Stereoselectivity in pharmacokinetics of rivoglitazone, a novel peroxisome proliferator-activated...

RESEARCH ARTICLE

Stereoselectivity in Pharmacokinetics of Rivoglitazone, A NovelPeroxisome Proliferator-Activated Receptor γ Agonist, in Ratsand Monkeys: Model-Based Pharmacokinetic Analysis andIn Vitro–In Vivo Extrapolation Approach

TAKASHI IZUMI,1 FUJIKO TSURUTA,2 TOMOKO ISHIZUKA,1 KOUICHI NAKAMURA,1 MASAKATSU KOTHUMA,2

MAKOTO TAKAHASHI1

1Drug Metabolism and Pharmacokinetics Research Laboratories, R&D Division, Daiichi Sankyo Company, Ltd., Tokyo, Japan

2Translational Medicine and Clinical Pharmacology Department, R&D Division, Daiichi Sankyo Company, Ltd., Tokyo, Japan

Received 13 March 2013; revised 12 April 2013; accepted 16 April 2013

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23586

ABSTRACT: Stereoselectivity in pharmacokinetics of rivoglitazone, a novel peroxisomeproliferator-activated receptor ( agonist, in rats and monkeys was examined. The pharma-cokinetic model involving chiral inversion explained well the plasma profiles of R-isomer andS-isomer after intravenous and oral administration of (R)-rivoglitazone or (S)-rivoglitazone torats and monkeys. The high stereoselectivity was evaluated in chiral inversion clearance (R/Sratio: 7.92), metabolic clearance (5.78), and volume of distribution (4.04) in rats; however, thesewere low (1.73, 1.31, and 1.06) in monkeys. The stereoselectivity in chiral inversion was alsoobserved in in vitro incubation studies in plasma, and the R/S ratio of chiral inversion showedhigh correlation with the R/S ratio of plasma unbound fraction. The metabolic clearance ofthe primary five metabolic pathways of rivoglitazone was evaluated from an in vitro–in vivoextrapolation approach using rat and monkey liver microsomes. The high stereoselectivity inmetabolic clearance in rat was evaluated (R/S ratio: 5.78), which was assumed to be becauseof the stereoselectivity in plasma unbound fraction, on the contrary, that in monkeys exhib-ited low stereoselectivity (0.774). Thus, the stereoselectivity in plasma unbound fraction wasestimated to be a major determinant of stereoselectivity in pharmacokinetics of rivoglitazonein rats and monkeys. ©2013 Wiley Periodicals, Inc. and the American Pharmacists AssociationJ Pharm SciKeywords: pharmacokinetics; protein binding; metabolic clearance; metabolism; distribu-tion; chiral inversion; stereoselectivity; in vitro–in vivo extrapolation (IVIVE); unbound fraction;intrinsic clearance

INTRODUCTION

Rivoglitazone, (RS)-5-{4-[(6-methoxy-1-methyl-1H-benzimidazol-2-yl)methoxy]benzyl}-1,3-thiazolidine-2,4-dione monohydrochloride (Fig. 1), is a novelthiazolidinedione (TZD) that selectively activates the

Abbreviations used: TZD, thiazolidinedione; IVIVE, invitro–in vivo extrapolation; IS, internal standard; LC–MS/MS, liq-uid chromatography–tandem mass spectrometry.

Additional Supporting Information may be found in the onlineversion of this article. Supporting Information

Correspondence to: Takashi Izumi (Telephone: +81-3-3492-3131; Fax: +81-3-5436-8567; E-mail: [email protected])Journal of Pharmaceutical Sciences© 2013 Wiley Periodicals, Inc. and the American Pharmacists Association

nuclear receptor peroxisome proliferator-activatedreceptor ( (PPAR().1 TZD-containing drugs, pioglita-zone and rosiglitazone, are a clinically important newclass of oral antidiabetic agents for the treatmentof type 2 diabetes mellitus, which increase insulinsensitivity in target tissues through interaction withPPAR(.2 Pharmacokinetics, metabolism, and dispo-sition studies of rivoglitazone after intravenous andoral administration of [C]rivoglitazone to rats andmonkeys have been reported.3,4 Rivoglitazone ex-hibits low clearance and small volume of distribution.The clearance is mainly the result of metabolism;and on the basis of the structures of the metabolites,five primary metabolic pathways consisting of fouroxidation pathways and one N-glucuronidation

JOURNAL OF PHARMACEUTICAL SCIENCES 1

2 IZUMI ET AL.

Figure 1. Proposed primary metabolic pathways of rivoglitazone in rats and monkeys.

pathway for rivoglitazone have been proposed:(1) O-demethylation to form O-demethyl rivoglita-zone (M12); (2) TZD ring opening to form TZD ring-opened mercapto amide (M22); (3) N-demethylationto form N-demethyl rivoglitazone (M17); (4) TZDring hydroxylation to form TZD ring 5-hydroxyrivoglitazone (M18); and (5) N-glucuronidation toform N-glucuronide (M13; Fig. 1).

Rivoglitazone is an equal mixture of two stereoiso-mers, arising from asymmetric carbons at the fifthposition of the TZD ring (Fig. 1). (S)-rivoglitazonehas been shown to bind PPAR( with greater than50-fold higher affinity than (R)-rivoglitazone.5 Thishigh stereoselectivity in the pharmacological effecthas also been reported for rosiglitazone.6 On the con-trary, the TZD-containing drugs have been observedas the chiral inversion at fifth position of TZD ringin vivo.6–9

In this study, the stereoselectivity in pharmacoki-netics of rivoglitazone was investigated from in vivoand in vitro studies in rats and monkeys, which wereused in safety and pharmacology assessment. For invivo studies, the pharmacokinetic model involvingchiral inversion was applied to the plasma profilesafter intravenous and oral dosing of each stereoiso-mer. For in vitro studies, protein binding studies, chi-ral inversion studies in plasma, and drug metabolismstudies for the five metabolic pathways have been ex-amined using liver microsomes for each stereoisomer.On the basis of the metabolism data, the stereoselec-

tivity in drug metabolism was evaluated using thein vitro–in vivo extrapolation (IVIVE) method andwas compared with the in vivo data. Finally, we eval-uated the predominant factors for the stereoseleciv-ity in pharmacokinetics of rivoglitazone in rats andmonkeys.

MATERIALS AND METHODS

Materials

(R)-rivoglitazone, (S)-rivoglitazone, its internal stan-dard (IS; R-121171), its metabolites (racemate): M12(R-239457), M22 (R-130712), M17 (R-125804), M18(R-14526), M13 (R-417680), and M9 (A200-8132),and IS for metabolites (R-252121, R-252122, andA203-5426) were synthesized by Daiichi SankyoCompany, Ltd. or Daiichi Sankyo RD AssocieCompany, Ltd. (Tokyo, Japan).3 Hydroxypropyl-$-cyclodextrin (HP-$-CD) was purchased from NihonShokuhin Kako Co., Ltd. (Tokyo, Japan). $-Nicotinamide adenine dinucleotide phosphate($-NADP) sodium salt, D-Glucose-6-phosphate (G-6-P) disodium salt hydrate, Glucose-6-phosphatedehydrogenase (G-6-PDH), and L-glutathione re-duced (GSH), uridine 5-diphosphoglucuronic acidtrisodium salt (UDPGA), and D-saccharic acid1,4-lactone monohydrate (D-saccharolactone) werepurchased from Sigma–Aldrich (St. Louis, Missouri).UDP-glucuronosyltransferase (UGT) Reaction Mix

JOURNAL OF PHARMACEUTICAL SCIENCES DOI 10.1002/jps

STEREOSELECTIVITY IN PHARMACOKINETICS OF RIVOGLITAZONE 3

Solution A (UDPGA Cofactor: 25 mM UDPGA inwater) and UGT Reaction Mix Solution B (5×-UGTBuffer Mix with Alamethicin: 250 mM Tris–HCl,40 mM MgCl2, and 0.125 mg/mL alamethicin in wa-ter) were purchased from BD Biosciences (San Jose,California). All other reagents and solvents usedwere commercially available and either of analyticalor high-performance liquid chromatography (HPLC)grade. Pooled Fischer 344 rat liver microsomes andpooled cynomolgus monkey liver were purchasedfrom XenoTech, LLC (Lenexa, Kansas). Pooled blankheparinized plasma from F344/DuCrlCrlj male ratsand male cynomolgus monkeys were purchased fromHamri Company, Ltd. (Ibaraki, Japan) and that fromhuman was purchased from Biopredic International(Rennes, France).

Animals

All animal studies were conducted with approvalin accordance with the guidelines of the Institu-tional Animal Care and Use Committee of DaiichiSankyo Company, Ltd. Male F344/DuCrlCrlj rats(148–184 g) were purchased from Charles RiverJapan, Inc. (Yokohama, Japan) and male cynomolgusmonkeys (3.79–4.36 kg) were purchased from the So-cialist Public of Vietnam. Animals were housed ina temperature-controlled environment with a 12-hlight/dark cycle and with free access to tap water anda laboratory diet. Before administration, the animalswere fasted overnight and fed at 8-h postdose for mon-keys. For administration, (R)-rivoglitazone and (S)-rivoglitazone were dissolved in the 20% (w/v) HP-$-CD 10 mM glycine–HCl buffer containing 0.32% NaCl(pH 2.0).

Methods

In Vivo Pharmacokinetic Studies in Rats

(R)-Rivoglitazone or (S)-rivoglitazone (0.5 mg/kg)were intravenously administrated via the tail veinand by oral gavage (n = 4 for each group). Approxi-mately 200 :L of blood was collected from the femoralvein at the following times under diethyl ether anes-thesia: predose and 0.083 (except for oral), 0.25, 0.5,1, 2, 4, 6, 8, and 24 h after dose. Each 100 :L plasmaobtained by centrifugation at 4◦C was mixed with100 :L of ice-cold 1.0 M citric acid solution storedat −80◦C until analysis.

In Vivo Pharmacokinetic Studies in Monkeys

The pharmacokinetic studies (intravenous dosinggroup and oral dosing group) were conducted with2-week washout periods. The monkeys (n = 4 foreach group) were dosed intravenously via a cephalicvein and orally using a catheter at a dose of (R)-rivoglitazone and (S)-rivoglitazone (0.5 mg/kg). Onemilliliter of blood was collected from the saphenous

vein at predose, 0.083 (except for oral), 0.25, 0.5, 1, 2,4, 6, 8, 24, and 48 h after dose. Each 300 :L plasmaobtained by centrifugation at 4◦C was mixed with300 :L of ice-cold 1.0 M citric acid solution and storedat −80◦C until analysis.

Protein Binding Study

The protein binding of rivoglitazone was measuredby the modified microscale ultracentrifugation tech-nique, which have developed and validated by Nakaiet al.10 Although we used this method without IS,we could confirm the accuracy and reproducibilityby repeated preexperiments (data not shown). Sevenmicroliter of (R)-rivoglitazone or (S)-rivoglitazonemethanol solution was added to 693 :L plasma andincubated on ice for 10 min. Five hundred microliter ofthe plasma sample was ultracentrifugated at 65,000gat 4◦C for 150 min. After cutting the tube containingchylomicron and very low-density lipoproteins witha tube slicer (Beckman Coulter, Inc., California),10

100 :L aliquot of the upper part of the middle layerwas mixed with 100 :L of 1 M citric acid solution forthe unbound concentration (Cup). The residual plasmasample (50 :L) was mixed with 50 :L of 1 M citric acidsolution for the total concentration (Ctp). The plasmaunbound fraction (fu) was calculated by Cup/Ctp.

Chiral Inversion Study

Blood was collected from rats and monkeys underfasting conditions (over 15 h) and that from humanswas collected in the morning on an empty stomach.For blood collection, the healthy Japanese male sub-jects agreed with informed consent and participatedvoluntarily in this study, which was approved bythe Institute of Human Ethical Committee of Dai-ichi Sankyo Company, Ltd. The heparinized bloodwas immediately centrifuged to obtain the plasma.After preincubation of 597 :L of plasma or phosphatebuffer saline (PBS, pH7.4) at 37◦C for 10 min, 3 :L of(R)-rivoglitazone or (S)-rivoglitazone in 0.1% trifluo-roacetic acid (TFA) methanol solution were mixed (fi-nal concentration: 200 ng/mL) and incubated at 37◦C.An aliquot (50 :L) was mixed with ice-cold 50 :L ofcitric acid (1.0 M) at 0.25, 0.5, 1, 2, 4, and 6 h duringincubation and stored at −80◦C until analysis.

Oxidation by Rat and Monkey Liver Microsomes

The oxidative metabolism was initiated by adding2 :L of (R)-rivoglitazone or (S)-rivoglitazonemethanol to 198 :L of reaction mixture: 1 mg pro-tein/mL microsomes, 100 mM potassium phosphatebuffer (pH 7.4), 2.5 mM $-NADP, 25 mM G-6-P, 0.5units/mL G-6-PDH, and 10 mM magnesium chloride,that had been preincubated at 37◦C for 5 min. Thereaction was terminated by mixing 50 :L of reactionmixture with ice-cold stop solution of 100 :L of ISsolution (R-252121 for M12 and R-252122 for M22,

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES

4 IZUMI ET AL.

M17, and M18 in 1% formic acid acetonitrile solu-tion) and 50 :L formic acid/acetonitrile/6 mM GSH(1:50:49, v/v/v). GSH was added to keep the stabil-ity of formed M22. The reaction mixture pretreatedby filtration using a Captiva 96-well filter plate (poresize 0.45 :m; Varian Inc., Palo Alto, California). Fivemicroliter of the cleaned-up sample was injected intothe liquid chromatography–tandem mass spectrom-etry (LC–MS/MS) system. An API5000 mass spec-trometer (AB SCIEX, Framingham, Massachusetts)connected to a Nanospace SI-2 (Shiseido Company,Ltd., Tokyo, Japan) was used as the LC–MS/MS sys-tem. Analyte detection was achieved with electro-spray ionization (ESI) in the positive ion mode usingmultiple reaction monitoring (MRM) of transitions ofm/z 384 → 162 for M12 and M17, m/z 414 → 176for M18, m/z 372 → 176 for M22, and m/z 389 →167 for IS (R-252121 and R-252122). The sample wasseparated on an Atlantis T3 column (2.1 × 50 mm2,3 :m; Waters Corporation, Milford, Massachusetts)and maintained at 40◦C with gradient elution usingmobile phase A (methanol/water/ammonium acetate;950:50:0.385, v/v/w) and mobile phase B (methanol/water/ammonium acetate/formic acid; 50:950:0.385:1,v/v/w/v). The elution was achieved using a linear gra-dient: 0 min at 40% of mobile phase A, hold for 1.0min, to 65% A at 1.1 min, hold for 1.4 min, to 90% Aat 2.6 min, hold for 1.4 min, and back to 40% A at 4.1min, hold for 2.9 min. The flow rate was 0.3 mL/min.

Grucuronidation by Rat and Monkey Microsomes

Because N-glucuronidation (M13) was subsequentlyhydrolyzed into the TZD ring opening form (M9),3,4

the velocity of N-glucuronidation was measured fromthe sum of the formation of M13 and M9. The glu-curonidation was initiated by adding 2 :L of (R)-rivoglitazone or (S)-rivoglitazone dimethyl sulfoxide(DMSO) solution to 198 :L of reaction mixture consti-tuting 1 mg protein/mL microsomes, 50 mM Tris–HClbuffer (pH 7.5), 25 :g alamethicin/mg protein,11

10 mM UDPGA, 5 mM D-saccharolactone,11 and 8mM magnesium chloride that was preincubated at37◦C for 5 min. After incubation at 37◦C for 3 min,the reaction was terminated by mixing 50 :L of ice-cold stop solution (1% formic acid acetonitrile solu-tion). Fifty microliter of the reaction mixture wasmixed with 50 :L of IS (A203-5462 in purified wa-ter) and deproteinized using sequential filtration us-ing a Captiva 96-well filter plate (Varian Inc.) andsubsequently treated with Oasis HLB :Elution plate(30 :m, Waters Corporation). Ten microliter of theextraction eluted with 50 :L of 3% formic acid in50% acetonitrile was injected into the LC–MS/MSsystem. An API3200 mass spectrometer (AB SCIEX)connected to a Prominence LC-20A (Shimadzu Cor-poration, Kyoto, Japan) was used as the LC–MS/MSsystem. Analyte detection was achieved with ESI in

the positive ion mode using MRM of transitions of m/z574→ 161 for M13, m/z 592 → 161 for M9, and m/z533 → 134 for IS (A203-5462). The sample was sepa-rated on an Atlantis T3 column (2.1 × 50 mm2, 3 :m;Waters Corporation) and maintained at 40◦C with amixture of mobile phase A (acetonitrile/water/formicacid; 950/50/1, v/v/v) and mobile phase B (acetonitrile/water/formic acid; 50/950/1, v/v/v) as 30:70 (v/v). Theflow rate was 0.2 mL/min.

Determination of Unbound Fractions in MicrosomalReaction Mixtures

The unbound fraction of R-isomer and S-isomer ofrivoglitazone in the microsomal reaction mixture wasdetermined by the ultracentrifugation method. Twomicroliter of (R)-rivoglitazone or (S)-rivoglitazonemethanol solution was added to the cofactor-free(without NADPH or UDPGA) microsomal suspensionconsisted of the same components for the oxidative re-action or glucuronidation (198 :L). Twenty microliterof the reaction mixture was taken as the total concen-tration sample (Ct,mic). After the remainder was ul-tracentrifuged at 105,000g and 4◦C for 60 min, 20 :Lof the supernatant (Cu,mic) and the reaction mixture(Ct,mic) was mixed with 100 :L of ice-cold 1% formicacid acetonitrile solution and 100 :L of IS solution(R-121171 in methanol).

Analysis of Stereoisomers of Rivoglitazone

R-isomer and S-isomer of rivoglitazone were de-termined by LC–MS/MS. The frozen samples werethawed in an ice bath. One hundred microliter of theplasma sample treated with 1.0 M citric acid solutionfor pharmacokinetic studies, protein binding studies(Ctp), and in vitro chiral inversion studies were mixedwith 800 :L of 0.5 M citric acid solution. Two hundredmicroliter of the supernatant treated with 1M citricacid solution for protein binding (Cup) was mixed with200 :L of 1 M citric acid solution and 500 :L of 0.5 Mcitric acid solution. These samples were further mixedwith 50 :L of 0.1% TFA methanol solution and 50 :Lof IS solution (R-121171 in 0.1% TFA methanol solu-tion) and pretreated using a Versa plate C2 extrac-tion plate (sorbent mass: 100 mg, Varian Inc.), andthe analyte was and eluted with 0.3 mL of 0.1% TFAmethanol solution. For measurement of unbound frac-tions in microsomal reaction mixtures, 20 :L of thesupernatant and reaction mixture was mixed with100 :L of 1% formic acid acetonitrile solution and100 :L of IS (R-121171) methanol solution and de-proteined by filtration using a Captiva 96-well filterplate (Varian Inc.).

For analysis by the LC–MS/MS system, an aliquotof 5 :L was injected onto the LC–MS/MS system,An API3200 (AB SCIEX) connected to a ProminenceLC-20A system (Shimadzu Corporation) for the sam-ples for rat pharmacokinetic studies and Quattro II



(Waters Corporation) connected to an Alliance 2795Separations Module (Waters Corporation) for thesamples of monkey pharmacokinetic studies, proteinbinding, chiral inversion, and unbound fractions inmicrosomal reaction mixtures. Analyte detection wasachieved with ESI in the positive ion mode MRM oftransition of m/z 398 → 176 for R-isomer and S-isomerof rivoglitazone, m/z 404 → 182 for IS (R-121171). Thesample was separated on a Chiralcel OJ–RH column(2.0 × 150 mm2, 5 :m; Daicel Chemical IndustriesLtd. Tokyo, Japan) and maintained at 35◦C with a mo-bile phase (acetonitrile/water/formic acid/ammoniumbicarbonate; 300/700/1/0.77, v/v/v/w). The flow ratewas 0.2 mL/min.

Pharmacokinetic Model Involving Chiral Inversion

Noncompartmental pharmacokinetic parameters, thearea under the concentration–time curve extrapo-lated to infinity (AUC0–inf), the maximum plasma con-centration (Cmax), the time to reach Cmax (Tmax), andterminal half-life (t1/2), and pharmacokinetic modelanalysis involving chiral inversion were calculatedusing WinNonlin Professional software (version 4.0.1;Pharsight Corporation, Mountain View, California).

After intravenous administration of (R)-rivoglitazone and (S)-rivoglitazone, R-isomer andS-isomer concentrations were analyzed using apharmacokinetic model involving reversible chiralinversion in accordance with the following assump-tions: (1) Because the plasma profiles in both animalsrepresent biphasic disappearance, two compartmentmodels involving central and peripheral compart-ments were set as the model. (2) Because chiralinversion between each stereoisomer was observedfrom in vitro incubation experiments in rat andmonkey plasma, the chiral inversion was assumed tooccur in the central compartment with a first-orderprocess. The first-order chiral inversion rate con-stants from R-isomer to S-isomer and from S-isomerto R-isomer are designed kRS and kSR, respectively.(3) The elimination of each stereoisomer is assumedto occur from the central compartment, and thefirst-order elimination rate constant for R-isomerand S-isomer is designed as keR and keS, respectively.(4) No interaction between both stereoisomers ineach process was assumed.

The mass-balance equations for the model inFigure 2a on the above assumptions are describedas follows:

dCR

dt= − (kRS + keR + k2) CR + kSRCS

(VS

VR

)+ k1

XR

VR

(1)

dCS

dt= − (kSR + keS + k3) CS + kRSCR

(VR

VS

)+ k4

XS

VS

(2)

Figure 2. Pharmacokinetic model involving chiral in-version of rivoglitazone after (a) intravenous administra-tion and (b) oral administration of (R)-rivoglitazone or(S)-rivoglitazone to rats and monkeys. CR and CS, plasmaconcentration of R-isomer and S-isomer in central compart-ment, respectively; VR and VS, apparent volume of distri-bution of R-isomer and S-isomer in the central compart-ment, respectively; XR and XS, amount of R-isomer andS-isomer in peripheral compartment, respectively; kRS andkSR, first-order chiral inversion rate constant from R-isomerto S-isomer and from S-isomer to R-isomer, respectively; keRand keS, first-order elimination rate constant from the cen-tral compartment for R-isomer and S-isomer, respectively;k1 and k2, transfer rate constant between central and pe-ripheral for R-isomer; k3 and k4, transfer rate constant be-tween central and peripheral for S-isomer; ka, first-orderabsorption rate; F, oral bioavailability; fR and fS, fractionof R-isomer entered into the system and that of S-isomer,respectively.

dXR

dt= −k1XR + k2CRVR (3)

dXS

dt= −k4XS + k3CSVS (4)

where CR and CS are the concentrations of R-isomerand S-isomer in the central compartment, respec-tively, VR and VS are the apparent volume of distri-bution of R-isomer and S-isomer in the central com-partment, respectively, XR and XS are the amount ofR-isomer and S-isomer in peripheral compartment,


6 IZUMI ET AL.

respectively, k1 and k2 are transfer rate constant be-tween central and peripheral for R-isomer, and k3and k4 are those for S-isomer, respectively. The meanplasma concentrations of R-isomer and S-isomer afterintravenous administration of (R)-rivoglitazone and(S)-rivoglitazone were simultaneously fitted to Eqs1–4.

The chiral inversion clearance for R-isomer and S-isomer (CLRS and CLSR) are defined by

CLRS = VRkRS (5)

CLSR = VSkSR (6)

Because rivoglitazone is mainly excreted asmetabolites, the total body clearance is assumed tobe metabolic clearance.3 The metabolic clearance forR-isomer and S-isomer (CLeR and CLeS) are definedby

CLeR = VRkeR (7)

CLeS = VSkeS (8)

After oral administration of (R)-rivoglitazone and(S)-rivoglitazone, the plasma concentrations of R-isomer and S-isomer were analyzed using a pharma-cokinetic model involving chiral inversion in accor-dance with the assumptions: (1) After oral dosing,R-isomer and S-isomer was absorbed into the cen-tral compartment via a first-order process. (2) Oralbioavailability (F) and the first-order absorption rate(ka) for R-isomer and S-isomer were assumed to bethe same. (3) After R-isomer and S-isomer reachedthe systemic circulation, the two compartment modelsinvolving reversible chiral inversion as described forintravenous administration was applied. The mass-balance equations for the model in Figure 2b on theabove-mentioned assumptions are described as fol-lows:

dCR

dt= ka

DoseFf R

VR

− (kRS + keR + k2) CR + kSRCS

(VS

VR

)+ k1

XR

VR

(9)

dCS

dt= ka

DoseF (1 − f R)VS

− (kSR + keS + k4) CS + kRSCR

(VR

VS

)+ k3

XS

VS

(10)

dCR

dt= ka

DoseF (1 − f S) f R

VR

− (kRS + keR + k2) CR + kSRCS

(VS

VR

)+ k1

XR

VR

(11)

dCS

dt= ka

DoseFf S

VS

− (kSR + keS + k4) CS + kRSCR

(VR

VS

)+ k3

XS

VS

(12)

dXR

dt= −k1XR + k2CRVR (13)

dXS

dt= −k4XS + k3CSVS (14)

where ka represents the first-order absorption rate, Frepresents the oral bioavailability, and fR and fS rep-resent the fraction of R-isomer entered into the sys-tem and that of S-isomer, respectively.12 The plasmaconcentrations of R-isomer and S-isomer after oraladministration of (R)-rivoglitazone were simultane-ously fitted to Eqs. 9, 10, 13, and 14 and those afteradministration of (S)-rivoglitazone were simultane-ously fitted to Eqs. 11–14. The values for kRS, kSR,keR, keS, k1, k2, k3, k4, VR, and VS were used as thedata obtained for intravenous administration.

Chiral Inversion Study

The kinetic model for chiral inversion of R-isomerand S-isomer in a closed incubation system was ap-plied. The corresponding differential equations forthis model are

dCR

dt= −kRS, in vitro × CR + kSR, in vitro × CS (15)

dCS

dt= −kSR, in vitro × CS + kRS, in vitro × CR (16)

where kRS,in vitro and kSR,in vitro represent the rate con-stant of chiral inversion from R-isomer to S-isomerand from S-isomer to R-isomer, respectively. CR andCS represent the relative proportion of R-isomer andS-isomer, which were calculated from the concentra-tion of each isomer divided by the sum of both isomerconcentrations at the sampling time, respectively. Therelative proportion data up to 6 h for both stereoiso-mers were simultaneously fitted to Eqs. 15 and 16 us-ing WinNonlin Professional software. The half-lives(t1/2) of R-isomer and S-isomer for chiral inversionwere calculated at 0.693 divided by kRS,in vitro andkSR,in vitro, respectively.



Enzyme Kinetic Analysis and IVIVE

The unbound fraction of each substrate in the mi-crosomal reaction mixture was calculated as Cu,mic/Ct,mic. The unbound substrate concentration (Cu) wascalculated by multiplying the nominal concentrationand unbound fraction. The initial velocity (v) was cal-culated by dividing each metabolite concentration bythe reaction time (5 min for oxidation and 3 min forglucuronidation) and the protein concentration. Us-ing the values of Cu, v, and v/Cu, the following threeplots were generated: (1) Direct plot (x-axis: Cu, y-axis: v), (2) Eadie–Hofstee plot (x-axis: v/Cu, y-axis: v),and (3) Clearance plot (x-axis: Cu, y-axis: v/Cu). Theseplots were used as diagnostic tools for determiningdeviations from simple Michelis–Menten kinetics andmethods to estimate the intrinsic metabolic clearance(CLuint).13 When the Eadie–Hofstee plot exhibits amarked curvature or “boomerang” shape, it indicatesthe cooperatively and suggests the presence of multi-ple binding sites. In this case, the sigmoidal kineticmodel (Eq. 17), which incorporates the Hill coefficient(n), substrate concentration resulting in 50% of Vmax(S50, analogous to the Michaels constant, Km) and themaximum velocity (Vmax), was applied.

v = VmaxCnu

Sn50 + Cn

u(17)

The maximum v/Cu was adopted as a maximummetabolic clearance (CLmax), alternative parameterto CLuint by Eq. 18.

CLmax = Vmax (n − 1)Sn

50n (n − 1)1/n (18)

When the Eadie–Hofstee plot exhibits two compo-nents, it indicates a combination of two or more con-tributing enzymes. In this case, a two-site Michaelis—Menten model was applied (Eq. 19)

v = Vmax,1Cu

Km,1 + Cu+ CL2Cu (19)

where Vmax,1, Km,1, and CL2 are the apparent max-imum velocity, the apparent Michaelis–Menten con-stant for the high-affinity site, and the clearance forthe low affinity site, respectively. For this case, theCLuint was calculated from the sum of Vmax,1/Km,1,and CL2. For all fitting analysis, WinNonlin softwarewas used. For other reactions, CLuint was calculatedas a mean value of the plateaued v/Cu values at lowsubstrate concentrations that were selected by theclearance plot, being observed as a nonsaturable andalmost linear increase of the v values with increasingCu values.

CLuint (mL/min/mg) was extrapolated to CLuint(mL/min kg) by physiological-based scaling factors(PBSF) of 1720.2 (mg protein/kg of body weight) forrats and 793.6 (mg protein/kg of body weight) formonkeys.14 Because blood clearance of rivoglitazonein rats and monkeys was very low compared with hep-atic blood flow (0.94%–0.95% for rats and 0.96%–1.1%for monkeys),3 the metabolic clearance (CLm) in bothanimals is assumed to be unbound plasma fraction(fu) × intrinsic metabolic clearance (CLuint).15

This assumption was applied to five metabolicpathways for each isomer

CLmR = f uRCLuint,R (20)

CLmS = f uSCLuint,S (21)

where CLmR, CLmS, and CLuint,R, CLuint,S aremetabolic clearance and intrinsic clearance for R-isomer and S-isomer for individual metabolic path-way, respectively, and fuR and fuS are unbound plasmafraction for R-isomer and S-isomer, respectively.

Total CLmR and total CLmS were calculated fromthe sum of CLmR and CLmS for the five metabolic path-ways, respectively. From the mean of CLmR and CLmSfor each metabolic pathway, the fraction metabolizedfor individual metabolic pathways (fm) was also cal-culated.

f m = Mean CLm of each pathwayTotal of mean CLm

(22)

RESULTS

Pharmacokinetic Profiles

After intravenous administration of (R)-rivoglitazoneor (S)-rivoglitazone to rats and monkeys, inversed S-isomer or R-isomer were observed in plasma and dis-posed parallel to administered R-isomer or S-isomer(Figs. 3a and 3b for rats and Figs. 3e and 3f for mon-keys). Noncompartment pharmacokinetic parametersof both isomers in rats and monkeys are shown in Ta-ble S1. In rats, the AUC0–inf of inversed S-isomer was4.46-fold higher than that of administered R-isomer,whereas that of inversed R-isomer was only 8.75% ofthat of administered S-isomer. The t1/2 of both isomersadministered and inversed was almost similar withthe range of 5.16–5.67 h. In monkeys, contrary to rats,the AUC0–inf of inversed S-isomer was 0.504-fold ofthat of administered R-isomer, and that of inversed R-isomer was 1.2-fold of that of administered S-isomer.The t1/2 of both isomers administered and inversedwas almost similar with the range of 7.18–7.56 h.


8 IZUMI ET AL.

Figure 3. Plasma profiles of R-isomer (closed circle) and S-isomer (open circle) of rivoglitazone after intravenous ororal administration of (R)-rivoglitazone or (S)-rivoglitazone to rats and monkeys. (a) Intravenous administration of (R)-rivoglitazone to rats; (b) intravenous administration of (S)-rivoglitazone to rats; (c) oral administration of (R)-rivoglitazoneto rats; (d) oral administration of (S)-rivoglitazone to rats; (e) intravenous administration of (R)-rivoglitazone to monkeys;(f) intravenous administration of (S)-rivoglitazone to monkeys; (g) oral administration of (R)-rivoglitazone to monkeys; and(h) oral administration of (S)-rivoglitazone to monkeys. Symbols and bars represent the mean ± SD of observed data (N = 4)for both animals. The lines represent the fitting curves to the pharmacokinetic model involving chiral inversion.



Table 1. Pharmacokinetic Parameters Based on a Model Involving Chiral Inversion After Intravenous and OralAdministration of (R)-Rivoglitazone or (S)-Rivoglitazone to Rats and Monkeys

Rats Monkeys

Parameters Estimate CV Estimate CV

kRS (h−1) 0.384 7.9 0.285 6.5kSR (h−1) 0.196 8.0 0.520 4.9keR (h−1) 0.238 11.0 0.150 8.8keS (h−1) 0.166 7.2 0.121 13.8VR (mL/kg) 326 7.9 169 5.5VS (mL/kg) 80.7 3.5 160 3.9k1 (h−1) 2.42 50.3 2.73 30.7k2 (h−1) 0.968 62.1 1.15 40.8k3 (h−1) 0.925 21.0 0.540 36.1k4 (h−1) 1.55 18.0 1.47 34.7CLRS (mL/min kg)a 2.09 0.803CLSR (mL/min kg)a 0.264 1.39CLeR (mL/min kg)b 1.29 0.423CLeS (mL/min kg)b 0.223 0.323

Oral dosing of (R)-rivogitazoneF 0.784 2.4 0.757 10.1ka (h−1) 1.67 7.6 0.192 16.1fR 0.963 1.0 1.02 3.0

Oral dosing of (S)-rivogitazoneF 0.836 1.5 0.781 13.0ka (h−1) 2.67 4.5 0.190 20.5fS 0.996 0.8 0.979 7.7

aCLRS = kRSVR, CLSR = kSRVS.bCLeR = keRVR, CLeS = keSVS.

The plasma profiles of R-isomer and S-isomer af-ter intravenous administration of (R)-rivoglitazoneor (S)-rivoglitazone to rats and monkeys were wellsimultaneously fitted to a two-compartment model in-volving reversible chiral inversion (Figs. 3a and 3bfor rats and Figs. 3e and 3f for monkeys). For thepharmacokinetic parameters related to chiral inver-sion, kRS was 1.96-fold higher than kSR and CLRSwas 7.92-fold larger than CLSR in rats (Table 1). Onthe contrary, kSR was 1.82-fold higher than kRS andCLSR was 1.73-fold larger than CLRS in monkeys. Formetabolic clearance, CLeR was 5.78-fold larger thanCLeS in rats, and 1.31-fold larger in monkeys. In bothanimals, CLSR and CLRS were larger than CLeR andCLeS, respectively, which suggested that a chiral in-version clearance was predominantly compared withthe metabolic clearance. The VR was 4.04-fold largerthan VS in rats, whereas the VR and VS was almostsimilar (VR/VS = 1.06) in monkeys. The plasma pro-files of R-isomer and S-isomer after oral adminis-tration of (R)-rivoglitazone and (S)-rivoglitazone torats and monkeys were well simultaneously fitted toa two-compartment model involving reversible chiralinversion (Figs. 3c and 3d for rats and Figs. 3g and3h for monkeys). F values for R-isomer and S-isomerwere 0.784 and 0.836 for rats, and 0.757 and 0.781 formonkeys, respectively. These values were lower thanoral bioavailability of rivoglitazone (95% for rats and90.4% for monkeys at dose of 0.493 mg/kg).3 The val-

ues of fR and fS were 0.936 and 0.996 for rats and 1.02and 0.979, which suggested that no chiral inversionoccurred before absorption in both animals.

Protein Binding

The plasma protein binding of both isomers of rivogli-tazone (1000−10,000 ng/mL) in rat, monkey, and hu-man was determined by ultracentrifugation at 4◦C toavoid chiral inversion during assay procedure. Themean fu for R-isomer and S-isomer in rat plasmawas 0.00894 ± 0.00134 (mean ± SD, n = 3) and0.00119 ± 0.00046, that in monkey plasma was0.00430 ± 0.00006 and 0.00530 ± 0.00024, and thatin human plasma was 0.00210 ± 0.00011 and 0.00572± 0.00008, respectively. Thus, there were species dif-ferences in the stereoselectivity of fu (fuR/fuS): 7.51 forrats, 0.811 for monkeys, and 0.367 for humans.

In Vitro Chiral Inversion

The time profiles of both the isomers in PBS, rat,monkey, and human plasma during the incubationof R-isomer or S-isomer were at 37◦C. The simul-taneous fitting of the profiles of both isomers toEqs. 15 and 16 exhibited a good fit to the observeddata (Fig. 5 and Table 2). For PBS, the chiral inver-sion rates for both isomers were almost similar andthat its racemization was observed. For rat plasma,the chiral inversion from R-isomer to S-isomer wasfaster than that from S-isomer to R-isomer (kRS,in vitro/


10 IZUMI ET AL.

Figure 4. Chiral inversion of R-isomer (closed circle) and S-isomer (open circle) of rivoglita-zone in PBS, rat, monkey, and human plasma during incubation at 37◦C; (a) and (b) addition of(R)- and (S)-rivoglitazone in PBS, respectively; (c) and (d), that in rat plasma, respectively; (e)and (f), that in monkey plasma, respectively; (g) and (h), that in human plasma.



Figure 5. Relationship between fuR/fuS and kRS,in vitro/kSR ,in vitro in vitro chiral inversion studies.

kSR,in vitro = 3.81). On the contrary, for monkey andhuman plasma, the chiral inversion from S-isomer toR-isomer was faster than that from S-isomer to R-isomer, kRS,in vitro/kSR,in vitro was 0.665 and 0.547, re-spectively. In addition, the chiral inversion rate inplasma was faster than that in PBS, except for S-isomer in rat plasma. The stereoselectivity of thechiral inversion rate (kRS,in vitro/kSR,in vitro) exhibitedgood correlation (r2 = 0.9994) to the stereoselectiv-ity of plasma unbound fraction (fuR/fuS) as shown inFigure 5.

In Vitro Metabolism and IVIVE

The unbound fractions of R-isomer and S-isomerin the microsomal oxidation reaction mixtures forboth species were comparably ranged between 0.440and 0.604 and 0.451 and 0.577, respectively, andthe species difference was not observed. On the con-trary, those in the microsomal glucuronidation reac-tion mixture were ranged between 0.608 and 0.735and 0.632 and 0.717 in rat microsomes, and between0.559 and 0.651 and 0.449 and 0.632 in monkey mi-crosomes, respectively.

The in vitro metabolism kinetic analysis for thefive metabolic pathways in rats and monkeys areshown in Figures S1 and S2, respectively, andTable S2. In rats, the CLuint,S for O-demethylationand N-demethylation was calculated from CLmax asan alternative parameter using the sigmoidal ki-

netic model because it clearly observed the curvateshapes in their Eadie–Hofstee plots. The CLuint ofboth isomers for TZD ring opening pathway andthe TZD ring hydroxylation pathway was calculatedfrom the Michaelis–Menten model incorporating twocontributing enzymes. Although the CLuint,R forO-demethylation, N-demethylation, and CLuint forboth isomers for N-glucuronidation were calculatedas the mean value of the plateaued v/Cu values inthe clearance plots because a nonsaturable and al-most linear increase of the v values with increasingCu values were observed.

For five metabolic pathways of both isomers in mon-keys, because a nonsaturable and almost linear in-crease of the v values with increasing Cu values wasobserved, the CLuint was calculated as the mean valueof the plateaued v/Cu values in the clearance plots.

As shown in Table 3, the rank order of CLuint,Rin rats was O-demethylation, TZD ring opening,TZD ring hydroxylation > N-demethylation, andN-glucuronidation, and that of CLuint,S was O-demethylation > TZD ring opening, TZD ring hydrox-ylation, and N-demethylation > N-glucuronidation,respectively. The highest stereoselectivity (CLuint,R/CLuint,S) was observed for N-demethylation (0.148),whereas other pathways ranged between 0.789and 1.21. The rank order of CLuint,R in mon-keys was N-glucuronidation, TZD-ring opening,and O-demethylation > TZD ring hydroxylation> N-demethylation and that of CLuint,S was O-demethylation > TZD ring opening, TZD ring hydrox-ylation > N-demethylation, and N-glucuronidation,respectively. The highest stereoselectivity in CLuintwas observed for N-glucuronidation (28.5) and themodest stereoselectivity was for O-demetylation(0.352) and TZD ring hydroxylation (0.507), and theother pathways ranged between 1.15 and 1.45.

The low CLm of rivoglitazone is assumed to be cal-culated from CLuint × fu. For rat, the high stere-oselectivity (R/S ratio) of CLm was observed for N-glucuronidation (9.06), TZD ring hydroxylation (7.59),TZD ring opening (6.61), and O-demethylation (5.93),whereas low stereoselectivity for N-demethylation(1.11). The total CLm,R and CLm,S was 0.336 and0.0606 mL/min kg, respectively, and the R/S ratiowas 5.55. For monkey, the high stereoselectivity ofCLm was observed for N-glucuronidation (23.2), O-demethylation (0.285), and TZD ring hydroxylation

Table 2. In Vitro Chiral Inversion Rate of Rivoglitazone Stereoisomers in PBS, Rat, Monkey, and Human Plasma

Incubation Medium kRS ,in vitro (h−1) kSR ,in vitro (h−1) kRS ,in vitro/kSR, in vitro t1/2, RS,in vitro (h) t1/2, SR,in vitro (h)

PBS 0.148 0.164 0.902 4.68 4.23Rat plasma 0.619 ± 0.004 0.163 ± 0.011 3.81 ± 0.27 1.12 ± 0.01 4.26 ± 0.30Monkey plasma 0.321 ± 0.014 0.483 ± 0.012 0.665 ± 0.016 2.16 ± 0.09 1.44 ± 0.03Human plasma 0.238 ± 0.008 0.436 ± 0.010 0.547 ± 0.009 2.91 ± 0.10 1.59 ± 0.04

Data represent the mean ± SD (n = 4).


12 IZUMI ET AL.

Table 3. IVIVE of Metabolic Clearance for Five Metabolic Pathways for Each Isomer of Rivoglitazone Using Rat and Monkey LiverMicrosomes

CLuint,R (:L/min mg) CLuint,S (:L/min mg) CLuint,Ra (mL/min kg) CLuint,S

a (mL/min kg) CLuint,R/CLuint,Sb

(A) RatsO-Demethylation 8.76 11.1 15.1 19.1 0.791TZD ring opening 5.91 6.72 10.2 11.6 0.879N-Demethylation 0.817 5.51 1.41 9.48 0.149TZD ring

hydroxylation5.90 5.84 10.1 10.0 1.01

N-Glucuronidation 0.457 0.379 0.786 0.652 1.21Total – – 37.6 50.8 0.740c

CLm,Rd (mL/min kg) CLm,S

d (mL/min kg) CLm,R/CLm,S Mean CLme (mL/min kg) fmf

O-Demethylation 0.135 0.0227 6.01 0.0789 0.398TZD ring opening 0.0912 0.0138 6.61 0.0525 0.265N-Demethylation 0.0126 0.0113 1.12 0.0120 0.0606TZD ring

hydroxylation0.0903 0.0119 7.59 0.0511 0.258

N-Glucuronidation 0.00703 0.000776 9.06 0.00390 0.0197Total 0.336 0.0605 5.55g 0.198 –

(B) MonkeysCLuint,R (:L/min mg) CLuint,S (:L/min mg) CLuint,R

h (mL/min kg) CLuint,Sh (mL/min kg) CLuint,R/CLuint,S

b

O-Demethylation 5.84 16.6 4.63 13.2 0.351TZD ring opening 7.97 5.51 6.32 4.37 1.45N-Demethylation 0.615 0.535 0.488 0.425 1.15TZD hydroxylation 2.78 5.48 2.21 4.35 0.508N-Glucuronidation 9.96 0.349 7.90 0.277 28.5

Total – – 21.5 22.6 0.951c

CLm,Ri (mL/min kg) CLm,S

i (mL/min kg) CLm,R/CLm,S Mean CLme (mL/min kg) fmf

O-Demethylation 0.0199 0.0700 0.284 0.0450 0.425TZD ring opening 0.0272 0.0232 1.17 0.0252 0.238N-Demethylation 0.00210 0.00225 0.933 0.00218 0.0206TZD ring

hydroxylation0.00950 0.0231 0.411 0.0163 0.154

N-Glucuronidation 0.0340 0.00147 23.1 0.0177 0.167Total 0.0927 0.120 0.773g 0.106 –

aCLuint,R or CLuint,S (mL/min kg) = 1720.214 × CLuint,R or CLuint,S (mL/min mg).bCLuint,R (mL/min kg)/CLuint,S (mL/min kg).cTotal of CLuint,R (mL/min kg)/total of CLuint,S (mL/min kg).dCLm,R = 0.00894 (fu,R) × CLuint,R, CLm,S = 0.00119 (fu,S) × CLuint,S.gTotal of CLm,R/total of CLm,S.eMean of CLm,R and CLm,S.ffm = Mean CLm for each pathway/total of mean CLm.hCLuint,R or CLuint,S (mL/min kg) = 793.614 × CLuint,R or CLuint,S (mL/min mg).iCLm,R = 0.00430 (fu,R) × CLuint,R, CLm,S = 0.00530 (fu,S) × CLuint,S.

(0.412), whereas low stereoselectivity was observedfor TZD ring opening (1.17) and N-demethylation(0.933). The total CLmR and CLmS was 0.0927 and0.120 mL/min kg, respectively, and the R/S ratio was0.774. On the basis of the values of fm, the mainmetabolic pathways in rat were O-demethylation,TZD ring opening, and TZD ring hydroxylation andthose in monkey were O-demethylation, TZD ringopening, and N-glucuronidation.

DISCUSSION

In this study, we investigated the stereoselectivityin the pharmacokinetics of the two stereoisomers ofrivoglitazone in a series of in vivo and in vitro experi-ments in rats and monkeys, which were used in safety

and pharmacology assessment. In particular, we fo-cused on the stereoselectivity in chiral inversion andmetabolic clearance of rivoglitazone. The chiral inver-sion of rivoglitazone occurs at the fifth position of theTZD ring, which has been reported for TZD deriva-tives such as rosiglitazone,6 troglitazone,16 and MK-0767.9 The pharmacokinetic model involving chiralinversion is comparable with a reversible metabolicsystem.17 The plasma profiles after intravenous andoral administration of R-isomer or S-isomer to ratsand monkeys were well fitted to this model (Fig. 2).Because the fraction of R-isomer and S-isomer en-tered into systemic from the gut was estimated to beless than 0.9 in both animals, it suggested that nochiral inversion occurred before absorption. No chiralinversion may be because of the oral administration



of both stereoisomers as acidic solution (pH 2), whichmight inhibit the chiral inversion during the absorp-tion process. For rivoglitazone, because S-isomer hashigher efficacy compared with R-isomer,5 the chiralinversion in the gut will have some effect on the ef-ficacy. It will be useful to know the chiral inversionbefore absorption for this pharmacokinetic model fororal administration.

The similar species difference of the stereoselectiv-ity in the chiral inversion rate was evaluated from amodel analysis and the in vitro incubation study; kRS/kSR = 1.96 for rats and 0.548 for monkeys from the invivo analysis and kRS, in vitro/kSR, in vitro = 3.81 for ratsand 0.669 for monkeys from the in vitro study.

The protein binding of stereoisomers of rivoglita-zone was measured by the modified microscale ultra-centrifugation technique at 4◦C to avoid chiral inver-sion during assay procedure. Although these valuesmay be different from those at 37◦C, the comparisonof fuR and fuS for pharmacokinetic analysis is thoughtto be effective because of the measurement being atthe same condition for both isomers and this methodbeing reported to be comparable with other methodsat 37◦C for other highly protein bound drugs.10

The relationships between chiral inversion and pro-tein binding were reported for many drugs.18 Forstereoselectivity in the chiral inversion, mexalactamexhibited the comparable relationship between theR/S ratio of chiral inversion (1.88) and that of theunbound fraction (1.55).19 Troglitazone, TZD ringderivative, showed stereoselectivity in the chiral in-version rate among the four stereoisomers, 1.47-folddifferences; however, the differences in the unboundfraction were small.15 Meanwhile, Aso et al.18 re-ported the retardation and acceleration of racemiza-tion and epimerization for the relationship betweenchiral inversion and protein binding. Indeed, rivogli-tazone exhibited 1.6–4.2-fold acceleration of the chi-ral inversion rate in rat, monkey, and human plasmacompared with PBS buffer, except for the chiral inver-sion from R-isomer to S-isomer in rat plasma. Trogli-tazone exhibited the acceleration of chiral inversionof 4.2–7.3-fold in ddY mouse plasma,16 whereas MK-0767 exhibited the retardation of chiral inversion infresh human plasma.20 Thus, because the relation-ships between chiral inversion and protein bindingseem to depend on the drug, the individual profileshould be investigated.

The stereoselectivity of chiral inversion clearance,calculated as kRSVR and kSRVS, was high (CLRS/CLSR= 5.78) in rats, which although was low (CLRS/CLSR= 1.31) in monkeys. The large species difference instereoselectivity of chiral inversion clearance will bebecause of the volume of distribution. VR/VS was 4.04for rats and that for monkeys was 1.06. The volumeof distribution for both stereoisomers were small andcomparable to the extracellular fluid (0.3 L/kg for rats

and 0.2 L/kg for monkeys),21 therefore, the tissue dis-tribution is assumed to be low. The stereoselectivityin the distribution volume seemed to be predominatedby the stereoselectivity in plasma protein binding (fuR/fuS = 7.51) for rats, whereas stereoselectivity in mon-keys exhibited minimally (fuR/fuS = 0.811) comparedwith rats.

The metabolic clearance evaluated from in vivopharmacokinetic model analysis for rats exhibitedhigh stereoselectivity (CLeR/CLeS = 5.78), but thatfor monkeys was low (CLeR/CLeS = 1.31). The bloodmetabolic clearance, calculated by CLeR and CLeS di-vided by RB (whole blood-to-plasma concentration ra-tio: 0.671 for rats and 0.810 for monkeys at 100 ng/mL of rivoglitazone),3 was 3.5% and 0.6% of the rathepatic blood flow (55.2 mL/min kg) and 1.2% and0.9% for monkey hepatic blood flow (43.6 mL/min kg),respectively. Therefore, the metabolic clearance forboth stereoisomers in both animals is assumed to beunbound plasma fraction (fu) × intrinsic metabolicclearance (CLuint).15 On the basis of this assumption,CLuint,R and CLuint,S in rats was estimated to be 166and 130 mL/min kg, respectively, and the stereoselec-tivity in CLuint (R/S ratio of 1.28) was estimated tobe low. Thus, the stereoselectivity in metabolic clear-ance in rats (CLeR/CLeS = 5.78) was estimated to bepredominated by the stereoselectivity in fu (fuR/fuS =7.51). For monkeys, CLuint was estimated to be 98.1mL/min kg for R-isomer and 62.1 mL/min kg for S-isomer. Thus, the stereoselectivity in metabolic clear-ance in monkeys (CLeR/CLeS = 1.31) was estimated toreflect the stereoselectivity in fu (fuR/fuS = 0.811) andCLuint (R/S ratio of 1.58).

This stereoselectivity in metabolic clearance wasevaluated using the IVIVE approach based on thein vitro metabolism study for the five pathways ofrivoglitazone. For rats, the sum of intrinsic clearancefor the five pathways of each isomer exhibited lowstereoselectivity (CLuintR/CLuintS = 0.739), but thatof metabolic clearance exhibited high stereoselectiv-ity (CLmR/CLmS = 5.55). These values were compa-rable with the stereoselectivity in CLuint (R/S ra-tio of 0.770) and metabolic clearance (CLeR/CLeS of5.78) evaluated from pharmacokinetic model anal-ysis. For monkeys, the sum of intrinsic clearanceand metabolic clearance for the five metabolic path-ways exhibited low stereoselectivity, CLuintR/CLuintS= 0.954 and CLmR/CLmS = 0.774, which were almostcomparable with those evaluated from the in vivomodel analysis, 1.62 and 1.31, respectively. However,in vitro metabolic clearance for each isomer was un-derestimated compared with in vivo metabolic clear-ance, 26%–28% for rats and 22%–37% for monkeys.Many studies have been published for the IVIVE,and high prediction accuracy of the hepatic clear-ance needs the correction of the plasma and micoroso-matic binding.14,22–27 The underpredicition of in vivo


14 IZUMI ET AL.

CL was reported for the low clearance and highlyprotein bound drugs.22,23,25 Although we evaluatedthe metabolic kinetic parameters after correction ofthe cofounder of binding by the ultracentrifugation,the features of rivoglitazone, low clearance and highprotein bound (>99%), may be the reasons for un-derprediction. In addition, the metabolic pathway ofrivoglitazone involves the N-glucuronidation path-way, and the prediction of metabolic clearance of glu-curonidation using the IVIVE method was reportedto be difficult.28–30 However, the evaluation of fm ofmajor metabolic pathways from the IVIVE methodexhibited relatively comparable to fm from in vivodata3; in rats, the fm of O-demethylation pathway was45.0% from in vitro and 43.1% from in vivo, in mon-keys, the fm of O-demethylation pathway was 42.5%from in vitro and 32.8% from in vivo, and the fm ofN-glucuronidation pathway was 15.4% from in vitroand 12.9% from in vivo, respectively. Thus, the con-tribution of the metabolic pathway from an IVIVEapproach will relatively reflect well that in vivo.

In this study, the high stereoselectivity wasobserved in CLuint of N-demethylation in rats(CLuint,R/CLuint,S = 0.149), but the stereoselectivityin metabolic clearance was low (CLm,R/CLm,S = 1.12)because of canceling out by opposite stereoselectivityin protein binding (fuR/fuS = 7.51). On the contrary,the stereoselectivity of the four pathways was low inCLuint (0.791–1.21), but that of metabolic clearancewas high (6.01–9.06-fold difference) because of con-tribution of high stereoselectivity in unbound frac-tion (fuR/fuS = 7.51). Thus, the high stereoselectiv-ity in the metabolic clearance of rat was dominatedby the stereoselectivity of unbound fraction, but notthat of intrinsic clearance. In monkeys, although thepathway of N-glucuronidation exhibited high stere-oselectivity in CLuint (CLuint,R/CLuint,S = 28.5) andCLm (CLm,R/CLm,S = 23.1), the other four pathwaysexhibited low stereoselectivity in CLuint (0.351–1.45)and CLm (0.281–1.17) and the contribution of N-glucuronidation was low (fm = 0.167), ultimately totalCLm showed low stereoselectivity (R/S ratio of 0.773).

CONCLUSIONS

The stereoselectivity in pharmacokinetics of rivogli-tazone in rats and monkeys was investigated usingmodel-based pharmacokinetic analysis and IVIVE ap-proach, and the stereoselectivity was evaluated in chi-ral inversion clearance, metabolic clearance, and vol-ume of distribution. Ultimately, the stereoselectivityin unbound fraction worked as a major determinantof stereoselectivity in pharmacokinetics of rivoglita-zone. Furthermore, the model based on the CLuintand fu for each stereoisomer, which are the importantfactors to determine the pharmacokinetics of bothstereoisomers of rivoglitazone, will be able to more

accurately evaluate these parameters to discuss thestereoselectivity. The approach combined with model-based pharmacokinetic analysis and the IVIVE ap-proach will be effective to correctively evaluate thestereoselectivity of this case involving chiral inver-sion process and drug metabolism process.

ACKNOWLEDGMENTS

The authors would like to thank Dr. MinoruUchiyama for giving metabolism information and Dr.Yuichi Aizawa for giving patent information.

REFERENCES

1. Kanda S, Nakashima R, Takahashi K, Tanaka J, Ogawa J,Ogata T, Yachi M, Araki K, Ohsumi J. 2009. Potent antidia-betic effects of rivoglitazone, a novel peroxisome proliferator-activated receptor-( agonist, in obese diabetic rodent models.J Pharmacol Sci 111:155–166.

2. Sood V, Colleran K, Burge MR. 2000. Thiazolidinediones: Acomparative review of approved uses. Diabetes Technol Ther2:429–440.

3. Uchiyama M, Iwabuchi H, Tsuruta F, Abe K, Takahashi M,Koda H, Oguchi M, Okazaki O, Izumi T. 2011. Pharmacoki-netics, metabolism, and disposition of rivoglitazone, a novelperoxisome proliferator-activated receptor ( agonist, in ratsand monkeys. Drug Metab Dispos 39:653–666.

4. Uchiyama M, Koda H, Fischer T, Mueller J, Yamamura N,Oguchi M, Iwabuchi H, Okazaki O, Izumi T. 2011. In vitrometabolism of rivoglitazone, a novel peroxisome proliferator-activated receptor ( agonist, in rat, monkey, and human livermicrosomes and freshly isolated hepatocytes. Drug Metab Dis-pos 39:1311–1319.

5. Ohsumi J, Wada K, Matsui Y, Ogata T. 2006. Optically activethiazolidinedione derivatives. Patent US5886014.

6. Parks D, Tomkinson N, Villeneuve M, Blanchard S, WillsonT. 1998. Differential activity of rosiglitazone enantiomers atPPAR-gamma. Bioorg Med Chem Lett 8:3657–3658.

7. Sohda T, Mizuno K, Kawamatsu Y. 1984. Studies on an-tidiabetic agents. VI. Asymmetric transformation of (±)-4-[4-(1-methyl cyclohexylmethoxy)benzyl]-2, 4-thiazolidinedione(ciglitazone) with optically active 1-phenylethylamines. ChemPharm Bull 32:4460–4465.

8. Clark DA, Goldstein SW, Volkmann RA, Eggler JE, HollandGF, Hulin B, Stevenson RW, Kreutter DK, Gibbs M, KruppMN, Merrigan P, Kelbaugh PL, Kappeler WH, Mcdermott RE,Hutson NJ, Johnson MR. 1991. Substituted dihydrobenzopy-ran and dihydrobenzofuran thiazolidine-2, 4,-diones and hy-poglycemic agents. J Med Chem 34:319–325.

9. Rippley RK, Yan KX, Matthews ND, Greenberg HE, HermanGA, Wagner JA. 2007. Human pharmacokinetics and intercon-version of enantiomers of MK-0767, a dual PPAR"/( agonist.J Clin Pharmacol 47:323–333.

10. Nakai D, Kumamoto K, Sakikawa C, Kosak T, Tokui T. 2004.Evaluation of the protein binding ratio of drugs by a micro-scale ultracentrifugation method. J Pharm Sci 93:847–854.

11. Fisher MB, Campanale K, Ackermann BL, Vandenbranden M,Wrighton SA. 2000. In vitro glucuronidation using human livermicrosomes and the pore-forming peptide alamethin. DrugMetab Dispos 28:560–566.

12. Xu J, Winkler J, Derendorf H. 2007. A pharmacokinetic/pharmacodynamic approach to predict total prednisolone con-centrations in human plasma. J Pharmacokinet Pharmacodyn34:355–372.



13. Houston JB, Kenworthy KE, Galetin A. 2003. Typical andatypical enzyme kinetics. In Drug metabolizing enzymes: Cy-tochrome P450 and other enzymes in drug discovery and de-velopment; Lee J, Ocach RS, Fischer MB, Eds. Boca Raton,Florida: CRC Press, pp 211–254.

14. Ring BJ, Chien JY, Adkison KK, Jones HM, Rowland M, JonesRD, Yates JWT, Ku MS, Gibson CR, He H, Vuppugalla R,Marathe P, Fischer V, Dutta S, Sinha VK, Bjornsson T, LaveT, Poulin P. 2011. PhRMA CPCDC initiative on predictivemodel of human pharmacokinetics. 3. Comparative assess-ment of prediction methods of human clearance. J Pharm Sci100:4090–4110.

15. Rowland M, Tozer TN. 2011. Elimination. In Clinical pharma-cokinetics and pharmacodynamics: Concepts and applications.Philadelphia, Pennsylvania: Lippincott Williams & Williams,pp 111–157.

16. Izumi T, Enomoto S, Hoshiyama K, Sasahara K, SugiyamaY. 1997. Pharmacokinetic stereoselectivity of troglitazone, anantidiabetic agent, in the KK mouse. Biopharm Drug Dispos18:305–324.

17. Cheng H, Jusko WJ. 1993. Pharmacokinetics of reversiblemetabolic systems. Biopharm Drug Dispos 14:721–766.

18. Aso Y, Yoshioka S, Takeda Y. 1990. Epimerization and racem-ization of some chiral drugs in the presence of human serumalbumin. Chem Pharm Bull 38:180–184.

19. Yoshikawa T, Oguma T, Ichihashi T, Kinoshita H, Hirano K,Yamada H. 1999. Epimerization of moxalactam by albuminand simulation of in vivo epimerization by a physiologicallybased pharmacokinetic model. Chirality 11:309–315.

20. Shen Z, Bakhtiar R, Kumuro M, Awano K, Taga F, Colletti A,Hora D, Feeney W, Iliff S, Franklin RB, Vincent S. 2005. Enan-tiomer ratio of MK-0767 in humans and nonclinical species.Rapid Commun Mass Spectrom 19:1125–1129.

21. Davies B, Morris T. 1993. Physiological parameters in labora-tory animals and humans. Pharm Res 10:1093–1095.

22. Obach RS. 1999. Prediction of human clearance of twenty-nine drugs from hepatic microsomal intrinsic clearance data:

An examination of in vitro half-life approach and nonspecificbinding to microsomes. Drug Metab Dispos 27:1350–1359.

23. Poulin P, ECAHop, Ho Q, Halladay JS, Haddad S, Kenny JR.2012. Comparative assessment of in vitro-in vivo extrapolationmethods used for predicting hepatic metabolic clearance ofdrugs. J Pharm Sci 101:4308–4326.

24. Poulin P, Kenny JR, ECAHop, Haddad S. 2012. In vitro-invivo extrapolation of clearance: Modeling hepatic metabolicclearance of highly bound drugs and comparative assessmentwith existing calculation methods. J Pharm Sci 101:838–851.

25. Chao P, Uss AS, Cheng KC. 2010. Use of intrinsic clearancefor prediction of human hepatic clearance. Expert Opin DrugMetab Toxicol 6:189–198.

26. Ito K, Houston JB. 2005. Prediction of human drug clearancefrom in vitro and preclinical data using physiologically basedand empirical approaches. Pharm Res 22:103–112.

27. Sternbeck AKS, Afzelius L, Prusis P, Neelissen J, HoogstraateJ, Johansson J, Floby E, Bengtsson A, Grissberg O, SternbeckJ, Petersson C. 2010. Evaluation of the human prediction ofclearance from hepatocyte and microsome intrinsic clearancefor 52 drug compounds. Xenobiotica 40:637–649.

28. Miners JO, Mackenzie PI, Knights KM. 2010. The predic-tion of drug-glucuronidation parameters in humans: UDP-glucuronosyltransferase enzyme-selective substrate and in-hibitor probes for reaction phenotyping and in vitro-in vivoextrapolation of drug clearance and drug–drug interaction po-tential. Drug Metab Rev 42:196–208.

29. Kilford PJ, Stringer R, Sohal B, Houston JB, Galetin A.2009. Prediction of drug clearance by glucuronidation fromin vitro data: Use of combined cytochrome P450 and UDP-glucuronosyltransferase cofactors in alamethicin-activatedhuman liver microsomes. Drug Metab Dispos 37:82–89.

30. Gill KL, Houston JB, Galetin A. 2012. Characterization ofin vitro glucuronidation clearance of a range of drugs in hu-man kidney microsomes: Comparison with liver and intestinalglucuronidation and impact of albumin. Drug Metab Dispos40:825–835.


Stereoselectivity in pharmacokinetics of rivoglitazone, a novel peroxisome proliferator-activated...

Documents

Transcript of Stereoselectivity in pharmacokinetics of rivoglitazone, a novel peroxisome proliferator-activated...