Abstract
Mass balance and metabolism of formoterol were investigated in six healthy men in an open study. Mean age was 49.7 years (range: 40–63). Simultaneous oral (mean dose 88.6 nmol, 49.3 MBq) and i.v. (mean dose 38.2 nmol, 21.4 MBq) doses of tritium-labeled formoterol were administered. The combination of these two administrations was aimed at simulating the fate of inhaled formoterol. Total radioactivity was monitored for 24 h in blood plasma and for at least 4 days in urine and feces. Formoterol and metabolites were determined using liquid chromatography plus radiodetection, directly after centrifugation in urine and after sample workup in blood plasma and feces. Metabolites were identified in urine, sampled from two subjects, using liquid chromatography-electrospray ionization mass spectrometry. Mean total recovery was 86% of the administered formoterol dose, 62% in urine and 24% in feces. Tritiated water was generated and because its in vivo turnover is slow, the terminal decline of total radioactivity was slow and dose recovery was incomplete during the sampling period. Formoterol was conjugated to inactive glucuronides and a previously unidentified sulfate. The phenol glucuronide of formoterol was the main metabolite in urine. Formoterol was also O-demethylated and deformylated. Plasma exposure to these pharmacologically active metabolites was low.O-demethylated formoterol was seen mainly as inactive glucuronide conjugates and deformylated formoterol only as an inactive sulfate conjugate. Intact formoterol and O-demethylated formoterol dominated recovery in feces. Mean recovery of unidentified metabolites was 7.0% in urine and 2.0% in feces.
Formoterol (Fig. 1) is a bronchodilator with prolonged duration of effect after inhalation. Formoterol fumarate dihydrate is used clinically as a racemic (RR, SS) mixture. The dose inhaled via Turbuhaler is 6 to 24 μg (corresponding to a delivered dose of 4.5–18 μg) twice daily.
There are several likely molecular sites for biotransformation of formoterol, e.g., O-demethylation of the methoxyphenyl group and hydrolysis of the formamide group (Gibson and Skett, 1994b). Conjugation of formoterol could be expected, too (Sasaki et al., 1982). The short-acting β2-agonists terbutaline, salbutamol, and fenoterol form phenol sulfates as their major metabolites in humans (Nilsson et al., 1972; Lin et al., 1977;Hildebrandt et al., 1994). However, these and other drugs with similar structures have been shown to conjugate also with glucuronic acid in humans (Shimizu et al., 1978;Tegnér, 1984; Hildebrandt et al., 1994). A glucuronide of formoterol has been found in humans after oral administration (Kamimura et al., 1982; Tasaka, 1986). Pharmacological inactivation of formoterol by sulfate conjugation has not been seen (Kamimura et al., 1982).
The objective of this study was to investigate the mass balance and the in vivo metabolism of formoterol when presented to humans via the lungs. However, inhalation of radiolabeled formoterol was not considered pharmaceutically feasible, because of the difficulties associated with the development of a new radiolabeled formulation for inhalation. Therefore, combined i.v. and oral dosing was applied to simulate the fate of inhaled formoterol. Pulmonary bioavailability of a drug after inhalation is up to 30% (Pauwels et al., 1997). In an attempt to mimic systemic uptake of inhaled formoterol and on the assumption that inhaled formoterol reaches the system intact via the lungs, 30% of the dose was infused. The remaining 70% was given orally. A supratherapeutic dose of 54 μg was chosen to facilitate bioanalysis. Total radioactivity, formoterol, and metabolites of formoterol were determined in plasma, urine, and feces after administration to healthy men.
Materials and Methods
The study was performed in accordance with the declaration of Helsinki and approved by local Ethics and Isotope Committees in Lund, Sweden. Subjects gave informed consent before enrollment.
Subjects.
Six healthy men with normal medical history, blood pressure, pulse, ECG signs, and clinical laboratory values participated. Demographic data are given in Table 1.
Test Formulations and Chemicals.
Racemic (RR, SS) mixtures of formoterol and [3H]formoterol (Fig. 1) were synthesized as fumarate dihydrates at AstraZeneca Pharmaceutical Productions, Södertälje and AstraZeneca R&D, Lund, Sweden, respectively.
Buffered (pH 6, citric acid and sodium phosphate) physiological 0.9% NaCl solutions of [3H]formoterol and nonlabeled formoterol for i.v. and oral administration (1.08 μg formoterol fumarate dihydrate/ml corresponding to 1.4 MBq/ml and 2.57 nmol/ml) were manufactured and dispensed at the Pharmaceutical Department, AstraZeneca R&D, Lund, Sweden. The maximum dose (i.v. + oral) was 54 μg (72 MBq). Radiochemical purity was 99%. Actual i.v.-administered formoterol solution was calculated by weighing the syringe before and after administration by use of a density value of 1.005 g/ml. The i.v. dose was corrected for 0.4% adsorption in the tubing used for the infusion. The outcome of the drug administrations is given in Table 1.
Reference compounds, D2536 for O-demethylated formoterol (Met1),1 D2537 for deformylated formoterol (Met2), and D2522 for racemic formoterol as such, were synthesized at AstraZeneca R&D, Lund, Sweden. All other chemicals were of analytical grade.
Clinical Procedures.
Subjects came fasted to the clinic in the morning, about 30 min before drug administration started (zero time: 8:00 AM ± 1 h). After bladder emptying and defecation (if possible), the subjects had indwelling catheters inserted into antecubital veins on both arms. One catheter was used for i.v. drug infusion, the other for blood sampling. Fasting (no food or beverages) continued with breaks for lunch (3 h) and a snack (8 h). Subjects stayed overnight at the clinic. Alcohol intake was not allowed and strenuous activity or physical exercise was avoided for at least 48 h before and during assessments in conjunction with drug administration.
The subjects drank the oral solution of formoterol, and the bottle in which the solution was stored was rinsed 4 times with 50 ml of tap water. The rinsing water was swallowed.
First order pulmonary input was simulated by use of exponentially declining rates of infusion for 30 min (eq. 1) using an infusion pump (IVAC P4000; Alaris Medical Systems, Basingstoke, UK).
Plasma samples were prepared from 10-ml blood portions sampled in tubes with anticoagulant, before and 0.083, 0.165, 0.33, 0.66, 1,1.5, 2, 3, 4, 6, 12, and 24 h after start of drug administration. The collected blood was centrifuged and the supernatant plasma transferred to cryotubes to which citric acid (final concentration 50 mM) was added to avoid enzymatically mediated hydrolysis of formoterol and conjugates. The bladder was emptied before start of drug administration. A blank urine sample was saved. Subsequently voided urine was collected quantitatively in tared polyethylene bottles (0–2 h, 2–4 h, 4–6 h, 6–12 h, 12-hourly up to 48 h, then daily up to 4 days in all subjects and up to 7 days in subjects 3–6). After weighing, three 10-ml portions of urine from each collection interval and the blank sample were saved. Blank samples of feces were saved if possible and subsequently defecated matter was collected quantitatively in canisters (daily up to 4 days in all subjects and up to 7 days in subjects 3–6). All urine and feces samples were stored refrigerated or frozen during the collection period.
Adverse events, vital signs (pulse and blood pressure), and 12-lead ECGs were recorded before and after drug administration.
Determination of Total Radioactivity.
Plasma (0.1–0.2 g) was added to 10 ml Ultima Gold (Packard Instrument Co., Downers Grove, IL). Total radioactivity was determined in plasma as such, as well as in plasma that had been dried for two days in a fume cupboard. The dried plasma residue was redissolved in water before the total radioactivity was determined.
Feces were homogenized in about 2 parts (w/w) of 0.1 mol/liter sodium citrate buffer pH 2.0 with a Polytron PT 3000 (Kinematica, Littau, Switzerland) for 2 min with the sample vial immersed in an ice-water bath. About 0.1 to 0.2 g of the homogenates, as well as of homogenates that had been dried, were combusted in triplicate in an Oxidizer 307 equipped with Oximate 80, a robotics system for sample processing (Packard). The tritiated water (3H2O) generated during combustion was automatically dissolved in 15 ml Monophase-S (Packard) and then its radioactivity was determined by liquid scintillation counting. The recovery of the combustion procedure was tested using weighed amounts of blank feces homogenates spiked with the test solution and found satisfactory (>99%).
Urine (0.1–0.2 g) was added to 10 ml Ultima Gold (Packard). No pretreatment was performed.
All determinations of total radioactivity were performed at least in duplicate by liquid scintillation counting with Tri-Carb Spectrometers (Packard). Quench correction was performed using external standard procedures.
Sample Workup for Determination of the Metabolite Pattern.
Plasma (2.0 g) was deproteinated using 2.0 ml of acetonitrile. After thorough mixing, the sample was centrifuged at 1370g (Wifug Doctor; Wifug, Bradford, UK) for 5 min. The supernatant was collected and the pellet was washed with 2.0 ml of acetonitrile: 0.1 mol/liter ammonium acetate, pH 5 (50:50, v/v) and centrifuged again. The combined supernatants were evaporated to dryness under vacuum (Speed Vac Concentrator A 290; Savant, Farmingdale, NY). The residue was dissolved in 1.5 ml of methanol: 0.1 mol/liter ammonium acetate, pH 5 (10:90, v/v). The solution was transferred into a microtube and centrifuged at 17,900g (Eppendorf 5415 C) for 2 min. About 0.75 ml of the supernatant was injected on a liquid chromatography (LC) system denoted “LC1” (cf. below). One part was saved for determining total radioactivity to calculate the recovery of the sample workup procedure. Any3H2O in the plasma was evaporated during sample workup. Therefore, the calculation of sample workup recovery was based on radioactivity in dried samples.
Urine was centrifuged at 1370g (Wifug Doctor) for 5 min before being analyzed. About 0.1 to 0.2 ml of urine was injected on the LC1 system.
Feces homogenate was treated similar to plasma but the precipitation procedure was scaled down to half the amounts. The dried residue was dissolved in 1.5 ml methanol: 0.1 mol/liter ammonium acetate, pH 5 (10:90, v/v) and centrifuged at 1370g for 5 min. About 0.7 ml of the supernatant was injected on the LC1 system.
LC1-Radiochromatography.
Analysis
The metabolite pattern of formoterol in plasma, urine, and feces was investigated by LC-radiochromatography, “LC1”. The LC equipment consisted of a solvent delivery system, an autosampler, and an absorbance detector (Varian 9010, 9100, and 9050; Varian, Harbor City, CA). A radioactivity detector (Flo-One, A525AX; Radiomatic Instruments & Chemical Co., Tampa, FL) was attached on-line. The chromatography system consisted of a precolumn and an analytical column. Both columns were 3 μm Supelcosil LC-18-DB, the precolumn 20 × 4.6 mm and the analytical column 150 × 4.6 mm (Supelco, Inc., Bellefonte, PA). A linear mobile phase gradient system was used, with a flow rate of 1 ml/min. Mobile phase A was 0.1 mol/liter ammonium acetate, pH 5, and mobile phase B was methanol: 0.1 mol/liter ammonium acetate, pH 5 (80:20, v/v). Initial condition was 88% A and 12% B, which changed to 72% A and 28% B during 30 min and further to 25% A and 75% B during the next 25 min. These conditions were then maintained for 15 min before returning to the initial conditions. The system was equilibrated for 7 min between the runs. The radioactivity detector was equipped with a 2.5-ml flow cell and, after establishing a quench correction curve, it was run in the dpm-mode. The flow rate of the scintillation cocktail was 3 ml/min (Quickszint flow 302; Zinsser Analytic, Ltd., Maidenhead, UK). For regular control of the analytical column and the chromatographic separation of formoterol and possible metabolites, a mixture of reference compounds for formoterol racemate, Met1, and Met2 dissolved in deionized water or authentic blank samples from plasma, urine, and feces was used. Single analyses were performed.
Quality Control Samples.
Authentic human urine (Subject 1, 0–2 h) at a total concentration of 71.7 pmol/g was used.
Authentic human plasma was not available for analytical quality control. Instead, test samples were blank human plasma to which urine from Subject 1 (2–4 h) had been added to total concentrations of 0.12 and 1.2 nM.
Preparation of control feces samples equal to authentic feces samples was not feasible. Testing of the stability of urinary metabolites added to feces homogenate showed that glucuronidated compounds were not stable. Therefore no feces quality control samples were used.
All quality control samples were kept as portions in a freezer (−20°C) for the time covering the duration of storage of the authentic samples in the study. One sample of plasma or urine was analyzed on every occasion that authentic samples were analyzed. The stability of the samples during the residence time in the autosampler in a run set-up was controlled by placing a duplicate of the first sample at the end of each run.
Calculations.
Total radioactivity determined in the samples was converted to total concentration of radioactive compounds by using the relation between radioactivity and formoterol concentration derived from the analysis of the test solution. Metabolite pattern was evaluated from radiochromatograms. Separated peak areas (dpm) were integrated and expressed as the percentage of total dpm in the chromatogram. The concentration of each radioactive compound was then calculated from this percentage times total concentration of radioactive compounds determined separately in urine samples and in the samples of plasma and feces that had been dried.
The LC1 chromatographic front peak in urine samples and the difference between total radioactivity in wet and dried feces samples were assigned recovery of 3H2O. The fate of 3H2O was not further addressed in plasma. Chromatographic deficit was calculated as total radioactivity on the chromatogram minus the sum derived from LC-quantitated peaks.
Identification of Metabolites using LC-Mass Spectrometry (MS) Techniques.
Frozen urine from two subjects was thawed and 2 ml was applied on a solid-phase extraction column (BondElut C18, 3 ml; Varian), conditioned with one volume of methanol and one volume of water. Light pressure was applied to push the volume through until the surface of the sorbent was reached. The filtrate was collected for liquid scintillation counting of radioactivity. The column was then washed with 1 ml of water, which was collected separately. Finally, the compounds of interest were eluted with two 1-ml portions of 95% methanol (aq), which were collected separately and evaporated in a vacuum centrifuge (Savant SC210A) with medium heating. The residue was redissolved in 100 μl of 10% methanol (aq). After centrifugation, 5 to 20 μl was injected on another LC system denoted “LC2”. A methanol gradient in aqueous formic acid was used. This LC system had a selectivity that was different from that of the LC1 system, which caused some chromatographic peaks to change positions. Therefore, to match chromatographic peaks in the different systems, some fractions were reanalyzed in LC1 radiochromatography. Similarly, some peaks from LC1 were collected and reanalyzed with LC2 MS.
Two Pharmacia LKB 2150 pumps (Pharmacia LKB, Bromma, Sweden), controlled by Autochrom software via a CIM114 interface (Autochrom, Milford, MA) and a Valco manual injector (50 μl loop; VICI AG, Schenkon, Switzerland) were used in line with the mass spectrometer. A precolumn (Applied Biosystems Brownlee Spheri-5 RP-8, 5 μm, 30 × 2.1 mm; Applied Biosystems, Inc., Foster City, CA) and an analytical column (YMC Octyl, 5 μm, 100 × 2 mm; YMC Inc., Wilmington, NC) were used with a mobile phase gradient of methanol/water, with 0.5% formic acid. Starting at 18.5% (v/v), the methanol concentration was increased linearly to 22% at 10 min, to 38% at 20 min, and to 95% at 22 min. After 4 min isocratically at 95%, the system was returned to initial values. The flow rate was 0.20 ml/min. During the first 3 min, including the chromatographic front, the flow was diverted and not let into the MS system. In some experiments, the LC eluent was directed to a fraction collector (Gilson 202; Gilson, Villiers LeBel, France). Four fractions per minute were collected during 20 min and the radioactivity in the fractions was determined by liquid scintillation counting.
A Finnigan TSQ700 mass spectrometer (Finnigan, San Jose, CA) was used, with a Finnigan electrospray ionization (ESI) interface for the LC system. Instrument control and data processing were performed on a DECstation 5000/125 with Finnigan ICL and ICIS II software (versions 7.0 and 7.4). Interface variables: Spray voltage 4 kV, sheath gas (N2) 90 psi, auxiliary gas (N2) 5 U, capillary temperature 190°C. For tuning and calibration of the quadrupoles, cluster ions ((HCOONa)nNa+) from injected 10 mM sodium salt solutions were used. The capillary and tube lens voltages of the ESI interface were optimized using formoterol racemate. Full scan positive ion mass spectra were recorded (within the mass range 200–600 in 1 s) scanning quadropole 1. MS/MS was performed with Ar gas (2 mtorr, 0.27 Pa) in the collision cell and collision energies of 15 to 30 eV. Selected reaction monitoring (SRM) of characteristic precursor and product ions was performed to detect and characterize the metabolites, using various ICL procedures. Generally, the expected proton adduct of the molecule (MH+) was selected as precursor ion and m/z 149 (compounds with an intact methoxy group) or m/z135 (O-demethylated compounds) were chosen as characteristic product ions of collision-induced fragmentation. Selected reaction profiles (with 3-point smoothing) and background-subtracted averaged mass spectra are reported. Chromatographic retention times and mass spectral properties of unknown moieties were compared with the analytical outcome using the described reference compounds dissolved in water. To obtain the 3H/1H ratio in the drug solution, the batch used for infusion was analyzed with LC-MS.
Pharmacokinetic Evaluation.
Urine and feces recoveries and plasma concentrations were used to estimate standard pharmacokinetic parameters (cf. abbreviations list) of total radioactivity and, when feasible, of formoterol and its metabolites. Evaluations of recoveries in urine and feces and concentrations in plasma have been based on observed data without extrapolation. Concentrations below 0.01 nM, particularly in plasma, were used although they had low precision and sometimes were biased (cf. Performance of LC1-Radiochromatography below). Not using such data would render the evaluation less apt for interpretation of the clinical pharmacokinetic implications of the study, especially with respect to plasma exposure to and renal clearance of metabolites. The terminal elimination rate constant of formoterol and its metabolites was calculated as ln(urinary excretion rate) regressed on midpoints of urine collection intervals.T½ was defined as ln(2) divided by the terminal elimination rate constant.
Results
Clinical Performance and Handling of Protocol Deviations.
The largest blood sampling time deviation was 3 min. Endpoints of urine collections deviated less than 3% up to 12 h and less than 6% after that, up to the end of collection. Feces collection times varied considerably between subjects. By accident, subject 3 contaminated feces with urine at 334 min after start of infusion. Radioactivity of metabolites normally not found in feces was subtracted and added to data for the urine collection covering the corresponding interval, whereas other radioactivity was considered to be of fecal origin.
Pulse rate, systolic blood pressure, and the heart rate corrected Q-Tc interval Q-Tc (Bazett, 1920) increased and diastolic blood pressure decreased after administration of formoterol. None of the changes were considered clinically relevant. Four subjects reported palpitations after drug administration (duration 5–28 min) and one of them was subject 6, who had a common cold. Reported adverse events were all mild.
Tritium Labeling of Formoterol.
The intention was to synthesize formoterol with tritium bound to the carbon in position 3 of the 1-formamide-2-hydroxy-phenyl part of formoterol (Fig. 1). However, a complementary NMR analysis revealed that 20% of the labeled molecules had an extra tritium bound to the carbonyl carbon of the formamide group. Bioanalytical evaluation was performed on the assumption that formoterol was monolabeled. This assumption does not affect the outcome of the analysis with respect to formoterol and metabolites with an intact formamide group, but deformylated metabolites are underestimated by approximately 20%.
Mass Balance.
Mean recovery was 86% of administered dose, 62% in urine and 24% in feces (Fig. 2). Only 70% of the dose was recovered from subject 3. His feces recovery did not deviate from the other subjects but urinary recovery was 20 percentage units lower than mean recovery. Except for the first interval, subject 3 always had the lowest recovery in urine. Probably, urine was not collected quantitatively by him. Linear interpolation of ln (urinary excretion rate) versus time was used to reassess recoveries from the two collections most likely to be incomplete, i.e., 2 to 4 and 12 to 24 h, and suggested that the dose recovery of subject 3 might have been underestimated by at least 10 percentage units.
Mean residual percentage of the dose in the form of3H2O remaining in the body was about 10%. In this calculation it was assumed that water accounts for 2/3 of the body weight and that3H2O was equally distributed in the body during the last sampling period.
Performance of LC1 Radiochromatography.
The recovery of the sample workup for analysis of authentic plasma samples was 89.6 ± 9.9% (n = 78) and for the feces samples 82.7± 2.2% (n = 22). The recoveries of quality control plasma samples were 87.7 ± 3.8 (n= 3) and 85.6 ± 1.8 (n = 10) at 0.12 and 1.2 nM, respectively.
The LC1 method was a compromise between selectivity of the chromatographic system and a reasonable analysis time. For example, the benzyl glucuronide of formoterol (metabolite FG2) was not separated at baseline from formoterol and the sulfate of formoterol (FS). Representative examples of LC1 radiochromatograms of plasma, urine, and feces are given in Fig. 3. Analysis of the quality control samples of plasma and urine showed that degradation of the compounds during storage or during the residence time in the autosampler did not occur. In plasma, which in general showed the lowest concentrations, the bioanalytical coefficient of variation for formoterol and metabolites was 99% at 0.001 pmol/g, 48% at 0.005 pmol/g, 22% at 0.007 pmol/g, and <6.4% at ≥0.08 pmol/g. Accuracy was concentration dependent; because of unfavorable backgrounds, low plasma concentrations were generally slightly overestimated. Limit of quantitation was set to 0.01 nM. Individual plasma concentrations of some formoterol metabolites were often [FS, glucuronide 1 of O-demethylated formoterol (Met1G1), and the sulfate of deformylated formoterol (Met2S)] or always (Met1) below this limit.
Identification of Formoterol Metabolites.
From the mass spectra, the3H/1H ratio in the MH+ ions (m/z 347 and 345) of formoterol in the drug solution was found to be approximately 1/1.7. This relation between labeled and unlabeled compound should be displayed also in metabolites and was used to show that the metabolites originated from the administered drug mixture. Doubly tritiated formoterol (m/z 349) was present, confirming the observation of this analog in the NMR analysis of the test compound.
In LC separation, four major radioactive peaks in urine eluted in the same order when a methanol gradient in either formic acid (LC2) or ammonium acetate buffer (LC1) was used as mobile phase; these peaks were the glucuronides of O-demethylated formoterol (Met1G1 and Met1G2), a phenol glucuronide of formoterol (FG1), and formoterol itself, all previously identified in vitro using human liver microsomes (G.H., personal communication, 1998). Because the urine extracts contained large amounts of innumerable matrix components, full spectrum MS data were not very informative and MS/MS was mainly used. A product ion spectrum of FG1, identical with that of the corresponding compound generated in vitro, is shown in Fig.4. SRM of dissociations occurring in unlabeled and labeled Met1G2, FG1, and formoterol are shown in Figs.5 and 6. Some additional dissociation was monitored for each compound (not shown) to further strengthen the assignment of the structures.
One additional peak appeared in the SRM profiles indicative of FG2, a new glucuronide of formoterol, isomeric to the previously found FG1 (Fig. 6). FG2 and FG1 have the molecular mass and several fragments in common, but FG2 did not fragment by loss of H2O from the protonated molecule (Fig.7), which FG1 did. This difference can be explained if the glucuronic acid position in FG2 is on the benzylic OH, assuming the phenolic position for FG1 (Figs. 4 and 7).
Attempts were made using LC2 MS to find Met1 and Met2, but their concentrations were too low to be verified by MS. Retention time of unlabeled reference substance of Met1 was used to assign the radioactive peak of Met1 (3 in Fig. 3). A small radiochromatographic peak was found with a shorter retention time than Met1G2, earlier identified as another glucuronide of Met1 (G.H., personal communication, 1998). By virtue of its shorter retention time it is denoted Met1G1. Met1G1 and Met1G2 fragmented partly by loss of water from the protonated molecule, suggesting that conjugation took place on either of the two aromatic rings.
A second new peak appeared in the SRM profiles intended to show formoterol. It was shown to be a sulfate of formoterol (FS), which partly decomposed to formoterol in the LC-MS interface. When SRM procedures were designed with protonated formoterol sulfate as the precursor ion, its relation to formoterol was verified (Fig.8). The proposed position of the sulfate is on the phenol group of formoterol.
Another new peak appeared in the SRM profiles intended to show Met2, but with a retention time different from that observed with the reference compound. Also in this case it could be shown that it was a sulfate (Met2S) which partly decomposed in the LC-MS interface to the unconjugated structure of Met2. In analogy with FS, the protonated Met2-sulfate ion (m/z 397) could be selected and fragmented in the mass spectrometer to the structures of Met2 (m/z 317) and its characteristic fragments (e.g.,m/z 299; Fig. 9). From these data alone we do not know whether the phenolic hydroxyl or the aromatic amine was sulfated, but the former structure is tentatively proposed. Metabolic pathways of formoterol are summarized in Fig. 10.
Recovery of Formoterol and Metabolites.
Urine
FG1 (the main metabolite), formoterol as such, one of the two glucuronides of O-demethylated formoterol (Met1G2) and formoterol glucuronidated on its benzylic hydroxyl group (FG2) constituted quantitatively the major part of urinary recovery, on average 48.1% of the dose (Fig. 11). The mean urinary recovery of FS plus Met2S was 4.8% and Met1G1 constituted only 2.0% of the dose. Unconjugated Met1 was about 1.0% of the dose. Met2 was not found unconjugated in urine. Formoterol constituted on average 30% of the urinary recovery of formoterol plus formoterol conjugates (F+FG1+FG2+FS), whereas Met1 constituted on average 8% of its total urinary recovery (Met1+Met1G1+Met1G2).3H2O and quantitated but unidentified metabolites (XTOT) accounted for 1.9 and 7.0% of the dose, respectively. The chromatographic deficit was 1.6% (individually 2.5% at the most).
Feces.
Formoterol and Met1 dominated quantitatively the dried feces recovery, on average 21% of the dose, whereas conjugates were scarce. The recovery of formoterol was about 3 times the recovery of Met1. The mean recoveries of 3H2O and XTOT were 0.1 and 2%, respectively. The chromatographic deficit was about 1% (individually 1.7% at the most).
Plasma Exposure to Formoterol and Metabolites.
The estimates of area under the curve of plasma concentration versus time up to 24 h (AUC0–24 h) for formoterol, FG1, FG2, and Met1G2 were based to at least 64% on plasma concentrations above the limit of quantitation, 0.01 nM, whereas this part of AUC0–24 h was smaller and therefore rendered the estimates for FS, Met1, Met1G1, and Met2S less reliable (Table 2).
The gross metabolic pattern in urine quantitatively reflected plasma exposure, which, accordingly, was dominated by the FG1, formoterol as such, one of the two glucuronides of O-demethylated formoterol (Met1G2) and formoterol glucuronidated on its benzylic hydroxyl group (Fig. 12). On average, these substances constituted 74.8% of the total AUC0–24 h. In contrast with urinary recovery, however, the mean exposure to Met1G1 (5.7%) was more pronounced than the exposure to sulfate conjugates of formoterol and Met2 (5.0%). The mean AUC0–24 h of unconjugated Met1 was only 0.53%. Unconjugated Met2 was not detected at all in plasma. Formoterol constituted 31% of the total plasma AUC0–24 hof formoterol plus formoterol conjugates (F+FG1+FG2+FS), which is close to the corresponding share of urinary recovery (30%). However, Met1 constituted only 2% of its total AUC0–24 h(Met1+Met1G1+Met1G2), which is lower than the corresponding share of urinary recovery (8%). Urinary recovery overestimated the plasma exposure to Met1 and sulfated Met2, but correlated well with AUC0–24 h of formoterol and other metabolites. The mean integrated chromatographic deficit (100% minus mean AUC0–24 h percentages of quantitated substances) in plasma was 4.0% (individually 6.4% at the most).
Pharmacokinetic Parameters (Table 2).
Mean plasma Cmax of total radioactivity (0.971 nM) was observed 40 to 80 min after start of infusion. MeanCmax of formoterol was 0.566 nM and was normally observed 5 min after start of infusion. The peaks of metabolites were lower and appeared later than the peak of formoterol.
The terminal decline of metabolites paralleled the decline of formoterol, as illustrated most clearly in plots of urinary excretion rates versus time (Fig. 13), where it is also made clear that generated3H2O was eliminated slowly. Mean terminal T1/2 of formoterol was 17 h; individual values varied about 3-fold. Mean terminalT1/2 values of identified formoterol metabolites were 11 to 16 h. The terminalT1/2 of total radioactivity was quite long (16–80 h) because of the slow elimination of3H2O. The decline of Met2S tended to be faster during the first 6 h after dosing compared with the terminal phase of elimination (Fig.14). Because the plasma concentration of Met1 generally was low, its renal clearance could not be assessed. The mean value for Met2S was 415 ml/min and the means for formoterol and other metabolites were 224 ml/min or less.
Discussion
Mass Balance and Decline of Total Radioactivity.
Mean total recovery was 86% of administered dose, 62% in urine and 24% in feces. 3H2O was formed from the labeled fraction of administered formoterol. The terminal T1/2 of body water has been estimated to 9.46 days (Richmond et al., 1962). This slow elimination was well illustrated in this study with the decline of urinary excretion rate of 3H2O (cf. Fig. 13). A consequence of the slow turnover of3H2O was that the terminal decline of total radioactivity was slow, too. Thus, radioactivity not accounted for could probably be explained by the fact that3H2O was generated in the body and retained beyond the sampling period.
In Vivo Aspects of Metabolism of Formoterol.
Formoterol, formoterol glucuronides, O-demethylated formoterol, and glucuronides of this metabolite were detected. Glucuronidation of formoterol occurred mainly at the phenolic position, but also the benzylic OH position was conjugated. This is similar to the adrenoceptor antagonist labetalol, where monoglucuronides were found at the benzylic and the phenolic positions (Niemeijer et al., 1991; Alton et al., 1994). The glucuronic acid conjugation in the benzylic OH position should further reduce the pharmacological activity of formoterol (Weiner, 1985). The β2-agonists ritodrine and ractopamine, compounds which are chemically similar to Met1, were conjugated with glucuronic acid (and sulfate) at either of their two phenolic positions (Brashear et al., 1990; Smith et al., 1995). Results from the present study suggested that Met1, which has one benzylic and two phenolic OH positions, preferably was glucuronidated at one of the two phenolic OH positions. Sulfate conjugates have previously not been detected after administration of formoterol (Kamimura et al., 1982). This study, however, showed that formoterol was sulfated to some extent and that Met2 was only found as a sulfate conjugate (Met2S).
After inhalation via a pressurized metered dose inhaler, urinary recovery of formoterol and glucuronides has been estimated to be 24% of the dose, using a method in which glucuronides were cleaved with β-glucuronidase before determination (Firkusny et al., 1990). After LC separation of the metabolites in this study, the mean recovery in urine of formoterol and its glucuronides was estimated to be 35%. Possibly, the infused fraction of the dose was larger than the fraction that reached the system intact via the lungs after the mentioned inhalation (Firkusny et al., 1990). Incomplete cleavage of glucuronides or a higher limit of quantitation of unlabeled formoterol in the previous study could also explain the difference.
FG1 was the main metabolite in urine, but formoterol and glucuronides of O-demethylated formoterol were recovered there too. Met1 has two phenolic hydroxyl sites available for conjugation, whereas only one is available at intact formoterol. It is therefore probably more susceptible to conjugation than formoterol. This might explain why only 8% of Met1 {100 × Met1/(Met1+Met1G1+Met1G2)} but 30% of formoterol {100 × F/(F+FG1+FG2+FS)} were found unconjugated in urine. Urinary recovery was related to AUC.
Little is known about dose recovery in human feces after inhalation of formoterol. Part of the oral formoterol dose might pass through the gut without being absorbed, or absorbed formoterol could be secreted back into the gut via the bile and the gut wall. Excretion into the gut via the bile may be important when the molecular mass is >300. Tasaka (1986) has suggested that formoterol might be excreted via the bile, which is reasonable because molecular masses of formoterol and its main metabolites, including conjugates, are >300.
Formoterol and Met1, but no conjugates, were found in feces. Stability tests indicated that several formoterol metabolites were unstable in feces (cf. Materials and Methods). Deconjugation may take place in the intestine (Gibson and Skett, 1994c), which might explain the difference in proportions of formoterol and metabolites in feces compared with urine and plasma. Both in urine and in feces, total recovery of formoterol and formoterol conjugates (F+FG1+FG2+FS) was approximately 3 times higher than total recovery of Met1 and its conjugates (Met1+Met1G1+Met1G2). Assuming that intestinal secretion rates of formoterol, Met1, conjugates of formoterol, and conjugates of Met1 are similar, this could indicate that formoterol quantitatively entered the system. Otherwise, relatively more formoterol should have been found in feces, because nonabsorbed formoterol would escape systemic biotransformation.
Clinical Pharmacokinetics.
Peak plasma concentration of formoterol appears within 15 min after inhalation (Lecaillon et al., 1997). In this study, formoterol normally peaked at the first assessment point (5 min after start of infusion), but further investigations are required to evaluate the relevance of the first order scheme of infusion rate decline to simulate first order absorption in the lungs. Total radioactivity in plasma peaked later than formoterol. This might have been an expression of oral absorption and presystemic metabolism, e.g., glucuronidation of formoterol (cf. Fig. 14). One case of combined presystemic phase I and phase II metabolism might be the formation of Met2S. Because unconjugated Met2 was not found in plasma, urine, or feces, one possibility is that both deformylation and sulfation of formoterol took place during absorption and first pass of the gut wall and liver. Plasma exposure to FS, Met1, Met1G1, and Met2S should be cautiously interpreted because substantial parts, particularly for Met1, were based on plasma concentrations below 0.01 nM, the limit of quantitation.
Formoterol and its equipotent O-demethylated metabolite (Jacob Sjöberg, personal communication, 1998) were recovered mainly as their glucuronic acid conjugates. The finding of low systemic plasma exposure (AUC0–24 h) to formoterol and its O-demethylated metabolite was encouraging because effects of inhaled formoterol should occur mainly in the lungs.
The capacity to glucuronidate is normally high (Gibson and Skett, 1994a). Thus, reduced capacity of O-demethylation might be metabolically compensated for, so that relatively more of the formoterol glucuronides are generated. The share of unconjugated formoterol in urine and plasma was higher than the share of unconjugated Met1, suggesting that the metabolite was more efficiently eliminated via conjugation than was the parent drug. Thus, the consequence could be increased effect if the capacity toO-demethylate formoterol was reduced.
Deformylation of formoterol is likely to be mediated by liver amidases (Gibson and Skett, 1994b). Furthermore, unspecific hydrolysis mediated by plasma cholinesterase cannot be excluded; therefore, pH was lowered by citric acid in plasma samples. A minor fraction of the administered formoterol was found as a sulfate conjugate of deformylated formoterol. Met2, retaining 10% of the pharmacological effect of formoterol (Jacob Sjöberg, personal communication, 1998), was not detected, possibly because sulfation took place before deformylation of formoterol.
The terminal T1/2 of inhaled formoterol has been estimated at about 10 h in healthy subjects (Lecaillon et al., 1997). The mean T1/2 in the present study was longer. Visually, elimination rates of metabolites looked similar to the elimination rate of formoterol, suggesting that metabolite elimination was rate-limited by the formation from formoterol. However, the calculated mean terminalT1/2 of formoterol was longer than the terminal T1/2 values of its main metabolites. Elimination of formoterol metabolites was probably determined by composite terminal T1/2values, on the one hand by short T1/2values of the metabolites formed during first-pass from the gut and on the other by the T1/2 of formoterol itself. This might explain why the initial decline of Met2S was faster than the terminal decline: a burst of this metabolite was probably generated presystemically. The fast initial decline might reflect a true elimination rate of Met2S whereas the terminal decline was prolonged, probably because elimination at that time was grossly rate-limited by formation from formoterol or the sulfate conjugate of formoterol (cf. above).
Conclusions.
Mean total recovery was 86% of the administered formoterol dose, 62% in urine and 24% in feces.3H2O was generated and because its in vivo turnover is slow, the terminal decline of total radioactivity was slow and dose recovery was incomplete during the sampling period. Formoterol was conjugated to inactive glucuronides and a previously unidentified sulfate. The phenol glucuronide of formoterol was the main metabolite in urine. Formoterol was alsoO-demethylated and deformylated. Plasma exposure to these pharmacologically active metabolites was low. O-demethylated formoterol was seen mainly as inactive glucuronide conjugates and deformylated formoterol only as an inactive sulfate conjugate. Intact formoterol and O-demethylated formoterol dominated recovery in feces. Mean recovery of unidentified metabolites was 7.0% in urine and 2.0% in feces.
Acknowledgments
We thank Inger Svensson and Marita Bengtsson for laboratory analyses, Ola Beckman and Vivian Kristoffersson for study coordination, Magnus Simonsson and Ingemo Grönberg for clinical management, Jenny Bendz for data management, and Lars Nyberg for scientific discussions.
Footnotes
-
Send reprint requests to: Johan Rosenborg, Experimental Medicine, AstraZeneca R&D, S-221 87 Lund, Sweden. E-mail:Johan.Rosenborg{at}draco.se.astra.com
-
Part of the study was presented by Gösta Hallström in a key lecture at The 10th Nordic Conference on Mass Spectrometry, August 22–25, 1998, Umeå, Sweden under the title “LC-ESIMS/MS of Formoterol Metabolites in Man”.
- Abbreviations used are::
- Met1
- O-demethylated formoterol
- 3H2O
- tritiated water
- FG1
- phenol glucuronide of formoterol
- FG2
- benzyl glucuronide of formoterol
- FS
- sulfate of formoterol
- Met1G1
- glucuronide 1 of O-demethylated formoterol
- Met 1G2
- glucuronide 2 of O-demethylated formoterol
- Met2
- deformylated formoterol
- Met2S
- sulfate of deformylated formoterol
- XTOT
- quantified unidentified metabolites
- LC
- liquid chromatography
- MS
- mass spectometry
- ESI
- electrospray ionization
- SRM
- selected reaction monitoring (MS/MS technique in which one or a few precursor ions, plus expected characteristic product ions after collision-induced dissociation, are selected and continuously monitored
- AUC0–24h
- area under the curve of plasma concentration versus time up to 24 h (a measure of plasma exposure, calculated using the linear trapezoidal rule)
- CLR(0–24h)
- renal plasma clearance (amount excreted in urine up to 24 h divided by AUC0–24h
- Received December 10, 1998.
- Accepted June 16, 1999.
- The American Society for Pharmacology and Experimental Therapeutics