Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Capecitabine (CAP¹, N⁴-pentyloxycarbonyl-5'-deoxy-5-fluorocytidine) is an orally available prodrug of 5'-deoxy-5-fluorouridine (5'-DFUR) with limited side effects, which has recently come into clinical practice for the treatment of breast and colorectal cancers (Di Costanzo et al., 2000; Marshall, 2001; Twelves et al., 2001; Wang et al., 2001). This fluoropyrimidine carbamate was designed to pass intact through the human intestinal mucosa and take advantage of the differential enzymatic levels in tumors to achieve better targeting of 5-fluorouracil (5-FU), the active moiety of the drug (Miwa et al., 1998; Shimma et al., 2000). An activation pathway of CAP has been proposed that requires three sequential steps of enzyme reactions (Bajetta et al., 1996). CAP is first converted to 5'-deoxy-5-fluorocytidine (5'-DFCR) by carboxylesterase and then to 5'-DFUR by cytidine deaminase, mainly in the liver (Budman et al., 1998; Miwa et al., 1998). Finally, systemic 5'-DFUR is converted to 5-FU by thymidine phosphorylase, the activity of which is increased in tumor tissue (Ishikawa et al., 1998; Verweij, 1999) (Fig. 1, in part).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Metabolic pathway of capecitabine. Novel metabolites of capecitabine are shown in boxes.

Most of the pharmacokinetic and metabolic studies on CAP have been carried out with liquid chromatography as the main analytical tool (Reigner et al., 1998). Since ¹⁹F NMR spectroscopy gives an overview of all fluorinated compounds present without any prerequisite separative process, we used this method to study the metabolism of CAP. Previous work has illustrated advantages of this method for the evaluation of metabolites of antineoplastic fluoropyrimidine drugs (Martino et al., 2000; Wolf et al., 2000). The aim of the present study is to reinvestigate the metabolism of CAP using ¹⁹F NMR as analytical method. Regarding the high variability of localization and activities of enzymes from species to species (Camenier and Smith, 1965; Ho, 1973; Shimma et al., 2000) involved in CAP activation, various animals and experimental models have been used. Because rats have a very low activity of cytidine deaminase, isolated perfused rat livers (IPRL) and healthy rats were used as simple models to evaluate primary process of CAP activation. We investigated fluorinated compounds present in liver, bile, perfusate medium from IPRL, and in liver, kidneys, bile, plasma, and urine from healthy rats. New metabolites of CAP were observed. A nonsystemic glucuronide of 5'-DFCR (5'-DFCR-G) has been detected in liver and bile. 5-Fluorocytosine (5-FC) and its hydroxylated derivative, 5-fluoro-6-hydroxycytosine (5-FCOH), were formed in liver and excreted in urine. Finally, a comparative study of the metabolites measured in urine from rats, mice, and humans was conducted to determine whether the rate-limiting process of CAP activation could be indirectly determined.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. 5-FU, 5-FC, fluoroacetate (FAC), hydroxypropylmethyl-cellulose 2910, phenolphtalein glucuronic acid, D-saccharic acid 1,4-lactone, -D-glucuronidase (EC 3.2.1.1.31) from bovine liver (type B-10) and from Helix pomatia (type H-1) were purchased from Sigma and chromium (III) acetylacetonate (Cr(acac)₃) from Aldrich (all from Sigma-Aldrich Chimie, Saint Quentin Fallavier, France). -Fluoro--alanine (FBAL) hydrochloride was provided by Tokyo Kasei Chemicals (Tokyo, Japan). 5,6-Dihydro-5-fluorouracil (5-FUH₂), 5'-DFUR, 5'-DFCR and 5-FCOH were generously supplied by F. Hoffmann-La Roche (Basel, Switzerland). CAP was a gift of Professor Walter Wolf, University of Southern California (Los Angeles, CA). -Fluoro--ureidopropionic acid (FUPA) was prepared by chemical opening of 5-FUH₂ pyrimidine ring in 1 M NaOH (Malet-Martino et al., 1986). 2-Fluoro-3-hydroxypropionic acid (FHPA) was synthesized as described by Arellano et al. (1998). All other chemicals were reagent grade and obtained from standard commercial sources.

IPRL Experiments. Male Wistar rats (Iffa Credo, Lyon, France) were used. The IPRL experiments have been described previously (Arellano et al., 1997). CAP was injected into the perfusate after 1 h of liver equilibration, and the experiments were continued for 3 h. At the end of the experiments (n = 5), the perfusate was freeze-dried, stored at 80°C, and resuspended in H₂O to a known volume close to 3 ml immediately before ¹⁹F NMR analysis. Bile samples were gathered (1.12 ± 0.26 ml) and diluted with H₂O to a known volume close to 2.5 ml, then stored at 80°C until analysis. Liver was weighed (13.6 ± 1.2 g), immersed in liquid nitrogen, powdered, and sequentially extracted with cold and hot 1 M perchloric acid (PCA) by using the method of Wain and Staatz (1973). The acid-soluble (AS) and acid-insoluble (AI) fractions thus obtained were lyophilized to dryness and stored at 80°C until analysis. The lyophilized materials were resuspended in a known volume of H₂O containing 30 mM EDTA close to 3 ml and centrifuged immediately before ¹⁹F NMR analysis. The pH of the supernatant was adjusted to 5.5.

In Vivo Experiments, Rat and Mice Urine. CAP was dissolved in an aqueous solution of 0.5% hydroxypropylmethyl cellulose immediately before per oral administration. Six rats simultaneously treated with 80 mg/kg B.W. of CAP (1 ml of the above-mentioned solution) were sequentially anesthetized per pair at 1, 2, or 3 h. The abdomen was then opened and whole blood collected. Bloodless liver and kidneys were excised and immersed in liquid nitrogen, then maintained at 80°C until PCA extractions were performed.

Human Urine. Eleven colorectal cancer patients received an intravenous injection of 200 to 250 mg/m² irinotecan (CPT-11) over 30 min followed 24 h later by CAP administered orally at a dose of 1000 to 1250 mg/m² twice daily at 12 h interval. Among these patients, four received a second treatment three months later, and the urine was included in the study. A part (33 ± 10 ml) of the total volume of urine (500 ± 300 ml) collected over 12 h after the first CAP dose was freeze-dried, stored at 80°C, and resuspended in H₂O to a known volume close to 3 ml immediately before ¹⁹F NMR analysis.

Control Experiments. To check that the formation of fluorinated compounds was a metabolic process rather than a chemical transformation of CAP taking place during the perfusion time, several control experiments were carried out. CAP or 5'-DFCR was added to a perfusate collected from a blank IPRL experiment and maintained under standard warming and gassing for 3 h. Only the signal arising from CAP or 5'-DFCR was detected when perfusates were analyzed.

¹⁹F NMR Spectroscopy and Quantification. ¹⁹F NMR spectra were recorded at 282.4 MHz with ¹H-decoupling on a Bruker WB-AM 300 spectrometer using 10-mm diameter NMR tubes (Bruker S.A., Wissembourg, France). The recording conditions were: probe temperature, 25°C; sweep width, 41,667 Hz; 32,768 data points zero-filled to 65,536; pulse width, 7 µs (flip angle 40° in bile and urine, 30° in PCA extracts, and 20° in concentrated perfusate); pulse interval, 1.4 s for quantification of perfusates, AS and AI extracts, or 3.4 s for quantification of bile and urine samples; number of scans, 10,000 to 50,000; line broadening caused by exponential multiplication, 5 Hz. The chemical shifts () were reported relative to the resonance peak of CF₃COOH (5% w/v aqueous solution) used as external chemical shift reference ( = 0 ppm).

Isolation and Purification of 5'-DFCR-G. Prior HPLC analysis, crude bile (3 ml) and liver AS extract (3 ml) samples of two in vivo experiments were separately pooled and each subjected to a cleaning procedure. In brief, three volumes of dichloromethane were added; after extraction, the aqueous phases were collected and reextracted with ethyl acetate. The aqueous phases were then lyophilized. The dry residues were diluted in 400 µl of H₂0 and then placed at the top of a 10 cm-length glass column of 0.8-cm diameter, which had been previously packed with Matrex Silica C₁₈ (50 µm particle size and 60 Å pore diameter; Millipore Corp., Danvers, MA). Elution of the column was carried out by water (5 × 1 ml fractions) and then by a water-methanol (80:20 v/v) mixture (6 × 1 ml fractions). Each fraction was analyzed by ¹⁹F NMR. Fractions 4 and 8 contained the unknown metabolite and 5'-DFCR, respectively.

Enzymatic Hydrolysis of 5'-DFCR-G. -D-Glucuronidase from bovine liver or H. pomatia (1000 units) was added to the purified compound from liver or bile dissolved in 0.2 ml of 0.2 M acetate buffer (pH 5.0). Incubations were run overnight in a shaking water bath at 37°C. At the end of the incubation period, remaining -glucuronidase activities were checked with phenolphthalein glucuronic acid. Parallel incubations were conducted in the absence of -D-glucuronidase (control), and in the presence of -D-glucuronidase previously inactivated by heating (20 min at 80°C), or in which D-saccharic acid 1,4-lactone was included in the incubation medium. The HPLC analysis showed that the unknown compound at R_T = 6.1 min disappeared when subjected to -D-glucuronidase. No change in the chromatogram was observed in assays without enzyme or with enzyme denatured by heating or inhibited by D-saccharic acid 1,4-lactone. In the presence of -D-glucuronidase, the peak at 6.1 min disappeared in favor of a compound giving a peak at R_T = 6.6 min identified as 5'-DFCR with authentic standard. Some chromatograms are presented in Fig. 2.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   HPLC chromatograms of the purified glucuronide of 5'-DFCR. A, 5'-DFCR glucuronide purified from bile; B, overnight incubation with -D-glucuronidase (bovine liver, type B-10); C, chromatogram of the standard 5'-DFCR.

LC-MS and ¹H NMR Analysis of 5'-DFCR-G. LC-MS was carried out in positive mode on a PerkinElmer Sciex API 365 triple quadrupole mass spectrometer (PerkinElmerSciex Instruments, Boston, MA) equipped with a turboionspray source operating at 450°C. Nitrogen served both as auxiliary and collision gas and air as nebulizer gas. The HPLC system used consisted of a PE series 200 LC pump, a Rheodyne 8125 injector valve (40-µl loop; Rheodyne, Rohnert Park, CA) and a PE 785 UV-Vis detector ( = 280 nm), both controlled by a PE Sciex MassChrom data system (version 1.1.1). HPLC conditions were identical to those reported above except that TFA was replaced by formic acid. MS analysis showed that the glucuronide with a R_T of 6.1 min exhibited a characteristic ion at m/z 422 [M + H]⁺, suggesting a molecular mass of 421 Da, namely that of a glucuronide of 5'-DFCR.

Statistical Analysis and Data Treatment. All results were expressed as mean ± S.D. When a compound was not detected in a sample, its value was assumed to be zero and was included in the mean value. When necessary, Student's t test was applied. Differences were considered significant when p value was <0.05, unless otherwise specified.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Isolated Perfused Rat Liver and in Vivo Experiments. IPRL were treated with CAP for 3 h at 80 mg/kg B.W., which corresponds to a mean value of 6.8 ± 0.5 µmol/g of liver wet weight. Typical ¹⁹F NMR spectra of perfusate, bile, and acid extract samples are presented in Fig. 3. Data of the quantitative analysis are reported in Table 1.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Typical 1H-decoupled 19F NMR spectra from an IPRL treated with capecitabine (80 mg/B.W.) for 3h: A, perfusate, pH = 8.2; B, acid-soluble extract, pH = 5.5; C, bile, pH = 8.4.  Region of interest surrounding the 5'-DFCR peak was expanded to discriminate the signal of 5'-DFCR and 5'-DFCR-G. S, doublet arising from the 13C-19F coupling.

View this table: [in this window] [in a new window]   TABLE 1 Metabolites quantification carried out by 19F NMR in perfusate, bile, AS and AI fractions from IPRL experiments (n = 5) Data are means ± S.D. expressed as % a.d. Concentrations of compounds in perfusate are expressed relative to the volume of perfusate recovered at the end of the IPRL experiment before lyophilization. Concentrations of compounds in liver extracts and bile are expressed as nanomoles per gram wet weight; for AS extracts, concentrations are also given as nanomoles per milliliter of tissue water (µM); see Materials and Methods section for further details. Significance of differences from the concentration measured in the perfusate, * p > 0.4;  p < 0.005.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Correlation between 5'-DFCR concentrations measured in liver versus plasma. Blood and liver from six rats submitted by pair to the same dose of CAP for 1, 2, or 3 h were taken to measure the amounts of 5'-DFCR in the plasma and AS extract, respectively. Values of 5'-DFCR concentration determined in the AS extract of liver are estimated in nanomoles per milliliter of tissue water (µM).

Analysis of Urine from Rats and Mice. Urine from rats and mice treated with CAP have been fractionally collected over 72 h. The dose was 80 mg/kg B.W. for rat and LD mice groups and 500 mg/kg B.W. for the HD mice group. Typical spectra of urine from the 6 to 24 h collections are presented in Fig. 5, A-C. CAP and its metabolites, 5'-DFCR and 5'-DFUR, were found in all urine samples. 5-FU and its first catabolite, 5-FUH₂, were detected in three of four experiments for LD mice and in all experiments for HD mice, whereas 5-FU was barely detected and 5-FUH₂ never detected in urine of rats. FUPA was barely detected in the 0 to 6 h urine collections of rats whereas it was detected in all urine collections of mice. FBAL was found in all urine samples whatever the species or the urine fraction. N-carboxy--fluoro--alanine (CFBAL) was detected mainly in urine of rats because they were alkaline (Martino et al., 1987). Indeed, urine collections of rats had a mean pH value of 8.2 ± 0.8 (n = 18), which was significantly higher than the mean pH values of 7.2 ± 0.8 (n = 15) and 7.3 ± 0.6 (n = 16) for urine of LD mice and HD mice groups, respectively (p < 0.001). FHPA was detected in one of the four experiments for rats or LD mice but in all experiments for HD mice. 5-FC and its hydroxylated metabolite, 5-FCOH, were found in urine of rats and mice. 5-FCOH was never detected when 5-FC was not excreted in urine. Two unknown compounds, one in rat urine and the other in mice urine, giving rise to weak signals at 107.2 ppm and 108.8 ppm, respectively, were also detected in some samples.

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Typical 1H-decoupled 19F NMR spectra of urine samples. A and B, fractions 6 to 24 h from rats and mice treated with oral capecitabine at 80 mg/B.W. A, rat (pH = 8.5); B, mice (pH = 6.25); C, fraction 6 to 24 h from mice treated with oral capecitabine at 500 mg/B.W. (pH = 6.8); and D, fraction 0 to 12 h of urine from a patient treated with a single oral dose of 1900 mg of capecitabine (pH = 5.5).

Analysis of Human Urine. Urine (n = 14; pH 5.7 ± 0.3) of patients collected over 12 h after the first daily dose of CAP were analyzed. Urinary recovery of CAP and its metabolites accounted for 71 ± 17% a.d. The major metabolite was by far FBAL accounting for 46 ± 4% a.d. The other main compounds were CAP (3 ± 1% a.d.), 5'-DFCR (10 ± 3% a.d.), 5'-DFUR (7 ± 2% a.d.), and FUPA (4 ± 1% a.d.). 5-FU (0.6 ± 0.2% a.d.) and catabolites, 5-FUH₂ (0.3 ± 0.2% a.d.), F (0.2 ± 0.3% a.d.), FHPA (0.3 ± 0.1% a.d.), and FAC (in 4 of 14 samples analyzed; (0.004 ± 0.002% a.d.)) were minor metabolites. 5-FC and 5-FCOH were observed in 4 over 14 urine samples accounting for 0.01 ± 0.002% a.d. and 0.02 ± 0.002% a.d., respectively. A spectrum of a concentrated urine where 5-FC and 5-FCOH were detected is presented in Fig. 5D.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The main metabolites of the activation pathway of CAP are 5'-DFCR and 5'-DFUR that finally yields 5-FU (Bajetta et al., 1996; Reigner et al., 2001). Up to now, 5'-DFCR has been described as a simple intermediary metabolite. However, the formation of 5-FC, 5-FCOH, and 5'-DFCR-G supports that several metabolites diverge from the CAP activation pathway at the level of 5'-DFCR.

5-FC and 5-FCOH are found in the liver and perfusion medium of IPRL experiments, in urine of rats and mice, and in some human urine samples. The presence of 5-FC raises the question of the mechanism responsible for its synthesis from 5'-DFCR. A spontaneous formation of 5-FC from 5'-DFCR can be excluded regarding the stability of 5'-DFCR (see Materials and Methods). The metabolism of 5'-DFCR into 5-FC is rather the result of a nonspecific liver oxidation system than that of the action of uridine nucleoside phosphorylase for which cytidine is not a substrate (Yamada, 1978). A participation of the intestinal flora to the formation of 5-FC is probably minimal as this metabolite is observed in IPRL experiments. Nevertheless, it cannot be ruled out as 5-FC is detected in urine. 5-FCOH is a direct catabolite of 5-FC as it is observed only if 5-FC is formed, therefore excluding a chemical or enzymatic process from 5'-DFCR. 5-FCOH was initially observed in urinary gravel, then in urine of patients treated with 5-FC (Williams et al., 1981; Vialaneix et al., 1987).

5'-DFCR-G is formed in rat liver and exclusively excreted in bile. Cellular leak inherent to damages induced during the perfusion is likely responsible for the small amount of 5'-DFCR-G found in the perfusate medium from IPRL experiments (Table 1). The low biliary excretion found in IPRL experiments (2.2% a.d.) and the low recovery of radioactive dose measured in the feces after administration of ¹⁴C-CAP to humans (2.6%) (Judson et al., 1999) suggest that, if glucuronidation occurs in humans, it would be a minor CAP detoxification pathway. Since cholestasis has not been described as a side effect of CAP treatment, it is likely that 5'-DFCR-G does not affect the physicochemical properties of bile in contrast to what has been observed with a biliary conjugate of FBAL (Sweeny et al., 1987).

The third activation step of CAP leads to the formation of 5-FU that occurs mainly via uridine phosphorylase in rodents and thymidine phosphorylase in humans (Ninomiya et al., 1990). FBAL, the main catabolite of 5-FU, is formed since dihydropyrimidine dehydrogenase (E.C. 1.3.1.2), the rate-limiting enzyme of the catabolic pathway of 5-FU, has a high activity in the liver. In healthy animals, 5-FU catabolites account for 10 and 26 to 28% a.d. in rats and mice treated with CAP, respectively. In contrast, 5-FU catabolites account for 50% a.d. in human urine. A higher cytidine deaminase activity in humans (Ho, 1973) combined with an enhancement of the thymidine phosphorylase level in tumor tissue (Miwa et al., 1998) could account for the high formation of 5-FU and catabolites observed in human urine. It should be pointed out that CAP, as 5-FU (Lemaire et al., 1996; Arellano et al., 1997), leads to the formation of low amounts of two FBAL metabolites, FHPA and FAC, even in humans.

Our results support the involvement of a plasma-membrane transporter for CAP and 5'-DFCR as the Na⁺-independent equilibrative nucleoside transporter, es-type transporter (Balimane and Sinko, 1999). Indeed, in vivo experiments confirm that 5'-DFCR is present at the same concentration in liver and plasma (Fig. 4). In contrast, we observed that 5'-DFCR concentrations 3 h after CAP administration were four to five times higher in kidneys than in plasma supporting another type of transporter specific to that organ allowing an efficient secretory function (Chen and Nelson, 2000).

In IPRL experiments, CAP is forced only toward the liver metabolism whereas under in vivo conditions CAP administered orally must pass the intestinal barrier to reach the liver and then spread out the body. Therefore, CAP excreted in the urine depends on gastrointestinal, hepatic, extra-hepatic metabolisms, and renal clearance efficiency. Recoveries of drug-related materials in the urine are not significantly different between rats and mice when administered at the same dose, 84 ± 10 and 86 ± 1% a.d., respectively (p > 0.4) (Table 3), which is consistent with a nearly complete and reliable absorption of CAP from the gastrointestinal tract in these species as already observed in other studies with animals or humans (Verweij, 1999; Schüller et al., 2000). Even for a six-fold higher dose (500 mg/kg), the mean urinary recovery is 73 ± 6% a.d. in mice (Table 3). The remaining of drug-related compounds could be eliminated via feces as intact drug, 5'-DFCR and 5'-DFCR-G since these compounds are detected in the bile from IPRL and in vivo experiments. CAP excreted in urine over 72 h is significantly higher in mice than in rats (p < 0.005). Discrepancies of susceptibility of CAP to carboxylesterase between tissues, particularly intestine and liver, and species which have been described (Shimma et al., 2000) obviously accounts for such a difference with likely a higher carboxylesterase activity in rats than in mice.

In rats, however, in spite of a higher initial activation process of CAP, the subsequent activation step is limited. 5'-DFCR accumulates in the perfusion medium of IPRL experiments whereas the amount of 5'-DFUR is weak, leading to a high 5'-DFCR/5'-DFUR concentration ratio in the liver (Table 1). The same phenomenon is observed in urine where this ratio is close to 20 over 72 h (Table 3). The high 5'-DFCR/5'-DFUR ratio due to a very weak concentration of 5'DFUR leads to a low reaction rate through the uridine phosphorylase, regarding the K_m value for uridine, which is 140 and 240 µM in mice and rat livers, respectively (Yamada, 1978; Naguib et al., 1987). As a result, a weak amount of FBAL, the main catabolite of 5-FU, is found in liver and urine of rats. Therefore, CAP activation appears to be blocked at the cytidine deaminase step. This confirms the low cytidine deaminase activity in rat liver (Camenier and Smith, 1965) and suggests a lack of an extra-hepatic cytidine deaminase activity in rat in agreement with previous studies on other 5'-DFCR-related compounds such as galocitabine (Ninomiya et al., 1990; Funaki et al., 1993). On the contrary, the high activity of cytidine deaminase is revealed by the low value of the 5'-DFCR/5'-DFUR ratio (<1) in mice treated with a LD or a HD of CAP. The 5'-DFCR/5'-DFUR ratio could therefore be used as an indirect indicator of cytidine deaminase activity.

Pretreatment of patients with CPT-11 does not sensibly affect the pattern of metabolites excreted in urine. Indeed, our results are in agreement with the sole ¹⁹F NMR study of CAP metabolism that reports urine analysis from 6 patients treated with a single oral dose of 2 g of CAP (Judson et al., 1999). The same metabolites (CAP, 5'-DFCR, 5'-DFUR, 5-FU and its classical metabolites FUH₂, FUPA and FBAL) in close percentages of a.d., except for FBAL, are detected. Our analysis is limited to a period of 12 h after CAP administration versus 48 h in the Judson's study. This can explain the difference between the amount of FBAL found here (46 ± 4% a.d.) compared with 57 ± 5% a.d. as FBAL has a long half-life time. Regarding schedule treatment (administration of CPT-11 and CAP at 24 h interval), and CPT-11 pharmacokinetic parameters (Rivory et al., 1997), a possibility of drug interaction seems limited. Moreover, although both drugs share an esterase-mediated activation via a carboxylesterase, there is no interference of CAP on the formation of SN-38, the active metabolite of CPT-11, in human liver microsomes (Charasson et al., 2002). In addition, the tissue distribution of carboxylesterase activities toward CPT-11 and CAP suggest that the enzymes responsible for CAP (Shimma et al., 2000; Tsukamoto et al., 2001) and CPT-11 (Guichard et al., 1999; Khanna et al., 2000) activations are different. However, although CAP has limited effect on the formation of the glucuronide of SN-38 (Charasson et al., 2002), a competition between drugs at the level of the glucuronidation reaction should be reinvestigated in the light of a possible formation of a glucuronide of 5'-DFCR.

This study illustrates the advantages of ¹⁹F NMR applied to fluorinated drug metabolism and disposition. It allows a direct study of any biological medium without prior treatment avoiding the problems encountered in extraction recovery and chemical derivatization. It leads to the simultaneous detection and quantification in a single run of all-fluorine-containing compounds even unexpected substances. This contrasts with chromatography that usually requires some prior knowledge of metabolite structure to optimize sample preparation and/or detection. Set against these advantages, the method presents a main drawback, which is its relatively low sensitivity compared with chromatographic and other spectroscopic techniques (2 µM with a currently available spectrometer). Moreover, the unequivocal elucidation of the structure of unknown metabolites requires their isolation if a nonhyphenated method such as HPLC-NMR is not employed.

In conclusion, several new metabolites of CAP have been detected using ¹⁹F NMR spectroscopy. Three of them are 5-FU catabolites already observed in the 5-FU metabolic pathway (F, FHPA, and FAC) (Martino et al., 2000). 5-FC and 5-FCOH are also minor CAP metabolites. A detoxification pathway, reported for the first time, leads to a glucuronide of 5'-DFCR in rat. A complemented metabolic pathway of CAP is thus proposed in Fig. 1. The extent of involvement and pharmacological consequences of these new catabolic ways on the treatment of patients remains to be clarified.