox, SmithKline Beecham Pharmaceuticals, The Frythe, Welwyn, Herts. AL6 9AR, UK. E-mail: Peter_J_Cox{at}sbphrd.com
-
↵2 Nomenclature of metabolites. During preclinical metabolism studies (Bolton et al., 1996), metabolite structure names were not based on the IUPAC numbering system. Thus, metabolites previously designated as 3-hydroxy and 5-hydroxy should more correctly, under the IUPAC system, be designated 5-hydroxy and 3-hydroxy, respectively, a source of potential confusion. In this study, metabolites containing the IUPAC substructure 2-aminopyridinyl-5-hydroxy are referred to aspara-hydroxy metabolites, reflecting hydroxylation at a position para to the amino side chain. Similarly, metabolites containing the pyridinyl-3-hydroxy substructure are referred to as ortho-hydroxy, reflecting hydroxylation at a position ortho to the amino side chain.
- Abbreviations used are::
- LC-MS
- liquid chromatography-mass spectrometry
- LSC
- liquid scintillation counting
- SPE
- solid-phase extraction
- Cmax
- maximum observed concentration
- Tmax
- the time to reach Cmax
- AUC0-inf
- area under the plasma concentration-time curve from time zero to infinity
- Received July 27, 1999.
- Accepted April 10, 2000.
- The American Society for Pharmacology and Experimental Therapeutics