A mathematical model of the kinetics of 5-fluorouracil and its catabolites in freshly isolated rat hepatocytes

Inter-relations between 3-hydroxypropionate and propionate metabolism in rat liver: relevance to disorders of propionyl-CoA metabolism

Propionate, 3-hydroxypropionate (3HP), methylcitrate, related compounds, and ammonium accumulate in body fluids of patients with disorders of propionyl-CoA metabolism, such as propionic acidemia. Although liver transplantation alleviates hyperammonemia, high concentrations of propionate, 3HP, and methylcitrate persist in body fluids. We hypothesized that conserved metabolic perturbations occurring in transplanted patients result from the simultaneous presence of propionate and 3HP in body fluids. We investigated the inter-relations of propionate and 3HP metabolism in perfused livers from normal rats using metabolomic and stable isotopic technologies. In the presence of propionate, 3HP, or both, we observed the following metabolic perturbations. First, the citric acid cycle (CAC) is overloaded but does not provide sufficient reducing equivalents to the respiratory chain to maintain the homeostasis of adenine nucleotides. Second, there is major CoA trapping in the propionyl-CoA pathway and a tripling of liver total CoA within 1 h. Third, liver proteolysis is stimulated. Fourth, propionate inhibits the conversion of 3HP to acetyl-CoA and its oxidation in the CAC. Fifth, some propionate and some 3HP are converted to nephrotoxic maleate by different processes. Our data have implications for the clinical management of propionic acidemia. They also emphasize the perturbations of the liver intermediary metabolism induced by supraphysiological, i.e., millimolar, concentrations of labeled propionate used to trace the intermediary metabolism, in particular, inhibition of CAC flux and major decreases in the [ATP]/[ADP] and [ATP]/[AMP] ratios.

Profiling dihydropyrimidine dehydrogenase deficiency in patients with cancer undergoing 5-fluorouracil/capecitabine therapy

Fluoropyrimidine drugs such as 5-fluorouracil (5-FU) and capecitabine are a mainstay in the treatment of numerous solid tumors, including colorectal cancers, alone or as part of combination therapies. Cytotoxic drugs such as 5-FU and oral capecitabine display narrow therapeutic indexes combined with high interpatient pharmacokinetic variability. As a result, severe toxicities often limit or delay the administration of successive, optimal chemotherapeutic courses, leading to unfavorable clinical outcome in patients with cancer. Catabolism and deactivation of fluoropyrimidine drugs depend on a single and exclusive enzymatic step driven by dihydropyrimidine dehydrogenase (DPD). Dihydropyrimidine dehydrogenase is prone to marked circadian rhythms, drug-drug interactions, and genetic polymorphisms; influence of its erratic activity on 5-FU pharmacokinetics and toxicity profile has been extensively investigated, and it is now well known that DPD deficiency leads to severe toxicities with 5-FU or possibly capecitabine exposure. With the ever-increasing number of patients with cancer likely to be treated with fluoropyrimidines, predicting and preventing the occurrence of such toxicities is now a major issue in clinical oncology. Early determination of DPD status in patients with cancer would allow identification of those at risk and help in subsequent dose adjustment or selection of other treatment modalities. Numerous methods, either genotypic or phenotypic, have been proposed to achieve this goal. This review covers a wide range of techniques available to establish DPD status in patients with cancer.

Nonlinear pharmacokinetics of 5-fluorouracil in rats

The effects of dose on the pharmacokinetics of 5-fluorouracil (5-FU) were investigated following intravenous administration of 5-FU at 10, 50, and 100 mg/kg to adult male Sprague-Dawley rats. Six rats were studied at each dose level. The dose-normalized area under the curve (AUC) was significantly higher after administration of 100 mg/kg (1.14 +/- 0.55 mg.h/L/mg; mean +/- SD) than after 50 mg/kg (0.50 +/- 0.18 mg.h/L/mg) or 10 mg/kg (0.43 +/- 0.11 mg.h/L/mg), indicating nonlinear elimination of 5-FU in rats. Dose- and time-average pharmacokinetic parameters were calculated by area/moment analysis. The systemic clearance of 5-FU following administration of 100 mg/kg was significantly lower (1.1 +/- 0.49 L/h/kg) than after 50 mg/kg (2.2 +/- 0.72 L/h/kg) or 10 mg/kg (2.4 +/- 0.67 L/h/kg). There was no significant difference in renal clearance values between the three doses (0.47 +/- 0.26 L/h/kg). However, nonrenal clearance was significantly lower after the 100-mg/kg dose (0.77 +/- 0.2 L/h/kg) than after the 50-mg/kg (1.65 +/- 0.49 L/h/kg) and 10-mg/kg (1.87 +/- 0.75 L/h/kg) doses. There was no significant difference between the steady-state volume of distribution values (0.91 +/- 0.36 L/kg) at the three doses. The lower nonrenal clearance following the 100-mg/kg dose compared with that after the lower doses of 5-FU suggested nonlinear metabolism of 5-FU in rats. A two-compartment pharmacokinetic model with parallel first-order (renal excretion) and Michaelis-Menten elimination from the central compartment was simultaneously fit to mean plasma 5-FU concentration versus time data for the three doses. The maximum volume (Vmax) and Michaelis constant (Km) values averaged (mean +/- SE) 8.3 +/- 2.3 mg/h and 31.6 +/- 11.9 mg/L, respectively. The information obtained in this study will be valuable for the evaluation of prodrugs of 5-FU that are designed to reduce toxicities and to improve oral bioavailability of the anticancer agent.

Analysis of cyclic feed intake in rats fed on a zinc-deficient diet and the level of dihydropyrimidinase (EC 3.5.2.2)

The body weight and feed intake of rats fed on a Zn-deficient diet for 28 d were reduced compared with those of control rats. The feed intakes of the Zn-deficient and control groups during the period were 10.2 (SE 0.3) and 15.7 (SE 0.2) g/d respectively. Cyclic variations in feed intake and body-weight changes were found in analysis not only of all the data for five rats but also that in each individual rat. Cosinor analysis revealed that the cyclical period of both the feed intake and body-weight change in the Zn-deficient rats was 3.5 (SE 0.1) d. The mesor and amplitude value of the feed intake in the Zn-deficient rats was 10.1 (SE 0.4) g/d and 3.5 (SE 0.5) g/d respectively, and that of body-weight change was 1.4 (SE 0.1) g/d and 7.9 (SE 1.3) g/d respectively. Among pyrimidine-catabolizing enzymes, dihydropyrimidinase (EC 3.5.2.2) activity showed significant retardation in the Zn-deficient rat liver with decrease of the enzyme protein. The ratio of apo-form to holo-form dihydropyrimidinase in the liver was not affected by the Zn-deficient diet.

Purification, characterization and inhibition of dihydropyrimidinase from rat liver

Dihydropyrimidinase (DHPase) was purified 564-fold over the initial rat liver extract, using heat, ammonium sulfate fractionation, DEAE-Sepharose CL-6B, carboxymethyl-Sepharose CL-6B, hydroxyapatite and Sephacryl S-300 chromatography. The purified enzyme was shown to be homogeneous by gel electrophoresis both in the presence and absence of SDS. Its molecular mass, determined by gel filtration, was 215 kDa and the subunit mass was 54 kDa. DHPase catalyzed the reversible cyclization of 5,6-dihydrouracil (H2Ura) to N-carbamoyl-beta-alanine or 5,6-dihydrothymine (H2Thy) to N-carbamoyl-beta-aminoisobutyric acid. Authentic 5-bromo-5,6-dihydrouracil (BrH2Ura) and commercially available H2Thy were racemic. However, these 5-substituted 5,6-dihydropyrimidines were hydrolyzed by over 96% and 98%, respectively, by DHPase. These results suggest that dihydropyrimidinase has no stereo specificities for 5-substituents of H2Ura. The addition of H2Ura and H2Thy competitively inhibited the enzyme activity against BrH2Ura. However, the addition of N-carbamoyl-beta-alanine or N-carbamoyl-beta-amino-isobutyric acid showed hyperbolic mixed-type inhibition, when BrH2Ura was used as the substrate. The values of the dissociation constants of BrH2Ura, N-carbamoyl-beta-alanine and N-carbamoyl-beta-aminoisobutyric acid were 17 microM, 0.38 mM and 0.38 mM, respectively. DHPase from the rat liver contains 4 mol Zn2+/mol active enzyme, presumably one atom/subunit. Zn2+ also inhibited the hydrolysis of BrH2Ura by the enzyme. The Ki for Zn2+ as an inhibitor of DHPase was 23 microM, and the maximum rate of inactivation was 0.057 min-1 at 37 degrees C. H2Ura and H2Thy protected the enzyme activity from Zn2+ inactivation.

Measurement of in vitro cellular pharmacokinetics of 5-fluorouracil in human and rat cancer cell lines and rat hepatocytes using a flow-through system

A flow-through system was used to study the cellular pharmacokinetics of 5-fluorouracil (5-FU) in four human cell lines (squamous-cell carcinoma HEp-2, colon carcinoma WiDr, hepatoma Hep G2, and breast carcinoma MCF-7) as well as in the rat hepatoma H35 cell line and in freshly isolated rat hepatocytes. The system made it possible to restrict the decrease in the concentration of 5-FU in the medium, to keep the volume in which the metabolites accumulated relatively small, and to study the dynamics of a response during and after a change in the composition of the eluent. Clearance of 5-FU from the eluent was achieved predominantly (greater than 95%) by its catabolism to dihydrofluorouracil in the tumor cell lines and to 2-fluoro-beta-alanine in the hepatocytes. Not only rat hepatocytes but also HEp-2 cells showed relatively high clearance values. A concentration-dependent 5-FU elimination was observed, indicating saturation of 5-FU elimination according to Michaelis-Menten kinetics (Km 14-22 microM). The maximal velocity (Vmax) values ranged from 0.025 to 0.13 nmol 5-FU/10(6) cells per minute. For HEp-2 cells, high-concentration pulse injections of 5-FU, thymine, uridine, or uracil immediately led to a reduction in 5-FU conversion, followed by recovery within 5 min. The flow-through system proved to be adequate for the study of the non-linear pharmacokinetics of 5-FU in different intact cells and for the comparison of various manipulations of these pharmacokinetics.

A mathematical model of the kinetics of 5-fluorouracil and its metabolites in cancer patients

A compartmental model of the kinetics of 5-fluorouracil (5-FU) and its catabolites in humans is proposed. This model was developed using data from a previous study in which plasma levels and urinary amounts of unchanged drug and metabolites were quantitated after i.v. bolus injection of 500 mg/m2 5-FU in ten patients. Biliary excretion was also quantified in two subjects. The different processes, biochemical transformations, and urinary and biliary excretion were adequately described by first-order kinetics. The technique of multiresponse modelling was used for global fitting of all data for each patient. Satisfactory agreement was achieved between measured and predicted values. This model enabled accurate evaluation of pharmacokinetic parameters that could not be adequately calculated using a model-free analysis. The total clearance and elimination half-life of 5-FU and its catabolites are reported for all subjects. The estimated mean half-life was 6.9 +/- 3.9 min for unchanged 5-FU and 225 +/- 352, 7.6 +/- 4, and 9.6 +/- 7.7 min, respectively, for the three measured catabolites dihydrofluorouracil (FUH2), alpha-fluoro-beta-ureidopropionic acid (FUPA), and alpha-fluoro-beta-alanine (FBAL). The percentage of anabolic, catabolic, urinary, and biliary elimination in total clearance was also quantitated. Anabolic clearance accounted for 39% +/- 14% of total 5-FU clearance, with substantial variation occurring among patients. Urinary clearance represented 6.5% +/- 3.2%, 0.8% +/- 0.9%, 13.2% +/- 4.7%, and 98.2% +/- 2.5% of total clearance for 5-FU, FUH2, FUPA, and FBAL, respectively. The model was also satisfactorily fitted to the data of a patient deficient in dihydropyrimidine dehydrogenase, an enzyme previously thought to be the rate-limiting step for 5-FU catabolism. In this case, catabolism was highly reduced and urinary excretion of 5-FU increased up to 64% of total drug clearance. This first global model of the kinetics of 5-FU and all of its catabolites in patients given an i.v. bolus infusion of 500 mg/m2 5-FU represents a further step toward detailed comprehensive modeling of the kinetics of this drug.

Metabolism of pyrimidine analogues and their nucleosides

The pyrimidine antimetabolite drugs consist of base and nucleoside analogues of the naturally occurring pyrimidines uracil, thymine and cytosine. As is typical of antimetabolites, these drugs have a strong structural similarity to endogenous nucleic acid precursors. The structural differences are usually substitutions at one of the carbons in the pyrimidine ring itself or substitutions at on of the hydrogens attached to the ring of the pyrimidine or sugar (ribose or deoxyribose). Despite the differences noted above, these analogues, can still be taken up into cells and then metabolized via anabolic or catabolic pathways used by endogenous pyrimidines. Cytotoxicity results when the antimetabolite either is incorporated in place of the naturally occurring pyrimidine metabolite into a key molecule (such as RNA or DNA) or competes with the naturally occurring pyrimidine metabolite for a critical enzyme. There are four pyrimidine antimetabolites that are currently used extensively in clinical oncology. These include the fluoropyrimidines fluorouracil and fluorodeoxyuridine, and the cytosine analogues, cytosine arabinoside and azacytidine.

Clinical pharmacology of 5-fluorouracil

5-Fluorouracil, first introduced as a rationally synthesized anticancer agent 30 years ago, continues to be widely used in the management of several common malignancies including cancer of the colon, breast and skin. This drug, an analogue of the naturally occurring pyrimidine uracil, is metabolised via the same metabolic pathways as uracil. Although several potential sites of antitumour activity have been identified, the precise mechanism of action and the extent to which each of these sites contributes to tumour or host cell toxicity remains unclear. Several assay methods are available to quantify 5-fluorouracil in serum, plasma and other biological fluids. Unfortunately, there is no evidence that plasma drug concentrations can predict antitumour effect or host cell toxicity. The recent development of clinically useful pharmacodynamic assays provides an attractive alternative to plasma drug concentrations, since these assays allow the detection of active metabolites of 5-fluorouracil in biopsied tumour or normal tissue. 5-Fluorouracil is poorly absorbed after oral administration, with erratic bioavailability. The parenteral preparation is the major dosage form, used intravenously (bolus or continuous infusion). Recently, studies have demonstrated the pharmacokinetic rationale and clinical feasibility of hepatic arterial infusion and intraperitoneal administration of 5-fluorouracil. In addition, 5-fluorouracil continues to be used in topical preparations for the treatment of malignant skin cancers. Following parenteral administration of 5-fluorouracil, there is rapid distribution of the drug and rapid elimination with an apparent terminal half-life of approximately 8 to 20 minutes. The rapid elimination is primarily due to swift catabolism of the liver. As with all drugs, caution should be used in administering 5-fluorouracil in various pathophysiological states. In general, however, there are no set recommendations for dose adjustment in the presence of renal or hepatic dysfunction. Drug interactions continue to be described with other antineoplastic drugs, as well as with other classes of agents.

Clinical applicability of current pharmacokinetic models: splanchnic elimination of 5-fluorouracil in cancer patients

What can be inferred from limited clinical data by using current models of hepatic elimination? We examined this question by analyzing previously published data on the steady-state uptake of the anticancer agent 5-fluorouracil (5-FU) in seven cancer patients in terms of the venous equilibration model, the undistributed and distributed forms of the sinusoidal perfusion model, and the convection-dispersion model. Because of appreciable extrasplanchnic removal of 5-FU, the value of the steady infusion rate was not used in our analysis. When the data from all patients were pooled by plotting the measured hepatic venous concentration against the measured hepatic arterial concentration, the high concentration data fell on a limiting straight line of slope 1, indicating that at high dose rates elimination of 5-FU in both the liver and gastrointestinal tract was close to saturation. The intercept of this line gave a model-independent estimate of Vmax/Q = 48.0 +/- 11.6 (SD) microM for the pooled data set, where Vmax is the maximum splanchnic elimination rate of 5-FU, and Q is the hepatic blood flow. The low concentration data points fell on a limiting straight line through the origin, from which model-dependent values of the Michaelis constant were determined. The venous equilibration model gave Km = 9.4 microM, while the undistributed sinusoidal perfusion model gave Km* = 26.5 microM. With these values of Km, both models fit the pooled data equally well. These methods were applied to analyses of the five individual data sets which contained sufficiently high concentration data points. The resulting mean values were Vmax/Q = 41.0 +/- 5.1 (sem) microM, Km = 8.4 +/- 1.3 microM and Km* = 23.2 +/- 3.2 microM. However, the splanchnic region is a highly heterogeneous organ system, for which an undistributed analysis provides no more than an upper bound on the Michaelis constant Km+ (Km+ less than or equal to Km*). A perfusion model distributed to represent total splanchnic elimination is developed in the Appendix. Using previous estimates of the degree of functional heterogeneity in the liver alone, this model yields Km+ values for individual patients which have a mean of 20.3 +/- 2.8 microM.