Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

There has been a growing interest in predicting in vivo metabolic drug-drug interactions from in vitro systems. In the case of inhibition-based interactions, there is no consensus on the methodology for accurate predictions of the extent of in vivo inhibition based on in vitro data (Schmider et al., 1999; Yamano et al., 1999; Kohl and Steinkellner, 2000; Komatsu et al., 2000; Yao and Levy, 2002). Several issues remain unsolved, such as estimations of inhibition constants in vitro and inhibitor concentration around the enzyme site in vivo. For example, studies on fluvoxamine inhibition of CYP1A2 have shown that in vitro K_i values varied with microsomal protein concentration (Yao et al., 2001), suggesting that the concentration of microsomal protein present in the incubation is a factor contributing to the variance in in vitro K_i. However, even after correction for nonspecific binding of fluvoxamine in microsomes, there was still a 10-fold difference between the in vitro inhibition constant and the corresponding in vivo inhibition constant based on unbound fluvoxamine concentration in plasma.

The 10-fold underprediction of fluvoxamine inhibition potency toward CYP1A2 activity in vivo, based on in vitro data, may be due to a number of factors. These include, but are not limited to, active uptake of fluvoxamine from plasma into hepatocytes, thereby increasing the amount of inhibitor available to the enzyme (partitioning), the presence of inhibitory metabolites of fluvoxamine in plasma, and/or environmental differences that may differentially affect enzyme behavior or affinity in the two systems. Should it be the case that some type of partitioning alone is the dominant factor governing the in vitro-in vivo difference, one might expect that the RK_i (the ratio of the in vitro K_i to the in vivo K_i based on unbound concentration) of any enzyme inhibited by fluvoxamine would be similar to that of CYP1A2. In essence, RK_i would be largely enzyme independent. However, if inhibitory metabolites or enzyme environment play important roles in promoting the RK_i difference, RK_i values might become enzyme-dependent. A suitable candidate enzyme was sought to test the hypothesis that RK_i values will be conserved for a single inhibitor among the family of P450¹ enzymes.

There is some in vivo evidence that fluvoxamine also inhibits the clearance of a number of drugs metabolized by CYP2C19, such as mephenytoin, chloroguanide, and diazepam (Perucca et al., 1994a; Xu et al., 1996; Jeppesen et al., 1997). In the case of the fluvoxamine-mephenytoin interaction, the 0- to 8-h urine S/R ratio of mephenytoin increased from 0.16 to 0.55 after 100 mg/day fluvoxamine treatment for 2 weeks (Xu et al., 1996). Fluvoxamine also inhibited the CYP2C19 catalyzed bioactivation of chloroguanide: the formation clearance of 4-chlorophenylbiguanide decreased from 97 to 11 ml/min in CYP2C19-extensive metabolizers treated with 100 mg/day fluvoxamine for 6 days (Jeppesen et al., 1997). Diazepam oral clearance decreased from 0.4 to 0.14 ml/min/kg after treatment with 100 to 150 mg/day fluvoxamine for 4 days (Perucca et al., 1994a). The extent of inhibition observed in vivo is underpredicted with the reported in vitro K_i values of fluvoxamine toward CYP2C19 (0.087-0.69 µM; Rasmussen et al., 1998; Olesen and Linnet, 2000), when they are used in conjunction with the unbound plasma concentration of fluvoxamine. Therefore, it seems that a gap exists between in vitro and in vivo inhibition potencies of fluvoxamine toward CYP2C19.

The present study used (S)-mephenytoin as a probe substrate for CYP2C19 (CDER, 1999; Ring and Wrighton, 2000) to test the hypothesis that RK_i is enzyme-independent. Both in vitro and in vivo K_i values were determined for fluvoxamine inhibition of CYP2C19 using the formation of (S)-4-hydroxy-mephenytoin (4-OH-meph) as a measure of CYP2C19 activity, after confirming that steady-state kinetics also apply for CYP2C19. In vitro inhibition studies for this enzyme examined the effects of nonspecific binding. Determination of the in vivo inhibition constant used three doses of fluvoxamine in healthy subjects to provide a wide range of fluvoxamine concentrations in blood to firmly establish the relationship between fluvoxamine concentration and its effect on enzyme activity in vivo.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Reagents. NADPH and 5-(4'-hydroxyphenyl)-5-phenylhydantoin [p-HPPH, internal standard (IS)] were purchased from Sigma-Aldrich (St. Louis, MO). 4-OH-meph was purchased from Sigma/RBI (Natick, MA). Nirvanol was kindly provided by Dr. Stephen Hall (Indiana University, Indianapolis, IN). Fluvoxamine maleate was obtained from Solvay Duphar (Marietta, GA) and norfluoxetine maleate (IS) was obtained from Eli Lilly & Co. (Indianapolis, IN). A BCA protein assay kit was purchased from Pierce Chemical (Rockford, IL). All organic solvents and other chemicals were of high-performance liquid chromatography grade.

Human Liver Microsomes (HLMs) and Recombinant CYP2C19. HLMs were prepared by a standard technique (Meier et al., 1983) from six nonfatty donor livers (HL105, HL114, HL134, HL135, HL152, and HL157) from the University of Washington Human Liver Bank. Microsomes containing cDNA-expressed CYP2C19 (Supersomes) were purchased from BD Gentest (Woburn, MA).

In Vitro Studies.

Nonspecific binding of fluvoxamine in human liver microsomes and Supersomes (f_u,mic) Fluvoxamine (0.3 µM) was incubated with various amounts of microsomal protein (0.5 and 1 mg/ml from pooled HLMs, n = 6; 0.15 mg/ml for Supersomes and 0.5 mg/ml from six individual HLMs) at 37°C for 30 min in 100 mM potassium phosphate buffer (pH 7.4). Each sample (1 ml) was transferred into a Centrifree Micropartition System unit (Amicon Division, W.R. Grae & Co., Beverly, MA) and centrifuged at 1000g for 15 min at 37°C. Fifty microliters of norfluoxetine (IS, 10 µg/ml) was added to 100 µl of the filtrates. The filtrates were injected directly onto the LC-MSD (positive electrospray ionization). The value of f_u,mic was calculated as the ratio of the filtrate concentration resulting from incubation with protein to the filtrate concentration resulting from incubation without protein. HLM protein concentrations were measured using the BCA protein assay kit, and Supersomes protein concentration was calculated from the product information supplied by the manufacturer.

Tests for time-dependent inhibition of CYP2C19. Fluvoxamine (0.5 µM) and pooled HLMs (1 mg/ml) were preincubated at 37°C with and without NADPH for 0, 5, 15, and 30 min. One hundred-microliter aliquots were removed and transferred into mixtures (900 µl) containing (S)-mephenytoin (0.25 mM), NADPH (1 mM), and EDTA (1 mM) in 0.1 M potassium phosphate buffer at 37°C. The (S)-mephenytoin metabolic reaction was carried out for 15 min. In another experiment designed to test the possibility of fluvoxamine time-dependent inhibition of CYP2C19, human liver microsomes (1 mg/ml) and (S)-mephenytoin (0.1 mM) were coincubated with fluvoxamine (0.6 µM) or without fluvoxamine. After 3-min preincubation, the reaction was initiated by the addition of NADPH at 1 mM and allowed to proceed for 5, 15, 20, 25, and 30 min. The reactions were terminated by transferring each incubation mixture into a tube containing 300 µl of ice-cold 2% zinc sulfate, followed by vortex mixing.

Inhibitory effect of fluvoxamine on the metabolism of (S)-mephenytoin in microsomal systems. The incubation mixture consisted of 1.0 or 0.5 mg/ml of total human liver microsomal protein, or 0.15 mg/ml (17 pmol of CYP2C19) of Supersomes, (S)-mephenytoin (15, 50, and 150 µM), and fluvoxamine (0, 0.1, 0.3, and 0.9 µM) in 0.1 M potassium phosphate buffer (pH 7.2). After preincubation for 5 min at 37°C, the reaction was initiated by the addition of NADPH (1 mM) and allowed to proceed for 30 min. The reaction was terminated by the addition of 300 µl of ice-cold 2% zinc sulfate and cooling in an ice bath. All incubations were carried out in duplicate. Incubation conditions (microsomal protein content, incubation time) were within the linear range for the metabolic rate (preliminary studies; data not shown).

In Vivo Study.

Human subjects The subjects (seven males and five females; 23-49 years) were nonsmokers, and the females were not pregnant. The study was approved by the Human Subjects Review Committee at the University of Washington and subjects gave written informed consent.

Study design. A randomized, four-period crossover trial was performed in 12 healthy volunteers. This study was divided into a control phase (phase 1, no fluvoxamine, 2 days) and three experimental phases (phase 2, fluvoxamine 37.5 mg/day; phase 3, fluvoxamine 62.5 mg/day; and phase 4, fluvoxamine 87.5 mg/day; 12 days each). In phase 4, fluvoxamine was given 50 mg/day for 2 days and then escalated to 87.5 mg/day). Each subject was requested to abstain from taking any drug throughout the study. In the control phase, subjects were to arrive after an overnight fast and were given a single oral test dose of 100 mg of (S)-mephenytoin at 8:00 AM. Subjects remained fasting until 10:00 AM. Serial blood samples (10 ml) were collected from a catheter placed in the antecubital vein at 0, 15, 30, 45, 60, and 90 min and 2, 4, 6, 9, and 12 h after the test dose. Additional blood samples were collected 24 and 48 h after the test dose by venipuncture. Total urine output was collected during the 0- to 12-, 12- to 24-, and 24- to 48-h time periods. The 0- to 12-h urine sample was analyzed for 4-OH-meph to identify any poor metabolizers. Subjects who did not excrete at least 25% of the (S)-mephenytoin dose as 4-OH-meph within the first 12 h were not allowed to continue in the study. Phases 2, 3, and 4 were assigned to subjects in a randomized order. Fluvoxamine (phases 2, 3, and 4) was administered orally once daily at 4:00 PM for 11 of the 12 days. After eight doses of fluvoxamine, i.e., on the morning of the 9th day of treatment, subjects took the test dose of (S)-mephenytoin at 8:00 AM and fluvoxamine was given 8 h after the test dose. Serial blood samples and urine were collected as described for phase 1. When taking fluvoxamine, additional blood samples were collected at 8, 14, 32, and 72 h after the test dose of (S)-mephenytoin to check whether steady-state concentration of fluvoxamine was achieved and to allow calculation of the fluvoxamine area under the plasma concentration-time curve (AUC)_(0-). One additional urine sample was collected for the 48- to 72-h time period.

Sample preparation and high-performance liquid chromatography analysis. (S)-Mephenytoin, its metabolites, and p-HPPH (internal standard) in microsomal incubations were extracted by addition of 4 ml of methylene chloride. After shaking for 10 min, the organic phase was removed and evaporated under a stream of nitrogen (N₂), and the residues were reconstituted with 100 µl of methanol/water (25:75). Ten-microliter aliquots of the reconstituted residues were injected onto a Zorbax Rx-C8 (Narrow-Bore 2.1 × 150-mm, 5-µm) column connected to a Hewlett Packard 1100 LC-MSD system equipped with Hewlett Packard ChemStation software (version 1.0) for data acquisition and analysis. The flow rate was 0.20 ml/min, and the mobile phase consisted of methanol/0.01 M ammonium acetate buffer, pH 3.3 (50:50). The ions monitored were m/z 219 [(S)-mephenytoin], m/z 235 (4-OH-meph), m/z 205 (nirvanol), and m/z 269 (p-HPPH). (S)-Mephenytoin and its metabolites in human plasma and urine, together with p-HPPH (internal standard), were extracted and measured by a modification of the method of Kupfer et al. (1980). In brief, a 0.5-ml aliquot of each plasma sample was extracted with ethyl acetate/diethyl ether (67:33, v/v) and evaporated under a stream of N₂. For urine, a 0.5-ml aliquot of each sample was subjected to enzymatic hydrolysis with 1000 units of -glucuronidase (containing sulfatase activity) for 16 h at 37°C in the dark and then extracted. The residues were reconstituted in 100 µl of methanol/water (25:75). Ten-microliter aliquots were injected onto a Zorbax Rx-C8 (Narrow-Bore 2.1 × 150-mm, 5-µm) column connected to the same LC-MSD system as described above.

Data Analysis.

Determination of enzyme kinetic parameters The mechanism of inhibition was determined by graphical analysis. The formation of 4-OH-meph from (S)-mephenytoin in microsomal systems was analyzed by fitting the data to the single-enzyme competitive inhibition Michaelis-Menten model with a least-squares nonlinear regression program (BMDP; BMDP Statistical Software, Inc., Los Angeles, CA). This procedure provided estimates of V_max, K_M, and apparent K_i values.

Pharmacokinetic analysis. All pharmacokinetic parameters were calculated using the pharmacokinetic software package WinNonlin (version 1.5; Scientific Consulting, Inc., Mountain View, CA). The calculations were based on noncompartmental analysis. The terminal elimination rate constant (_Z) and apparent volume of distribution (V_d/F, where F is the bioavailability) of (S)-mephenytoin, were determined by linear regression of the terminal portion of the log concentration-time profile. The terminal elimination half-life (t_1/2) of (S)-mephenytoin was calculated as 0.693/_Z. The AUC was determined by the trapezoidal rule. (S)-Mephenytoin oral clearance was calculated as dose/AUC_inf (AUC_inf represents AUC extrapolated to infinity). The formation clearance of the (S)-mephenytoin metabolite 4-OH-meph was calculated as amount of 4-OH-meph in urine_{(0-72 h)}/AUC_{(0-72 h)}. The renal clearance of (S)-mephenytoin was calculated as amount of (S)-mephenytoin in urine_{(0-72 h)}/AUC_{(0-72 h)}. The average steady-state concentration of fluvoxamine was calculated as AUC_{(0-24 h)}/24 h.

Calculation of K_iiv values for CYP2C19. The approach used to calculate K_iiv is based on the competitive inhibition model used previously (Adedoyin et al., 1987; Shaw and Houston, 1987; Kunze and Trager, 1996; Tran et al., 1997; Yao et al., 2001). The model includes the following assumptions: 1) product formation is described by Michaelis-Menten kinetics with competitive inhibition; 2) concentrations of (S)-mephenytoin are much lower than its K_M; 3) only one enzyme mediates the formation of 4-OH-meph; 4) after oral administration, the operating clearance of (S)-mephenytoin is its intrinsic clearance because (S)-mephenytoin has a relatively large clearance (Ito et al., 1998); and 5) fluvoxamine does not affect the elimination of 4-OH-meph. The K_iiv was calculated from the following equation:

(1)

where CL_f and CL_fi represent formation clearances of 4-OH-meph associated with CYP2C19 in the absence and presence of the inhibitor fluvoxamine, respectively, and [I] represents the average steady-state plasma concentration of fluvoxamine. Plots of the ratio of uninhibited/inhibited formation clearances versus plasma concentration of inhibitor are expected to be linear with an intercept equal to 1 and a slope equal to 1/K_iiv. The tests of linearity and intercept of eq. 1 require measurements of multiple formation clearances and simultaneous measurements of inhibitor concentration. The overall K_iiv was determined based on linear regression with weight of 1/(formation clearance ratio)².

Calculation of RK_i. The ratio of the mean (n = 6) in vitro K_i to mean (n = 12) in vivo K_i based on unbound inhibitor concentration was calculated as K_{i,ub (in vitro)/} K_i,ub (in vivo). This ratio is defined as RK_i.

Statistical Analysis. The pharmacokinetic parameters of (S)-mephenytoin are reported as mean ± S.D. Comparisons between treatment and control periods were assessed by paired t test. The Wilcoxon rank sum test was performed for the comparison of RK_i values of fluvoxamine toward CYP1A2 and CYP2C19, assuming that the K_i values for CYP1A2 and CYP2C19 have the same distribution. Statistical significance was assumed if p < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

In Vitro Studies.

Nonspecific binding of fluvoxamine A previous study established that the free fraction of fluvoxamine in the HLM incubation system is reasonably constant up to 1 µM fluvoxamine (Yao et al., 2001). The free fractions of fluvoxamine in the microsomal preparations used in this study (protein concentration of 0.5 mg/ml) from six nonfatty human livers are shown in Table 1. The mean (± S.D.) free fraction was 0.33 ± 0.04. The free fraction in Supersomes was 0.81 ± 0.03 at 0.15 mg/ml protein.

Tests of time-dependent inhibition of fluvoxamine. In the preincubation study, CYP2C19 activity was not sensitive to the time of exposure to fluvoxamine, nor to the absence or presence of NADPH (Fig. 1A). In the coincubation study, 4-OH-meph was formed linearly with time in the incubations with or without fluvoxamine, suggesting that there are neither inhibitory metabolites formed nor time-dependent inhibition within the time frame used (Fig. 1B). These results validate the use of steady-state kinetics for evaluating the affinity of fluvoxamine for CYP2C19.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Tests of time-dependent inhibition of fluvoxamine toward CYP2C19. A, measurement of CYP2C19 activity after preexposure of HLMs to fluvoxamine up to 30 min in the presence or absence of NADPH. B, formation of 4-OH-meph from (S)-mephenytoin as a function of time in human liver microsomal incubation in the presence or absence of fluvoxamine.

Effect of nonspecific binding on the determination of in vitro K_i values. Lineweaver-Burk analysis indicated that fluvoxamine is a competitive inhibitor of CYP2C19 activity (Fig. 2). Apparent K_i values based on nominal (total added) fluvoxamine concentration (K_i,total) were determined based on the single-enzyme competitive inhibition model at protein concentrations of 1.0 and 0.5 mg/ml in pooled liver microsomes, giving apparent K_i,total values of 400 ± 37 and 226 ± 19 nM, respectively. K_i,total in the Supersomes system at 0.15 mg/ml protein was 96 ± 14 nM (Table 2). Because f_u,mic was independent of fluvoxamine concentration within the range used in the determination of in vitro K_i, K_i,ub was calculated by multiplying K_i,total by f_u,mic. After applying this correction, fairly uniform values of K_i,ub were obtained (around 70-80 nM; Table 2). K_i,total values were measured in microsomal preparations from six human livers at 0.5 mg/ml protein (Table 1) to determine interindividual variability. The mean K_i,total was 235 ± 51 nM and the mean K_i,ub was 76 ± 7.1 nM. The mean values of K_M and V_max for the formation of 4-OH-meph were 34 ± 7.9 µM and 0.19 ± 0.19 nmol/min/mg protein, respectively. These values are in agreement with those previously reported (Meier et al., 1985; Hickman et al., 1998). The values of K_M and V_max for the formation of 4-OH-meph were 15 ± 2.1 µM and 0.39 ± 0.012 nmol/min/nmol CYP2C19 for the Supersomes.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Inhibitory effect of fluvoxamine on the formation of 4-OH-meph from (S)-mephenytoin in Supersomes (0.15 mg/ml protein). Data points represent the average of duplicate incubations.

In Vivo Studies. Pharmacokinetic parameters of (S)-mephenytoin.

Pharmacokinetic parameters of fluvoxamine. Fluvoxamine average steady-state concentrations for each subject are shown in Table 4. These values are within the reported concentration ranges for these doses and consistent with the nonlinear disposition of this drug (Perucca et al., 1994b; Spigset et al., 1998). Fluvoxamine oral clearance decreased as the dose increased: oral clearances in the three phases were 94.7 ± 78.7, 60.9 ± 36.7, and 48.0 ± 28.9 l/h, respectively. The nonlinear disposition of fluvoxamine emphasizes the need to measure the plasma concentration of fluvoxamine in drug interaction studies.

Determination of K_iiv for fluvoxamine inhibition of CYP2C19. K_iiv for each subject was calculated from the plot of the ratio of 4-OH-meph formation clearances (control/inhibited) versus fluvoxamine average steady-state plasma concentration. Individual plots are shown in Fig. 3, A to D. The mean K_iiv values based on total and unbound plasma concentrations were 13.5 ± 5.6 and 1.9 ± 1.1 nM (f_p = 0.14 ± 0.02 from a previous study; Yao et al., 2001) (Table 4). A summary plot of ratios of 4-OH-meph formation clearances (control/inhibited) versus total plasma concentrations of fluvoxamine for 12 subjects is shown in Fig. 4. The overall K_iiv was 13.9 nM with an intercept of 0.85 (r² = 0.88, p < 0.001).

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Representative individual plots of ratios of 4-OH-meph formation clearances (control/inhibited) as a function of total plasma concentration of fluvoxamine at steady state used to determine in vivo inhibition constants.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Overall plot of ratios of 4-OH-meph formation clearances (control/inhibited) as a function of the total plasma concentration of fluvoxamine at steady state for 12 subjects. The line was fit by linear regression with weight of 1/(formation clearance ratio)2.

Comparison of the ratios of in vitro to in vivo inhibition constant of fluvoxamine toward CYP1A2 and CYP2C19. RK_i is used to allow a comparison between different enzymes affected by an inhibitor. The in vitro K_i,_ub values of fluvoxamine toward CYP1A2 (Yao et al., 2001) and CYP2C19 are 38 ± 6.7 nM (n = 8) and 76 ± 7.1 nM (n = 6), respectively, whereas the corresponding in vivo K_i values based on unbound plasma concentration are 3.6 ± 1.4 nM (n = 8) and 1.9 ± 1.1 nM (n = 12). Therefore, the mean RK_i values for CYP1A2 and CYP2C19 are 10 and 38, respectively. The Wilcoxon rank sum test showed that the difference in RK_i values was significant (p < 0.05).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

A primary goal of this study with CYP2C19, and the companion study with CYP1A2, was to unambiguously determine the absolute difference in potency of fluvoxamine toward these enzymes in vivo and in vitro, and to confirm that the kinetic models required to analyze the results were appropriately applied. To that end, studies were carried out in both systems with both enzymes to explore the relationship between fluvoxamine concentration and enzyme activity over a wide range of inhibitor concentrations. The results of in vitro studies indicate that the use of steady-state kinetics to determine inhibitor affinity is appropriate and that K_i values should be calculated based on the concentrations of unbound fluvoxamine rather than nominal concentrations. Similarly, in vivo, the relationship between fluvoxamine concentrations in plasma and their effect on enzyme selective metabolite formation clearances was found to be well behaved in a manner consistent with the standard kinetic model, allowing for the determination of an in vivo K_i based on steady-state fluvoxamine plasma levels.

The present study showed that the mean K_iiv of fluvoxamine toward CYP2C19 is 13.5 nM, a value which is well below the fluvoxamine therapeutic range (100-700 nM), suggesting that fluvoxamine is a potent inhibitor of CYP2C19 in vivo. The plots of ratios of 4-OH-meph formation clearance versus fluvoxamine plasma concentration (Figs. 3 and 4) showed good linearity with intercept values around 1 for individuals as well as the study population. The linearity and intercept values suggested that a good estimation of K_iiv values can be obtained using only a single dose of inhibitor. Based on this observation, the results of a published study of fluvoxamine inhibition of the CYP2C19 catalyzed bioactivation of chloroguanide (Jeppesen et al., 1997) were used to calculate a K_iiv of 25 nM (assuming that fluvoxamine plasma concentration at 100 mg/day is 199.3 nM; Spigset et al., 1998) and that only CYP2C19 catalyzes the formation of 4-chlorophenylbiguanide). This K_iiv estimate is similar to the values obtained in the present study, which used a different probe of CYP2C19 and three doses of fluvoxamine.

A second goal of these studies was to test the hypothesis that RK_i values for different enzymes subject to inhibition by fluvoxamine will be the same even though the RK_i values are both significantly higher than 1. The observed difference in RK_i values between CYP1A2 and CYP2C19 (10 versus 38) is intriguing and could be explained by the following possibilities: 1) the presence of one or more metabolites inhibiting only CYP2C19. So far, 11 metabolites of fluvoxamine have been found in urine. About 30 to 60% of the metabolites seem to be produced by oxidative demethylation of the methoxy group, whereas 20 to 40% seem to be formed by degradation at the amino group or by removal of the entire ethanolamino group (Spigset et al., 2001); and 2) the sensitivity of enzyme activity to environmental differences between in vitro and in vivo, such as pH and ionic strength, might be enzyme-dependent. Previous data from our laboratory (Yao and Levy, 2002) have shown a parallel decrease in K_i values for both enzymes (3-fold decrease in CYP1A2 versus 2.5-fold for CYP2C19) as pH increased from 7.2 to 7.6. Also, there was a 2-fold decrease in K_i for CYP1A2 and virtually no change for CYP2C19 as the concentration of potassium phosphate buffer (pH 7.4) increased from 0.01 to 0.1 M. Thus, these preliminary studies suggest a differential effect on enzyme affinity by ionic strength but not pH.

It is not possible at this time to evaluate the relative importance of environmental differences and partitioning effects. However, further work with other enzymes also inhibited by fluvoxamine may help to further discriminate between these and other factors contributing to the differences between the in vitro and in vivo systems. In view of the findings of RK_i values larger than 1 for CYP1A2 and CYP2C19, we searched the literature for similar data on other enzymes inhibited by fluvoxamine, namely, CYP2C9, CYP2D6, and CYP3A4. Reported in vitro inhibition potencies of fluvoxamine toward CYP2C9, CYP2D6, and CYP3A4 also seem to be inconsistent with corresponding in vivo inhibition effects. In vitro K_i values for CYP2C9, CYP2D6, and CYP3A4 are 2.2 to 13.3 µM (Hemeryck et al., 1999; Olesen and Linnet, 2000), 3.9 to 8.2 µM (Crewe et al., 1992; Skjelbo and Brosen, 1992), and 10 to 40 µM (von Moltke et al., 1994; Rasmussen et al., 1995), respectively, and these values are much higher than the fluvoxamine therapeutic range of 0.1 to 0.7 µM (0.014-0.1 µM based on unbound plasma concentration). Yet, recent studies (Kashuba et al., 1998; Madsen et al., 2001) suggest that fluvoxamine significantly inhibits the metabolism of tolbutamide (CYP2C9), dextromethorphan (CYP2D6), and midazolam (CYP3A4) in vivo. These comparisons suggest that, for all five P450s inhibited by fluvoxamine, the inhibition potency is greater in vivo than predicted from in vitro data. It remains to be seen whether RK_i values will also be in the range seen with CYP1A2 and CYP2C19 in this study. In particular, large changes in RK_i values for other enzymes would support a primary role for enzyme environment, whereas no differences in RK_i values would support partitioning effects (elevated fluvoxamine concentrations available to the enzyme relative to free concentrations in plasma).