Evaluation of flurbiprofen urinary ratios as in vivo indices for CYP2C9 activity

Abstract

Aims

We investigated flurbiprofen pharmacokinetics in 12 volunteers to develop a phenotypic trait measure that correlates with the fractional clearance to 4′-hydroxyflurbiprofen. The effect of the CYP2C9 inhibitor fluconazole on flurbiprofen metabolism was also evaluated.

Methods

Flurbiprofen pharmacokinetics were evaluated before and after the first and seventh doses of fluconazole. The urinary recovery ratio was calculated as FLRR = 4′-OHF/ [4′-OHF + F_tot] and the urinary metabolic ratio was calculated as FLMR = 4′-OHF/F_tot, where 4′-OHF and F_tot represent total (conjugated and unconjugated) amounts recovered in urine.

Results

There was a statistically significant relationship between the 4′-OHF formation clearance (4OHCLf) and both the 8-h FLRR and the 8-h FLMR with and without administration of fluconazole. The flurbiprofen apparent oral clearance (CL/F) was decreased by 53% [90% confidence interval (CI) −58, −48] and 64% (90% CI −69, −59), respectively, after administration of one and seven doses of fluconazole when compared with administration of flurbiprofen alone; similarly, the 4OHCLf decreased by 69% (90% CI −74, −64) and 78% (90% CI −83, −73), the 8-h FLRR decreased by 35% (90% CI −41, −29) and 40% (90% CI −46, −35) and the 8-h FLMR decreased by 61% (90% CI −65, −58) and 67% (90% CI −70, −63). The magnitude of decrease in CL/F and 4OHCLf was greater after seven doses compared with after one dose of fluconazole (P < 0.005).

Conclusions

This study provides strong evidence that both the 8-h FLRR and the 8-h FLMR are suitable phenotypic indices for CYP2C9 activity.

Introduction

The cytochrome P450s (CYP) are a family of enzymes involved in the oxidative metabolism of many xenobiotics, including drugs. The expression and activity of CYP enzymes are controlled by both genetic and environmental factors. Consequently, large interindividual variability exists in the expression and activity of these enzymes resulting in wide differences in the rate and extent of metabolism of drugs between individuals [1]. The CYP superfamily consists of individual enzymes that are organized into subfamilies based on similarities in their amino acid sequences [2]. The individual members havebeen shown to be products of individual genes and exhibit substrate selectivity with some overlap [3]. This substrate selectivity has enabled the identification of selective probes for the activities of individual enzymes in the development of approaches to estimate the activity both in vitro and in vivo [4].

The CYP2C9 enzyme accounts for approximately 18% of the CYP protein content in human liver and is responsible for the metabolism of approximately 20% of drugs approved by the Food and Drug Administration [5]. These include warfarin [6], phenytoin [7], many nonsteroidal anti-inflammatory drugs (NSAIDs) such as diclofenac [8] and many sulphonylureas such as glyburide and glimepiride [9]. There are several known genetic polymorphisms associated with the CYP2C9 enzyme, the most well studied and clinically relevant being CYP2C9*2 and CYP2C9*3 [10], which have been shown to a have a significant role in contributing to interindividual variability in flurbiprofen metabolism [11]. Although it is often more cumbersome than genotyping, phenotyping provides the most clinically relevant information about enzyme activity, because it is a measure of the combined effects of genetic, environmental and endogenous factors on CYP activity [12]. Since CYP2C9 exhibits such a wide interindividual variability, there is a need for a reliable phenotypic measure of its activity.

One possible strategy for the estimation of the activity of an enzyme that catalyses a particular drug metabolism pathway is a full pharmacokinetic study of the enzyme-specific drug, and estimation of the metabolite formation clearance associated with the involved pathway. However, this is impractical for many situations, especially in epidemiological studies, where estimation of the enzyme activity in many individuals is desirable. Consequently, simpler and less intensive approaches involving limited sampling and a phenotypic measure that correlates with the metabolite formation clearance obtained from the full pharmacokinetic study is preferable [13].

Such an approach has been utilized for the CYP2C9 enzyme, mainly using the following probe drugs: tolbutamide, phenytoin, warfarin and losartan; however, each substrate is associated with limitations that make these drugs less than ideal as a probe drug for CYP2C9. For example, tolbutamide is associated with a risk of hypoglycaemia that can be mitigated by dose reduction but necessitates more sensitive quantification techniques [14]; phenytoin has a narrow therapeutic index and its renal clearance is dependent on urine flow [15]; warfarin is associated with a risk of bleeding, its CYP2C9-dependent metabolism is stereoselective and it has a long half-life [16]; and losartan pharmacokinetics do not segregate by CYP2C9 genotype [11].

Flurbiprofen, 2-(2-fluoro-4-biphenyl propionic acid), is a member of the NSAID class of drugs. It exists as a chiral compound with stereoselective disposition in humans [17]. It is almost completely available (∼92%) after oral administration and is extensively (>99%) bound to plasma proteins. Between 20 and 25% of the administered dose of flurbiprofen is excreted in urine as an acyl glucuronide and 2–3% is eliminated as unchanged drug; flurbiprofen renal elimination is independent of urine flow rate [18]. The remainder of the dose is excreted as metabolites (4′-hydroxy, 40–47%; 3′-hydroxy-4′-methoxy, 20–25%; and 3′,4′-dihydroxy, 5%). Approximately 60–70% of the dose is eliminated after glucuronidation, predominantly at the carboxyl group [19]. CYP2C9 has been shown to be exclusively involved in the formation of 4′-hydroxyflurbiprofen, thus raising the possible candidacy of flurbiprofen as a selective probe for this enzyme [20–22]. Furthermore, it has been shown that there is a lack of stereoselectivity in the formation of 4′-hydroxyflurbiprofen, which supports the use of the currently marketed racemate as a probe drug [21, 23].

An additional strategy for determining CYP enzyme participation in drug disposition involves the assessment of metabolism after selective manipulation of CYP activity, either by enzyme induction or inhibition [24]. Fluconazole is an azole antifungal drug that has been shown to be a potent inhibitor of CYP2C9-mediated reactions and a modest inhibitor of metabolism by CYP3A4 [25]. It has been used in multiple in vitro as well as in vivo studies, all showing a significant reduction of CYP2C9 activity with different doses and lengths of exposure to fluconazole [5, 25–27]. The effect of fluconazole on the pharmacokinetics of established probe drugs for CYP2C9, such as tolbutamide [28], phenytoin [29], warfarin [30–32] and losartan [33], has been examined. A recent study showed a 43% reduction in flurbiprofen oral clearance after single-dose administration of fluconazole 200 mg, but the effect on 4′-hydroxyflurbiprofen formation clearance and urinary phenotypic trait measures was not evaluated [34].

The purpose of this study was to test the hypothesis that flurbiprofen can be used as a probe drug to determine the in vivo activity of CYP2C9, and to validate a reliable urinary phenotypic marker for the CYP2C9-mediated 4′-hydroxylation of flurbiprofen. Furthermore, in order to provide further evidence for the use of flurbiprofen as a probe for CYP2C9, we evaluated the inhibitory effect of both single and seven daily doses of fluconazole on the in vivo metabolism and pharmacokinetics of flurbiprofen.

Methods

Patient selection and clinical protocol

Twelve healthy volunteers (age ± SD, 37 ± 3.1 years; body mass index ± SD, 22.59 ± 0.96 kg m⁻²; seven females and five males; 10 Whites and two African-Americans; seven nonsmokers and five smokers) participated in the study after informed consent was obtained. The study protocol was approved by the Institutional Review Board of the University of Pittsburgh. The subjects underwent screening tests based on history, physical examination and routine biochemical and urinalysis tests. Subjects were excluded if there was any evidence of underlying disease or concurrent drug administration, including NSAIDs, and women must not have been pregnant at the time of the study. All subjects had to abstain from alcohol (of any form) for 8 days and avoid consuming any caffeine-containing products and grapefruit juice for 48 h before and during the period of the study. Subjects were specifically questioned about the last time they consumed alcohol or any caffeine-containing products.

The pharmacokinetics of flurbiprofen were evaluated before (visit A) and after the first and seventh doses of fluconazole 400 mg (visits B and C, respectively). Visits A and B (before and after fluconazole) were separated by at least 2 weeks. Patients were specifically asked about compliance with intake of the 7-day fluconazole course. Racemic-Flurbiprofen (Ansaid®) 50 mg was administered orally after an overnight fast on each visit and administered 2 h after the single or seventh dose of fluconazole (two tablets of Diflucan® 200 mg once daily). Prior to drug administration, 40 ml of blood was obtained. Following drug administration, blood samples (20 ml) were obtained at 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 10, 12 and 24 h. Total voided urine was collected before and at the following intervals after flurbiprofen administration: 0–2, 2–4, 4–6, 6–8, 8–10, 10–12 and 12–24 h. Subjects stayed at the University of Pittsburgh General Clinical Research Center for 12 h during each study visit, and returned the next morning to complete the overnight 12–24-h urine collection and provide the 24-h blood sample. Plasma was obtained from the blood samples by centrifugation and, together with aliquots of the urine samples, was stored at −20°C until analysis.

Analytic methods

Plasma flurbiprofen and urine 4′-hydroxyflurbiprofen and flurbiprofen concentrations were quantified by a high-performance liquid chromatographic (HPLC) assay previously developed and described by Hutzler et al. [35]. Briefly, urine samples (50 µl) were prepared by adding 400 µl of purified water and 100 µl 6 m hydrochloric acid, followed by vortex mixing and incubation at 90°C for 30 min to facilitate glucuronide cleavage via acid hydrolysis. After incubation, 500 µl acetonitrile containing the internal standard (1000 ng ml⁻¹ of 2-fluoro-4-biphenylacetic acid) was added and the samples were vortex-mixed. The samples were then centrifuged at 14 000 g for 5 min and 150 µl of sample was transferred to autosampler vials and 10 µl was injected onto the HPLC system for analysis. This approach uses acid hydrolysis to facilitate glucuronide cleavage, since NSAIDs such as flurbiprofen are subject to acyl migration of the glucuronide and, thus, resistant to β-glucuronidase cleavage; therefore, the urinary 4′-hydroxyflurbiprofen and flurbiprofen that we have measured are total concentrations comprised of both conjugated and nonconjugated forms (4′-hydroxyflurbiprofen is predominately conjugated on the carboxylic acid group with much less conjugation occurring on the phenolic hydroxy group). Plasma samples (100 µl) were prepared by adding 100 µl acetonitrile, 200 µl internal standard (500 ng ml⁻¹ of 2-fluoro-4-biphenylacetic acid in acetonitrile), 40 µl of half-strength phosphoric acid, followed by vortex mixing. The samples were then centrifuged at 14 000 g for 5 min, and 150 µl of sample transferred to autosampler vials and 10 µl was injected onto the HPLC system for analysis. Only unchanged flurbiprofen was measured in plasma as we did not do the deconjugation step (acid hydrolysis). The coefficients of variation for intraday, interday, freeze-thaw, and in process stability were consistently <14%. The lower limit of quantification was 0.25 µg ml⁻¹ for 4′-hydroxyflurbiprofen and flurbiprofen in urine and flurbiprofen in plasma. Flurbiprofen (racemic) was obtained from Sigma (St Louis, MO, USA); 4′-hydroxyflurbiprofen and 2-fluoro-4-biphenyl acetic acid (internal standard) were gifts from the former Pharmacia (Kalamazoo, MI, USA).

The plasma concentrations of fluconazole were determined by liquid chromatography–mass spectrometry (LC/MS). Plasma (0.1 ml) was combined with glycine buffer, pH 11.0 (0.1 ml) and ketoconazole (internal standard; 50 ng), briefly vortex-mixed and then extracted with water-saturated ethyl acetate (3 ml). Samples were shaken on low speed for 10 min and then centrifuged at 3000 g for 10 min. The organic layer was transferred to a clean tube and evaporated under a stream of nitrogen at 40°C. The residue was reconstituted in 0.2 ml of water, acetonitrile and formic acid [73 : 27 : 0.1 (v/v/v)], transferred to autosampler vials and then a portion (5 µl) was injected onto the HPLC system. The LC/MS system consisted of a Waters 2690 Alliance separation module (Waters, Milford, MA, USA) coupled to an aQa single quadrupole mass spectrometer (ThermoFinnigan, San Jose, CA, USA) equipped with an electrospray ionization (ESI) source. The mass spectrometer was operated in positive ESI mode with a probe voltage of 3.75 kV; detection was by selected ion monitoring of 307 m/z and 531 m/z for fluconazole and ketoconazole, respectively. Analytes were chromatographed on a BetaBasic C18 analytical column (3.0 × 150 mm, 5 µm; Thermo Keystone, Bellefonte, PA, USA) with a gradient mixture of (A) 0.1% formic acid in water and (B) 0.1% formic acid in acetonitrile that was delivered at a flow rate of 0.5 ml min⁻¹. The starting conditions were 40% mobile phase B, which was held for 1.2 min and then increased over 1 min to 70% mobile phase B, which was maintained for 1.3 min until returning over 0.25 min to the starting conditions for 2 min until the next injection. Signal output was captured with Xcalibur software version 1.2 (Thermo). The lower limit of quantification was 0.1 µg ml⁻¹ (0.33 µmol l⁻¹). Intra- and interday assay coefficients of variation for fluconazole were <10% at quality control sample concentrations of 0.75, 7.5 and 12.5 µg ml⁻¹. The assay was accurate to within 4% of the nominal concentration over the standard curve concentration range of 0.1–15 µg ml⁻¹ (0.33–49.0 µmol l⁻¹).

Data analysis

Plasma concentration–time data were analysed by noncompartmental methods. Pharmacokinetic parameters including area under the concentration–time curve (AUC) were estimated using WinNonlin (Pharsight, Palo Alto, CA, USA). The area under the concentration–time curve for flurbiprofen (AUC_∞) was determined using the trapezoidal rule with extrapolation to infinity. Flurbiprofen apparent oral clearance was calculated as CL/F = Dose/AUC. The 4′-hydroxyflurbiprofen formation clearance was calculated as 4OHCLf = total amount of 4′-hydroxyflurbiprofen recovered in urine (24 h)/flurbiprofen AUC₀₋₂₄. The flurbiprofen glucuronide formation clearance was estimated as FGCLf = total amount of flurbiprofen (glucuronide + unchanged) recovered in urine (24 h)/flurbiprofen AUC₀₋₂₄. The FGCLf calculated is only an estimate because the amount of unchanged flurbiprofen eliminated in urine was not quantified. Therefore, the amount recovered in urine reflects conjugated and unconjugated flurbiprofen, the latter of which would account for approximately 10% of the total amount of flurbiprofen recovered in urine after acid hydrolysis. The urinary flurbiprofen recovery ratio was calculated as FLRR = 4′-OHF/ [4′-OHF + F_tot] and the urinary flurbiprofen metabolic ratio was calculated as FLMR = 4′-OHF/F_tot, where 4′-OHF is the total (conjugated and unconjugated) amount of 4′-OHF and F_tot is the total amount of flurbiprofen recovered in urine as the acyl glucuronide and unchanged flurbiprofen (after acid hydrolysis).

Genotyping

Two unique identifying single nucleotide polymorphisms (SNP) referred to by the db SNP reference number were genotyped: CYP2C9*2, rs1799853 and CYP2C9*3, rs1057910. Variant alleles were screened using commercially available TaqMan® allele discrimination assay kits (Applied Biosystems, Foster City, CA, USA). Blinded duplicate sample analyses were performed for all genotyping assays. An additional 10% of samples were repeated to verify the reproducibility of the assay. All results were independently interpreted by two laboratory personnel.

Statistical analysis

Data generated from visit A were used to validate a phenotypic index for the activity of CYP2C9. The relationships between the phenotypic descriptors (such as the urine FLRR and urine FLMR) and the formation clearance of flurbiprofen by 4′-hydroxylation, the gold standard measure of enzyme activity evaluated from the full pharmacokinetic study, were evaluated. A priori, we defined that a suitable phenotypic measure would be one that shows a strong correlation to the partial clearance by Spearman rank correlation coefficient (r > 0.7) with a significant P-value (P < 0.05). When more than one descriptor matched our definition, we chose the one for which the time point fit more with the already established Pittsburgh protocol (e.g. the 0–8-h time point) [36]. We also preferred the urinary phenotypic indices, because urine sample handling would be more convenient for sample analysis of parent and metabolite [35]. This is important for future applications of the CYP2C9 phenotypic index.

The primary end-point under consideration in this study is the flurbiprofen AUC. Sample size calculations based on an estimated within-subject CV of 20% for flurbiprofen AUC indicate that a sample size of N = 12 subjects will allow for the detection of an effect size of 1.0 (paired, two-sided t-test), with 80% power and 5% type I error. This translates into the ability to detect a within-subject change in AUC of ≥20%. In order to investigate the inhibitory role of fluconazole on the CYP2C9-mediated 4′-hydroxylation of flurbiprofen, data generated from all three visits were compared by analysis of variance followed by the Sidak's multiple comparison test. We also tested for the presence of interaction based on a 90% confidence interval (CI) for the ratio of geometric means (visit B/visit A, and visit C/visit A) for the flurbiprofen AUC_∞ and the maximum observed concentration (C_max). A pharmacokinetic interaction was inferred if the 90% CI for the geometric mean ratio (expressed as a percent) fell outside the 80–125% interval for both AUC_∞ and C_max [37, 38]. Clearance values were log transformed prior to analysis. Differences were regarded as statistically significant when P-values were <0.05. All analyses were performed using Stata 9 statistical software (Stata Corp. LP, College Station, TX, USA).

Results

Subjects reported no significant adverse effects after administration of flurbiprofen, fluconazole, or the combination of both drugs. Four individuals were heterozygous carriers of CYP2C9*2, and there was no obvious change in the apparent enzyme activity. Table 1 lists the pharmacokinetic parameter estimates from each study visit and the flurbiprofen plasma concentration–time profiles are shown in Figure 1. Data from visit A (control phase) show that flurbiprofen was readily absorbed, with plasma concentration peaking at 2.3 h and a mean half-life of 4.3 h. The amounts of flurbiprofen (acyl glucuronide and unchanged) recovered in urine at 8 h were (in mg) 5.2 ± 1.8, 6.8 ± 2.2 and 6.3 ± 2.4; and for 4′-OHF 9.9 ± 2.8, 5.1 ± 1.7 and 4.0 ± 1.4 (mean ± SD) at visits A, B and C, respectively. The amounts recovered in urine at 24 h were (in mg) 7.1 ± 2.7, 10.7 ± 3.1 and 11.1 ± 3.2; and for 4′-OHF 15.0 ± 3.6, 10.1 ± 3.9 and 9.1 ± 3.6 (mean ± SD) at visits A, B and C, respectively. The 4OHCLf was decreased by 69% (90% CI −74, −64) and 78% (90% CI −83, −73) and the FGCLf was decreased by 28% (90% CI −36, −21) and 43% (90% CI −50, −36), after administration of one and seven doses of fluconazole, respectively.

Table 1.

Mean flurbiprofen kinetic parameter estimates^*

Variable	Units	Visit A	Visit B	Visit C
AUC∞	µg l−1 h−1	50.6 ± 17.6	109.3 ± 34.4†	141.9 ± 34.2†
CL/F	l h−1	1.09 ± 0.35	0.50 ± 0.16†	0.38 ± 0.11†
Cmax	µg ml−1	9.4 ± 2.5	11.7 ± 2.3	12.9 ± 2.8‡
Tlag	h	0.31 ± 0.47	0.13 ± 0.13	0.19 ± 0.19
Tmax	h	2.3 ± 1.0	1.9 ± 0.8	1.9 ± 0.6
V/F	l	6.12 ± 1.11	5.45 ± 1.06	5.00 ± 0.92‡
T1/2	h	4.3 ± 1.6	8.0 ± 2.0†	9.5 ± 1.6†
4OHCLf	l h−1	0.369 ± 0.190	0.120 ± 0.068†	0.091 ± 0.048†
FGCLf		0.144 ± 0.042	0.104 ± 0.041†	0.084 ± 0.038†
8-h FLRR		0.652 ± 0.109	0.432 ± 0.146†	0.396 ± 0.127†
8-h FLMR		2.315 ± 1.731	0.912 ± 0.686‡	0.744 ± 0.481†

Data are mean values ± SD. Visit A, flurbiprofen 50 mg alone; visit B, flurbiprofen 50 mg + fluconazole 400 mg; and visit C, flurbiprofen 50 mg after 7 days fluconazole 400 mg daily. AUC_∞, area under the plasma concentration–time curve extrapolated to infinity; CL/F, total clearance; 4OHCLf, 4′-hydroxyflurbiprofen formation clearance; FGCLf, sum of flurbiprofen glucuronide formation clearance and flurbiprofen renal clearance; FLRR, flurbiprofen recovery ratio = 4′-OHF/4′-OHF+ F; FLMR, flurbiprofen metabolic ratio = 4′-OHF/F.

^*There were no statistically significant differences in the comparison between visit B and visit C.

^†Comparison between visit B or visit C and visit A: P < 0.001.

^‡Comparison between visit B or visit C and visit A: P < 0.05.

Figure 1.

Mean (±SE; N = 12) plasma flurbiprofen concentration–time profiles after oral administration of flurbiprofen (50 mg) alone (squares) and after the first dose (triangles) and seventh dose (circles) of fluconazole 400 mg

The data from the control phase (visit A) of this pharmacokinetic study were used to evaluate the relationship between 4′-hydroxyflurbiprofen formation clearance and various single point measures that could serve as a phenotypic index of CYP2C9 activity. There was a significant relationship between formation clearance and the flurbiprofen plasma concentration at 10 h (Spearman correlation r = −0.827, P = 0.0017) after drug administration, but there was no significant relationship for any of the other time points. When data points for all three study sessions were included, there was a significant correlation at 6 h (−0.531, P = 0.0009), 8 h (−0.611, P = 0.0001), 10 h (−0.513, P = 0.0022) and 12 h (−0.659, P < 0.0001) after drug administration. None of these correlation values exceeded our defined threshold and so plasma-based trait measures were not evaluated further.

The urinary ratios (FLRR and FLMR) were calculated for each urine collection interval starting at zero and ending at 2, 4, 6, 8, 12 and 24 h after drug administration (i.e. 0–2, 0–4, etc.). The results showed similar high correlation values (Spearman r > 0.76, P < 0.006) for each collection interval and each ratio. When the data from all three study visits were included, there were also high correlation values (Spearman r > 0.88, P < 0.0001) for each collection interval and each ratio. The relationships between the 8-h FLRR and FLMR urinary ratios and formation clearance for all visits are depicted in Figure 2.

Figure 2.

Relationship between the formation clearance of 4′ hydroxyflurbiprofen (4OHCLf) and the urinary flurbiprofen recovery ratio (FLRR; left panel) and the urinary flurbiprofen metabolic ratio (FLMR: right panel) when administered alone (visit A; squares), when administered 2 h after one dose of fluconazole (visit B; triangles), and when administered after the seventh dose of fluconazole (visit C; circles). The Spearman rank correlation coefficient for both was 0.878, P < 0.0001

The effects of single-dose and 7-day pretreatment with fluconazole are also provided in Table 1 (visit B and visit C, respectively). There were statistically significant changes in many of the parameters, including FLRR, FLMR and flurbiprofen 24-h plasma concentrations (Figures 1 and 3). The flurbiprofen apparent oral clearance (CL/F) was decreased by 53% and 64% after coadministration of one and seven doses of fluconazole, respectively, when compared with administration of flurbiprofen alone; similarly, the 4OHCLf was decreased by 69% and 78%, the 8-h urinary flurbiprofen recovery ratio was decreased by 35% and 40% and the 8-h urinary FLMR was decreased by 61% and 67%. The same trend was observed when the change in CL/F, 4OHCLf and FLRR values was compared on an individual level.

Figure 3.

Percent change in flurbiprofen apparent oral clearance (CL/F), 4′-hydroxyflurbiprofen formation clearance (4OHCLf), flurbiprofen 8-h urinary recovery ratio (FLRR) and flurbiprofen 8-h metabolic ratio (FLMR) after administration of flurbiprofen alone, following coadministration of flurbiprofen and one dose of fluconazole (triangles), or following administration of fluconazole for 7 days (circles)

The mean plasma concentration–time profiles of fluconazole after one and seven doses are shown in Figure 4. The C_max, average concentration (C_avg) and AUC₀₋₁₂ of fluconazole were 16.6 ± 3.3 µmol l⁻¹ (mean ± SD), 12.9 ± 2.5 µmol l⁻¹ and 154.4 ± 30.3 µmol l⁻¹ h⁻¹ after single-dose administration and 37.7 ± 8.9 µmol l⁻¹, 30.6 ± 7.3 µmol l⁻¹ and 367.2 ± 88.1 µmol l⁻¹ h⁻¹ after multiple-dose administration. There was a significant correlation between the C_avg of fluconazole and the percent change in flurbiprofen CL/F (Spearman r = −0.644, P = 0.0007), 4OHCLf (r = −0.533, P = 0.0074) and the FLMR (r = −0.436, P = 0.033) with greater inhibition seen at higher fluconazole concentrations.

Figure 4.

Mean (±SE; N = 12) 12-h plasma fluconazole concentration–time profiles after the first (triangles) and seventh dose (circles) of fluconazole 400 mg

There was a significant increase in the flurbiprofen AUC_∞ (geometric mean ratio, 222% for visit B and 299% for visit C; P < 0.001); and the 90% CI of the geometric mean ratio between the two treatment periods was not included in the 80–125% bioequivalence range: 190–244% for visit B vs. visit A and 237–336% for visit C vs. visit A. There was a moderate increase in C_max (geometric mean ratio, 130% for visit B and 145% for visit C), which was significant at only visit C; furthermore, the 90% CI of the geometric mean ratio between the two treatment periods was not entirely included within the 80–125% bioequivalence range: 107–144% for visit B vs. visit A and 117–161% for visit C vs. visit A. The magnitude of inhibition after seven doses of fluconazole (visit C) was significantly greater (P < 0.005) than after a single dose (visit B).

Discussion

Fluconazole is a potent inhibitor of CYP2C9 activity and in this study produced concentration-dependent inhibition of flurbiprofen 4′-hydroxylation, which is a specific marker of CYP2C9 activity. This study provides evidence that both the 8-h urinary FLRR and the 8-h urinary FLMR can be used as single-point phenotypic indices for the fractional clearance of flurbiprofen to 4′-hydoxyflurbiprofen, and by implication to the enzymatic activity of CYP2C9. Furthermore, inhibition of the 4′-hydroxylation of flurbiprofen by fluconazole provided additional evidence on the importance of CYP2C9 in the in vivo metabolism of flurbiprofen.

Validation of flurbiprofen as a selective probe of CYP2C9

There was a good relationship between both the urinary FLRR and FLMR and 4′-hydroxyflurbiprofen formation clearance for each urine collection interval evaluated. Therefore, for practical reasons we chose to use the 0–8-h urine collection interval, because this interval was already being used in our cocktail approach. Although both the 8-h urinary FLRR and the urinary FLMR were significantly correlated with the 4′-hydroxyflurbiprofen formation clearance, the FLRR may be a more robust phenotypic index for the activity of CYP2C9 due to its insensitivity to phenotypic extreme metabolizers [39]. The recovery ratio was originally proposed for use with debrisoquine by Inaba et al. on the basis that it was less variable than the metabolic ratio [40]. Lee et al. have previously used flurbiprofen as a probe for measurement of the activity of CYP2C9, and compared the resultant 0–12-h urine metabolic ratio (MR0–12) with that of tolbutamide and losartan [41]. Although the flurbiprofen MR0–12 was significantly correlated with the formation clearance of its metabolite (r = 0.71), they concluded that tolbutamide is a better probe drug for CYP2C9 because it exhibited a higher correlation coefficient (r = 0.84). However, these investigators did not report the urinary recovery ratio and, thus, direct comparisons of the methods are difficult. Note that we have calculated the FLMR as described by Lee et al., whereby the formula is reciprocal to the commonly used CYP2D6 metabolic ratio.

We used flurbiprofen as a CYP2C9 probe drug based on the fact that it is exclusively and nonstereoselectively metabolized by CYP2C9 in vitro and CYP2C9 genotype is a significant predictor of flurbiprofen total clearance and 4′-hydroxyflurbiprofen formation clearance in humans [22, 41]. Flurbiprofen has a wider therapeutic index than warfarin, phenytoin or tolbutamide, and hence is potentially a safer probe drug to be administered [15, 16, 19]; in contrast to phenytoin, its elimination does not depend on urine flow [18, 19]; and more flurbiprofen is excreted unchanged in urine compared with tolbutamide, which facilitates reliable urinary concentration measurements [41].

Inhibition of CYP2C9 by fluconazole

It is well recognized that fluconazole, when administered at the doses used in the present study, is a potent CYP2C9 inhibitor. Although it has been shown by Black et al. that fluconazole inhibits the in vitro activity of CYP3A4 and other undefined isoforms in addition to CYP2C9, this would not affect the results of our study because flurbiprofen hydroxylation is exclusively mediated by CYP2C9 [22, 31]. Our results are consistent with those observed after a single dose of fluconazole 200 mg and collectively confirm that fluconazole is a significant inhibitor of the metabolism of flurbiprofen in vivo [34]. The inhibitory effect of fluconazole occurs with the first dose and is increased slightly with continuous administration of the drug due to accumulation to steady-state concentration values. The extent of this inhibition provides additional support for the concept that flurbiprofen is a selective probe drug for CYP2C9 activity. Furthermore, the relationships observed between flurbiprofen formation clearance and FLRR and FLMR using all data points give more evidence for the utility of these ratios as phenotypic indices for CYP2C9 activity, as they maintain robustness over a wide range of enzymatic activities.

The flurbiprofen parameters observed in visit A (absence of inhibitor), mainly CL/F, volume of distribution (V/F), T_max and half life, are comparable to those reported by others [19, 42–45]. Inhibition with fluconazole led to significant changes in the parameters that reflect metabolism. Although there was a significant increase in the AUC_∞ with the coadministration of fluconazole, C_max only changes moderately, and the 90% CI of the geometric ratio results suggests that this change is not clinically significant. It appears that flurbiprofen is a drug with intermediate intrinsic clearance. If the intrinsic clearance had been high, we would expect a marked increase in C_max with fluconazole coadministration, suggesting loss of the first-pass hepatic effect; on the other hand, if the intrinsic clearance had been low, we would expect no change in C_max. In practice, we have observed a 30% and a 45% increase in C_max with one and seven doses of fluconazole, respectively. This is supported by the findings of Tracy et al., who have recently shown that the metabolism of flurbiprofen by CYP2C9 follows the standard Michaelis–Menten pharmacokinetic model with an intrinsic clearance for the wild-type genotype of CYP2C9 of 0.49 µl min⁻¹ pmol⁻¹ P450 [46]. The V/F was significantly smaller on visit C, which may be due in part to a change in first-pass metabolism leading to an increase in bioavailability and a decrease in V/F. Fluconazole is known to selectively inhibit glucuronidation by UGT2B7 [47], which is an enzyme responsible for flurbiprofen conjugation and is present in both the intestine and liver [48, 49]. Thus, the inhibition of CYP2C9 and UGT2B7 by fluconazole may have decreased first-pass metabolism. The FGCLf, which reflects the sum of flurbiprofen renal clearance and flurbiprofen glucuronide formation clearance, was decreased by fluconazole administration. Because little flurbiprofen is eliminated unchanged in urine (<3% of the dose), this largely reflects a decrease in UGT2B7-mediated flurbiprofen conjugation. Additional mechanisms that may also contribute to the decrease in V/F include an interaction of fluconazole with drug transporters, although fluconazole does not affect digoxin transport in vitro by P-glycoprotein-expressing MDCK cells; or an effect on flurbiprofen protein binding, but this is unlikely to occur given that fluconazole is only minimally bound to plasma proteins (11–12%) and does not alter warfarin plasma protein binding [25, 32].

In this study, flurbiprofen hydroxylation after one and seven doses of fluconazole was studied to explore the concentration response relationship of inhibition by fluconazole. The average total fluconazole concentrations were 12.9 ± 2.5 µm and 30.6 ± 7.3 µm after the first and seventh doses, which are similar to concentration values reported previously. These fluconazole concentrations are in the range of the K_i values reported for CYP2C9-mediated (S)-warfarin 7-hydroxylation both in vitro (7–8 µm) and in vivo (∼20 µm). Neal et al. have shown that a plot of the ratio (control/inhibited) of metabolite formation clearances vs. inhibitor steady-state plasma concentrations can be used to estimate the in vivo K_i, as the relationship is expected to be linear with an intercept equal to 1 and a slope equal to 1/(in vivo K_i) [50]. Therefore, in this study, based on the average fluconazole concentrations and ratio of 4′-hydroxyflurbiprofen formation clearance values, the in vivo K_i is estimated to be 10 µm. These results are consistent with the findings of Blum et al., who evaluated the inhibitory effect of 200 mg fluconazole on the metabolism of phenytoin at 4 and 21 days [29], Kaukonen et al., who studied the combination of one dose of fluconazole 400 mg with losartan [33], and Black et al., whereby a 6-day course of 400 mg of fluconazole inhibited the CYP2C9-mediated metabolism of warfarin in vivo [31].

Limitations of the study

A limitation of this study is that blood and urine were collected for only 24 h, even after inhibition by fluconazole. Therefore, the pharmacokinetic parameter estimates are not precise, as the mean half-lives of flurbiprofen under inhibited conditions ranged from 8 to 9.5 h. The percent difference between AUC_∞ and AUC_{24 h} went up from 8.1% at baseline to 14.6% and 23.5% with concomitant single dose and seven doses of fluconazole, respectively.

Another limitation is that we did not measure the amount of 4′-hydroxyflurbiprofen in plasma, and hence we were not able to calculate plasma metabolic ratios, which have been shown to have certain advantages over urinary ratios [51]. Finally, we have used acid hydrolysis to facilitate glucuronide cleavage in urine, and hence the urinary flurbiprofen that we have measured is a total concentration comprised of both the conjugated and nonconjugated forms.

Conclusion

This study confirms that fluconazole is a potent in vivo inhibitor of CYP2C9-mediated metabolism and may modify the dose requirements for CYP2C9 substrates. It also provides strong evidence that both the 8-h flurbiprofen recovery ratio (FLRR) and the 8-h urinary flurbiprofen metabolic ratio (FLMR) can be used as phenotypic indices for the activity of CYP2C9.