Oral first-pass elimination of midazolam involves both gastrointestinal and hepatic CYP3A-mediated metabolism*

eP “‘HERAPEUTICS A VOLUME59 NUMBER5 MAY 1996

l?HARMACOIL!NETICS AND DRUG DISPOSITION

Oral first-pass elimination of midazolam involves both gastrointestinal and hepatic CYl?3A-mediated metabolism

Objective: To determine in humans the relative roles of intestinal and hepatic metabolism in the oral first-pass elimination of a CYP3A substrate using midazolam as a model compound. Methods: Midazolam was administered intravenously ( 1 mg) or orally (2 mg) to 20 healthy young subjects ( 10 men and 10 women) in a random fashion, and the disposition of the drug and its 1’-hydroxy metabolite were determined. In separate in vitro studies, the CYP3A-mediated formation of 1’-hydroxymidazolam by human hepatic and intestinal microsomes was investigated. Results: No gender-related differences were noted in either the systemic (370 + 114 ml/min [mean + SD]) or oral (1413 + 807 ml/min) clearance values of midazolam. Despite complete oral absorption, measured oral bioavailability was on average about 50% less than that predicted on the assumption that only the liver contributed to first-pass metabolism. Pharmacokinetic estimation of the intestinal component indicated an extraction ratio (0.43 f 0.24) that was similar to that of the liver (0.44 + 0.14). l’-Hydroxymidazolam was extensively but variably formed in vitro by both hepatic and intestinal microsomes and, although the intrinsic clearance (V,,,/&) was higher in the liver preparations (540 f 747 versus 135 + 92 pl/min/mg protein), this difference was not statistically significant. Conclusions: These results show that the small intestine can be a major site for presystemic, CYP3A- mediated metabolism after oral administration. Moreover, it appears that this represents a true first-pass effect. In addition, intestinal and hepatic metabolism may be important factors in interindividual variability in disposition after oral administration of midazolam and similar CYP3A substrates. Finally, intestinal localization of CYP3A may be significant in metabolism-based drug-drug interactions. (Clin Pharmacol Ther 1996;59:491-502.)

Kenneth E. Thummel, PhD, Diarmuid O’Shea, MB, ChB, Mary F. Paine, BS, Danny D. Shen, PhD, Kent L. Kunze, PhD, James D. Perkins, MD, and Grant R Wilkinson, PhD Seattle, Wash., and Nashville, Term.

From the School of Pharmacy, University of Washington, Seattle, and the Department of Pharmacology, Vanderbilt University, Nashville.

Supported in part by grants GM-31304, GM-32165, and RR- 00095 from the US. Public Health Service and by Roche Laboratories, Nutley, N.J.

Received for publication July 6, 1995; accepted Oct. 17, 1995. Reprint requests: Grant R. Wilkinson, PhD, Department of Phar-

macology, Vanderbilt University, Nashville, TN 37232-6600. Copyright 0 1996 by Mosby-Year Book, Inc. 0009-9236/96/$5.00 + 0 1311169993

491

492 Thumvtzel et al.

A large number of orally administered drugs exhibit low systemic availability because of extensive first-pass metabolism catalyzed by the microsomal cytochrome P450 superfamily of enzymes. Many of these drugs are substrates of CYP3A,* a major constitutive enzyme(s) in liver and intestine.’ For example, nifedipine,* midazolam,3 etoposide,4 lidocaine,’ terfenadine,‘j cocaine,7 and cyclosporines are all CYP3A4 substrates with less than 50% oral bioavailability,’ despite being well absorbed. The site of such first-pass oxidative metabolism has usually been considered to be the liver because of its size, its relatively high level of P450 activity, and its ana- tomic location relative to the site of absorption. However, recent studies of cyclosporine disposition indicate that CYP3A metabolic activity in the intestinal mucosa may substantially contribute to the overall first-pass effect. ‘o-1’ Such findings raise the question of whether other orally administered CYP3A substrates are also subject to significant intestinal first-pass extraction.

Midazolam is metabolized almost exclusively by enzymes of the CYP3A subfamily,3”3,14 and in vivo such metabolism accounts for essentially all of the administered dose. The drug is rapidly eliminated after intravenous administration to a single major metabolite, 1’-hydroxymidazolam, and urinary recovery of this metabolite, as its glucuronide, accounts for at least 70% of an administered dose.15 Formation of a minor primary metabolite, 4-hydroxymidazolam, and a minor second- ary metabolite 1’,4-dihydroxymidazolam, are also catalyzed by CYP3A and together, as glucuronide conjugates, comprise another 4% to 6% of an administered dose. Plasma midazolam concentrations after usual therapeutic doses are well below the Michaelis constant (I&,) for the 1’-hydroxylation and 4-hydroxylation pathways,14 consistent with the linear pharmacokinetic behavior of midazolam in vivo.15

In the current study, midazolam was administered to 20 healthy volunteers by the intravenous and oral routes to determine the relative roles of the intestinal epithelium and liver in the first-pass effect of the

*The human CYP3A subfamily is composed of at least four different genes; CYP3A3, CYP3A4, CYP3A5, and CYP3A7. Gene products of CYP3A3 and 3A4 appear to be allelic vari- ants with virtually identical catalytic functions, and the latter protein is the predominant hepatic and intestinal CYP3A isoform. CYP3A5 enzyme exhibits catalytic activity similar to CYP3A4 but is expressed at detectable levels in only about 25% of the adult population, at least in liver. CYP3A7 is the predominant fetal liver 3A isoform but is not expressed at any significant level in the adult liver or intestine.

drug and to evaluate the feasibility of using this drug as a probe for intestinal CYP3A activity. The in vitro Michaelis-Menten kinetics of midazolam l’- hydroxylation in microsomes isolated from the liver and the small intestinal mucosa of organ donors were also characterized to assess the relative intrinsic activities of CYP3A in these two tissues.

MATERIAL. AND METHODS Chemicals and reagents. Midazolam, l’-hydroxy-

midazolam, diazepam, temazepam, and l’-[*H2]-l’- hydroxymidazolam were provided by Drs. William Garland and Bruce Mica, Roche Laboratories (Nutley, N.J.). N-Methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide (MTBSTFA) was purchased from Pierce Chemical Co. (Rockford, Ill.), and NADPH (reduced form, tetrasodium salt) and B-glucuronidase (G-7770) were obtained from Sigma (St. Louis, MO.). All other chemicals were of reagent grade or better.

Clinical protocol. The study was performed in healthy subjects as indicated by medical history, physical examination, and routine biochemical test values, including those reflective of hepatic and re- nal function. The 10 men were aged from 26 to 42 years (mean t SD, 32 + 5 years) and weighed from 73 to 100 kg (83 ? 9 kg), whereas the age of the 10 women ranged from 24 to 44 years (31 ? 6 years) and their weight was between 48 and 118 kg (70 ? 17 kg). With the exception of one individual, the subjects were all nonsmokers and were not regularly taking any medication, including known or sus- pected inducers or inhibitors of CYP3A activity. All subjects abstained from ethanol and caffeine inges- tion for 48 hours before the study period.

After an overnight fast, each subject randomly received either a 2 mg oral midazolam (Versed, Roche Laboratories, Nutley, N.J.) diluted in 50 ml apple juice or 1 mg intravenous midazolam by injection over 15 to 30 seconds. Ethylenediamine tet- raacetic acid (EDTA)-blood samples (7 ml) were then obtained through an indwelling catheter at 0, 15,30, 60, 120, 180,240, 300, and 360 minutes; after intravenous injection a 5-minute sample was also obtained. Urine was collected for 24 hours after drug administration. The harvested plasma and a urine aliquot were stored at -20” C until analyzed. Between 1 and 6 weeks later, each subject was crossed over to the other route of administration.

Analysis of midazolam and I ‘-hydroxymidazoiam in plasma and urine. Midazolam and l’-hydroxymidazolam in plasma were determined by gas

CLINICAL PHARhIACOLOGY & ‘THEKAPEUTICS VOLUhlE 59. NUMRt.K 5 Thumtzmel et al. 493

chromatography/selective ion mass spectrometric analysis, as described previously.i6 The same ana- lytic method was used to determine the compounds in urine after treatment of the sample with p-glucuronidase. For l’-hydroxymidazolam, this involved diluting 40 pl urine to 1 ml with deionized water, adding 1 ml of 100 mmol/L sodium acetate, pH 5.0, that contained 200 units of P-glucuronidase, and incubating at 37” C for 22 hours. A standard curve was prepared from blank urine spiked with known amounts of l’-hydroxymidazolam. The internal standard, temazepam (10 kg), was added to all samples and standards after the hydrolysis step, fol- lowed by the addition of 1 ml 50 mmol/L sodium carbonate, pH 12.8. The mixture was then extracted twice with 2 ml of 5% isoamyl alcohol: 95% toluene. The organic extracts were combined and prepared for gas chromatography/mass spectrometric analysis as described previously for plasma samples. For the measurement of unchanged midazolam in urine, 1 ml of 50 mmol/L sodium carbonate, pH 12.8, and 20 ng diazepam (internal standard) were added to 1 ml urine and extracted twice with 2 ml of 5% isoamyl alcohol : 95% toluene. The organic solvent from the combined extracts was evaporated under nitrogen at room temperature. The residue was dissolved in 20% MTBSTFA in acetonitrile (volivol) and analyzed by gas chromatography/mass spectrometry as described.‘”

The mass spectrometer was a VG model Trio- 1000 interfaced to a Hewlett-Packard 5890A gas chromatograph equipped with a Hewlett-Packard 7376A autoinjector and a DB-5 fused capillary column (30 m x 0.32 mm internal diameter, 0.25 micron film thickness; J & W Scientific, Ventura, Calif.). Helium carrier gas was used at a head pres- sure of 5 psi, and 2 ~1 injections were made in the splitless mode. The injector and transfer line tem- peratures were held at 250” C and 290” C, respectively. The oven temperature was held at 110” C for 1 minute, increased at 25” per minute to 290” C, held for 5 minutes, and increased at 40°C per minute to 300” C and held for a final 3 minutes. Methane was used as the reagent gas and the source temperature was held at 200” C.

Plasma protein binding. The fraction of midazolam bound to plasma proteins was determined by equi- librium dialysis. ” Plasma (1 ml), obtained after cen- trifugation of EDTA-treated blood collected before midazolam administration, was spiked with 200 ng midazolam and dialyzed against 1 ml of 67 mmol/L sodium phosphate buffer, pH 7.4, at 37” C, for 4

hours. Plasma and buffer cells were separated by a 14,000 molecular weight cutoff membrane (Spectra Por 2, Spectrum, Houston, Texas). At the end of the equilibration period, 0.8 ml aliquots of plasma and buffer were removed, spiked with 20 ng of the diazepam internal standard, and analyzed for midazolam as described above. The free fraction was calculated as the ratio of the midazolam concentration in buffer to that in the plasma.

Pharmacokinetic analysis. To estimate midazolam’s terminal half-life (t&, the plasma concentration-time data after intravenous administration were analyzed by use of a biexponential equation and the nonlinear least-squares regression program RSTRIP (version 5.0, MicroMath Scientific Soft- ware, Salt Lake City, Utah). The zero and first statistical moments (AUC and AUMC, respectively) were obtained by the trapezoidal method with extrapolation to infinite time with use of the regression estimate of the terminal t,,,.ls Systemic clearance from plasma (CL,) and the steady-state volume of distribution were calculated from the intravenous data by standard pharmacokinetic methods.‘s The trapezoidal method was also used to calculate the AUC after oral administration. In this case, the last three to four plasma concentration data were used to calculate the terminal t,,,. The extrapolated area for each of the 20 subjects after both routes of administration ranged from 6.8% to 23.6% of the respective total AUC. Oral clearance was estimated from the ratio of the administered dose to the AUC after oral dosing, and the observed absolute bioavailability of midazolam (Fobs) was determined according to the relationship:

F <,bs = (AUC”,,,/AUC,) * (DoseiV/Dose,,,,) (1)

The hepatic midazolam bioavailability (FH) was predicted from the observed systemic clearance,i” assuming that extrahepatic midazolam metabolism is negligible after intravenous administration (equation 2):

F, = 1 - E, = 1 - CLs/QH+m,,a (2)

in which liver plasma flow (QH,rlasma) = (1 - hematocrit) X QH,blood, and liver blood flow (Q H,b,ood) = body weight (kg) X 0.0216 L/min/kg.20 The hematocrit was assumed to be 0.48 for all subjects, and partitioning of midazolam into red blood cells was considered to be negligible at the observed in vivo midazolam concentrations.21

The overall observed oral bioavailability (Fobs) reflects the continuous product of the individual

494 Thzwwnel et al. CLINICAL I’HAPJdACOLOGY &THERAPEUTICS

MAY 1996

availabilities of each in-series organ involved in the first-pass effect,22 in this case assumed to be only the gastrointestinal epithelium (FGI) and the liver (FH), that is, Fobs = Foi * F,. Accordingly, the fraction of the absorbed dose extracted by the gastrointestinal tract (Ear) was estimated by equation 3:

E,, = 1 - F,, = 1 - Fobs/F, (3)

In vitro hepatic and intestinal microsomal metabolism. Samples of human liver and small intestine were obtained from organ donors through the Uni- versity of Washington Transplant Program and the Northwest Organ Procurement Agency. Consent for research was obtained in all cases. Four nonfatty livers and four segments of small intestine were removed from eight different individual donors after in situ vascular perfusion with cold (4” C) University of Wisconsin solution. The excised livers were kept on ice for between 12 and 24 hours before they were cut into approximately 10 gm amounts and snap-froze in liquid nitrogen. Small intestine (duodenum and/or je- junum) was removed during procurement surgery and immediately frozen on dry ice. Both liver and intestinal samples were stored at - 80” C. Microsomes were prepared from liver tissue as described previously.23 For intestinal microsomes, 50 to 100 gm of whole tissue was thawed in cold 0.9% sodium chloride that contained 0.1 mmol/L phenylmethylsulfonyl fluoride (PMSF), pH 7.4. The intestinal sac was opened longi- tudinally and the mucosa removed by gently scraping the luminal surface with a glass slide. The mucosal mass (10 to 20 ml) was diluted fivefold (vovVo1) with a 10 mmol/L potassium phosphate, pH 7.4, buffer that contained 0.25 mol/L sucrose, 1 mmol/L EDTA, and 0.1 mmol/L PMSF, and homogenized in a glass Wheaton tube with use of a motor-driven Teflon- tipped pestle (8 to 10 strokes). The microsomal pellet was then isolated by standard differential ultracentrif- ugation methods. Liver microsomes were stored in 0.25 mol/L sucrose and 1 mmol/L EDTA at -80” C. Microsomes from the intestinal mucosa were stored in a similar solution that also contained 0.1 mmol/L PMSF. Protein concentrations were determined by the method of Lowry et a1.24 using bovine serum albumin as the reference standard. An average yield of 25 mg microsomal protein was obtained from 50 gm of whole intestine.

Hepatic and intestinal microsomes from each do- nor were incubated in duplicate with 0,0.5, 1,2,4,6, and 8 pmol/L midazolam. Microsomes (20 to 100 kg protein) were thawed and suspended in cold potassium phosphate buffer (0.1 mol/L, pH 7.4) and mi-

dazolam in a total volume of 0.9 ml. Incubation mixtures were preincubated in a Dubnoff metabolic shaking incubator at 37” C for 4 minutes. The reac- tion was initiated by the addition of 0.1 ml NADPH (1 mmol/L final concentration in 1 ml) and termi- nated after 5 minutes by the addition of 1 ml 0.1 mol/L sodium carbonate (final pH about 11). Alka- linized incubation samples were spiked with 100 ng 1’-[2H2]-1’-hydroxymidazolam and extracted with 5 ml ethyl acetate. After extraction, solvent was removed by evaporation under nitrogen, and the resulting residue was dissolved in 100 l.~l of derivatiz- ing reagent (20% MTBSTFA in acetonitrile). The samples were transferred to autoinjector vials and sealed before they were heated at 80” C for 2 hours prior to analysis.

Quantitation of 1’-hydroxymidazolam in microsomal incubates was also accomplished by selective ion, gas chromatography/negative chemical ioniza- tion-mass spectrometry (GC/NCI-MS) similar to that previously described. In this case 1’-[2H2]-1’- hydroxymidazolam was used as the internal standard and both it and the unlabeled metabolite eluted at a retention time of 14 minutes and were monitored at m/z 327 and m/z 323, respectively. Concentrations of 1’-hydroxymidazolam were quantified by comparison of the peak area ratio from unknown samples to a standard curve (area ratio versus concentration) prepared by the addition of known amounts of the metabolite and 100 ng 1’-[2H2]-1’-hydroxymidazolam to human liver microsomes suspended in phosphate buffer. The assay sensitivity was 0.5 ng in a 1 ml incubation, with an interday variability of 8.6% and 6.3% at 1 rig/ml and 50 @ml, respectively.

Initial estimates of K, and V,, values for the l’- hydroxylation of midazolam were calculated for each microsomal sample by application of the linear Eadie-Hofstee equation for a single enzyme system to the product formation rate data. Final parameter values were obtained by nonlinear least-squares regression analysis (PCNONLIN, version 4.2, SC1 Software, Lexington, KY.). Coefficients of variation for K, and V,,, parameters, calculated from the analysis of unweighted data, were less than 13% in all but one tissue preparation; for the K, of intestinal sample HD-01, the variation was 23%. The unbound intrinsic clearance (CL:“,) for each microsomal sample was calculated as the ratio: V,,,JK,,,.‘9

Statistical analysis. Comparisons of the free fraction of midazolam in plasma and the urinary midazolam and l’-hydroxymidazolam recoveries were made by use of a Student paired t test. Gender-

CLINICAL PHARMACOLOGP & THERAPEUTKS VOLUME 59, NUMBER 5 Thzzmmel et al. 495

Subject 4

intravenous 1007

Time (min)

Subject 8

Time (min)

100; intravenous oral

10:

200

Time (min)

100 2w

Time (min)

300 )O

Fig. 1. Representative midazolam and l’-hydroxymidazolam plasma concentration-time profiles after intravenous and oral administration. Subject 4 exhibited an oral bioavailability close to the predicted hepatic availability (Fobs = 0.47; 1 - E, = 0.42), whereas subject 8 showed a markedly lower oral bioavailability compared with the predicted hepatic availability (Fobs = 0.19; 1 - E, = 0.68). Solid symbols correspond to midazolam and open symbols indicate l’-hydroxymidazolam.

related comparisons of the pharmacokinetic parameters of midazolam were made with use of a two- tailed unpaired t test with pooled variances. Linear regression analysis was used for comparisons of CL, or CL,/F,,, to the plasma l’-hydroxymidazolam/ midazolam concentration ratio. For each regression, a t test was used to test the null hypothesis that the correlation coefficient (r) was equal to zero. Com- parisons of V,,, and K, parameters obtained from liver and intestinal microsomal incubations were made with use of an unpaired t test.

RESULTS In vivo midazolam disposition. Plasma concentra-

tions of midazolam declined rapidly and in a biexponential fashion after intravenous administration (Fig. 1). In all subjects, measurable amounts of the

1’-hydroxy metabolite appeared in plasma within 5 minutes-the first collection time-and the maxi- mum concentration was achieved within 15 to 30 minutes. Subsequently, the l’-hydroxymidazolam level declined in parallel with that of parent drug, which exhibited a terminal t,,, ranging from 70 to 352 minutes (Table I). The systemic clearance of midazolam varied by almost threefold, and similar interindividual variability was also present in the volume distribution of the drug (Table I). The absorption of midazolam was rapid after oral administration and peak concentrations of both unchanged drug and its l’-hydroxy metabolite were achieved within 15 to 60 minutes (Fig. 1). The sub- sequent elimination was monoexponential, with a t,,* similar to or modestly longer than that observed after intravenous administration. Considerably

496 Thummel et al. CLINICAL PHAPJvlACOLOGP &THERAPEUTICS

MAY 19%

Table I. Pharmacokinetic parameters of midazolam disposition in healthy subjects after intravenous and oral administration

Subject No. Weight

fkl

Intravenous Oral

CL, V t,, CLsIF,,s (mllmin) (6 (min) (mllmin) F ohs Predicted FH

Women 1 75 2 66 3 118 4 68 5 61 6 70 7 64 8 80 9 65

10 61 Mean 2 SD 73 ? 17

Men 11 82 12 75 13 84 14 75 15 99 16 80 17 100 18 78 19 82 20 80 Mean k SD 84 ? 9

p Value* 0.094

Overall mean 2 SD 78 ? 14

360 127.9 352 992 0.36 0.57 252 79.0 262 641 0.39 0.67 302 48.3 145 798 0.38 0.78 440 78.4 160 936 0.47 0.42 281 60.1 175 820 0.34 0.59 386 49.6 99 978 0.39 0.49 282 37.8 228 1942 0.14 0.61 289 44.6 122 1514 0.19 0.68 239 36.4 110 830 0.29 0.67 412 70.3 169 1676 0.24 0.40

324 i 70 63.2 2 27.6 182 ? 78 1113 2 437 0.32 2 0.10 0.59 +- 0.12

292 66.1 183 577 0.50 0.68 503 68.1 102 3339 0.15 0.40 376 36.0 70 3086 0.12 0.40 32.5 44.8 104 1294 0.25 0.62 330 66.3 152 1106 0.30 0.70 473 51.2 79 1213 0.39 0.48 583 60.7 78 1742 0.33 0.48 665 98.3 137 2797 0.24 0.24 350 40.0 82 990 0.35 0.62 258 54.2 185 980 0.26 0.71

416 2 134 58.6 -c 18.0 117 ? 44 1712 5 972 0.29 k 0.11 0.53 5 0.16

0.073 0.66 0.034

370 2 114 60.9 k 22.8 150 -t 70

0.097 0.54 0.39

1413 2 807 0.30 k 0.10 0.56 ? 0.14

CL,, Systemic plasma clearance; Vss, steady-state distribution volume; t,,,, terminal half-life; CL,/F,,,, oral clearance; Fobs, observed absolute bioavailability. *Women versus me*.

more intersubject variability was noted in the plasma concentration-time profiles of both midazolam and its metabolite after oral dosing than after intravenous injection (Fig. 1). Substratification of the subjects according to gender suggested differences in the disposition parameters between men and women. For example, the systemic and oral clearances of midazolam, as well as its volume of distribution, tended to be greater in men (Table I). However, these differences were not statistically significant and furthermore were not apparent after normalization for body weight. The terminal t,,, of midazolam was longer in women than men, but because of the wide interindividual variability, a level of statistical difference was not reached (Table I).

There was no difference in the extent of the plasma binding of midazolam after intravenous and oral administration (fU = 1.92% -I 0.69% versus

2.12% ? 1.28%, respectively). There was also no gender-related difference in such binding. A correlation was observed (r = 0.70, p < 0.001) between the oral clearance of midazolam and the l’- hydroxymidazolamlmidazolam plasma concentration ratio measured 60 minutes after administration. A similar statistically significant but weaker correlation (r = 0.43,~ < 0.05) was also noted between systemic clearance and this plasma concentration ratio measured 30 minutes after intravenous administration.

Less than 1% of the administered dose was excreted in urine over 24 hours as unchanged drug, regardless of route of administration (Table II). The 1’-hydroxymidazolam urinary recovery, represented a substantial fraction of the administered dose: mean values of 69.3% and 72.9% for the intravenous and oral doses, respectively. After intravenous midazolam administration, women excreted slightly more 1’-hydroxymidazolam than men (p = 0.049)

CLINICAL PHAlU4ACOLOGY & THERAPEUTICS VOLUME 59, SUhlRER 5 Thumzmel et al. 497

Table II. Midazolam and I’-hydroxymidazolam disposition after intravenous and oral administration

Subject No.

Intravenous Oral

Urinary excretion AUC Urinary excretion AUC

l’-Hydroxy- I’-Hydrony- 1’ -Hydroxy- I’-Hydroq Midazolam midazolam midazolam Midazolam midazolam midazolam

(9% dose) (70 dose) (ng * ml-’ * hr) (% dose) (5% dose) (ng * ml-’ * hr)

Women 1 2 3 4 5 6 7 8 9

10 Mean L SD

Men 11 12 13 14 15 16 17 18 19 20 Mean -C SD

p Value*

Overall mean -C SD

0.38 49.3 398 0.50 71.4 859 0.20 58.2 450 0.26 83.3 758 0.29 51.6 152 0.34 71.2 229 0.28 48.1 290 0.34 94.6 455 0.26 69.8 389 0.26 70.3 767 0.30 66.2 218 0.23 95.1 677 0.30 58.2 241 0.21 52.5 237 0.29 90.0 647 0.28 69.8 757 0.20 77.8 384 0.38 76.6 548 0.18 64.4 394 0.29 75.7 664

0.27 + 0.06 63.4 2 13.3 356 2 140 0.31 t- 0.08 76.0 2 12.6 595 -+ 2221-

0.43 71.6 450 0.62 50.2 699 0.14 50.7 285 0.31 59.0 496 0.33 76.7 270 0.60 73.6 288 0.18 63.2 407 0.31 67.6 614 0.37 85.2 346 N.D. 82.4 484 0.32 88.1 302 0.67 76.1 462 0.24 82.2 269 0.51 53.3 303 0.33 71.0 289 0.18 81.0 422 0.22 76.0 395 0.45 76.2 545 0.23 87.3 392 0.57 78.2 465

0.28 ? 0.09 75.2 ? 11.7 340 5 66 0.47 i 0.17 69.8 ? 11.7 478 k 1267

0.75 0.049 0.70 0.017

69.3 -e 13.6 348 ? 107 0.38 t 0.15

0.26 0.16

0.27 t 0.07 72.9 -c 12.3 536 -c 186-f

Midazolam and I’-hydroxymidazolam are expressed as a percentage of the intravenous dose (1 mg) or oral dose (2 mg) recovered in 24.hour urine. *Women versus me*. tp < 0.01, intravenous versus oral.

but this difference was not observed after oral administration. However, there was a 52% greater mean recovery of unchanged midazolam in men compared with women @ = 0.017) after oral administration. All other differences in mean urinary recoveries between genders were minor and did not reach statistical significance. Although the mean fractional urinary recovery of 1’-hydroxymidazolam after intravenous and oral administration did not differ, the systemic plasma AUC of metabolite was 54% higher after the oral dose (p = 0.0004) compared with the intravenous dose (Table II). This difference was independent of gender.

The oral bioavailability of midazolam was not different in men and women and varied fourfold (12% to 50%) in the 20 subjects, with an overall mean value of 30% t 10% (Table I). Except in two

subjects, oral bioavailability was substantially less than the hepatic bioavailability (FH) predicted from the systemic clearance of the drug, assuming that the liver was the only organ involved in the presystemic metabolism (Table I). On the basis of these values, estimates were obtained of the relative contribu- tions of the gastrointestinal epithelium and liver to the overall first-pass effect (Table I and Fig. 2). The estimated gastrointestinal extraction ratio ranged from 0% to 77% (mean, 43% + 24%) whereas that for hepatic extraction was 22% to 76% (mean, 44% ? 14%). The correlation between these two extraction values was significant but relatively weak (r = -0.49,p < 0.05).

In vitro metabolism of midazolam. Microsomes prepared from both the liver and the intestinal epithelium metabolized midazolam to its l’-hydroxy me-

498 Thummel et al. CLINICAL PHARMACOLOGY & THERAPEUTICS

MAY 1996

Intestine subject Liver

100 7.5 50 25 0 0 25 50 75 1

Calculated Extraction Ratio (%)

Fig. 2. Estimated intestinal (left panel) and hepatic (right panel) first-pass extraction ratios of midazolam after oral administration in the 20 study subjects.

tabolite. Such formation was well described by Michaelis-Menten type kinetics, assuming a single enzyme site (Fig. 3). The estimated K, value for 1’-hydroxymidazolam formation was similar in both tissues (Table III) and varied only modestly+efficients of variation were only 11.3% (liver) and 15.1% (intestine). However, greater variability was noted in the V,, values: the coefficients of variation were 43.9% and 61.3% for liver and intestine, respectively. Thus, although the mean V,, for the intestinal microsomes was 54% lower than the comparable value from the liver preparations (463 + 284 versus 1004 2 441 pmol/min/mg protein), this difference was not statistically significant. A similar situation also applied to the estimated CLynt values, that is, considerable interindividual variability and a higher but not statistically significantly different rate of metabolism by the liver (540 + 247 versus 13.5 ? 92 kl/min/mg protein).

DISCUSSION The observed disposition characteristics of mida-

zolam and its l’-hydroq metabolite are similar to those previously reported in healthy subjects after intravenous and oral administration.‘5,21 In particular, the low oral availability (mean, 30% + lo%), indicative of an extensive first-pass effect, has been noted earlier, along with the large interindividual variability in this parameter.25W27 Several factors determine oral availability, including the fraction of

the administered dose that is absorbed intact from the gastrointestinal lumen (Fabs), the fraction of the absorbed drug that is not metabolized during pas- sage through the gastrointestinal epithelium (FGI), and the fraction of the absorbed dose delivered to the liver through the portal vein that is not subsequently lost during first-pass hepatic metabolism (FH). In principle, first-pass metabolism by the lungs should also be considered, but there is little evi- dence of substantial constitutive CYP3A in this organ.28 Incomplete oral absorption of midazolam is also not likely to be a determinant because the urinary recovery of the major metabolite, 1’-hydroxy midazolam, was similar after both intravenous and oral administration (Table II). Also, more than 90% of an oral dose of radioactively labeled dose of midazolam was reported to be excreted in urine.29 By contrast, the potential for significant metabolism of midazolam in both the intestinal epithelium and the liver is strongly suggested by the in vitro findings.

As previously noted,3,13,14,16 human hepatic microsomes were found to exhibit a high level of midazolam 1’-hydroxylase activity. Moreover, a good correlation has been noted between the formation of this metabolite in vivo and the in vitro hepatic CYP3A microso- ma1 content.‘6’30 The immunochemical demonstration of CYP3A in the human gastrointestinal tract, especially the duodenum and jejenum,10,11,31,32 would therefore infer that midazolam would be l’- hydroxylated by this tissue. Accordingly, the present in

CLINICAL PHARMACOLOGY 8; THERAPEUTICS VOLUME 59, NUMBER 5

Table III. Kinetics of hepatic and intestinal microsomal midazolam l’-hydroxylation

V max CGI, (pmoliminlmg (~iminlmg

Donor ~p%L~ protein) protein)

Liver HL-102 3.05 562 814.3 HL-103 2.77 737 266.1 HL-106 2.85 1177 413.0 HL-129 2.31 1540 666.7

Small intestine HD-01 3.30 519 157.2 HD-02 4.38 222 50.7 HD-03 3.44 268 77.9 HD-10 3.20 842 256.1

K, and V,, parameters were obtained by nonlinear regression analysis of velocity concentration data. The unbound intrinsic clearance (CL:“,) was calculated as the ratio: V,,,/K,.

vitro results with intestinal epithelial microsomes are entirely consistent with this expectation and also con- firm a recent similar observation based on fresh biopsy sampling of the small intestine.33 However, a surpris- ing finding was the relatively high level of midazolam l’-hydroxylase activity in the intestinal microsomes compared with that in hepatic microsomes. The mean intestinal CL;, was 34% of the hepatic value; however, there was considerable overlap in the two sets of estimates.

Recently the kinetic parameters of the midazolam 1’-hydroxylase activity determined in vitro using human hepatic microsomes have been suc- cessfully extrapolated to the in vivo situation.16 With this approach, the experimental CL:,,, values (Table II) provided an estimate of midazolam’s clearance of between 213 and 536 ml/min. By comparison, the systemic clearance of the drug measured after intravenous administration ranged from 239 to 665 ml/min. The close agreement between these two values would suggest that metabolic elimination is essentially limited to the liver after intravenous administration, that is, circulat- ing drug does not gain ready access to CYP3A in the intestinal epithelium. Such a notion is consistent with the localization of CYP3A in terminally differentiated enterocytes toward the villus tip,31 which probably have relatively limited perfusion.34 By extension, presystemic metabolism at this site would represent a true first-pass effect after oral drug administration, that is, a fraction of the absorbed dose is metabolized only during the absorption process.

” (pmol/minhng)

i!lwnzmel et al. 499

6W- Km = 2.8 pt.4 “mm = 737 500. pmo”mhhg

400.

04 0 2 4 6 s II

S (PM)

1 Km.438pM I

OY . , I 0 2 4 6 s 10

S tuMJ

Fig. 3. In vitro 1’-hydroxylation of midazolam in microsomes isolated from human liver and intestinal mucosa. Representative plots of I’-hydroxymidazolam formation velocity (V) against substrate concentration in human liver (upper panel) and small intestinal (lower panel) microsomes. A linear Eadie-Hofstee transformation of the data is shown as an inset.

It is difficult to extrapolate the in vitro data on midazolam’s l’-hydroxylation by intestinal microsomes to the in vivo situation. First, unlike the liver, the intestine is a highly heterogeneous tissue, and therefore the currently available organ clearance models” necessary for such extrapolation might not apply. Also, the area of the small intestine to which midazolam is exposed and the longitudinal distribution of CYP3A activity along the intestine are not well established in humans. Accordingly, the mass of available enzyme to which a drug substrate may be exposed is unknown and only a rough qualitative estimate can be made of the ability of the intestine to metabolize midazolam in vivo based on the in vitro findings.

However, the measured clearance values after both oral and intravenous administration allow quantitative estimation of intestinal metabolism with respect to its relative contribution, as well as that of the liver, to midazolam’s overall first-pass

500 Thummel et al. CLINICAL PHARMACOLOGY&THERAPEL7TICS

h4AY 1996

effect. If the liver is assumed to be the only site of metabolism and if the estimated hepatic blood flow rate accurately reflects the actual value in each subject, then equation 2 permits calculation of the mag- nitude of the extraction ratio of this organ (Fig. 2) and the resulting bioavailability (Table I). In all but two subjects, the predicted bioavailability was substantially greater-on average twofold-than the observed value (Table I), indicating that first-pass metabolism was in fact occurring in the gastrointestinal epithelium. The extent of such metabolism was then estimated by equation 3, with the finding (Fig. 2) that the intestinal extraction ratio ranged from zero to 0.77 (mean, 0.43 -+ 0.24). Comparative values for the liver were 0.24 to 0.76 (mean, 0.44 -+ 0.14). Such estimates are of course dependent on the validity of the assumptions that any intraindividual variability in midazolam’s metabolism is small and that systemic clearance after intravenous administration reflects only hepatic elimination. The current findings, vide supra, support the latter assumption; moreover, is should be noted that if intestinal metabolism does contribute to systemic clearance, then the estimated intestinal extraction ratio would in fact be underestimated. It also appears that the systemic clearance of midazolam exhibits little variability within a subject over a period similar to that of this study (Thummel KE and Kharasch ED, unpublished observation, 1995). There was a weak negative correlation between the calculated intestinal and hepatic extraction ratios, suggesting that CYP3A-mediated metabolic activity in the two organs is not tightly coregulated. A similar conclusion was reached by Lown et a1.,33 who observed a lack of correlation between the erythromycin breath test, an in vivo measure of (hepatic) CYP3A activity,35 and either midazolam l’- hydroxylation or CYP3A content in biopsy samples of small intestinal mucosa.

A significant quantitative role for CYP3A- mediated first-pass metabolism in the intestinal epithelium is also supported by studies with cyclosporine, another substrate of the enzyme. For example, it was observed that after duodenal instillation of the immunosuppressant during the anhepatic phase of liver transplantation, significant blood levels of its oxidative metabolites were present in the portal vein.” Also, the differential effect of rifampin induction on the disposition of cyclosporine after oral and intravenous administration suggested an intestinal extraction ratio of 49% to 82%.i2 Accord- ingly, it is not unreasonable to speculate that all

CYP3A substrates undergo intestinal first-pass metabolism, the extent of which may be quite marked, depending on the importance of the mediated path- way in the drug’s overall metabolism, its affinity for the enzyme(s), and the amount of CYP3A enzyme present in each individual subject. If this is the case, then the consequences of modulating intestinal CYP3A activity, especially by local inhibition, may be pronounced, as probably exemplified by the interaction between terfenadine and ketoconazole,“6 midazolam and ketoconazole,37 and midazolam and erythromycin.38

The present findings also extend and support the use of midazolam as an in vivo probe for the determination of CYP3A activity in individual subjects and factors that modulate this.14,30 Mea- surement of the formation clearance of l’- hydroxymidazolam after oral and intravenous administration is currently the best way of accomplishing this but requires extensive blood sampling and urine collection. Previously, we observed in liver transplant patients3’ that the l’- hydroxymidazolamimidazolam plasma concentration ratio measured 30 minutes after an intravenous administration was well correlated to the clearance of midazolam (r = 0.84). Although statistically significant, the correlation in the present study of healthy subjects was much less (r = 0.43). The major pharmacokinetic difference between these two studies was the greater variability in the plasma concentration ratio in the patients (12-fold) compared with the healthy in- dividuals (threefold). Thus the simpler phenotypic trait measurement based on a single plasma sample may have significant limitations when applied to pop- ulations in which interindividual variability in midazolam metabolic clearance is small.

In summary, an analysis of intravenous and oral midazolam kinetics in healthy young male and fe- male subjects suggests that there is a significant contribution by the intestine to the overall first-pass elimination of this CYP3A substrate after oral administration. The mean intestinal extraction ratios calculated for liver and intestine were comparable. Estimates of the unbound intrinsic clearance (V,,.J K,) for midazolam l’-hydroxylation, measured in microsomes from human liver and small intestine, were not significantly different, although the mean value for intestine was approximately one-fourth the value for liver. Both the in vivo and in vitro midazolam data suggest that the small intestine can be a major source of interindividual variability in oral

Thzwzuzel et al. 50 1

bioavailability. If this is borne out by future studies with other CYP3A substrates, the small intestine may well be an important site for metabolically based drug-drug interactions.

References 1. dewaziers I, Cugnenc PH, Yang CS, Leroux JP,

Beaune PH. Cytochrome P450 isoenzymes, epoxide hydrolase and glutathione transferases in rat and human hepatic and extrahepatic tissues. J Pharmacol Exp Ther 1990;253:387-94.

2. Guengerich FP, Martin MV, Beaune PH, Kremers P, Wolff T, Waxman DJ. Characterization of rat and human liver microsomal cytochrome P-450 forms involved in nifedipine oxidation, a prototype for genetic polymorphism in oxidative drug metabolism. J Biol Chem 1986;261:5051-60.

3. Kronbach T, Mathys D, Umeno M, Gonzalez FJ, Meyer UA. Oxidation of midazolam and triazolam by human liver cytochrome P450IIIA4. Mol Pharmacol 1989;36:89-96.

4. Relling MV, Nemec J, Schuetz EG, Schuetz JD, Gonzalez FJ, Korzekwa KR. 0-Demethylation of epi- podophyllotoxins in catalyzed by human cytochrome P450 3A4. Mol Pharmacol 1994;45:352-8.

5. Bargetzi MJ, Aoyama T, Gonzalez FJ, Meyer UA. Lidocaine metabolism in human liver microsomes by cytochrome P45OIIIA4. Clin Pharmacol Ther 1989;46: 521-7.

6. Yun CH, Okerholm RA, Guengerich FP. Oxidation of the antihistaminic drug terfenadine in human liver microsomes: role of cytochrome P-450 3A(4) in N-dealkylation and C-hydroxylation. Drug Metab Dispos 1993;21:403-9.

7. LeDuc BW, Sinclair PR, Shuster L, Sinclair JF, Evans JE, Greenblatt DJ. Norcocaine and N-hydroxy- norcocaine formation in human liver microsomes: roles of cytochrome P-4503A4. Pharmacology 1993;46:294- 300.

8. Combalbert J, Fabre I, Fabre G, et al. Metabolism of cyclosporin A; IV: purification and identification of the rifampin-inducible human liver cytochrome P-450 (cyclosporin A oxidase) as a product of P450IIIA gene subfamily. Drug Metab Dispos 1989; 17:197-207.

9. Benet LZ, 8ie S, Schwartz JB. Design and optimiza- tion of dosage regimens; pharmacokinetic data. In: Hardman JG, Limbird LE, Molinoff PB, Ruddon RW, Gilman AG, eds. Goodman and Gilman’s The Pharmacologic Basis of Therapeutics. New York: Per- gamon Press, 1995:1707-92.

10. Kolars JC, Awni WM, Merion RM, Watkins PB. First-pass metabolism of cyclosporin by the gut. Lan- cet 1991;338:1488-90.

11. Kolars JC, Schmeidlin-Ren P, Scheutz JD, Fang C,

Watkins PB. Identification of rifampin-inducible P450IIIA4 (CYP3A4) in human small bowel enterocytes. J Clin Invest 1992;90:1871-8.

12. Hebert MF, Roberts JP, Prueksaritanont T, Benet LZ. Bioavailability of cyclosporine with concomitant rifampin administration is markedly less than predicted by hepatic enzyme induction. Clin Pharmacol Ther 1992;52:453-7.

13. Fabre G, Rahmani R, Placidi M, et al. Characteriza- tion of midazolam metabolism using human hepatic microsomal fractions and hepatocytes in suspension obtained by perfusing whole human livers. Biochem Pharmacol 1988;37:4389-97.

14. Gorski JC, Hall SD, Jones DR, VandenBranden M, Wrighton SA. Regioselective biotransformation of midazolam by members of the human cytochrome P450 3A (CYP3A) subfamily. Biochem Pharmacol 1994;47:1643-53.

15. Dundee JW, Halliday NJ, Harper KW, Brogden RN. Midazolam: a review of its pharmacologic properties and therapeutic use. Drugs 1984;28:519-43.

16. Thummel KE, Shen DD, Pod011 TD, et al. Use of midazolam as a human cytochrome P450 3A probe; I: in vitro-in vivo correlations in liver transplant patients. J Pharmacol Exp Ther 1994;271:549-56.

17. Semmes RLO, Shen DD. Nonlinear binding of valproic acid (VPA) and E-A’-valproic acid to rat plasma proteins. Pharm Res 1990;7:461-7.

18. Gibaldi M, Perrier D. Pharmacokinetics. 2nd ed. New York: Marcel Dekker, 1982.

19. Wilkinson GR. Clearance approaches in pharmacology. Pharmacol Rev 1987;39:1-47.

20. Pirttiaho HI, Sotaniemi EA, Pelkonen RO, Pitkanen U. Hepatic blood flow and drug metabolism in patients on enzyme-inducing anticonvulsants. Eur J Clin Pharmacol 1982;22:441-5.

21. Smith MT, Eadie MJ, O’Rourke Brophy T. The pharmacokinetics of midazolam in man. Eur J Clin Phar- macol 1981;19:271-8.

22. Pond SM, Tozer TN. First-pass elimination-basic concepts and clinical consequences. Clin Pharmaco- kinet 1984;9:1-25.

23. Thummel KE, Kharasch ED, Pod011 T, Kunze K. Human liver microsomal enflurane defluorination catalyzed by cytochrome P450 2El. Drug Metab Dis- pos 1993;21:350-7.

24. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265-75.

25. Allonen H, Ziegler G, Klotz U. Midazolam kinetics. Clin Pharmacol Ther 1981;30:653-61.

26. Klotz U, Ziegler G. Physiologic and temporal variation in hepatic elimination of midazolam. Clin Phar- macol Ther 1982;32:107- 12.

27. Mandema JW, Tuk B, van Steveninck AL, Breimer DD, Cohen AF, Danhof M. Pharmacokinetic-

502 Thamamel et al. CLINICAL PHAlMACOLOGY & THERAPEUTICS

MAY 1996

pharmacodynamic modeling of the central nervous system effects of midazolam and its main metabolite cw-hydroxymidazolam in healthy volunteers. Clin Phar- macol Ther 1992;51:715-28.

28. Murray GI, Barnes TS, Sewell HF, Ewen SWB, Melvin WT, Burke MD. The immunocytochemical localisation and distribution of cytochrome P-450 in normal human hepatic and extrahepatic tissues with a monoclonal antibody to human cytochrome P-450. Br J Clin Pharmacol 1988;25:465-75.

29. Heizmann P, Ziegler WH. Excretion of 14C- midazolam in humans following oral dosing. Arznei- mittelforschung 1981;31:2220-3.

30. Thummel KE, Shen DD, Pod011 T, et al. Use of midazolam as a human cytochrome P450 3A probe; II: characterization of inter- and intraindividual hepatic P450 3A variability after liver transplantation. J Pharmacol Exp Ther 1994;271:557-66.

31. Kolars JC, Lown KS, Schmiedlin-Ren P, et al. CYP3A gene expression in human gut epithelium. Pharmaco- genetics 1994;4:247- 59.

32. Watkins PB, Wrighton SA, Schuetz EG, Molowa DT, Guzelian PS. Identification of glucocorticoid-

inducible cytochromes P-450 in the intestinal mucosa of rats and man. J Clin Invest 1987;80:1029-36.

33. Lown KS, Kolars JC, Thummel KE, et al. Interpatient heterogeneity in expression of CYP3A4 and CYP3A5 in small bowel: lack of prediction by the erythromycin breath test. Drug Metab Dispos 1994;22:947-55.

34. Minchin RF, Ilett KF. Presystemic elimination of drugs: theoretical considerations for quantifying the relative contribution of gut and liver. J Pharm Sci 1981;71:458-60.

35. Watkins PB. Non-invasive tests of CYP3A enzymes. Pharmacogenetics 1994;4:171-84.

36. Honig PK, Wortham DC, Zamani K, Conner DP, Mullin JC, Cantilena LR. Terfenadine-ketoconazole interaction: pharmacokinetic and electrocardio- graphic consequences. JAMA 1993;269:1513-8.

37. Olkkola KT, Backman JT, Neuvonen PJ. Midazolam should be avoided in patients receiving the systemic antimycotics ketoconazole or itraconazole. Clin Phar- macol Ther 1994;55:481-5.

38. Olkkola KT, Aranko K, Luurila H, et al. A potentially hazardous interaction between erythromycin and midazolam. Clin Pharmacol Ther 1993;53:298-305.

~ Availability of Journal Back Issues As a service to our subscribers, copies of back issues of Clinical Pharmacology & 7Eerapeuticsfor the preceding

5 years are maintained and are available for purchase from the publisher, Mosby-Year Book, Inc., at a cost of $12.50 per issue. The following quantity discounts are available: 25% off quantities of 12 to 23, and 33% off quantities of24 or more. Please write to Mosby-Year Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, or call (800)453-4351 or (314)453-4351 for information on availability of particular issues. Ifunavailable from the publisher, photocopies ofcomplete issues are available from UMI, 300 N. Zeeb Rd., Ann Arbor, MI 48106 (313)761-4700.

Documents

Oral first-pass elimination of midazolam involves both gastrointestinal and hepatic CYP3A-mediated metabolism*