Differentiation of intestinal and hepatic
`cytochrome P450 3A activity with use of
`midazolam as an in vivo probe: Effect of
`ketoconazole
`
`Background: The cytochrome P450 3A (CYP3A) isoforms are responsible for the metabolism of a major-
`ity of therapeutic compounds, and they are abundant in the intestine and liver. CYP3A activity is highly
`variable, causing difficulty in the therapeutic use of CYP3A substrates. A practical in vivo probe method
`that characterizes both intestinal and hepatic CYP3A activity would be useful.
`Objectives: To determine the intestinal and hepatic contribution to the bioavailability of midazolam with
`use of the CYP3A inhibitor ketoconazole.
`Methods: The pharmacokinetics of midazolam was assessed in nine (six men and three women) healthy indi-
`viduals after single doses of 2 mg intravenous and 6 mg oral midazolam (phase I). These pharmacokinetic
`values were compared with those obtained after single doses of 2 mg intravenous and 6 mg oral midazo-
`lam and three doses of 200 mg oral ketoconazole (phase II).
`Results: After ketoconazole therapy, area under the concentration versus time curve of midazolam increased
`5-fold after intravenous midazolam administration (P ≤ .001) and 16-fold after oral midazolam adminis-
`tration (P ≤ .001). Intrinsic clearance decreased by 84% (P = .003). Total bioavailability increased from
`25% to 80% (P < .001). The intestinal component of midazolam bioavailability increased to a greater extent
`than the hepatic component (2.3-fold [P = .003] and 1.5-fold [P ≤ .001], respectively). In the control
`phase, female subjects had greater midazolam clearance values than the male subjects.
`Conclusions: Ketoconazole caused marked inhibition of CYP3A activity that was greater in the intestine
`than the liver. Administration of single doses of oral and intravenous midazolam with and without oral
`ketoconazole exemplifies a practical method for differentiating intestinal and hepatic CYP3A activity. (Clin
`Pharmacol Ther 1999;66:461-71.)
`
`Shirley M. Tsunoda, PharmD, Rebecca L. Velez, PharmD, Lisa L. von Moltke, MD,
`and David J. Greenblatt, MD Boston, Mass
`
`The cytochrome P450 3A (CYP3A) subfamily is the
`most important of the cytochrome P450 superfamily
`because of its abundance in the liver and intestine and
`
`From the School of Pharmacy, Bouvé College of Health Sciences,
`Northeastern University, and the Department of Pharmacology and
`Experimental Therapeutics, Tufts University School of Medicine.
`Supported by the Burroughs Wellcome Fund and the American Foun-
`dation for Pharmaceutical Education and by grants M01 RR00054,
`MH-01237, and MH-34223 from the National Institutes of Health
`(Bethesda, Md).
`Received for publication April 23, 1999; accepted Aug 17, 1999.
`Reprint requests: Shirley M. Tsunoda, PharmD, School of
`Pharmacy, Bouvé College of Health Sciences, 206 Mugar Life
`Sciences Building, Northeastern University, Boston, MA 02115.
`E-mail: s.tsunoda@nunet.neu.edu
`Copyright © 1999 by Mosby, Inc.
`0009-9236/99/$8.00 + 0 13/1/102335
`
`its ability to metabolize a wide variety of therapeutic
`compounds. The activity of CYP3A is highly variable
`both between and within individuals, which makes the
`dosing and therapeutic use of many CYP3A substrates
`difficult, especially those with narrow therapeutic
`ranges. The CYP3A subfamily consists of three known
`isoforms
`in humans—CYP3A4, CYP3A5,
`and
`CYP3A7. The CYP3A4 isoform is thought to be the
`dominant form in humans1; however, recent evidence
`reports the presence of CYP3A5 in 10% to 25% of adult
`human livers2 and CYP3A7 in 50% of adult liver sam-
`ples.3 It has not been established whether the presence
`or absence of CYP3A5 and/or CYP3A7 in the intestine
`or liver contributes to a significant difference in meta-
`bolic capacity.4-8 Therefore, to maintain consistency,
`we will use the general term “CYP3A” to refer to the
`
`461
`
`1
`
`TEVA1007
`
`

`

`462 Tsunoda et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`NOVEMBER 1999
`
`subfamily of enzymes when appropriate. The CYP3A
`activity of an individual is probably determined both
`by genetics and by environmental factors that can mod-
`ulate the activity. Many factors may potentially influ-
`ence the activity of CYP3A, including age,9 gender,10
`menopausal status,11 hormones,12 and numerous xeno-
`biotics.13 However, in general these factors do not have
`a predictable effect on CYP3A activity. In addition, the
`CYP3A activities of an individual in the intestine and
`the liver do not appear to be coordinately regulated14,15;
`therefore it is likely that these many factors may dif-
`ferentially modulate intestinal and hepatic activity.
`Because of the importance of the CYP3A subfamily
`and because of the many factors that influence CYP3A
`activity, a practical method to determine the CYP3A
`activity of an individual is desirable. There have been
`many attempts to use various CYP3A substrates as
`probe compounds to predict CYP3A activity, including
`dapsone, 6-b -hydroxylation of testosterone, nifedipine,
`6-b -hydroxycortisol, and the erythromycin breath test.
`Although the erythromycin breath test has shown cor-
`relations with cyclosporine (INN, ciclosporin) clear-
`ance,16 OG 37-325,17 and midazolam clearance,18 there
`are limitations to this test, including the administration
`of radioactivity, the variability in correlating radioac-
`tive exhaled carbon dioxide and N-demethylation of
`erythromycin, the role of CYP3A in erythromycin
`metabolism,1 and the lack of prediction of intestinal
`CYP3A activity. Therefore an optimal CYP3A probe
`has yet to be identified.
`Midazolam is a benzodiazepine that is used clini-
`cally for conscious sedation. It has pharmacokinetic
`properties that make it an attractive in vivo probe: it is
`specifically metabolized by CYP3A19 to one predomi-
`nant metabolite (1¢ -hydroxymidazolam); it has a short
`half-life (t1⁄2), so estimation of total area under the con-
`centration versus time curve (AUC) is easily measur-
`able and it can be given intravenously and orally (with
`the injection preparation). In addition, hepatic CYP3A
`content measured in vitro has been shown to be highly
`correlated with midazolam intravenous clearance (r =
`0.93; P < .001),20 and plasma 1¢ -hydroxymidazo-
`lam/midazolam ratio.21
`Ketoconazole is one of the most potent inhibitors of
`CYP3A in clinical use. Its estimated in vitro inhibition
`constant (Ki) with use of human liver microsomes for
`1-hydroxymidazolam formation is less than 0.01
`m mol/L (1 m mol/L = 0.53 m g/mL).22 Administration of
`the CYP3A substrate midazolam intravenously (when
`the contribution of intestinal CYP3A activity is
`assumed to be negligible) and orally both in the absence
`and presence of the CYP3A inhibitor ketoconazole
`
`allows the contributions of intestinal and hepatic
`CYP3A to the bioavailability of midazolam to be esti-
`mated. Our study population consisted of healthy indi-
`viduals; therefore potential confounding variables such
`as abnormal liver or intestinal function and concomi-
`tant medications were minimized.
`
`METHODS
`Nine healthy nonsmoking human subjects (six men
`and three women) were enrolled in the study after each
`gave written informed consent. Their mean age was 26
`years (age range, 19 to 41 years). The mean body
`weight of male subjects was 77.5 kg, and the mean
`body weight of female subjects was 59.7 kg. All sub-
`jects were white. The study was approved by the
`Human Investigation Review Committee at New Eng-
`land Medical Center and Tufts University School of
`Medicine and by the Institutional Review Board at
`Northeastern University. The study was conducted in
`the General Clinical Research Center at New England
`Medical Center. All subjects were healthy adults with
`no evidence of medical disease. No subjects took any
`medications. Female subjects were not taking oral con-
`traceptives.
`The study was conducted in two phases. In phase I,
`baseline oral and intravenous midazolam pharmacoki-
`netic parameters were established. Subjects received a
`single dose of 2 mg intravenous midazolam (Versed, 2
`mg/mL for intravenous injection, Hoffmann-LaRoche
`Inc, Nutley, NJ) or 6 mg oral midazolam as 3 mL of the
`parenteral preparation diluted with 30 mL water.
`Although midazolam bioavailability has been estimated
`to average 41%,9 we chose the conservative estimate of
`33% to ensure adequate measurement of plasma con-
`centrations. In phase II, subjects received single doses
`of 2 mg intravenous or 6 mg oral midazolam with 200
`mg ketoconazole (Nizoral, Janssen Pharmaceutica,
`Titusville, NJ). The first dose was administered 12
`hours before the midazolam dose and subsequent doses
`were administered every 12 hours for a total of three
`doses to maintain plasma ketoconazole levels in excess
`of the Ki. The order of treatments was randomized; how-
`ever, treatments remained constant between phase I and
`II for each subject. Before each study day, subjects were
`instructed to fast from midnight until they were given
`a standardized light breakfast approximately 90 min-
`utes before midazolam administration. All subsequent
`meals during each study day were standardized. Alco-
`hol, caffeine, and grapefruit juice were prohibited dur-
`ing each study day. Blood samples were collected
`before and at 15, 30, and 45 minutes and 1, 11⁄2, 2, 21⁄2,
`3, 4, 5, 6, and 8 hours after midazolam dosing during
`
`2
`
`

`

`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 66, NUMBER 5
`
`Tsunoda et al 463
`
`Table I. Intravenous and oral midazolam pharmacokinetic parameters: Noncompartmental analysis
`
`Control
`
`With ketoconazole
`
`Intravenous
`
`Oral
`
`Intravenous
`
`Oral
`
`Pharmacokinetic parameters
`AUC (ng · h/mL)
`CL (mL/min/kg)
`CLint (mL/min/kg)
`Varea (L/kg)
`Elimination t1⁄2 (h)
`Cmax (ng/mL)
`tmax (h)
`F (%)
`FH (%)
`FABS · FG (%)
`Ratios (ketoconazole/control)
`AUC
`CLint
`F
`FH
`FABS · FG
`
`70.2 ± 25.8
`7.6 ± 2.9
`13.3 ± 8.9
`1.9 ± 0.8
`3.4 ± 2.0
`—
`—
`—
`—
`—
`
`54.2 ± 28.0
`34.2 ± 18.7
`
`—
`19.1 ± 8.1
`0.8 ± 0.2
`25.3 ± 9.6
`64.6 ± 13.4
`40.1 ± 14.7
`
`354 ± 185*
`1.6 ± 0.8*
`1.8 ± 1.0†
`1.6 ± 0.5
`14.0 ± 8.2†
`—
`—
`—
`—
`—
`
`5.1 ± 1.9
`0.16 ± 0.06
`
`738 ± 191*
`2.1 ± 0.6*
`
`—
`81.0 ± 28.9*
`0.8 ± 0.6
`80.0 ± 31.6*
`92.4 ± 3.7*
`87.8 ± 38.5†
`
`16.0 ± 6.1
`—
`3.4 ± 1.7
`1.5 ± 0.3
`2.3 ± 0.8
`
`Data are mean values ± SD.
`AUC, Area under the concentration versus time curve; CL, clearance; CLint, intrinsic clearance; Varea, volume of distribution; t1⁄2
`tration; tmax, time to reach Cmax; F, total bioavailability; FH, hepatic bioavailability; FABS · FG, intestinal bioavailability.
`*P ≤ .001.
`†P = .003.
`
`, half-life; Cmax, maximum concen-
`
`phase I, with additional samples taken at 12 and 24
`hours after midazolam dosing for phase II.
`Plasma was analyzed for midazolam concentrations
`by electron capture gas chromatography according to a
`previously published method.23 In addition, plasma keto-
`conazole concentrations were determined by HPLC.24
`Pharmacokinetic analysis was conducted by use of
`two methods. The first method used noncompartmen-
`tal methods in which midazolam plasma AUC values
`were calculated by use of the log-linear and linear
`trapezoidal methods for decreasing and increasing
`concentrations, respectively, with extrapolation from
`the last measured concentration to infinity. Estimation
`of the terminal elimination rate constant was per-
`formed by linear regression of at least the last three
`concentration–time points. Oral and intravenous clear-
`ance rates (CLoral and CLIV, respectively) were calcu-
`lated by dividing the respective dose by the AUC after
`oral and intravenous administration, respectively.
`Intrinsic clearance (CLint), defined as the clearance
`attributed solely to hepatic metabolism, was calculated
`with the equation:
`
`(CLIV · QH)/(QH – CLIV)
`in which QH is the liver blood flow. Liver blood flow
`was assumed to be 1500 mL/min/70 kg, normalized for
`body weight of each individual subject. Volume of dis-
`
`tribution (Varea) was calculated with use of the area
`method. The second method used nonlinear regression
`with a linear sum of two exponential terms to analyze
`plasma midazolam concentrations after intravenous
`administration. Coefficients and exponents from the fit-
`ted function were used to estimate total AUC, Varea,
`elimination t1⁄2, and total clearance. After oral adminis-
`tration, AUC until the last detectable concentration was
`calculated by use of the linear trapezoidal method and
`extrapolated to infinity by addition of the final concen-
`tration divided by b
`. The maximum concentration
`(Cmax) and the time to reach Cmax (tmax) were deter-
`mined by the highest measured concentration and
`respective time.
`Bioavailability (F) was determined from dose-
`corrected AUC values, assuming dose-independent
`pharmacokinetics. The components of bioavailability
`were further predicted on the basis of the equation:
`
`F = FABS · FG · FH
`in which FABS is the fraction of the dose absorbed from
`the gut lumen, FG is the fraction of the dose not metab-
`olized by intestinal metabolic enzymes, and FH is the
`fraction of the dose absorbed into the hepatic portal
`vein that escapes first-pass liver metabolism. FH was
`defined as follows:
`
`1 – ERH
`
`3
`
`

`

`464 Tsunoda et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`NOVEMBER 1999
`
`Fig 1. Mean ± SEM plasma concentration versus time curves after intravenous and oral midazo-
`lam in the control phase versus with ketoconazole.
`
`Table II. Intravenous and oral midazolam pharmacokinetic parameters: Nonlinear regression with a linear sum of
`two exponential terms
`
`Control
`
`With ketoconazole
`
`Intravenous
`
`Oral
`
`Intravenous
`
`Oral
`
`Pharmacokinetic parameters
`AUC (ng · h/mL)
`CL (mL/min/kg)
`CLint (mL/min/kg)
`Varea (L/kg)
`Elimination t1⁄2 (h)
`Cmax (ng/mL)
`tmax (h)
`F (%)
`FH (%)
`FABS · FG (%)
`Ratios (ketoconazole/control)
`AUC
`CLint
`F
`FH
`FABS · FG
`
`Data are mean values ± SD.
`*P ≤ .001.
`
`63.4 ± 18.7
`8.1 ± 2.8
`14.6 ± 9.4
`1.3 ± 0.4
`1.9 ± 0.5
`—
`—
`—
`—
`—
`
`58.2 ± 31.1
`32.7 ± 19
`
`1.7 ± 0.4
`19.1 ± 8.1
`0.8 ± 0.2
`29.3 ± 11.3
`62.4 ± 13.0
`47.5 ± 16.6
`
`300 ± 104*
`1.7 ± 0.5*
`1.9 ± 0.6*
`1.3 ± 0.4
`9.7 ± 4.1*
`—
`—
`—
`—
`—
`
`4.8 ± 1.4
`0.15 ± 0.06
`
`719 ± 181*
`2.1 ± 0.6*
`
`9.9 ± 3.3*
`81.0 ± 28.9*
`0.8 ± 0.5
`83.8 ± 21.0*
`92.0 ± 2.4*
`91.4 ± 24.4*
`
`15.1 ± 6.7
`—
`3.2 ± 1.3
`1.5 ± 0.4
`2.1 ± 0.7
`
`in which ERH is the hepatic extraction ratio defined
`as clearance of intravenously administered midazo-
`lam divided by liver blood flow: CLIV/QH. Non-
`hepatic contributions to clearance were assumed to
`
`be negligible. Therefore, assuming linear pharmaco-
`kinetics, the product FABS · FG was determined and
`compared with and without the presence of the
`enzyme inhibitor ketoconazole. It was assumed that
`
`4
`
`

`

`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 66, NUMBER 5
`
`Tsunoda et al 465
`
`ketoconazole does not significantly affect liver blood
`flow or FABS.
`Statistical analysis was conducted by use of the
`paired Student t test, with a = .05.
`
`RESULTS
`Midazolam pharmacokinetic parameters before (con-
`trol) and with ketoconazole calculated by use of
`noncompartmental analysis are shown in Table I; param-
`eters calculated by use of nonlinear regression with a
`linear sum of two exponential terms are shown in Table
`II. In the presence of ketoconazole, intravenous mid-
`azolam AUC increased 5-fold and oral midazolam AUC
`increased 16-fold (P < .001). The comparatively larger
`increase in oral midazolam AUC versus intravenous
`midazolam AUC is shown graphically in Fig 1. Accord-
`ingly, midazolam clearance after intravenous and oral
`administration decreased from 7.6 ± 2.9 to 1.6 ± 0.8
`(P < .001) and 34.2 ± 18.7 to 2.1 ± 0.6 mL/min/kg
`(P < .001) for control versus administration with keto-
`conazole, respectively. Consistent with the presumed
`effect of ketoconazole on CYP3A metabolism, the
`CLint decreased dramatically after ketoconazole admin-
`istration: 13.3 ± 8.9 versus 1.8 ± 1.0 mL/min/kg (P =
`.003; Tables I and II). The comparative declines in all
`three clearance values are shown in Fig 2.
`Ketoconazole caused total measured bioavailability
`to increase from 25% to 80% (P < .001). Ketoconazole
`also caused both the hepatic and intestinal components
`of total bioavailability to be increased, although the
`intestinal component was increased to a greater extent
`(Tables I and II). There was a lack of correlation
`between FH and FABS · FG (r2 = 0.13; Fig 3, A).
`The Cmax of oral midazolam increased 5-fold with
`ketoconazole, consistent with the large gut effect. The
`tmax was not changed with ketoconazole (0.78 ± 0.21
`versus 0.82 ± 0.57 hour, control versus with ketocona-
`zole, respectively; P = NS). The t1⁄2 increased approxi-
`mately 5-fold in the presence of ketoconazole (P =
`.003; Fig 1).
`Interestingly, in the control phase, the three female
`subjects had a mean midazolam clearance that was
`greater than the six male subjects when corrected for
`weight (10.2 ± 3.2 versus 6.3 ± 1.8 mL/min/kg; P =
`.047). However, this difference disappeared during the
`ketoconazole phase (1.8 ± 1.3 versus 1.5 ± 0.5
`mL/min/kg; P = .64). The FH was significantly lower
`in the women compared with the men in the control
`phase only (52.6% ± 14.8% versus 70.7% ± 8.4%; P =
`.047), which was consistent with the change in clear-
`ance. Women also had a larger weight-adjusted Varea
`than men that reached statistical significance during the
`
`Fig 2. Mean ± SEM plasma clearance rates in the control
`phase versus with ketoconazole. Intravenous and oral clear-
`ance values were measured. Intrinsic clearance was calcu-
`lated as described in the text.
`
`ketoconazole phase (2.17 ± 0.25 versus 1.32 ± 0.34 L/kg;
`P = .007), when calculated with noncompartmental
`methods. It should be noted that there were no changes
`in weight between the control and ketoconazole phases.
`However, using the two-compartment method, females
`had larger weight-adjusted Varea compared to males in
`both the control (1.65 ± 0.49 versus 1.08 ± 0.20 L/kg,
`P = .03) and ketoconazole (1.70 ± 0.68 versus 1.10 ±
`0.23 L/kg, P = .025) phases. In addition, there were no
`gender-related differences in clearance or the elimina-
`tion rate constant during the ketoconazole phase. There
`were no other gender-related differences in any other
`pharmacokinetic parameter.
`Ketoconazole plasma concentrations ranged from
`0.077 to 4.852 m g/mL, thereby exceeding the in vitro
`Ki at most time points (Fig 4).
`
`DISCUSSION
`The CYP3A subfamily is clearly important in deter-
`mining the pharmacokinetics and subsequent pharmaco-
`
`5
`
`

`

`466 Tsunoda et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`NOVEMBER 1999
`
`A
`
`C
`
`B
`
`D
`
`Fig 3. A, Lack of correlation between midazolam hepatic bioavailability (FH) and intestinal bioavail-
`ability (FABS · FG ). B, Lack of correlation between total bioavailability (F) in the control phase
`(without ketoconazole) and the relative change with ketoconazole. C, Correlation between FH in
`the control phase and the relative change in FH with ketoconazole. D, Lack of correlation between
`FABS · FG in the control phase and the relative change with ketoconazole.
`
`dynamics of many compounds that encompass a variety
`of therapeutic areas. In addition, the intestinal contri-
`bution of CYP3A is also important in mechanisms
`of drug-drug interactions and the determination of
`the bioavailability of a number of CYP3A sub-
`strates.20,21,25-30 It appears that the intestinal and
`hepatic CYP3A activities of an individual are regulated
`
`independently14,15; therefore an ideal in vivo probe
`must quantitate both. Our study used a CYP3A com-
`pound that can be administered intravenously and orally
`in the absence and presence of a potent CYP3A
`inhibitor to partition intestinal and hepatic activity.
`In the control phase, the midazolam pharmacokinetic
`parameters clearance and total bioavailability were
`
`6
`
`

`

`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 66, NUMBER 5
`
`Tsunoda et al 467
`
`Fig 4. Mean ± SEM plasma concentration versus time curve after intravenous and oral midazolam.
`
`comparable to those previously reported. Clearance val-
`ues ranged from 4.7 to 8.1 mL/min/kg and total
`bioavailability values ranged from 27 to 40%9,28,30-32
`in healthy volunteers.
`Midazolam pharmacokinetic parameters changed
`dramatically during the administration of ketoconazole,
`which was expected. The magnitude of change was
`consistent with the results of Olkkola et al33 in which
`7.5 mg oral midazolam was administered with 400 mg
`ketoconazole for four doses in nine healthy volunteers.
`Administering a 20% higher dose of midazolam and a
`2-fold higher dose of ketoconazole gave a mean mid-
`azolam AUC that increased 16-fold compared with our
`mean oral AUC, which also increased 16-fold. The inhi-
`bition of CYP3A by ketoconazole does not appear to
`be dose- or concentration-dependent at this dosing
`range; and it is likely that a dose of 200 mg causes
`nearly complete CYP3A inhibition. Total systemic
`clearance decreased by 78% in the presence of keto-
`conazole. This magnitude is slightly greater than that
`in the study by Gorski et al,30 who found a 64%
`decrease in midazolam clearance in the presence of
`clarithromycin. The inhibitory effect of ketoconazole
`on CYP3A metabolism is more accurately illustrated
`by the 84% decrease in CLint. Because midazolam is
`an intermediate extraction ratio compound (mean
`extraction ratio, 35%), CLint more accurately reflects
`the contribution of clearance attributable to drug metab-
`olism, without the influence of hepatic blood flow. In
`
`the presence of ketoconazole, CLoral, CLIV, and CLint
`fall to a similar level (approximately 2 mL/min/kg),
`again illustrating the potent inhibitory effect of keto-
`conazole on CYP3A-mediated intestinal and hepatic
`metabolism (Fig 2). Consistent with the intestinal
`effect, ketoconazole caused a 5-fold increase in Cmax
`without any change in the tmax. We also found an almost
`5-fold lengthening of the elimination t1⁄2 with ketocona-
`zole, similar to the 310% increase in elimination t1⁄2
`found by Olkkola et al,33 and the almost 3-fold increase
`found by Gorski et al.30
`Although several midazolam studies have failed to
`show a statistically significant gender-related effect in
`young subjects,9,28,34,35 we found a gender-related
`effect in the control midazolam phase of 60% greater
`weight-corrected clearance in the young women versus
`the young men (P = .05). This is qualitatively consis-
`tent with that shown with other CYP3A substrates such
`as erythromycin, methylprednisolone, tirilazad, and
`verapamil, as recently reviewed.10 In addition, on the
`basis of blood concentrations, Gorski et al30 reported a
`statistically significant 27% higher midazolam clear-
`ance in young healthy women versus men. They also
`found weight-corrected CLoral to be 50% higher in
`young women versus men, with a corresponding
`decrease in the AUC.30 Our data are also consistent
`with that of Gorski et al30 with the CYP3A inhibitor
`clarithromycin in that the gender-related effect disap-
`peared during the inhibition phase. Some investigations
`
`7
`
`

`

`468 Tsunoda et al
`
`CLINICAL PHARMACOLOGY & THERAPEUTICS
`NOVEMBER 1999
`
`imply that intestinal metabolism may be different in
`men versus women because gender-related differences
`have been found in either CLoral or bioavailability with
`oral triazolam,36 oral cyclosporine,14 and oral verap-
`amil.37 On the other hand, other studies with oral
`triazolam38 and other oral CYP3A substrates, such as bu-
`spirone,39 alprazolam,40 and trazodone,41 have not found
`a gender-related difference. The studies that have found
`a gender-related difference have shown that young
`females have greater CYP3A activity compared with
`young males. Our data with midazolam are consistent
`with this; however, confirmation in a larger population
`is required because we enrolled a small number of
`female subjects. The gender issue is complex, may
`involve other intestinal proteins, and may not prove to
`be consistent with all CYP3A substrates. Our data illus-
`trated a larger weight-adjusted Varea in women com-
`pared with men, which is consistent with other reports
`in the literature.9,28,34
`Ketoconazole decreased the variability in hepatic
`bioavailability. It has been suggested that inhibition of
`CYP3A causes a “normalization” of activity across the
`study population; that is, subjects with the greatest
`CYP3A activity exemplified by high clearance rates or
`low bioavailability show the greatest change. This phe-
`nomenon has been shown with grapefruit juice and
`felodipine42 and with clarithromycin and midazolam.30
`We found a correlation between FH in the control phase
`and the relative increase in FH during ketoconazole
`inhibition (r2 = 0.96; Fig 3, C). A lack of correlation
`was found in these respective relationships with total
`bioavailability (r2 = 0.22) and FABS · FG (r2 = 0.14; Figs
`3, B and D).
`Total measured bioavailability increased by a factor
`of 3.4 in the presence of ketoconazole. The intestinal
`availability increased 2.3-fold, whereas the hepatic
`availability increased 1.5-fold in the presence of keto-
`conazole. These results are qualitatively consistent with
`other CYP3A inhibition studies reported in the litera-
`ture in that the intestinal component is affected to a
`greater extent than the hepatic component. Gorski et
`al30 found a 197% increase in midazolam intestinal
`availability and a 120% increase in hepatic availability
`after 7 days of 500 mg clarithromycin twice daily. Sim-
`ilarly, Gomez et al26 demonstrated a 2.3-fold increase
`in intestinal availability and a 1.15-fold increase in
`hepatic availability of cyclosporine after 6 to 10 days
`of 200 mg ketoconazole therapy. Midazolam is readily
`absorbed,30 so these effects are likely the result of an
`intestinal metabolism effect. These effects are proba-
`bly not modulated by P-glycoprotein because midazo-
`lam is not a P-glycoprotein substrate.43 It is likely that
`
`higher local concentrations of the inhibitor, in this case
`ketoconazole, are achieved in the intestine, leading to
`greater CYP3A inhibition compared with that in the
`liver. In addition, the dynamics and variability of gas-
`tric motility probably contribute to the complexity of
`the interaction. However, we did not test this directly.
`Nonetheless, the potentially greater effect on intestinal
`CYP3A inhibition has important clinical implications
`for drug-drug and drug-food interactions with oral com-
`pounds in that separating the timing of administration
`of potentially interacting oral CYP3A substrates may
`lessen or even ameliorate an interaction.44
`Consistent with other reports that have suggested that
`intestinal and hepatic CYP3A activity are regulated
`independently, we found that there was no correlation
`between FH and FABS · FG (r2 = 0.13; Fig 3, A).
`Our conclusions are based on the assumption that
`midazolam exhibits dose-independent pharmacokinet-
`ics, which is a reasonable assumption because it has
`been shown by several studies in young healthy indi-
`viduals over a varied dosing range.9,34,45 We did not
`measure the protein binding of midazolam; however,
`because the extraction ratio of midazolam (estimated
`from our study to be 35% ± 13% [mean ± SD]) is
`much greater than the estimated percent fraction
`unbound, which has been reported to be 2.4% to
`3.7%,9,28 protein binding cannot completely restrict
`hepatic uptake. We also assumed that enzyme activity
`and blood flow remained constant between oral and
`intravenous midazolam administration. Although
`simultaneous administration of oral and intravenous
`substrate may be theoretically preferable, the practi-
`cal issues, such as the need for isotope-labeled drug
`and the high midazolam dose required, precluded this
`method. In addition, because all subjects were healthy
`and had no longer than a 7-day washout period
`between study days, variability in these processes is
`not likely to have influenced the results. A recent
`investigation showed insignificant interday variability
`in hepatic CYP3A4 metabolism in young men by use
`of midazolam intravenous pharmacokinetics.46 We
`used plasma rather than blood concentrations for cal-
`culation of pharmacokinetic parameters, which may
`lead to overestimation of AUC values. In addition,
`rather than measuring hepatic blood flow directly, we
`assumed it to be 1500 mL/min/70 kg. Based on lido-
`caine clearance,47 this is a reasonable assumption;
`however, measured hepatic blood flow would have
`been more accurate. Finally, fluctuations in ketocona-
`zole concentrations may have complicated the inter-
`action with midazolam (Fig 4); however, in general,
`plasma concentrations exceeded the in vitro Ki.
`
`8
`
`

`

`CLINICAL PHARMACOLOGY & THERAPEUTICS
`VOLUME 66, NUMBER 5
`
`Tsunoda et al 469
`
`The results of several studies6,48,49 show that mid-
`azolam is predominantly and virtually exclusively
`metabolized by enzymes of the CYP3A subfamily and
`that measuring midazolam clearance in vivo correlates
`well with content and activity in vitro. In vitro studies
`with hepatic microsomes that contain CYP3A4 were
`found to hydroxylate midazolam to its two primary
`metabolites, 1¢ -hydroxymidazolam and 4¢ -hydroxy-
`midazolam.6,22 Total midazolam clearance has been
`shown to be highly correlated with hepatic CYP3A con-
`tent(r = 0.93; P < .001) and cyclosporine clearance
`(r = 0.81; P < .001)20 but less strongly correlated with
`the erythromycin breath test result (r = 0.52; P = .03).18
`In addition, the apparent maximum rate (Vmax) for 1¢ -
`hydroxymidazolam formation was weakly correlated
`with total hepatic CYP3A4 content (r2 = 0.31; P < .05),48
`suggesting the importance of the intestinal contribution.
`Finally, 43% of an intraduodenally administered dose of
`midazolam was shown to be metabolized through the
`intestinal mucosa during the anhepatic phase of 10 liver
`transplant patients.49 These data6,18,20,48,49 support the
`use of midazolam as an in vivo probe, although a num-
`ber of the studies20,49 used invasive methods. Therefore
`the method used in our study with use of relatively small
`doses of midazolam intravenously and orally in the pres-
`ence and absence of the inhibitor ketoconazole exempli-
`fies a less invasive model for partitioning intestinal and
`hepatic CYP3A activity.
`
`We gratefully acknowledge the assistance of the following indi-
`viduals: Anna-Liza Durol for plasma analysis of midazolam and keto-
`conazole, Jerold S. Harmatz for pharmacokinetic and statistical analy-
`ses, Richard I. Shader, MD, for advice and collaboration, and the
`staff of the General Clinical Research Center, New England Medical
`Center.
`
`References
`1. Shimada T, Yamazaki H, Mimura M, Inui Y, Guengerich
`FP. Interindividual variations in human liver cytochrome
`P-450 enzymes involved in the oxidation of drugs, car-
`cinogens and toxic chemicals: studies with liver micro-
`somes of 30 Japanese and 30 Caucasians. J Pharmacol
`Exp Ther 1994;270:414-23.
`2. Wrighton SA, Ring BJ, Watkins PB, Vandenbranden M.
`Identification of a polymorphically expressed member of
`the human cytochrome P-450III family. Mol Pharmacol
`1989;36:97-105.
`3. Schuetz JD, Beach DL, Guzelian PS. Selective expres-
`sion of cytochrome P450 CYP3A mRNAs in embryonic
`and adult human liver. Pharmacogenetics 1994;4:11-20.
`4. Gibbs MA, Thummel KE, Shen DD, Kunze KL. Inhibi-
`tion of cytochrome P-450 3A (CYP3A) in human intesti-
`nal and liver microsomes: comparison of Ki values and
`impact of CYP3A5 expression. Drug Metab Dispos 1999;
`27:180-7.
`
`5. Gillam EM, Wunsch RM, Ueng YF, Shimada T, Reilly
`PE, Kamataki T, et al. Expression of cytochrome P450
`3A7 in Escherichia coli: effects of 5¢ modification
`and catalytic characterization of recombinant enzyme
`expressed
`in bicistronic
`format with NADPH–
`cytochrome P450 reductase. Arch Biochem Biophys
`1997;346:81-90.
`6. Gorski JC, Hall SD, Jones DR, Vandenbranden M,
`Wrighton SA. Regioselective biotransformation of mid-
`azolam by members of the human cytochrome P450
`3A (CYP3A) subfamily. Biochem Pharmacol 1994;47:
`1643-53.
`7. Paine MF, Khalighi M, Fisher JM, Shen DD, Kunze KL,
`Marsh CL, et al. Characterization of interintestinal and
`intraintestinal variations in human CYP3A-dependent
`metabolism. J Pharmacol Exp Ther 1997;283:1552-62.
`8. Tateishi T, Watanabe M, Moriya H, Yamaguchi S, Sato T,
`Kobayashi S. No ethnic difference between Caucasian
`and Japanese hepatic samples in the expression frequency
`of CYP3A5 and CYP3A7 proteins. Biochem Pharmacol
`1999;57:935-9.
`9. Greenblatt DJ, Abernethy DR, Locniskar A, Harmatz
`JS, Limjuco RA, Shader RI. Effect of age, gender, and
`obesity on midazolam kinetics. Anesthesiology 1984;
`61:27-35.
`10. Harris RZ, Benet LZ, Schwartz JB. Gender effects in
`pharmacokinetics and pharmacodynamics. Drugs 1995;
`50:222-39.
`11. Harris RZ, Tsunoda SM, Mroczkowski P, Wong H, Benet
`LZ. The effects of menopause and hormone replacement
`therapies on prednisolone and erythromycin pharmacoki-
`netics. Clin Pharmacol Ther 1996;59:429-35.
`12. Tsunoda SM, Harris RZ, Mroczkowski PJ, Benet LZ.
`Preliminary evaluation of progestins as inducers of
`cytochrome P450 3A4 activity in postmenopausal
`women. J Clin Pharmacol 1998;38:1137-43.
`13. Thummel KE, Wilkinson GR. In vitro and in vivo dr