In vitro drug interaction between diflunisal and indomethacin via glucuronidation in humans

BIOPHARMACEUTICS & DRUG DISPOSITIONBiopharm. Drug Dispos. 27: 267–273 (2006)

Published online in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/bdd.507

In vitro Drug Interaction between Diflunisal andIndomethacin via Glucuronidation in Humans

Yuji Mano*, Takashi Usui and Hidetaka KamimuraDrug Metabolism Research Laboratories, Astellas Pharma Inc., 1-8, Azusawa 1-Chome, Itabashi-ku, Tokyo, 174-8511, Japan

ABSTRACT: It was reported that the plasma concentration of indomethacin was increased withconcomitant oral dosages of diflunisal in humans. Both indomethacin and diflunisal areglucuronidated in humans. The effects of diflunisal on the indomethacin glucuronidation werethus investigated in vitro using human liver microsomes (HLM) and human intestine microsomes(HIM) in order to assess the drug–drug interaction. The glucuronidation of indomethacin in HLMshowed atypical kinetics with Km and Ksi values of 210 and 89.5 mm, respectively, while HIMexhibited Michaelis–Menten kinetics with a Km value of 17.4 mm. Diflunisal inhibited theindomethacin glucuronidation in HLM with IC50 values ranging from 100 to 231 mm. In HIM,inhibition of the indomethacin glucuronidation by diflunisal was more potent with IC50 values of15.2–48.7 mm. When the clinical dose of diflunisal (250 mg b.i.d.) is taken into consideration, it isexpected that the diflunisal concentration in the intestine would be higher than the IC50 values forindomethacin glucuronidation in the intestine. These findings suggest that the clinical drug–druginteraction between diflunisal and indomethacin may be at least partly attributable to the inhibitionof indomethacin glucuronidation by diflunisal in the intestine. Copyright # 2006 John Wiley &Sons, Ltd.

Key words: drug interaction; indomethacin; diflunisal; human; glucuronidation

Introduction

Drug–drug interactions due to metabolism havefrequently been reported in cytochrome P450(CYP) metabolism, but significant drug interac-tions via glucuronidation are rarely seen. Plasmalevels of indomethacin in humans increased toapproximately twice the normal level withconcomitant dosages of diflunisal, a non-steroi-dal anti-inflammatory drug (NSAID) [1], andindomethacin was subjected to glucuronidationin humans [2]. Diflunisal inhibited UDP-glucur-onosyltransferase (UGT) 1A1-catalysed estradiol3b-glucuronidation in human liver microsomes

(HLM) [3], 4-methylumbelliferone glucuronida-tion in recombinant UGT1A9 [4] and mycophe-nolic acid glucuronidation in HLM [5]. Plasmaconcentrations of diflunisal reached approxi-mately 80 mg/ml after repeated oral dosages of250 mg to humans [6], and the protein bindingof diflunisal was >98% [7,8]. In addition, it isexpected that oral dosages of diflunisal wouldmake the concentration of diflunisal in thegastrointestinal tract substantially high. Whenthe inhibition constant for diflunisal in indo-methacin glucuronidation, which remains to bedetermined, is lower than the diflunisal concen-tration in tissues involved in the glucuronidationof indomethacin, it might be that the drug–druginteraction between indomethacin and diflunisalis at least partly attributable to glucuronidationinhibition. Glucuronidation in the liver has been

*Correspondence to: Drug Metabolism Research Laboratories,Astellas Pharma Inc., 1-8, Azusawa 1-Chome, Itabashi-ku, Tokyo,174-8511, Japan. E-mail: [email protected]

Received 12 September 2005Revised 6 February 2006

Accepted 7 February 2006Copyright # 2006 John Wiley & Sons, Ltd.

extensively investigated, however, recently theexploration of extra-hepatic glucuronidation,such as that in the intestine, has begun [9]. TheUGTs in the intestine could contribute signifi-cantly to the bioavailability of drugs and thedetoxication of xenobiotics. Thus, in this study,the possible inhibition of indomethacin glucur-onidation by diflunisal was evaluated in vitrousing HLM and human intestine microsomes(HIM).

Materials and Methods

Chemicals and reagents

Indomethacin and diflunisal were purchasedfrom Sigma (St Louis, MO, USA) and ICNBiomedicals (Aurora, Ohio, USA), respectively.Pooled HLM and HIM were purchased from BDBiosciences (Woburn, MA, USA). All otherchemicals were of the highest grade or highperformance liquid chromatography (HPLC)grade.

Glucuronidation of indomethacin

Indomethacin was incubated in reaction mixturesof 0.25 ml of 50 mm Tris-HCl buffer (pH 7.5)containing 8 mm MgCl2, 25 mg/ml alamethicin,10 mm saccharic acid 1,4-lactone, 2 mm UDP-glucuronic acid (UDPGA) and either HLM orHIM. The microsomal concentration in HLM andHIM was 0.2 mg protein/ml. The indomethacinwas dissolved in dimethyl sulfoxide (DMSO) andits concentrations in kinetic studies were 2–500 mm and 2–100 mm in HLM and HIM, respec-tively. In the inhibition studies, the indomethacinconcentrations were 20, 100 and 500 mm in HLM,and 20, 50 and 100 mm in HIM. Diflunisal, used asan inhibitor for indomethacin glucuronidation,was dissolved in DMSO, and its concentrationswere 10–500 mm and 2–500 mm in HLM and HIM,respectively. After pre-incubation of the reactionmixtures at 37 8C for 5 min, the reaction wasstarted by adding UDPGA. The reaction mixtureswere then incubated for the designated times at37 8C. The reaction time was 20 min in HLM forkinetic and inhibition studies, while it was 40 and20 min in HIM for kinetic and inhibition studies,

respectively. The kinetic analysis was conductedunder conditions of protein concentrations andincubation times yielding linear product forma-tion. The reactions were terminated by theaddition of acetonitrile (0.04 ml), followed by0.01 ml of 10% formic acid (v/v), the mixtureswere then centrifuged at 1870� g for 5 min toobtain the supernatant. Aliquots (0.12 or 0.15 ml)of the supernatant were injected into the HPLCequipped with an ultraviolet detector.

Assay

The peak area of the indomethacin glucuronidewas analysed using reverse-phase HPLC (Shi-madzu, Kyoto, Japan). The HPLC system con-sisted of LC-10AS pumps, a SCL-10A systemcontroller, a SIL-10A autosampler, a SPD-10AVultraviolet detector and a C-R4AX integrator.Chromatographic separation was achieved usingan Inertsil Ph column (4.6 mm� 150 mm, 5 mm,GL Sciences, Tokyo, Japan) with an ultravioletwavelength of 300 nm. The mobile phase con-sisted of 0.1% formic acid:acetonitrile (6:4, v/v)and was delivered at a flow rate of 1.0 ml/min.The retention times of indomethacin glucuronideand indomethacin were approximately 7 and16 min, respectively. The peak at the retentiontime of 7 min was confirmed to be a glucuronideof indomethacin: this peak was observed only inthe presence of UDPGA in HLM and HIM. Theelectrospray ionization mass spectrum of thepeak in the positive and negative ion mode hadions at m/z 534.0 and 532.1, and the product ionspectrum in the positive ion mode showed aprotonated aglycon ion at m/z 358.0 (data notshown). Because the authentic standard forindomethacin glucuronide was not available,glucuronide in the reaction mixtures was quanti-fied using an indomethacin standard curve. Thepeak area of indomethacin glucuronide that wasformed was comparable to the reduced peak areaof indomethacin during the incubation of indo-methacin in the presence of UDPGA.

Kinetic analysis

The kinetics of indomethacin glucuronidation inHLM and HIM were fitted to substrate inhibitionkinetics (Equation (1), [10]) and Michaelis–Men-ten kinetics (Equation (2)), respectively, in order

Y. MANO ET AL.268

Copyright # 2006 John Wiley & Sons, Ltd. Biopharm. Drug Dispos. 27: 267–273 (2006)DOI: 10.1002/bdd

to estimate the Michaelis constant (Km), substrateinhibition constant (Ksi) and the maximumvelocity (Vmax). These calculations were carriedout using Prism Ver. 3.02 (Graph Pad Software,San Diego, CA, USA). The CLint, the intrinsicglucuronidation clearance value, was calculatedfrom Vmax/Km.

V ¼ VmaxS=ðKm þ Sþ S2=KsiÞ ð1Þ

V ¼ VmaxS=ðKm þ SÞ ð2Þ

where V and S represent the glucuronidationvelocity and the substrate concentration, respec-tively.

The IC50 value, representing the inhibitorconcentration that inhibits 50% of the controlactivity, was estimated using Prism Ver. 3.02.

Results

Indomethacin glucuronidation in HLM was acharacteristic of substrate inhibition with Km andKsi values of 210� 75.1 and 89.5� 33.5 mm,respectively, while HIM exhibited Michaelis–Menten kinetics with a Km value of17.4� 1.72 mm (Figure 1). The Vmax values were6330� 1834 and 591� 20.1 pmol/min/mg pro-tein in HLM and HIM, respectively. The CLint

values were 30.1 and 34.0 ml/min/mg protein, inHLM and HIM, respectively.

The inhibitory potency of diflunisal on indo-methacin glucuronidation was investigated inHLM and HIM. The IC50 values of diflunisal for

indomethacin glucuronidation in HLM were193� 42.5, 231� 64.6 and 100� 11.7 mm at in-domethacin concentrations of 20, 100 and 500 mm,respectively (Figure 2). In HIM, diflunisal in-hibited indomethacin glucuronidation more po-tently with IC50 values of 15.2� 2.20, 39.3� 3.04and 48.7� 5.10 mm at indomethacin concentra-tions of 20, 50 and 100 mm (Figure 3).

Discussion

This paper describes the inhibitory effects ofdiflunisal on indomethacin glucuronidation inHLM and HIM. Various UGT isozymes areinvolved in the glucuronidation of indomethacin:UGT1A1, 1A3, 1A7, 1A8, 1A9, 2B4 and 2B7, andamong these, UGT1A9 had the highest catalyticactivity, on a protein concentration basis, of therecombinant UGT isozymes [11]. UGT1A1, 1A3,1A8, 1A9 and 2B7 are expressed in the intestine[9], thus, the catalysis of indomethacin glucur-onidation is believed to occur in the intestine aswell as in the liver. HIM, however, glucuroni-dated indomethacin with a Km value of 17.4 mm,which was lower than that in HLM (210 mm),suggesting that the main UGT isozymes involvedin indomethacin glucuronidation for HIM aredifferent from those for HLM. The UGT isozymesresponsible for troglitazone glucuronidation inthe liver and intestine were also reported to bedifferent [12]: UGT1A1 is the main isozymeinvolved in troglitazone glucuronidation in theliver, while UGT1A8 and 1A10 seems to beresponsible in the intestine.

0 100 200 300 400 5000

500

1000

1500

2000

Indomethacin (µM) Indomethacin (µM)

Glu

curo

nid

atio

n v

elo

city

(pm

ol /

min

/ mg

pro

tein

)

0 20 40 60 80 1000

250

500

750

Glu

curo

nid

atio

n v

elo

city

(pm

ol /

min

/mg

pro

tein

)

(B)(A)

Figure 1. Indomethacin glucuronidation in human liver microsomes (A) and human intestine microsomes (B). Indomethacin(2–500 or 100mm) was incubated with liver and intestine microsomes (0.2 mg protein/ml) for 20 and 40 min, respectively. Eachincubation was performed in duplicate and data represent the mean

DRUG INTERACTION VIA GLUCURONIDATION 269


1 10 100 10000

20

40

60

80

100

120

Diflunisal ( µM) Diflunisal ( µM)

Diflunisal ( µM)

% o

f co

ntr

ol

% o

f co

ntr

ol

% o

f co

ntr

ol

1 10 100 10000

20

40

60

80

100

120

1 10 100 10000

20

40

60

80

100

120

(A) (B)

(C)

Figure 2. In vitro inhibition of diflunisal on indomethacin glucuronidation in human liver microsomes. Indomethacin atconcentrations of 20 (A), 100 (B) and 500 (C) mm, was incubated with microsomes (0.2 mg protein/ml) for 20 min in the absenceand presence of diflunisal (10–500mm). Each incubation was performed in duplicate and data represent the mean

1 10 100 10000

20

40

60

80

100

120

Diflunisal (µM) Diflunisal (µM)

Diflunisal (µM)

% o

f co

ntr

ol

% o

f co

ntr

ol

% o

f co

ntr

ol

1 10 100 10000

20

40

60

80

100

120

1 10 100 10000

20

40

60

80

100

120

(A) (B)

(C)

Figure 3. In vitro inhibition of diflunisal on indomethacin glucuronidation in human intestine microsomes. Indomethacin atconcentrations of 20 (A), 50 (B) and 100 (C) mm, was incubated with microsomes (0.2 mg protein/ml) for 20 min in the absence andpresence of diflunisal (2–500mm). Each incubation was performed in duplicate and data represent the mean

Y. MANO ET AL.270


Kuehl et al. reported that the Km and Vmax

values of indomethacin glucuronidation inHLM were 30.0 mm and 376.6 pmol/min/mgprotein [11]. These values are different fromthose found in this study. The reason for thediscrepancy remains to be determined, althougha possible explanation is the difference inincubation conditions, such as the microsomalprotein concentrations (1.0 vs 0.2 mg protein/ml)and the incubation buffer used (potassiumphosphate buffer vs Tris-HCl buffer). A previousreport on glucuronidation activity and theeffects of incubation conditions [13] supportthis supposition. In addition, the kineticequations used to estimate the kinetic para-meters are different between the two (Michaelis–Menten kinetics vs substrate inhibitionkinetics).

Diflunisal inhibited human UGT1A1-catalysedestradiol 3b-glucuronidation in HLM [3], 4-methylumbelliferone glucuronidation by recom-binant UGT1A9 [4] and mycophenolic acidglucuronidation in HLM [5]. However, sincemost NSAIDs had broad inhibitory potenciesagainst UGT isozymes in addition to broadsubstrate-specificities, it is likely that diflunisalinhibits other UGT isozymes. The UGT isozymesresponsible for the glucuronidation of diflunisalremain to be investigated, although UGT1A8,which is expressed in the intestine, couldglucuronidate diflunisal [14].

Indomethacin was subjected to CYP-mediatedmetabolism along with glucuronidation. Indo-methacin O-demethylation is one of the mainmetabolic pathways for CYP-mediated indo-methacin elimination [15]. The CLint value forindomethacin O-demethylation is 0.42 ml/min/mg protein in HLM [15], which is lower than thatfor indomethacin glucuronidation in HLM(30.1 ml/min/mg protein). This finding suggeststhat glucuronidation plays an important role inthe elimination of indomethacin in humans. TheCLint value for indomethacin glucuronidation onthe microsomal concentration basis in HIM (34.0ml/min/mg protein) is similar to that in HLM.When it is considered that the microsomalprotein contents per tissue weight in the intestineare lower than those in the liver, the CLint valueon a tissue weight basis (e.g. ml/min/g liver orintestine) is higher in the liver than in the

intestine. However, it should be noted that theregional differences in glucuronidation activity[16] should be considered in the extrapolation ofin vitro intrinsic glucuronidation activity from theunit of ml/min/mg protein to ml/min/g tissue. Inthe case of midazolam, intestinal metabolism issignificantly involved in the CYP3A4-mediatedmetabolism in vivo [17], although, CLint in HIM issmaller than that in HLM, even on a microsomalprotein concentration basis [17]. These findingsled us to speculate on the importance of theintestine as the site for indomethacin glucuroni-dation in humans. Other NSAIDs are alsoglucuronidated in the intestine. Ketoprofen,naproxen and etodolac, NSAIDs containing acarboxyl acid moiety, were glucuronidated inHIM as well as HLM [16], and naloxone wasalso glucuronidated in HIM [18]. Since manyNSAIDs are administered orally to humans,NSAIDs levels in the intestine are expected tobe higher, although drug concentrations at theglucuronidation site in the intestine are difficultto estimate.

The bioavailability of indomethacin isreported to be 90% or more in adults [19],however, when interpreting this value, entero-hepatic circulation should be taken intoconsideration [20]. Kwan et al. estimated thatapproximately 50% of the dose undergoesbiliary recycling [20]. The bioavailability of indo-methacin in infants has been shown to be20% with low enterohepatic circulation [21].Additionally, indomethacin glucuronide mightdeconjugate in plasma since it was not detectedthere, in spite of the abundant amount presentin the urine [1]. These factors might precludethe precise determination of the bioavailabilityof indomethacin.

Even when enterohepatic circulation is takeninto consideration, achieving a 2-fold increasein plasma indomethacin levels with concomitantdosages of diflunisal might not seem to befeasible. Processes that may contribute toincreased plasma levels of indomethacin areas follows: inhibition of urinary excretion,biliary excretion, CYP-mediated metabolismand glucuronidation. The urinary [22] andbiliary [23] excretion of indomethacin as theunchanged drug was minimal (less than 5%and 1–2% of the dose, respectively). As discussed



above, the CYP-mediated metabolism ofindomethacin is less than the glucuronidationrate, thus the impact of CYP-mediated meta-bolism inhibition on the plasma indomethacinlevels is considered to be small. This studydemonstrated the inhibitory effects on theglucuronidation rate in the intestine as well asin the liver.

The high plasma protein binding rate fordiflunisal (>98%, [7,8]) makes the unboundplasma diflunisal concentration lower than theIC50 values for the liver (100–231 mm), while a250 mg dose of diflunisal may make the intestinaldrug levels relatively high. These results maysuggest that inhibition of indomethacin glucur-onidation in the intestine, but not the liver,contributes to drug–drug interaction mainly viaglucuronidation when dosed concomitantly withdiflunisal and indomethacin. The small alterationof the plasma elimination half-life of indometha-cin, with and without diflunisal [1], may supportthis speculation. If indomethacin metabolism inthe liver is significantly inhibited, the plasmaelimination half-life tends to increase in thecase of low clearance drugs such as indo-methacin [19].

To summarize, the inhibitory potencies ofdiflunisal on indomethacin glucuronidation wereevaluated using HLM and HIM. Diflunisalinhibited indomethacin glucuronidation withthe IC50 values lower than the expected diflunisalconcentration in the intestine at clinical dosages.The drug–drug interactions between diflunisaland indomethacin may be at least partiallyattributable to the inhibitory effect of diflunisalon the intestinal glucuronidation of indo-methacin.

References

1. Van Hecken A, Verbesselt R, Tjandra-Maga TB, DeSchepper PJ. Pharmacokinetic interaction between indo-methacin and diflunisal. Eur J Clin Pharmacol 1989; 36:507–512.

2. Helleberg L. Clinical pharmacokinetics of indomethacin.Clin Pharmacokinet 1981; 6: 245–258.

3. Mano Y, Usui T, Kamimura H. In vitro inhibitory effects ofnon-steroidal antiinflammatory drugs on UDP-glucuro-nosyltransferase 1A1-catalysed estradiol 3b-glucuronida-

tion in human liver microsomes. Biopharm Drug Dispos2005; 26: 35–39.

4. Mano Y, Usui T, Kamimura H. In vitro inhibitoryeffects of non-steroidal anti-inflammatory drugs on4-methylumbelliferone glucuronidation in recombinanthuman UDP-glucuronosyltransferase 1A9}potentinhibition by niflumic acid. Biopharm Drug Dispos 2006;27: 1–6.

5. Vietri M, Pietrabissa A, Mosca F, Pacifici GM. Mycophe-nolic acid glucuronidation and its inhibition by non-steroidal anti-inflammatory drugs in human liver andkidney. Eur J Clin Pharmacol 2000; 56: 659–664.

6. Addison RS, Parker-Scott SL, Eadie MJ, Hooper WD,Dickinson RG. Steady-state dispositions of valproate anddiflunisal alone and coadministered to healthy volun-teers. Eur J Clin Pharmacol 2000; 56: 715–721.

7. Lin JH. Species differences in protein binding of difluni-sal. Drug Metab Dispos 1989; 17: 221–223.

8. Vietri M, De Santi C, Pietrabissa A, Mosca F, Pacifici GM.Inhibition of human liver phenol sulfotransferase bynonsteroidal anti-inflammatory drugs. Eur J Clin Pharma-col 2000; 56: 81–87.

9. Tukey RH, Strassburg CP. Human UDP-glucuronosyl-transferases: metabolism, expression, and disease. AnnuRev Pharmacol Toxicol 2000; 40: 581–616.

10. Mano Y, Usui T, Kamimura H. Effects of b-estradiol andpropofol on the 4-methylumbelliferone glucuronidationin recombinant human UGT isozymes 1A1, 1A8 and 1A9.Biopharm Drug Dispos 2004; 25: 339–344.

11. Kuehl GE, Lampe JW, Potter JD, Bigler J. Glucuronidationof nonsteroidal anti-inflammatory drugs: identifying theenzymes responsible in human liver microsomes. DrugMetab Dispos 2005; 33: 1027–1035.

12. Watanabe Y, Nakajima M, Yokoi T. Troglitazone glucur-onidation in human liver and intestine microsomes: highcatalytic activity of UGT1A8 and UGT1A10. Drug MetabDispos 2002; 30: 1462–1469.

13. Cheng Z, Radominska-Pandya A, Tephly TR. Studies onthe substrate specificity of human intestinal UDP-glucur-onosyltransferases 1A8 and 1A10. Drug Metab Dispos 1999;27: 1165–1170.

14. Soars MG, Ring BJ, Wrighton SA. The effect ofincubation conditions on the enzyme kinetics of UDP-glucuronosyltransferases. Drug Metab Dispos 2003; 31:762–767.

15. Nakajima M, Inoue T, Shimada N, Tokudome S, Yama-moto T, Kuroiwa Y. Cytochrome P450 2C9 catalyzesindomethacin O-demethylation in human liver micro-somes. Drug Metab Dispos 1998; 26: 261–266.

16. Sabolovic N, Heydel JM, Li X, et al. Carboxyl nonsteroidalanti-inflammatory drugs are efficiently glucuronidated bymicrosomes of the human gastrointestinal tract. BiochimBiophys Acta 2004; 1675: 120–129.

17. Thummel KE, O’Shea D, Paine MF, et al. Oral first-passelimination of midazolam involves both gastrointestinaland hepatic CYP3A-mediated metabolism. Clin PharmacolTher 1996; 59: 491–502.

18. Soars MG, Burchell B, Riley RJ. In vitro analysis ofhuman drug glucuronidation and prediction of in vivo

Y. MANO ET AL.272


metabolic clearance. J Pharmacol Exp Ther 2002; 301:382–390.

19. Alvan G, Orme M, Bertilsson L, Ekstrand R, Palmer L.Pharmacokinetics of indomethacin. Clin Pharmacol Ther1975; 18: 364–373.

20. Kwan KC, Breault GO, Umbenhauer ER, McMahon FG,Duggan DE. Kinetics of indomethacin absorption,elimination, and enterohepatic circulation in man.J Pharmacokinet Biopharm 1976; 4: 255–280.

21. Bhat R, Vidyasagar D, Fisher E, et al. Pharmacokinetics oforal and intravenous indomethacin in preterm infants.Dev Pharmacol Ther 1980; 1: 101–110.

22. Hucker HB, Zacchei AG, Cox SV, Brodie DA, CantwellNHR. Studies on the absorption, distribution and excre-tion of indomethacin in various species. J Pharmacol ExpTher 1966; 153: 237–249.

23. Terhaag B, Hermann U. Biliary elimination of indometha-cin in man. Eur J Clin Pharmacol 1986; 29: 691–695.



Documents

In vitro drug interaction between diflunisal and indomethacin via glucuronidation in humans