Abstract
Liposomes loaded with the nonionic iodinated contrast agent iodixanol were injected i.v. into monkeys, rats, and dogs, and liver samples were analyzed by HPLC and mass spectrometry. Two metabolites (M1 and M2), with UV spectra identical to those of the iodixanol isomers (exo and endo) and with a mass increase of 162 compared with iodixanol, were detected. Incubations of iodixanol-liposomes or iodixanol in rat liver homogenates resulted in large amounts of iodixanol metabolites, whereas no metabolites were formed in other organ or tissue homogenates. Four groups of unidentified HPLC peaks were detected: M1 and M2 with a relative retention similar to the metabolite peaks of the in vivo samples, and in addition the minor M3 and M4. UV spectrum analysis indicated that M1 and M3 were structurally related to the iodixanol exo-isomer, whereas M2 and M4 were related to the endo-isomer. Mass spectrometry techniques indicated that the metabolites were conjugates containing one or two hexose residues, which by carbohydrate analysis and experiments with concanavalin A-Sepharose and α- and β-glucosidase were shown to be glucose residues bound to iodixanol throughO-α1-glycoside-like linkages. Metabolites with similar mass increments also were detected for several other nonionic contrast agents after in vitro incubations in liver homogenates. In conclusion, M1 and M3 are conjugates of the iodixanol exo-isomer with one and two glucose adducts, respectively. M2 and M4 are similar conjugates of the iodixanol endo-isomer. This is the first report on hepatic biotransformation of this class of X-ray contrast agents.
The low-osmolar nonionic iodinated X-ray contrast agents are extremely well tolerated with very few serious adverse events, taking the high-administrated doses into consideration. This can be explained by the high water solubility of these agents, causing a rapid and almost complete renal excretion and no significant binding to plasma proteins (Jacobsen et al., 1995, and references therein). Internalization of these agents into intact cells occurs only to a very low degree both in vivo (Dean and Plewes, 1984; Nordby et al., 1990) and in vitro (Nordby et al., 1995), probably by liquid-phase pinocytosis (Dobrota et al., 1995); kidneys, liver, and a few other glandular organs are thus reported to retain small amounts of these agents (Dean and Plewes, 1984; Heglund et al., 1995). Despite this cellular uptake, data only on the lack of any metabolites have been published so far (Almen and Golman, 1987; Lorusso et al., 1994; Speck, 1995; Jacobsen et al., 1995).
Identifying new agents for imaging of organs and tissues, especially for computer tomography of the liver, have been a topic of research for many years. Most of the new exploratory agents have been based on the ability of the Kupffer cells in liver, or cells of the reticuloendothelial system in general, to internalize particles. Several reports on the development of particulate X-ray contrast agents have been published (Caride, 1985; Fischer, 1990; Krause et al., 1995). Liposomes loaded with water-soluble iodinated contrast agents were described in 1981 (Havron et al., 1981) and several preclinical studies have been performed with liposomes loaded with water-soluble X-ray contrast agents (Krause et al., 1991; Seltzer et al., 1995;Schuhmann-Giampieri et al., 1995). So far, no metabolism of these substances has been reported despite a very effective uptake in liver cells, and no studies designed to reveal any biotransformation of these liver contrast agents have been published. The lack of metabolism has been claimed for one liposome-encapsulated water-soluble X-ray contrast agent after in vivo hepatic uptake, although no experimental data were shown to verify this statement (Schuhmann-Giampieri et al., 1994; Niendorf et al., 1995).
Iodixanol, a dimeric and nonionic X-ray contrast agent that may be formulated isotonic to blood at all concentrations relevant for imaging (Almen, 1995), is well suited for liposome encapsulation, and i.v. injection of such liposomes gives liver images of high quality (Dick et al., 1996; Leander, 1996). The chemical and biological stability of the iodixanol molecule is maintained by primarily two features: the capability of the aliphatic carbamoyl side chains (Fig.1) to withdraw electrons from the benzene ring and hence stabilize the iodine binding (Eloy et al., 1991), and the high hydrophilicity that prevents protein binding. Moreover, the secondary and tertiary amides localized in the side chains and in the bridge of this dimeric molecule are biologically rather stable (Testa and Jenner, 1976; Damani, 1982; Heymann, 1982). The iodixanol molecule (Fig. 1) gives rise to a complex mixture of stereoisomers and rotational isomers (Fossheim et al., 1995). Three main forms of the molecule can be separated by reversed-phase HPLC: the exo-exo, exo-endo, and endo-endo rotational isomers that constitute ∼60, 37, and 3%, respectively, at equilibrium at room temperature. The exo-endo and endo-endo isomer peaks, however, very often fuse, resulting in the appearance of two peaks with an ∼60:40% ratio (Priebe et al., 1995) as shown in Fig. 2A. These peaks are later referred to as the “exo-isomer” and the “endo-isomer”, respectively.
In this article, it is shown that iodixanol, when incorporated into liposomes and injected i.v., is taken up by the liver and partly metabolized by a conjugation reaction in monkeys, rats, and dogs. Furthermore, we report the in vitro biotransformation of iodixanol and other X-ray contrast agents in rat liver homogenates and describe the purification of the iodixanol metabolites by HPLC and elucidation of the molecular structures of these metabolites. This is thus the first report showing biotransformation of this class of X-ray contrast agents.
Materials and Methods
Reagents and Substances.
Iodixanol (Visipaque), liposomes loaded with iodixanol (iodixanol-liposomes), iohexol (Omnipaque), and iopentol (Imagopaque) were obtained from Nycomed Imaging AS (Oslo, Norway); iopromide (Ultravist) was obtained from Schering AG (Berlin, Germany), and ioversol (Optiray) was obtained from Mallinckrodt Inc. (St. Louis, MO). Concanavalin A (Con A)1-Sepharose was obtained from Pharmacia LKB Biotechnology Inc. (Uppsala, Sweden). The Monosaccharide Composition Analysis Kit-PMP for chemical hydrolysis and analysis, including chemicals, reagents, vials, and the CHO-C18 HPLC column, was obtained from Perkin-Elmer Co., Applied Biosystems Division (Foster City, CA); α-glucosidase (Type I, G5003, from baker’s yeast) and β-glucosidase (G0395, from almonds) were obtained from Sigma Chemical Co. (St. Louis, MO). Acetonitrile was of HPLC grade (Merck, Darmstadt, Germany). Saline (154 mM NaCl in water) was obtained from B. Braun Melsungen AG (Melsungen, Germany). Water was purified by reversed osmosis and ion exchange chromatography with a Milli-Q system (Millipore Corp., Bedford, MA). All other chemicals were of analytical grade quality.
In Vivo Samples.
Selected samples of plasma, urine, and liver (details given inResults) from several animal safety studies were used: maleCynomolgus monkeys that had been given iodixanol-liposomes i.v. in single doses of 100 mg of encapsulated iodine/kg b.wt.; one monkey that had been given 100 mg of encapsulated iodine/kg b.wt. daily for 5 consecutive days; and male Sprague-Dawley rats and beagle dogs given doses of iodixanol-liposomes in the range from 100 to 500 mg of encapsulated iodine/kg b.wt.
Monkey plasma was prepared with EDTA as anticoagulant. The samples were diluted 1:1 with water before HPLC analysis with on-line dialysis for sample preparation (see below). Urine samples were diluted 1:10 with water before HPLC analysis (on-line dialysis was not used). The liver samples were homogenized in saline (1 ml of saline/g liver) with a Diax 600 homogenizer equipped with a 20 G knife (Heidolph Elektro GmbH, Kelheim/Donau, Germany). It was assumed that the contrast agent was homogeneously distributed in the organ. Further preparation of the liver samples was performed by two alternative procedures.
Extraction.
To 0.8 ml of the liver homogenate, 2 ml of methanol, 2 ml of chloroform, and 1.2 ml of water were added successively. The tubes were stirred by whirlmixer after each addition, and the phases were separated by centrifugation at 2000g for 10 min at room temperature. The aqueous phase was transferred to HPLC vials and analyzed (on-line dialysis was not used).
Ultrafiltration.
The homogenates were centrifuged at 13,000g for 20 min at 4°C. The supernatants were transferred to ultrafiltration tubes with a molecular weight cutoff at 20,000 (Centrisart I, SM 13249; Sartorius GmbH, Göttingen, Germany) and centrifuged at 2,500g for 2 h at room temperature. The filtrates were analyzed by HPLC (both with and without on-line dialysis).
In Vitro Samples.
Samples of frozen or fresh liver from Sprague-Dawley rats were cut in pieces by a scalpel and homogenized with a Diax 600 homogenizer equipped with a 20 G knife (Heidolph Elektro GmbH) for 30 s. The homogenizations were performed both with and without the addition of phosphate-buffered saline (10 mM Na,K-phosphate, 2.7 mM KCl, and 137 mM NaCl; pH 7.4); dilutions from 1 + 1 to 1 + 9 (micrograms of liver + milliliters of phosphate-buffered saline) were used. Incubations of iodixanol with homogenates were performed with a stock solution of 10 mg/ml iodixanol, giving final concentrations of 0.5 to 5.0 mg of iodixanol (0.25–2.5 mg of iodine)/g of incubate. Similar concentrations of total iodixanol were used for the incubations with suspensions of iodixanol containing liposome; the fraction of encapsulated iodixanol was ∼50% of the total iodixanol in these suspensions. The incubation times were from 0 min to 48 h at temperatures of 4, 25, 40, and 50°C and the iodixanol-related substances were isolated by chloroform-methanol extraction. Incubations with iohexol, iopentol, iopromide, and ioversol were performed at 25°C for 20 h with 1.0 mg of iodine/g of incubate.
To remove lipids and proteins, 1 volume of liver incubate was added 2 volumes of methanol, 2 volumes of chloroform, and 1.2 volumes of water in a glass centrifuge tube. The tube was stirred with whirlmixer between each addition and finally centrifuged at 2000g for 10 min. The aqueous (upper) phase was collected and flushed for some minutes at room temperature with nitrogen to remove dissolved chloroform.
On-Line Dialysis.
Sample preparation by dialysis on-line to the HPLC was performed by a Gilson automated sequential trace enrichment of dialysates (Gilson Medical Electronics, Villiers-le-Bel, France), an automatic sample preparation method that has been shown to be very suitable for analysis of iodixanol and other contrast agents in biological fluids (Andresen et al., 1992). The system consisted of a Gilson 231 sample injector, two Gilson 401 dilutors, a dialyser unit with a 100-μl donor channel volume and 175-μl recipient channel volume, a cellulose (Cuprophane) dialysis membrane with a cutoff at a molecular weight of 15,000 and a Gilson Prelute (5 × 1.6 mm Hypersil ODS, 10 μm) trace enrichment column. The aspired sample volume transferred to the donor channel was 110 μl, slightly exceeding the channel volume and kept static in the channel during dialysis. The recipient solution (water) volume (i.e., the dialysate) was 4000 μl and was pulsed in volumes of 175 μl through the recipient channel and into the trace enrichment column. These volumes and a net recipient flow rate of 300 μl/min (Pump programming: Asp1 speed = 4 and Disp1 speed = 1) were optimal for the concentration range of the analytes.
HPLC-UV Detection.
For all HPLC analyses, except the liquid chromatography (LC)-mass spectrometry (MS)-MS analysis and the carbohydrate analysis that are described in separate sections, the analytical column was a Brownlee Spheri-5, RP-18, 5 μm, 250 × 4.6 mm operated at ambient temperature, with a Brownlee RP-18 Newguard, 7 μm, 15 × 3.2-mm precolumn (Applied Biosystems, San Jose, CA). The HPLC systems consisted of SP8800 ternary pump and SP 8500 dynamic mixer (Spectra Physics/Thermo Separation Products, Freemont, CA) (for the in vivo sample analyses), or a P4000 pump and a SCM1000 degasser (Spectra Physics/Thermo Separation Products) (for the in vitro sample analyses) and a Gilson 232–401 automatic sample processor and injector (Gilson Medical Electronics) equipped with a Rheodyne 7010 injection valve (Rheodyne Inc., Cotati, CA) and 10-μl injection loop. When on-line dialysis was used for sample preparation (in vivo samples), the Gilson 232–401 sample injector was replaced by the Gilson ASTED. The detector was a Spectra FOCUS (Spectra Physics/Thermo Separation Products) with a 6-mm, 9-μl Kel-F LC flow cell. When single wavelength detection at 244 nm (the approximate absorbance maximum for iodixanol) was used, the data acquisition, calculation, and reporting were performed by an Access*Chrom gas chromatograph-LC data system (Perkin-Elmer Nelson Systems Inc., Cupertino, CA). When using multiwavelength detection to enable UV spectrum analysis of the chromatographic peaks, the Spectra Focus detector was programmed to scan a wavelength range from 220 to 280 nm in 5-nm intervals at a rate of 96 data points/s. Data acquisition and handling were performed by the Spectra FOCUS computer software (PC1000). The mobile phase was normally 7.5% (v/v) acetonitrile in water for analysis of the in vivo samples. Variations in organic phase strength and additions of methanol or tetrahydrofuran were tested to optimize the separations (details given inResults). If not otherwise described, the mobile phase was 3 to 8% (0–21 min), 8% (21–35 min), 8 to 3% (35–37 min), and 3% (37–40 min) acetonitrile in water, with linear gradients, for analysis of the in vitro samples. The flow rates were 1 ml/min.
Isolation of Metabolites.
Extracts from the in vitro incubations with iodixanol were concentrated with nitrogen flushing at room temperature to about one-forth of the initial volume. Aliquots of 80 μl of the concentrated extracts were subjected to HPLC analysis and 0.5-ml fractions were collected automatically by a Gilson model 201 fraction collector after UV monitoring. After repeated HPLC runs, the central peak fractions of the different metabolites (M1, M2, M3, and M4) were combined and reanalyzed for quantitation and inspection of the purity of the metabolites. The mobile phase used for these HPLC runs was 7.5% (v/v) acetonitrile in water (isocratic system), with flow rate 1 ml/min.
HPLC-MS.
The in vivo samples were analyzed by a single quadrupole LC-MS system equipped with a Hewlett Packard 1050 HPLC system (Hewlett Packard Company) with UV detection at 244 nm on-line before the MS. The HPLC column was a Brownlee Spheri-5, RP-18, 5 μm, 220 × 2.1 mm. The mobile phase was isocratic acetonitrile/water at 7.5:92.5 (v/v), with a flow rate of 0.3 ml/min without splitting before the MS. The MS detection was performed with a Platform single quadrupole MS equipped with an electrospray ion source (Fisons Instruments, Manchester, UK). The source temperature was 130°C with a cone voltage set to 45 V. The nitrogen drying gas and nitrogen nebulizer gas was ∼300 and 15 l/h, respectively. Data acquisition was obtained by scanning from 150 to 2000 Da in 6-s scans. The instrument was operated in the positive ion mode and was previously calibrated with NaI in the described mass range.
HPLC-Tandem MS.
The in vitro samples were analyzed by a double quadrupole LC-MS-MS system equipped with a Hewlett-Packard 1050 Ti HPLC system with diode array UV detection on-line before the MS. The HPLC column was a Brownlee Spheri-5, RP-18, 5 μm, 4.6 × 250 mm. The mobile phase was 3 to 30% (0–20 min), 30 to 3% (20–21 min), and 3% (21–30 min) acetonitrile in water. The flow rate was 1.0 ml/min and the injection volume was 20 μl. The diode array detector was set to obtain data in the 220–280-nm wavelength region with 2-nm intervals. A post-UV split (Valco T with 1:5 split) was connected to reduce the liquid flow into the mass spectrometer ion source. The mass spectrometry data were recorded on a Quattro II mass spectrometer (Micromass, Manchester, UK) equipped with an electrospray source. The source temperature was maintained at 130°C and the cone voltage was set to 30 V. The capillary voltage was set to 3 kV and the mass spectrometer was operated in the positive ion mode at unit mass resolution. LC-MS-MS product ion scan (fragment analysis of the molecular ion) was performed with argon as collision gas at 15-eV collision energy. Data were obtained by scanning from m/z = 1000 tom/z = 2000 in 3 s in both LC-MS mode and LC-MS-MS mode.
Fourier Transform (FT)-Ion Cyclotron Resonance (ICR)-MS.
FT-ICR-MS was performed with an Apex 47e (Bruker Instruments, Inc., Billerica, MA). To produce molecular ions of the analytes, electrospray ionization (ESI; Analytica of Branford, Branford, CT) with 1% acetic acid in 1:1 (v/v) water/methanol was used. The samples were introduced by continuous infusion into the ion source.
Carbohydrate Analysis.
Iodixanol metabolite M1 (purified by HPLC; aliquots of 0.7 or 1.4 nmol, quantified by comparing HPLC peak area with standard iodixanol peak area), iodixanol (1.5 nmol), and glucose (4 nmol) were added to 670 μl of trifluoroacetic acid and 20 μl of talose (5 nmol) as an internal standard in hydrolysis tubes (included in the monosaccharide analysis kit). Distilled water was added to a total volume of 1 ml and the hydrolysis tubes were sealed and incubated at 123°C for 1 h. The samples were then transferred to 1-ml Pierce Reacti-Therm vials and dried in a Speed Vac Concentrator Mod. 100H (Savant Instruments Inc., Farmingdale, NY) at 6 to 7 mbar for 2.5 h, before 500 μl of isopropanol was added to the vials and the samples dried in the Speed Vac Concentrator at 6 to 7 mbar for 1.5 h to remove residual trifluoroacetic acid. We then added 25 μl of 0.5 M 1-phenyl-3-methyl-5-pyrazolone (in methanol), 15 μl 0.5 N NaOH, and 10 μl of distilled water to the dried samples, mixed them on a whirl mixer, and incubated them at 70°C for 2 h. The solutions were then neutralized by adding 20 μl of 0.5 N HCl and 500 μl of extraction solvent (butyl ether) was added. The samples were vortexed for 5 s, centrifuged for 1 min at 800g, and the organic phase was carefully removed and discarded. The extraction procedure was repeated twice. We added 160 μl of distilled water to the samples and transferred them to autosampler vials for HPLC analysis on a Hewlett Packard-HP1050 system equipped with a 1050 diode array detector. An Applied Biosystem CHO C18 column, 5 μm, 200 × 2.1 mm, was used. The flow rate was 0.2 ml/min, and UV detection was performed at 245 nm. The mobile phase was 100 mM ammonium acetate (pH 5.5) with a gradient from 17.75 to 25% (v/v) acetonitrile (0–45 min), 25% (v/v) acetonitrile (45–50 min), and 17.75% (v/v) acetonitrile (50–55 min). The injection volumes were from 5 to 40 μl. A standard solution containing 10 nmol each of mannose, glucosamine, talose, galactosamine, glucose, galactose, xylose, and fucose was treated the same way as the samples and 5 nmol glucose was used as a control sample.
Affinity Chromatography.
Con A-Sepharose was suspended in 20 mM Tris·HCl, 0.5 M NaCl, pH 7.4, and sedimented to 3 cm in height in a 5 cm × 1 cm (i.d.) column. A mixture of iodixanol and a M1/M2 (see Results for description of sample) was added to an equal volume of the Tris buffer and 1.0 ml of this solution was applied to the column. After 90 min, unbound material was eluted with 5 ml of Tris buffer followed by elution with 10 ml 0.2 M α-d-methylmannoside for the release of bound substances, as recommended by the manufacturer of the Con A-Sepharose. Fractions of 0.5 ml were collected during the entire elution (both Tris buffer andd-methylmannoside) and analyzed by HPLC.
Incubation with Glucosidases.
The α-glucosidase was prepared in 50 mM phosphate buffer, pH 6.8, and the β-glucosidase in 50 mM sodium acetate adjusted to pH 5.0 with acetic acid (Kamimura et al., 1988). Both enzyme concentrations were 100 U/ml. Samples (500 μl) containing ∼30 μg/ml of a mixture of metabolite M1 and iodixanol were evaporated under nitrogen gas to dryness, 300 μl of α- or β-glucosidase was added, and the samples were incubated at 25°C for 20 h and analyzed by HPLC. Samples with a mixture of metabolite M1 and iodixanol (as described above) but with no glucosidase added, were used as control experiments to look for any effect due to experimental conditions other than that due to the enzymes. In addition, samples with only iodixanol were incubated with the glucosidases to look for any effect of the enzymes on the contrast agent itself.
X-Ray Fluorescence Spectrometry.
The determination of total iodine concentration in in vivo samples was performed by a Renalyzer PRX 90 X-ray fluorescence spectrometer (Provalid AB, Lund, Sweden). Samples of minimum 2.5 ml were analyzed three times for 60 s. Body fluids and tissue homogenates were analyzed without any further treatment. If necessary, the samples were diluted with water to obtain iodine concentrations within the linear detection range, which was 0.03 to 5.0 mg/ml.
Results
In Vivo Metabolism in Monkey.
An HPLC chromatogram of iodixanol with its exo- and endo-isomers is shown in Fig. 2A. Analysis of liver samples from monkey obtained 24 h after injection of iodixanol-liposomes revealed two additional peaks, named M1 and M2, one in front of each of the iodixanol isomers (Fig. 2B). The same results were obtained whether the samples were prepared by extraction or by ultrafiltration. On-line dialysis for further removal of high-molecular-weight components after ultrafiltration resulted in fewer endogenous front peaks of the chromatograms, but the iodixanol isomers and the M1 and M2 peaks appeared to be unchanged (data not shown).
The rather broad M1 and M2 peaks and the splitting of M2 indicated overlapping peaks. A series of HPLC analyses of the same liver sample, with variations of the mobile phase composition, revealed peak splitting of either M1 or M2, but distinct peak splitting could not be obtained for both metabolites by using only one mobile phase (Fig.3).
Liver samples from control animals given saline revealed no peaks corresponding to M1 and M2, which indicates that these compounds were not of endogenous origin. No peaks corresponding to M1 and M2 could be detected in samples of liver from control animals homogenized in the presence of iodixanol solution or iodixanol-liposomes and prepared in the same way as samples from treated animals.
The normalized UV spectra of M1 and M2 were identical with the normalized spectra of the exo- and endo-iodixanol isomers, respectively, in the region of 235 to 280 nm (Fig.4). The exo- and endo-iodixanol isomer spectra were different only by an ∼1-nm shift of maximum wavelength (Fig. 4, inset). No variations in the spectra within the broad/overlapping peaks of either M1 or M2 (Fig. 3) could be detected (data not shown), which indicated that both M1 and M2 represented substances of identical or closely related structure. None of the endogenous liver compounds showed spectra even slightly similar to that of iodixanol (data not shown).
The molecular weight of iodixanol is 1550. When iodixanol was measured by MS with electrospray ionization mainly two signals were detected:m/z = 1551, which can be assigned the protonated molecular ion [iodixanol + H]+, andm/z = 1573 for the sodiated ion [iodixanol + Na]+ formed during the MS ionization of iodixanol (Fig. 5B). The M1 peak showed signals in the MS spectrum (Fig. 5A) of m/z= 1713 ([metabolite + H]+) andm/z = 1735 ([metabolite + Na]+) (i.e., mass increments of 162 for both ions, when assuming that the charge z was equal to 1). Signals at m/z = 1713 andm/z = 1735 were absent in the spectrum of pure iodixanol (Fig. 5B).
Figure 6 shows the UV chromatogram of monkey liver sample (Fig. 6A) and the corresponding single-ion chromatograms of m/z = 1735 (Fig. 6B) andm/z = 1573 (Fig. 6C). The signals of the sodiated ions were used in this figure because their intensities were higher than those of the protonated molecular ions. The mass chromatogram of m/z = 1735 completely matched with M1 and M2 of the UV chromatogram. The irregularities of these two peaks are most likely caused by the baseline noise of the MS signals as a consequence of the rather low concentration of the metabolites.
The amounts of metabolites (sum of M1 and M2) relative to the total of iodixanol-related substances in liver samples obtained 24 h after injection of iodixanol-liposomes were determined by comparing peak areas of chromatograms acquired at 244 nm. The metabolites were calculated to represent 2 to 3% of the administrated iodixanol doses. The calculations were based on the recovery of total iodixanol-related compounds in the liver measured as total iodine by X-ray fluorescence spectrometry. It also was assumed that the iodixanol isomers and the metabolites all had similar extinction coefficients, which seems to be a valid assumption due to the similarities of the UV spectra (Fig. 4).
Analysis of monkey plasma samples obtained 5 min and 24 h after injection of iodixanol-liposomes showed only the endogenous plasma components and the iodixanol isomers (data not shown). Neither did the urine samples show any detectable amount of metabolites at any of the sampling time points (24, 48, 96, and 168 h). However, the detection of metabolite amounts <5% of the total iodixanol-related compounds was not possible, due to interference from urine matrix components and low iodixanol concentrations.
In Vivo Metabolism in Rat and Dog.
Analysis of liver samples from rats and dogs that had been injected iodixanol-liposomes revealed HPLC chromatograms similar to that shown for monkey in Fig. 2B. The UV spectra of the two metabolites were identical with the spectra of the subsequent-eluting iodixanol isomer, and the two metabolites were shown by MS to have molecular masses of 162 above that of iodixanol. As described for the monkey samples, the single-ion chromatograms of the signals atm/z = 1573 [iodixanol + Na]+ and m/z = 1735 [metabolite + Na]+ perfectly matched the UV peaks of the iodixanol isomers and the metabolites, respectively (results obtained with rat and dog samples are not shown).
In Vitro Biotransformation.
Incubation of rat liver homogenates with either iodixanol or iodixanol-liposomes resulted in the appearance of new chromatographic peaks in front of and between the iodixanol exo- and endo-isomer peaks (Fig. 7). Baseline separation could not be obtained between the iodixanol isomer peaks and the new peaks, or between the new peaks. Four groups of metabolite peaks could be detected (M1, M2, M3, and M4), all more or less complex with nonseparated components. The relative retention times for M1 and M2 were similar to the in vivo metabolites. The two additional peaks, M3 and M4, eluted earlier than M1 and M2, respectively. All four peak groups were absent in the chromatograms of both pure iodixanol and blank liver samples. Neither did incubations of iodixanol nor iodixanol-liposomes with homogenates of kidney, spleen, lung, or muscle tissue result in any detectable amounts of these metabolites or of other nonidentified peaks (data not shown).
The UV spectra of peak M1 and M2 after in vitro incubation (data not shown) were identical with the spectra obtained for the in vivo metabolites (Fig. 4). As observed for the M1 and M2 peaks, the spectra of the M3 and M4 peaks were almost identical with the spectra of the iodixanol exo- and endo-isomers, respectively (data not shown).
Incubations performed with different time and temperature showed that the metabolite formation was approximately proportional with incubation times from 6 to 48 h at 4°C, from 2 to 24 h at 25°C, and from 1 to 6 h at 40°C, and that up to ∼50% of the iodixanol was converted in these experiments (Fig.8). The metabolites could be detected after a 30-min incubation at 25°C and 40°C, whereas no distinct metabolite related peaks could be detected before 6 h of incubation at 4°C.
Isomer Equilibrium.
Pure metabolite fractions could not be isolated due to incomplete peak separation on the HPLC system, but the enrichment of metabolites M1, M2, and M3 during the chromatographic separation was sufficient for further characterization of these metabolites. M4 was not characterized further due to a high content of the iodixanol exo-isomer in this fraction. When the M1, M2, and M3 fractions were analyzed after a 1-day storage at 2–8°C, small but distinct peaks of M2 in the M1 sample, M1 in the M2 sample, and M4 in the M3 sample were detected. After 4 months storage of these samples at 2–8°C, the peak area ratios for M1:M2, and M3:M4 were close to 60:40 (Table1), which is equal to the ratio of the iodixanol exo- and endo-isomers at equilibrium (Priebe et al., 1995). Incubating freshly isolated metabolite fractions for 20 h at 25°C also resulted in pairs of metabolites in ratios close to 60:40. These data show that the metabolites are two pairs of equilibrium isomers: M1⇌M2 and M3⇌M4; in the following these metabolites are therefore called M1/M2 and M3/M4.
Mass Determination.
Extracts of iodixanol incubated with liver homogenate (0.5 mg iodixanol/ml homogenate, 20 h at 25°C) were analyzed by LC-MS to correlate the measured molecular-related m/zvalues to the different peaks (Fig. 9; note that the relative retention times of the peaks are different from that of Fig. 7 due to different experimental conditions and the UV peak of M4 is hidden behind the exo-iodixanol peak). The mass chromatogram of protonated iodixanol [M + H]+ atm/z = 1551 (B) correlated well with the exo- and endo-iodixanol peaks in the UV chromatogram (A). The mass chromatogram of m/z = 1713 (C) correlated well with peaks M1 and M2 in the UV trace showing identicalm/z values for the two metabolites. Furthermore, the mass chromatogram of m/z = 1875 (D) correlated in the same way with M3 and M4 (hidden behind the exo-iodixanol peak) in the UV chromatogram, but with less intensity due to the lower concentration of these metabolites.
Fragment Analysis.
Fragment analysis was done by product ion scan of the main metabolite. Figure 10 shows the LC-MS-MS product ion scan of the M1 peak at m/z = 1713 giving an intense iodixanol signal at m/z = 1550.7.
Exact Mass Determination.
Samples of metabolite M1/M2, metabolite M3/M4, and iodixanol purified by HPLC were analyzed with the very accurate technique FT-ICR-MS to estimate the elemental compositions of the substances. M1/M2 showed a mass increase of 162.063 relative to iodixanol, with a mass deviation of ±0.0170 based on internal calibration with iodixanol. M3/M4 showed a mass increase of 324.088 (±0.0400 in mass deviation) relative to iodixanol. This mass increase is exactly twice that observed for M1/M2 (within the mass deviation), which shows that M3/M4 are conjugates with two groups attached to iodixanol, each group giving a mass increase of 162.063 ± 0.0170.
A wide range of theoretically potential elemental compositions could be deduced within the mass deviation of ±0.0170 for the determined mass increase of 162.063. When elements not likely to occur covalently bound in biological substances were excluded, the most probable elemental composition having any relation to metabolic reactions was C6H10O5(mass 162.0528). This formula is consistent with the composition of a hexose sugar (mass 180.0634) with loss of H2O (mass 18.0106) after forming a glycoside binding.
Carbohydrate Analysis.
Glucose, galactose, and xylose were all identified in the M1/M2 fraction, the blank samples (with water substituting the metabolite fraction), and the pure iodixanol samples; the data for M1/M2 were thus corrected for the content in the blank samples. When doing this correction for three separate experiments, the amount of released glucose from fraction M1/M2 was approximately equal to the amount of metabolite in that sample [a glucose/metabolite ratio of 0.95 ± 0.14 (mean ± S.D., n = 3) was obtained]. The galactose and xylose contents in fraction M1/M2 were close to that observed for the blank samples and the iodixanol sample [the corrected ratio of galactose to metabolite was 0.057 ± 0.032 (mean ± S.D., n = 3) and the corrected ratio of xylose to metabolite was −0.033 ± 0.030 (mean ± S.D.,n = 3)].
Affinity Chromatography.
A mixture of pure iodixanol and metabolite M1/M2 in an ∼1:1 ratio, as estimated from the HPLC peaks, was applied to a Con A-Sepharose column. Iodixanol did not show any affinity for Con A because 90% of the applied iodixanol was recovered in fractions 3 to 7. M1/M2 did, however, bind to Con A because only ∼5% of the applied amount of M1/M2 was recovered in the first 10 fractions (in fractions 6–10). Attempts to elute M1/M2 with 0.2 M α-d-methylmannoside, which normally releases bound glucose and other sugars from Con A-Sepharose, did not cause any measurable release of M1/M2 from the column.
Glucosidase Treatment.
Treatment of metabolite fraction M1/M2 with α-glucosidase resulted in a decrease of the M1 and M2 peaks and an increase of the iodixanol exo- and endo-isomer peaks. The area percent of the sum of M1 and M2 was reduced from 83 to 38% when incubated with α-glucosidase, whereas incubation with β-glucosidase did not show any effect (Table2). Both the control samples containing M1/M2 but no enzymes added and the control samples containing iodixanol plus enzymes remained unchanged during the incubation (data not shown).
Discussion
HPLC analysis of liver samples from monkeys that had received iodixanol-liposomes showed the presence of two unknown peaks (M1 and M2) in addition to the iodixanol exo- and endo-isomer peaks (Fig. 2), whereas liver samples from control animals or liver samples from control animals with added iodixanol (to reveal any sample preparation artifacts) showed no unknown peaks. M1 and M2 had UV spectra very similar to their subsequent-eluting iodixanol isomers (Fig. 4). The difference in the λ maximum of the spectra of the iodixanol isomers was also characteristic for M1 and M2. These data strongly indicated that M1 is a metabolite of the exo-isomer, whereas metabolite M2 is a metabolite of the endo-isomer, both probably somewhat more hydrophilic than their subsequent-eluting iodixanol isomers as judged by their shorter retention times.
MS analysis revealed (Fig. 5) that the M1 substance had a molecular weight of 162 above that of iodixanol, given by the mass differences (Δm/z values) both for the molecular ions (m/z = 1713 versusm/z = 1551) and sodiated ions (m/z = 1735 versusm/z = 1573). HPLC-MS analysis revealed that M1 and M2 had identical molecular weight, visualized by the single-ion chromatogram of m/z = 1735 for the sodiated ion of the metabolites (Fig. 6).
In liver samples obtained 24 h after injection these metabolites constituted some 2 to 3% of the injected encapsulated iodixanol dose. The absence of detectable metabolites in urine and serum indicated that only a small fraction of the iodixanol molecules taken up by liver cells was metabolized. This biotransformation is probably exclusively caused by the extensive liver uptake of the liposomes, giving a very high intracellular concentration of the contrast agent. There is, however, no indication on metabolism of iodixanol after injection of only the water-soluble substance for vascular use.
The low concentration of metabolites in the in vivo samples made it difficult to determine the molecular structure of these metabolites and an in vitro method for production of large amounts of the metabolites was therefore developed. Large and similar amounts of metabolites were obtained when either iodixanol or iodixanol-liposomes was incubated with rat liver homogenates at 25, 40, or 50°C, showing that the reaction was independent of the liposomal phospholipids. The HPLC retention times, the UV spectra, and the molecular masses indicated that the metabolites M1 and M2 formed during the in vitro incubation of rat liver homogenates were identical with the metabolites found in vivo in monkey, dog, and rat liver samples. Two additional metabolites (M3 and M4) were detected after in vitro incubation; M3 and M4 had UV spectra identical with the iodixanol exo- and endo-isomers, respectively. The four metabolites showed to be two pairs of isomers because the relative peak area ratio at equilibrium between M1 and M2, as well as between M3 and M4, was equal to 60:40, which is the ratio between the iodixanol isomers (Priebe et al., 1995).
The m/z mass signals from the metabolites are clearly related to iodixanol as shown by the LC-MS-MS product ion spectrum of M1 (Fig. 10). This implies that the metabolic changes of iodixanol involve formation of a conjugate with an increase in net molecular weight of 162. With high-resolution FT-ICR-MS, the mass increase of M1/M2 was more accurately measured to be 162.063. The only conjugation product consistent with this mass increase and within the error of the mass determination was a glycoside formed with a hexose. This very accurate mass technique also verified that the mass increase of M3/M4 was exactly twice the increase of M1/M2 (i.e., that double conjugation was obtained in the in vitro experiments).
Three additional analyses were carried out to verify the metabolites as conjugates of iodixanol and hexose and to identify the hexose. First, carbohydrate analysis revealed that the hydrolyzed sugar in the metabolite M1/M2 sample was glucose, in a molar amount very close to the initial molar amount of M1/M2. No significant amounts of other carbohydrates were detected. Second, when a mixture of iodixanol and metabolite M1/M2 was passed through a column containing the lectin Con A, which is highly selective for α-glucose and α-mannose residues, the metabolites, but not iodixanol, were retained on the column. Third, incubation of a mixture of iodixanol and metabolites with α-glucosidase, which is selective for terminal O-linked glucose residues with α1 configuration (α1–4 if to another sugar), resulted in a decreased amount of the metabolite and a corresponding increase of iodixanol. Similar experiments with β-glucosidase did not cause any changes in the ratio of metabolites and iodixanol in the sample. Thus, M1 and M3 most likely are conjugates of the iodixanol exo-isomer with one and two α-glucose residues bound to iodixanol, respectively. Similarly, M2 and M4 are conjugates of the iodixanol endo-isomer with one and two α-glucose residues bound to iodixanol, respectively. Moreover, the glucose units seem to be bound to iodixanol by O-α1-glycoside-like linkages.
Glucosidation is a rare metabolic pathway of xenobiotics in mammals. Some cases of O-linked and N-linked glucosides in β-conformation have been reported (Tjørnelund et al., 1989; Tang, 1990; Soine et al., 1990, 1991; Egestad and Sjøberg, 1992), whereas α-glucoside conjugates are extremely rare for exogenous compounds (Kamimura et al., 1988; Muto et al., 1991). Enzymes of the α-glucosidase class, including acidic (lyzosomal) and neutral (microsomal) α-glucosidases of the liver, which both are essential for normal glycogen metabolism, have for some years been known to have transglucosidation activity (Muto et al., 1991). Thus, Kamimura et al. (1988) isolated glucoside conjugates in urine that were labile to α-glucosidase treatment but not to β-glucosidase, after p.o. administration of indeloxazine. In vitro formation of these metabolites was later demonstrated by incubating indeloxazine with a high-molecular-weight fraction of cytosol from rat liver homogenates and with glycogen granules, maltose or maltotriose as the source of glucose (Kamimura and Matsui, 1989). Several cellular fractions were later investigated and it was suggested that both neutral (microsomal) α-glucosidase and acid (lysosomal) α-glucosidase were involved in the formation of the indeloxazine–glucoside (Kamimura et al., 1992).
The in vivo uptake of the iodixanol liposomes by the macrophages of the liver is very high and the particles most probably end up in lysosomes or lysosomal-like vacuoles, where iodixanol may be exposed to the acidic α-glucosidases. One major difference between in vivo and in vitro conditions is the accessible concentration of glycogen, the most probable source of glucose. This concentration is very high after in vitro homogenization compared with the in vivo situation, where the storage glycogen and iodixanol are mainly located in separate cell compartments. However, some glycogen may be present in the lysosomes in vivo (Geddes and Stratton, 1977; Calder and Geddes, 1989;Konishi et al., 1990) and thus be the source of the observed in vivo glucose conjugates. The small and variable amounts of the in vivo conjugates, in contrast to the very large amounts produced in vitro, may reflect the low lysosomal glycogen concentration and/or limited access to the enzymes involved.
The in vivo metabolism may represent a minor metabolic process, but it seems to be species-independent because similar results were obtained in rat, dog, and monkey. Furthermore, preliminary results have shown that the in vitro reaction is not specific for iodixanol but rather general for structurally related nonionic contrast agents. Thus, incubation of rat liver homogenates with iohexol, iopentol, iopromide, and ioversol all resulted in new HPLC peaks with UV spectra similar to that of the contrast agent, and LC-MS analyses showed for all these substances, metabolites with a mass increase of 162 compared with the contrast agent itself (data not shown).
In conclusion, the O-α-glucosidation demonstrated in the present work is the first report showing in vivo and in vitro metabolism of this class of X-ray contrast agents.trans-Glucosylation by α-glucosidase is one possible mechanism of this conjugate formation, but it remains to confirm experimentally the enzyme activity that is responsible for the reaction and also to determine to which OH– group(s) of iodixanol the glucose unit(s) are bound. It is not known whether the complexity of the chromatographic peaks (Fig. 7) indicates that each of the four metabolites represent mixtures of iodixanol molecules with glucose bound to different OH– groups, or if complex isomerism is the cause of the irregular peaks.
Acknowledgments
We thank Einar Uggerud at the University of Oslo for performing the FT-ICR-MS analysis, Inger Oulie for doing the carbohydrate analysis, Mari Tøftum for handling the affinity chromatography, and Heidi Kristin Meyer-Mørch for expert technical assistance.
Footnotes
-
Send reprint requests to: Petter Balke Jacobsen, Nycomed Imaging AS, Department of Pharmacology, P.O. Box 4220 Torshov, N-0401 Oslo, Norway. E-mail: pbj{at}nycomed.com
- Abbreviations used are::
- Con A
- concanavalin A
- LC
- liquid chromatography
- FT-ICR-MS
- Fourier-transform ion-cyclotron resonance mass spectrometry
- MS
- mass spectrometry
- M1–M4
- metabolites 1–4
- Received January 28, 1999.
- Accepted June 14, 1999.
- The American Society for Pharmacology and Experimental Therapeutics