Preclinical pharmacokinetics of a novel HIV-1 attachment inhibitor BMS-378806 and prediction of its...

Received 1 March 2005Revised 23 June 2005

Accepted 27 June 2005Copyright # 2005 John Wiley & Sons, Ltd.

BIOPHARMACEUTICS & DRUG DISPOSITIONBiopharm. Drug Dispos. 26: 387– 402 (2005)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bdd.471

Preclinical Pharmacokinetics of a Novel HIV-1 AttachmentInhibitor BMS-378806 and Prediction of its HumanPharmacokinetics

Zheng Yanga,*, Lisa Zadjuraa, Celia D’Arienzoa, Anthony Marinoa, Kenneth Santonea, Lewis Klunka, DouglasGreenea, Pin-Fang Linb, Richard Colonnob, Tao Wangc, Nicholas Meanwellc and Steven Hansela,y

a Department of Metabolism and Pharmacokinetics, Bristol-Myers Squibb Co., Wallingford, CT 06492, USAb Department of Virology, Bristol-Myers Squibb Co., Wallingford, CT 06492, USAc Department of Chemistry, Bristol-Myers Squibb Co., Wallingford, CT 06492, USA

ABSTRACT: BMS-378806 is a prototype of novel HIV attachment inhibitors that block the gp120and CD4 interaction, the first step of HIV-1 entry into cells. The present work investigated thepharmacokinetics of BMS-378806 in rats, dogs and monkeys and assessed its in vitro permeabilityand metabolism. BMS-378806 exhibited species-dependent oral bioavailability which was 19%–24%in rats and monkeys and 77% in dogs. In rats and monkeys, absorption was prolonged, with anapparent terminal half-life of 2.1 and 6.5 h, respectively. In rats, linear pharmacokinetics wasobserved between i.v. doses of 1 and 5 mg/kg and between p.o. doses of 5 and 25 mg/kg. The totalbody clearance was intermediate in rats and low in dogs and monkeys. The steady-state volume ofdistribution was moderate (0.4–0.6 l/kg), contributing to a short half-life (0.3–1.2 h) after i.v. dosing.Studies in bile-duct cannulated rats together with intraportal infusion studies revealed that therenal and hepatic clearance each accounted for 30% and 70% of the total elimination in rats, with thehepatic clearance largely being oxidative metabolism. In vitro, BMS-378806 was not highly proteinbound (44%–73%). The Caco-2 permeability was modest (51 nm/s) and confounded by P-glycoprotein mediated efflux transport. Both of these may contribute to the low brain penetrationobserved in rats (brain/plasma AUC ratio ¼ 0:06). In human liver microsomes BMS-378806 wasequally metabolized by cytochrome P450 1A2, 2D6 and 3A4 and did not inhibit major drug-metabolizing enzymes to a significant extent. Based on in vitro and animal data, a mechanisticapproach that factors in absorption and first-pass metabolism was employed to predict the humanoral bioavailability of BMS-378806 (ca 20%). This, together with the complex Dedrick plot method,was used to simulate human oral profiles and to project an efficacious dose. These study resultsoffer a comprehensive assessment of the developability of BMS-378806 and provide importantguidance to improving absorption and half-life of future compounds in the series. The currentstudies also demonstrate the value and approaches of understanding pharmacokinetic properties inthe early stage of drug discovery. Copyright # 2005 John Wiley & Sons, Ltd.

Key words: pharmacokinetics; HIV attachment inhibitor; animal; in vitro; prediction

Introduction

Human immunodeficiency virus (HIV) infectionis a disease that affects millions of peopleworldwide. Despite advancement in combinationtherapies involving nearly 20 marketed anti-HIVdrugs, viral resistance, tolerance/toxicity and

*Correspondence to: Bristol-Myers Squibb Co., Rt. 206 andProvince Line Road, Princeton, NJ 08543, USA.E-mail: [email protected] address: Department of Pharmacokinetics, Dy-namics and Metabolism, Pfizer Inc., Ann Arbor, MI 48105,USA.

patient compliance remain major issues [1–3],requiring new classes of HIV-1 inhibitors forsuccessful disease intervention in the future.

BMS-378806 (Figure 1) is a prototype of anewly discovered series of novel HIV-1 inhibitorsthat block viral attachment [4,5]. It blocked HIV-1infection by binding directly to the viral gp120envelope protein and preventing the interactionbetween gp120 and cellular CD4 receptors on Tcells, the very first step of the viral entry process.Because of the known heterogeneity of HIV-1envelope glycoproteins, subtype B viruses ap-peared to be the most susceptible to BMS-378806,although activity was also observed against othersubtypes including A, C and D viruses. Theantiviral activity (median EC50) against 42 sub-type B clinical isolates was 62 nm [4]. BMS-378806was highly selective for HIV-1, having nosignificant activity against HIV-2 and showingno cytotoxicity in a variety of cell types(CC50>225 mm). The aqueous solubility of thecompound was 0.17 mg/ml and was constantbetween pH 4 and 8. Because of the zwitterionicnature of the compound, the solubility increasedto 1.3 and 3.3 mg/ml at pH values of 2.1 and 11,respectively [5].

The present work investigated the pharmaco-kinetics of BMS-378806 in rats, dogs and mon-keys and assessed its in vitro metabolism inanimal species and humans. In order to predicthuman pharmacokinetics, the study also evalu-ated absorption potential in Caco-2 cells, estab-

lished in vitro–in vivo correlation in clearance,delineated elimination pathways and first-passmetabolism in rats, and assessed metabolism-based drug–drug interaction potential usingin vitro systems. Finally, multiple approacheswere used, including allometry, an in vitromethod, and the complex Dedrick plot method,to predict human pharmacokinetic parametersand profiles of BMS-378806. These studies are ofgreat importance in discovering and developinga novel anti-HIV agent, since the current anti-retroviral therapy involves a combination ofmultiple agents, which is heavy with a pillburden and prone to metabolism-based drug–drug interactions. A thorough understanding ofthe absorption and disposition of BMS-378806 inthe preclinical setting is critical to assess thedevelopability of the compound. Furthermore,benchmarking the potential human pharmacoki-netic properties of BMS-378806 to marketed anti-HIV drugs provides vital guidance to the designof future compounds in the series, so that theywill be orally bioavailable, with an adequate half-life for once-daily dosing and low potential fordrug–drug interactions.

Materials and Methods

Chemicals and reagents

BMS-378806 was synthesized at Bristol-MyersSquibb Co. (Wallingford, CT). Furafylline, sulfa-phenazole, quinidine, ketoconazole, reducedb-nicotinamide adenine dinucleotide phosphate(b-NADPH), glucose-6-phosphate and glucose-6-phosphate dehydrogenase were purchased fromSigma-Aldrich Co. (St Louis, MO). 3-Cyano-7-ethoxycoumarin (CEC), 7-methoxy-4-trifluoro-methylcoumarin (MFC), 3-[2-(N,N-diethyl-N-methylamino)ethyl]-7-methoxy-4-methylcoumar-in (AMMC), benzoylresorufin (BzRes) and7-benzyloxy-4-trifluoromethylcoumarin (BFC)were obtained from BD Biosciences (Woburn,MA). All other solvents and chemicals were ofHPLC or analytical grade.

In vitro studies

Caco-2 cell permeability. Caco-2 cells, obtainedfrom the American Type Culture Collection

NN

OO

N

O

N

O

Figure 1. Structure of BMS-378806

Copyright # 2005 John Wiley & Sons, Ltd. Biopharm. Drug Dispos. 26: 387–402 (2005)

Z. YANG ET AL.388

(Rockville, MD), were seeded onto a collagencoated polycarbonate filter membrane at adensity of 80 000 cells/cm2 and were usedbetween 14 and 21 days in culture. The mediumfor permeability studies was modified Hank’sbalanced salt solution containing either 10 mm N-2-hydroxyethylpiperazine-N0-2-ethanesulfonic acid(HEPES) or 25 mm 2-(N-morpholino)ethane-sulfonic acid (MES) depending on the intendedpH (e.g. pH 5.5, 6.5 and 7.4 at the apical side andpH 7.4 at the basolateral side). BMS-378806 wasadded to either the apical or basolateral side ofthe monolayer at an initial concentration of200 mm, and then incubated at 378C for 4 h. Toassess the role of P-glycoprotein (P-gp), a knownP-gp substrate/inhibitor verapamil (50 mm) wasadded to both sides of the monolayer initially.Samples were collected from both the receiverand donor compartments. The apparent Caco-2permeability coefficient (Pc) was calculated usingthe equation Pc ¼ ðdA=dtÞ=ðS � C0Þ, where dA/dtis the rate of permeation across the monolayer, Sis the surface area of the monolayer (0.33 cm2),and C0 is the initial concentration in the donorcompartment.

Blood stability and blood-to-plasma concentrationratios. The blood stability of BMS-378806 wasevaluated in heparinized blood freshly collectedfrom rats, dogs, monkeys and humans. Thecompound was spiked in blood at a concentra-tion of 2.5 mm, and the mixture was incubated at378C for 2 h. Serial blood samples (0.5 ml each)were taken at 0, 0.17, 0.5 and 2 h. A half volumeof the blood sample was immediately placed intoa microtainer (Becton Dickinson, Franklin Lakes,NJ) and centrifuged to harvest plasma. The otherhalf of the sample was transferred to the tubecontaining 0.125 ml water and vortexed to hemo-lyse blood cells. Samples were stored at �208Cuntil analysis.

Plasma protein binding. The extent of plasmaprotein binding of BMS-378806 was determinedat 378C using an ultrafiltration method. Thecompound was spiked in heparinized plasmafreshly collected from rats, dogs, monkeys andhumans, and the mixture was prepared induplicate at 0.25 and 2.5 mm. After incubatingthe mixture at 378C for 30 min, an aliquot of

0.5 ml was transferred to the Amicon Centrifreemicropartition device (Millipore, Bedford, MA)followed by centrifugation at 2000 g for 15 min togenerate ultrafiltrate (Jouan MR 1822, Winche-ster, VA). Negligible nonspecific binding todevices was found for BMS-378806. The unboundfraction in plasma was determined from the ratioof ultrafiltrate to plasma concentrations.

Microsomal stability. The oxidative metabolism ofBMS-378806 was evaluated in pooled livermicrosomes of rats, dogs, monkeys and humans.The human and rat liver microsomes werepurchased from BD Biosciences (Woburn, MA)and In Vitro Technologies (Baltimore, MD),respectively. The dog and monkey liver micro-somes were obtained from XenoTech LLC (Kan-sas City, KS). The concentrations of cytochromeP450 (CYP) enzymes in these preparations were0.48, 0.79, 0.89 and 1.4 nmol/mg protein inhumans, rats, dogs and monkeys, respectively.

The incubation mixture (3 ml) was prepared intriplicate for each species in a 0.1m potassiumphosphate buffer (pH 7.4) containing 1 mm BMS-378806, 0.5 mm CYP enzymes, 1.8 mm glucose-6-phosphate and 0.4 units/ml glucose-6-phosphatedehydrogenase. The concentration of organicsolvents in the incubation mixture was 0.1%, ofwhich acetonitrile accounted for 98% and di-methyl sulfoxide (DMSO) was 2%. After a 5 minpreincubation at 378C, the reaction was initiatedby adding b-NADPH (0.15 mm final concentra-tion). Aliquots of 0.25 ml were taken at 0, 2, 5, 10,20 and 30 min, and placed into 3 volumes of ice-cold acetonitrile (containing an internal standard)to terminate the reaction. The samples werevortexed and centrifuged at 10 000 g for 5 min tocollect supernatant for sample analysis. Theturnover rate constant of oxidative metabolismin microsomes (k) was determined by nonlinearregression of the percentage of the compoundremaining (y)-time (t) curve using the equation ofy ¼ y0 � expð�k � tÞ, where y0 is the percentage ofthe compound remaining at time zero, asestimated from the equation.

Inhibition of human CYP enzymes. The potential ofBMS-378806 to inhibit major human cytochromeP450 (CYP) enzymes was evaluated using arecombinant CYP system (BD SupersomesTM


PRECLINICAL PHARMACOKINETICS OF BMS-378806 389

Enzymes, BD Biosciences, Woburn, MA). TheIC50 values of BMS-378806 for each CYP isoformwas determined from the inhibition of deethyla-tion of CEC (CYP1A2 and CYP2C19), dealkyla-tion of MFC (CYP2C9), demethylation of AMMC(CYP2D6), and dealkylation of BzRes (CYP3A4)and of BFC (CYP3A4) by fluorescence detectionin a 96-well format. The studies were conductedin duplicate over a 45-min incubation period,with substrate concentrations equal to the appar-ent Km (BFC concentration below the Km) andeight inhibitor concentrations ranging from 18 nmto 40 mm. The IC50 values were estimated usingXlfitTM curve-fitting software (ID Business Solu-tions Limited, UK). For assessment of time-dependent inhibition, the incubation intervalsfor the IC50 evaluation were done in 5, 15, 30, 45and 60 min.

Reaction phenotyping in human liver microso-mes. The role of CYP enzymes in the metabolismof BMS-378806 was studied in human livermicrosomes (HLM) using isoform-selective che-mical inhibitors. The inhibitors were furafylline(20 mm) for CYP1A2, sulfaphenazole (30 mm) forCYP2C9, quinidine (10 mm) for CYP2D6 andketoconazole (3 mm) for CYP3A4. The stocksolutions of inhibitors and BMS-378806 wereprepared in methanol or acetonitrile, with a finalorganic concentration of 0.2%–0.3% in the in-cubation mixture. The studies were conducted intriplicate for each inhibitor along with the vehiclecontrols, using the incubation procedure de-scribed above. For furafylline, it was preincu-bated in the incubation mixture for 15 min in thepresence of b-NADPH, with the reaction initiatedby the addition of BMS-378806 (1 mm finalconcentration). For other inhibitors, they wereadded along with BMS-378806 for a 5-minpreincubation, with the reaction initiated by theaddition of b-NADPH. Aliquots of samples weretaken at 0 and 30 min. The extent of inhibitionwas determined by comparing the percentage ofBMS-378806 remaining at 30 min in the presenceand absence of inhibitors.

In vivo studies

The animal testing methods used in the studiesdescribed below were approved by the Bristol-

Myers Squibb Institutional Animal Care and UseCommittee and adhered to the ‘Principles ofLaboratory Animal Care’ (NIH publication 85-23,revised in 1985). The animals were housed in a12-h light cycle, with free access to food andwater.

Pharmacokinetics in rats. Male Sprague-Dawleyrats (300–350 g) with catheters placed in thejugular vein, portal vein and bile duct werepurchased from Hilltop Lab Animals, Inc. (Scott-dale, PA) and used in the pharmacokineticstudies of BMS-378806. In the studies describedbelow, the compound was dosed in a PEG400/ethanol solution (90/10, v/v) unless notedotherwise. Plasma samples were harvested fromblood (0.2–0.3 ml) collected from the jugular veinand stored at �208C until analysis.

In an intravenous (i.v.) study, a dose of 1 mg/kg was administered to three animals over 30 s,and serial plasma samples were collected beforedosing and 0.033, 0.17, 0.5, 1, 1.5, 2, 4, 6 and 8 hafter dosing. In an oral (p.o.) study, 5 and 25 mg/kg doses were administered to fasted animals(n ¼ 3 per dose), and plasma samples wereobtained before dosing and 0.17, 0.5, 1, 1.5, 2, 4,6 and 8 h after dosing. In addition, a p.o. dose of25 mg/kg was given to bile duct-cannulated(BDC) rats ðn ¼ 3Þ, where plasma, urine and bilesamples were collected on ice up to 24 h andstored at �208C until analysis.

In an intraportal study, three animals receiveda dose of 0.85 mg/kg given in a 100% PEG400solution. The compound was administered via a40-min infusion at a rate of 7.5 mg/min peranimal (volume infused ¼ 25 ml=min per animal),with the infusion rate set to mimic the rate of oralabsorption into the portal vein following p.o.administration of the 5 mg/kg dose. Serialplasma samples were taken before infusion,0.17, 0.33, 0.5 and 0.67 h during infusion, and0.083, 0.17, 0.5, 1, 1.5 and 2 h after infusion.

In a brain uptake study, six animals were givenan i.v. dose of 5 mg/kg and euthanized at 0.033,0.17, 0.5, 1, 1.5 and 2 h after dosing (n ¼ 1 animalper time point). At each time point, brain tissuesand plasma samples were collected. Brain sam-ples were blotted dry, weighed and stored at�208C until analysis.


Z. YANG ET AL.390

Pharmacokinetics in dogs. Three male beagle dogs(BW 9.0� 0.47 kg, Marshall Farms USA, Inc.,North Rose, NY) were fasted overnight and usedin the oral bioavailability studies with BMS-378806 in a crossover study design. In the firstphase, a p.o. dose of 3.4 mg/kg was given to twodogs and an i.v. dose of 0.67 mg/kg wasadministered to the third dog. After a 1-weekwashout, the dosing order was reversed in thesecond phase. BMS-378806 was administered in aPEG400/ethanol solution (90/10, v/v). For i.v.administration, the dose was given via thecephalic vein as a 5-min infusion at a rate of0.1 ml/min/kg; samples were taken before dos-ing and 0.083, 0.17, 0.33, 0.5, 1, 1.5, 2, 4, 6, 8 and24 h after starting the infusion. For p.o. dosing,samples were taken prior to dosing and 0.17,0.33, 0.5, 1, 1.5, 2, 4, 6, 8 and 24 h post dosing.Plasma samples were harvested from blood(0.5 ml) drawn from the femoral artery andstored at �208C until analysis.

Pharmacokinetics in monkeys. Three male cyno-molgus monkeys (BW 8.7� 1.0 kg, Charles RiverBiomedical Research Foundation, Houston, TX)were fasted overnight and used in the oralbioavailability studies with BMS-378806 in acrossover study design. In the first phase, a p.o.dose of 3.4 mg/kg was given to two monkeysand an i.v. dose of 0.67 mg/kg was administeredto the third monkey. After a 1-week washout, thedosing order was reversed in the second phase.BMS-378806 was administered in a PEG400/EtOH solution (90/10, v/v). For i.v. dosing, thedose was given via the femoral vein as a 5-mininfusion at a rate of 0.1 ml/min/kg; the plasmasamples were taken before dosing and 0.083, 0.17,0.33, 0.5, 1, 1.5, 2, 4, 6, 8 and 24 h after startinginfusion. For p.o. dosing, the samples were takenprior to dosing and 0.17, 0.33, 0.5, 1, 1.5, 2, 4, 6, 8and 24 h post dosing. Plasma samples wereharvested from blood (0.5 ml) drawn from thefemoral artery and stored at �208C until analysis.

Sample analysis. Samples from in vitro and in vivostudies were analysed by LC/MS/MS. Theywere treated with 3 volumes of acetonitrilecontaining an IS. After centrifugation to removeprecipitated proteins, 10 ml supernatant was in-jected onto a column for analysis. Rat brain

samples were prepared as 25% brain homoge-nates. Urine and bile samples were diluted withwater and injected directly.

The HPLC system consisted of two ShimadzuLC10AD pumps (Columbia, MA), a Perkin ElmerSeries 200 autosampler (Norwalk, CT) and aHewlett Packard Series 1100 column compart-ment (Palo Alto, CA). A Keystone C18 column(2 mm� 20 mm, 3 mm particles, Bellefonte, PA)was used at 608C. The mobile phase consisted ofsolvent A (10 mm ammonium acetate and 25%methanol in water, pH adjusted to 5.5 withglacial acetic acid) and solvent B (10 mm ammo-nium acetate in methanol) at a flow rate of0.2 ml/min. The initial mobile phase compositionwas 95% A/5% B. After sample injection, themobile phase was changed to 15% A/85% B over0.1 min and held at that composition for anadditional 1 min. The mobile phase was thenreturned to initial conditions and the column re-equilibrated for 0.9 min. The total analysis timewas 2 min.

The HPLC system was interfaced to a Micro-mass Quattro LC tandem mass spectrometer(Beverly, MA) equipped with an electrosprayinterface. Ions representing the (M+H)+ speciesfor both the analyte and IS were selected in MS1and collisionally dissociated with argon at apressure of 2� 10�3 torr to form specific productions, which were subsequently monitored byMS2. The transitions monitored were m/z 407.4->174.8 for BMS-378806 and m/z 362.6->143.9 forthe IS. The lower limit of quantification (LLQ)was 9.6 nm, with measured concentrations of atleast two thirds of quality control samples within20% of nominal values.

Data analysis. Data are expressed as mean�standard deviation (SD). Differences are consid-ered significant when p50.05. For multiple-group comparisons, a one-way ANOVA wasused followed by Duncan’s multiple range test.For a two-group comparison, a two-sided non-paired Student’s t-test was employed.

The clearance of BMS-378806 in various specieswas predicted from microsomal data based onthe disappearance of the parent drug. Thepredictions were made by assuming the follow-ing for BMS-378806: linear kinetics at theconcentration studied; the liver acting as the



well-stirred model; hepatic oxidative metabolismas a major route of elimination; rapid equilibriumbetween blood and the liver; similar proteinbinding between blood and microsomes; andsimilar unbound intrinsic clearance betweenin vitro and in vivo [6]. Based on the equationdescribed by Obach et al. [7], the hepatic intrinsicclearance was calculated using the turnover rateconstant k from the liver microsomes, from whichthe hepatic blood clearance was predicted usingthe well-stirred model. The hepatic blood flowused in the calculations for rats, dogs, monkeysand humans was 70, 35, 44 and 20 ml/min/kg,respectively [8,9].

The pharmacokinetic parameters of BMS-378806 were estimated from its plasma concen-tration-time data using the non-compartmentalanalysis (KineticaTM software v.3.0, InnaPhaseCo., Philadelphia, PA). The peak concentration(Cmax) and time for Cmax (Tmax) were recordeddirectly from experimental observations. Thetotal areas under the concentration-time curve(AUC0�1) and under the first moment curvewere calculated from time zero to infinity using acombination of linear and log-linear trapezoidalsummations. The total body clearance fromplasma (CLtot), steady-state volume of distribu-tion (Vss), terminal half-life (T1/2), and meanresidence time (MRT) were also estimated fromi.v. administration [10]. The total body clearancefrom blood (CLtot,b) was calculated as the CLtot

divided by the blood-to-plasma concentrationratio. The renal clearance was estimated as thecumulative amount of unchanged drug in theurine divided by plasma AUC. The oral bioavail-ability (expressed as percentage) was estimatedas the ratio of dose-normalized AUC valuesbetween p.o. and i.v. doses. The variance of theoral bioavailability in rats was approximatedusing the first-order Taylor series expansion of aratio [11]. The percentage of the dose excretedunchanged in the urine and bile was calculatedas the cumulative amount of the unchanged drugin the respective fluid divided by the doseadministered.

To estimate the fraction of dose absorbed (fa),minimal gut wall first-pass metabolism wasassumed. The fa was calculated as the oralbioavailability divided by the fh, where fh is thefraction of the dose reaching the liver that

escapes the liver elimination. In rats, the fh wasequal to the bioavailability from the intraportalinfusion study. In dogs and monkeys, the fh wascalculated as 1 minus the hepatic extraction ratio,where the extraction ratio was estimated fromCLtot,b divided by the hepatic blood flow, assum-ing that the liver is a major organ of elimination.For the marketed drugs metoprolol, timolol,acebutalol and cimetidine, human pharmacoki-netic data (i.e. non-renal clearance and oralbioavailability) were obtained from Goodmanand Gilman’s Pharmacological Basis of Thera-peutics [12]. The fa was estimated using the sameapproach described above by assuming theblood-to-plasma concentration ratio of unityand non-renal clearance equal to hepatic clear-ance. To assess the rate of absorption, theabsorption rate constant (ka) was determined asthe reciprocal of the mean absorption time(MAT), assuming a first-order rate of absorption.MAT was calculated as the difference in MRTbetween p.o. and i.v. routes.

To predict human oral bioavailability, theindividual processes that govern bioavailabilitywere considered. The fa was estimated byconsidering both Caco-2 Pc values and animaldata. The fh was determined from human livermicrosomal results. Assuming minimal gut wallfirst-pass metabolism for BMS-378806, its humanoral bioavailability was predicted as fa � fh:

For clearance prediction in humans, fourallometric methods were used [13,14], in additionto scaling of in vitro liver microsomal data. Thefirst method was simple allometry that linearlyregresses the logarithm of clearance against thelogarithm of body weight. In the second ap-proach, the product of the maximum life-spanpotential and clearance was used instead ofclearance alone. The third method used theproduct of brain weight and clearance. The lastapproach directly scaled up the rat clearance tothe human clearance using the exponent of 0.66as described by Chiou et al. [14]. For predictionsof human Vss and T1/2, simple allometry wasemployed.

To predict human pharmacokinetic profiles ofBMS-378806, the complex Dedrick plot methodwas used [15]. The plasma concentration datafollowing i.v. administration to rats, dogs andmonkeys were normalized by (dose/body


Z. YANG ET AL.392

weight0.93) and plotted against (time/bodyweight0.37), where 0.93 and 0.37 are powerexponents determined from allometric scalingof Vss and T1/2, respectively. The data were thenfitted with a two-compartment model usingSAAMII (v1.2, Seattle, WA), from which humanpharmacokinetic profiles following i.v. adminis-tration were derived. The human oral profiles ofBMS-378806 were then simulated using predictedhuman i.v. data and oral bioavailability. The rateof absorption was described by the ka valueobtained in animals.

Results

Caco-2 cell permeability

The average apical-to-basolateral (A-to-B) Pc ofBMS-378806 was 51� 6.2 nm/s ðn ¼ 12Þ, 3.5-foldlower than that from the basolateral to apical(B-to-A) side (180� 32 nm/s, n ¼ 12; p50.05,Table 1). In the presence of 50 mm verapamil, thedirectional difference in permeability disap-peared, indicating that efflux transporters suchas P-gp are involved in transporting BMS-378806out of cells (Table 1).

Varying pH values at the apical (donor) sidefrom 5.5 to 7.4 appeared to have a minimal effecton the permeability of BMS-378806. The A-to-BPc at pH 5.5, 6.5 and 7.4 was 51� 2.1, 48� 8.7and 43� 3.1 ðn ¼ 3Þ, respectively. Although therewas a trend towards decreasing Pc values withincreasing pH, it was not statistically significant(p>0.05). These results are consistent with the fact

that BMS-378806 contains a weak base ðpKa ¼2:8Þ and a weak acid ðpKa ¼ 9:6Þ in its structure(Figure 1), and does not exist in an ionized formto any significant extent in the pH range tested[5].

Blood stability and blood-to-plasma concentra-tion ratios

BMS-378806 was stable in human, rat, dog andmonkey blood at 378C during a 2-h incubation.The blood-to-plasma concentration ratios in hu-mans, rats, dogs and monkeys were 1.1� 0.18,0.77� 0.053, 1.2� 0.079 and 0.92� 0.076 ðn ¼ 3Þ,respectively, suggesting that the compound isdistributed to approximately the same extentbetween plasma and blood cells.

Plasma protein binding

BMS-378806 exhibited a low extent of plasmaprotein binding in various species. At 2.5 mm, itwas 73% bound in humans, and 63%, 44% and54% bound in rats, dogs and monkeys, respec-tively. The extent of protein binding was con-centration-independent between 0.25 and 2.5 mm.The low extent of protein binding is consistentwith the fact that in vitro antiviral activity ofBMS-378806 was not altered significantly in thepresence of 40% human serum [4]. This is also incontrast with many protease inhibitors (PI) andnon-nucleoside reverse transcriptase inhibitors(NNRTI) that are highly protein bound [16].

Table 1. Bi-directional permeability of BMS-378806 in Caco-2 cells

Pc (nm/s)a

Study No. A-to-Bb B-to-Ac Efflux ratio

1d 56� 2.1 185� 27 3.32d 48� 8.7 135� 11 2.83d 51� 5.0 208� 12 4.14d 49� 7.0 193� 8.0 3.9with 50 mm Verapamil4 147� 10 153� 10 1.0

a n ¼ 3 in each study.b The apical pH value for studies 1, 2, 3 and 4 was 6.5, 6.5, 7.4 and 7.4, respectively.c The basolateral pH value was 7.4 for all studies.d The A-to-B Pc is significantly different from the B-to-A Pc (p50.05).



Microsomal stability

Figure 2 depicts the percentage of compoundremaining-time profiles of BMS-378806 afterincubations in human, rat, dog and monkey livermicrosomes. The rank order of the species, basedon the predicted hepatic intrinsic clearance fromliver microsomes, was monkey > rat > human >dog (Table 2). A good in vitro–in vivo correlationexisted between the observed and predictedclearance values in animal species (Table 2). Thehuman clearance of BMS-378806 predicted frommicrosomes was 9.2 ml/min/kg (46% of thehepatic blood flow), indicating an intermediateclearance.

Inhibition of human cytochrome P450

Unlike many of the approved PIs and NNRTIs[17,18], BMS-378806 exhibited weak potential of

CYP inhibition. The IC50 values for CYP1A2, 2C9,2C19, 2D6, 3A4 were >100, >100, 23, 20, 39–81 mm,respectively. No significant time-dependent in-hibition was observed in the inhibition ofCYP2C19 and 2D6, two isoforms for whichBMS-378806 had the lowest IC50 values.

The data, in conjunction with potentially lowtherapeutic concentrations (protein binding ad-justed EC50 ¼ 230 nm), suggest that BMS-378806may have a low potential of altering the clearanceof other drugs that are metabolized by CYPenzymes.

Reaction phenotyping in human liver microsomes

With the use of CYP isoform-selective chemicalinhibitors, BMS-378806 (1 mm) was found to bemetabolized by multiple CYP isoforms in HLM.CYP1A2, 2D6 and 3A4 each accounted for31%–36% of the metabolism of the compound(Table 3), suggesting that the elimination of BMS-378806 may not be affected significantly by drugsthat inhibit one of the enzymatic pathways.

Pharmacokinetics in rats

The pharmacokinetic parameters of BMS-378806following i.v. (1 and 5 mg/kg), p.o. (5 and 25 mg/kg), and intraportal (0.85 mg/kg) administrationto rats are summarized in Table 4. Following i.v.administration, the plasma concentrations ofBMS-378806 displayed a mono-exponential de-cline, with a terminal T1/2 of 0.30 h (Figure 3a).The Vss value in rats (0.55 l/kg) was lower thanthe volume of total body water (0.67 l/kg) butgreater than the volume of extracellular fluid(0.30 l/kg) [9], suggesting extravascular distribu-tion. The CLtot,b was 59% of the hepatic bloodflow, indicating an intermediate clearance in rats.

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30 35

HumanMonkeyDogRat

%C

ompo

und

Rem

aini

ng

Time (min)

Figure 2. Oxidative metabolism of BMS-378806 in rat, dog,monkey and human liver microsomes

Table 2. In vitro–in vivo correlation in the clearance of BMS-378806

Parameter Rat Dog Monkey Human

Turnover rate constant of oxidative metabolismin liver microsomesa (� 10�3 1/min/nmol CYP)

9.4� 0.26 1.2� 0.20 8.7� 1.6 13� 1.7

Predicted hepatic intrinsic clearance (ml/min/kg) 40 4.8 53 17Predicted hepatic blood clearance (ml/min/kg) 26 4.2 24 9.2Observed hepatic blood clearance (ml/min/kg) 31 4.8b 11b }

a n ¼ 3.b Total body blood clearance.


Z. YANG ET AL.394

At the 5 mg/kg i.v. dose which was administeredin the brain uptake study, plasma pharmacoki-netic parameters were similar to those at 1 mg/kg (Table 4), indicating linear pharmacokineticsbetween two doses. The brain penetration, how-ever, was low for BMS-378806, with a brain-to-plasma AUC ratio of 0.06 (Figure 4).

The oral bioavailability of BMS-378806 at 5 and25 mg/kg doses was 19� 5.1% and 27� 2.7%,respectively (Table 4). The corresponding Cmax

was 0.22� 0.086 and 1.8� 0.30 mm, respectively.The AUC and Cmax values were not significantlydifferent between two doses when comparedafter dose normalization (p>0.05). Absorptionwas prolonged, with an apparent terminal T1/2

after p.o. administration 7-fold longer than thatafter i.v. dosing (p50.05, Figure 3a and Table 4).The fa in rats was estimated to be 35%–49%,indicating incomplete oral absorption.

In BDC rats at a 25 mg/kg p.o. dose, about5.6% and 0.52% of dose were excreted unchanged

in the urine and bile, respectively. Comparedwith intact animals at 25 mg/kg, the oral bioa-vailability, Cmax, and AUC were significantlyreduced (p5 0.05, Table 4). However, there wasno statistically significant difference in the MRTand terminal T1/2 between BDC and intact rats. Itis possible that bile secretions in the intactanimals may help to solubilize the compoundin the gastrointestinal tract, thus enhancing theabsorption of BMS-378806. The renal clearance inrats was estimated be 9.9 ml/min/kg, which wassignificantly higher than the product of the freefraction and glomerular filtration rate in the rat(1.9 ml/min/kg), indicating that the compoundis actively secreted in the kidney.

In the intraportal infusion study, the intrapor-tal bioavailability of BMS-378806 was 55% (Table4), with a terminal T1/2 similar to that after i.v.dosing (Figure 5). The hepatic blood clearancein rats was estimated to be 31 ml/min/kg, whichis similar to the non-renal blood clearance

Table 3. Inhibition of BMS-378806 metabolism in human liver microsomes by chemical inhibitors

Inhibitor Inhibitor concentration (mm) CYP isoform inhibited % inhibition vs control ðn ¼ 3Þ

Furafylline 20 1A2 31� 8.5a

Sulfaphenazole 30 2C9 6.6� 15Quinidine 10 2D6 32� 8.4a

Ketoconazole 3 3A4 36� 6.0a

a Significantly different from zero (p50.05).

Table 4. Pharmacokinetic parameters of BMS-378806 in rats

Parameter i.v. Intraportal p.o.

Dose (mg/kg) 1 5a 0.85 5 25 25-BDC ratsBioavailability (%) } } 55� 3.3 19� 5.1 27� 2.7c 18� 0.80c

ka (1/h) } } } 0.28� 0.071 0.26� 0.054 0.18� 0.059Cmax (mm) } } 1.1� 0.11 0.22� 0.086 1.8� 0.30c 1.1� 0.31c

Tmax (h) } } 0.67� 0.0 2.7� 1.2 1.5� 0.50 1.5� 0.87AUC0�1 (mm.h) 1.3� 0.075 6.2 0.61� 0.052 1.3� 0.59 8.9� 1.5c 6.0� 0.36c

CLtot (ml/min/kg) 32� 1.8 33 } } } }

CLtot,b (ml/min/kg) 42 43 } } } }

Vss (l/kg) 0.55� 0.023 0.66 } } } }

T1/2 (h) 0.30� 0.032b 0.27 0.33� 0.067 2.1� 0.56b 2.5� 1.4b 4.7� 0.77b

MRT (h) 0.30� 0.014 0.33 0.67� 0.018 4.1� 0.98 4.3� 0.95 6.4� 2.2

a Composite data.b Significantly different between i.v. and p.o. routes (p50.05).c Significantly different between intact and BDC rats at 25 mg/kg (p50.05).



(29 ml/min/kg) determined from the BDC ratstudy, suggesting that the non-renal clearance inrats is primarily due to the hepatic clearance.Furthermore, the hepatic blood clearance pre-dicted from microsomal data was about 26 ml/min/kg (Table 2), which is similar to the hepaticblood clearance observed in vivo. This along withthe insignificant biliary clearance found in ratssuggests that the rat hepatic clearance of BMS-378806 is largely due to oxidative metabolismmediated by CYP enzymes.

10

100

1000

104

0 2 4 6 8 10

IV (1 mg/kg)PO (5 mg/kg)PO (25 mg/kg)

Pla

sma

Con

cent

rati

on (

nM)

10

100

1000

104

Pla

sma

Con

cent

rati

on (

nM)

1

10

100

1000

104

105

Pla

sma

Con

cent

rati

on (

nM)

Time (hr)

0 2 4 6 8 10Time (hr)

IV (0.67 mg/kg)

PO (3.35 mg/kg)

0 5 10 15 20 25

IV (0.67 mg/kg)PO (3.35 mg/kg)

Time (hr)

(a)

(b)

(c)

Figure 3. Plasma concentration-time profiles of BMS-378806after i.v. and p.o. administration to rats (a), dogs (b) andmonkeys (c). In rats, plasma samples were taken up to 8 h. Indogs and monkeys, plasma samples were taken up to 24 h(data not shown are below LLQ)

10

100

1000

104

105

0 0.5 1 1.5 2 2.5

Plasma

Brain

Con

cent

rati

on (

nM)

Time (hr)

Figure 4. Plasma and brain concentration-time profiles ofBMS-378806 after i.v. dosing of 5 mg/kg to rats

10

1

100

1000

104

0 0.5 1 1.5 2 32.5Time (hr)

Intraportal (0.85 mg/kg)IV (1 mg/kg)

Pla

sma

Con

cent

rati

on (

nM)

Figure 5. Plasma concentration-time profiles of BMS-378806after intraportal infusion and an i.v. bolus to rats


Z. YANG ET AL.396

Pharmacokinetics in dogs

The pharmacokinetic parameters of BMS-378806following i.v. (0.67 mg/kg) and p.o. (3.4 mg/kg)administration to dogs are summarized inTable 5. After i.v. administration, the plasmaconcentrations of BMS-378806 exhibited a bi-exponential decline, with a terminal T1/2 of 1.2 h(Figure 3b). The CLtot,b was 14% of the hepaticblood flow, indicating a low clearance in dogs.Similar to rats, the Vss in dogs (0.47 l/kg) waslower than the volume of total body water(0.60 l/kg) but greater than the volume of theextracellular fluid (0.28 l/kg) [9].

In dogs, BMS-378806 was rapidly absorbed,with an oral bioavailability of 77� 1.1% (Table5). The average Tmax was 1 h, with an averageCmax of 7.4 mm. The oral plasma concentration-time profile was parallel to that after i.v. dosing.The fa in dogs was 89%, indicating complete oralabsorption.

Pharmacokinetics in monkeys

Table 5 also summarizes the pharmacokineticparameters of BMS-378806 following i.v.(0.67 mg/kg) and p.o. (3.4 mg/kg) administrationto monkeys. After i.v. administration, the plasmaconcentrations of BMS-378806 displayed a bi-exponential decline, with a terminal T1/2 of 0.90 h(Figure 3c). The CLtot,b was low (25% of thehepatic blood flow). Similar to the other twospecies, the Vss in monkeys (0.39 l/kg) was

greater than the volume of extracellular fluid(0.21 l/kg) but less than the volume of total bodywater (0.69 l/kg) [9].

The oral bioavailability of BMS-378806 inmonkeys was 24� 11% (Table 5). The averageTmax was 1.5 h, with an average Cmax of 0.51 mm.Similar to rats, BMS-378806 exhibited prolongedoral absorption in monkeys; the apparent term-inal half-life after p.o. administration was 7-foldlonger than that after i.v. dosing (p50.05, Figure3c and Table 5). Consistent with the findings inrats, the fa of BMS-378806 in monkeys was 32%,suggesting incomplete oral absorption.

Discussion

BMS-378806 represents a novel class of HIV-1inhibitors that block the attachment of viralgp120 to cellular CD4 receptors, the first step ofHIV-1 entry into cells. Various preclinical phar-macokinetic and in vitro permeability and meta-bolism studies were conducted on BMS-378806 inorder to provide a comprehensive profile of thisprototypic compound. Knowledge of these al-lowed us to take multiple and mechanisticapproaches to predict human pharmacokineticparameters and profiles of BMS-378806 andbenchmark its properties with marketed anti-HIV agents. The efforts are critical to the leadoptimization of a discovery program, so that a

Table 5. Pharmacokinetic parameters of BMS-378806 in dogs and monkeys

Dog Monkey

Parameter i.v. p.o. i.v. p.o.

Dose (mg/kg) 0.67 3.4 0.67 3.4Bioavailability (%) } 77� 1.1 } 24� 11ka (1/h) } 1.3� 0.29 } 0.15� 0.026Cmax (mm) } 7.4� 1.0 } 0.51� 0.18Tmax (h) } 1.0� 0.50 } 1.5� 0.50AUC0�1 (mm.h) 4.8� 0.42 19� 1.4 2.7� 0.28 3.2� 1.1CLtot (ml/min/kg) 5.7� 0.48 } 10� 1.1 }

CLtot,b (ml/min/kg) 4.8 } 11 }

Vss (l/kg) 0.47� 0.065 } 0.39� 0.13 }

T1/2 (h) 1.2� 0.063 1.4� 0.18 0.90� 0.22a 6.5� 1.2a

MRT (h) 1.4� 0.11 2.2� 0.32 0.63� 0.17 7.6� 1.0

a Significantly different between i.v. and p.o. routes (p50.05).



novel anti-HIV agent could be identified that isorally bioavailable, with an adequate half-life foronce-daily dosing and low potential for drug–drug interactions.

In this work, PEG400 was used as a vehicle inboth i.v. and p.o. administration. Several lines ofevidence suggest that PEG400 does not signifi-cantly affect the pharmacokinetics of BMS-378806. First of all, the in vivo clearance observedwas consistent with the predicted clearance fromliver microsomes. Secondly, the hepatic clearancedetermined from the intraportal infusion studies,where 100% PEG400 was used as a solutionformulation, was similar to that predicted fromliver microsomes and the non-renal clearancedetermined from the BDC rat study. Lastly, theapparent terminal half-life of BMS-378806 in rats,when given orally in a PEG400 solution formula-tion, was not significantly different from thatgiven in an aqueous suspension (data notshown). Collectively, these data suggest thatPEG400 does not significantly affect the pharma-cokinetics of BMS-378806 in vivo.

In the present study, a mechanistic approachwas developed to predict the human oralbioavailability of BMS-378806 by understandingthe individual processes (i.e. absorption and liverfirst-pass metabolism) that govern oral bioavail-ability from in vitro and animal data. In makingthe prediction of human oral bioavailability,minimal gut wall first-pass metabolism wasassumed for BMS-378806. The assumption ap-peared to be reasonable in animal species. Indogs and monkeys, the total body clearance waslow, thus any significant first-pass gut wallmetabolism would not be expected. In rats, thenon-renal clearance was primarily due to thehepatic clearance, also indicating minimal gutwall metabolism.

To assess absorption potential in humans,Caco-2 cells were employed to study the intest-inal membrane permeability of BMS-378806. Aninitial concentration of 200 mm was chosen in thestudy based on the assumption that a 100 mgdose ðMW ¼ 400Þ and 1 l of intestinal fluid inhumans would result in a concentration of 250 mmin the intestinal lumen. In the present study, theefflux ratio, measured as the ratio of B-to-A Pc

relative to A-to-B Pc, was 3.5. Such a differencedisappeared in the presence of 50 mm verapamil.

Verapamil is a known P-gp substrate/inhibitor[19]. In addition to P-gp, multidrug resistance-associated protein 2 (MRP2) and breast cancerresistance protein (BCRP) are also major effluxtransporters expressed at the apical side of theintestine [20]. Recent studies by Prime-Chapmanet al. [21] showed that verapamil at 100 mm didnot inhibit the apical efflux transport of calcein inCaco-2 cells mediated by MRP2, suggesting nointeractions between MRP2 and verapamil. Also,verapamil did not appear to be a BCRP substrateor inhibitor [22,23]. These results suggest that P-gp is likely involved in transporting BMS-378806from the B to A direction, thus potentiallylimiting the oral absorption of BMS-378806 inhumans. In addition, P-gp may contribute topoor brain penetration of BMS-378806 and isinvolved in the active secretion of the compoundin the kidney.

In this study, metoprolol, a completely ab-sorbed standard (100% absorbed in humans) [12],had an A-to-B Pc value of 132 nm/s andmannitol, a poorly absorbed standard (16%absorbed in humans) [24], had an A-to-B Pc

value of 15 nm/s. Compounds that had similarA-to-B Pc values to BMS-378806 (51 nm/s) inCaco-2 cells showed a range of percentages ofdose absorbed in humans. For example, timolol,acebutalol and cimetidine had an A-to-B Pc valueof 47, 48 and 49 nm/s (data not shown),respectively, and were 71%, 46% and 73%absorbed in humans [12]. Interestingly, likeBMS-378806, timolol, acebutalol and cimetidineare also subject to P-gp mediated efflux transportin the intestine [25–27]. Given these similarities,the percentage of dose absorbed in humans forBMS-378806, based on the A-to-B Caco-2 Pc

value, is predicted to be in the range 40%–80%.In animals, the percentage of dose absorbed for

BMS-378806 varied among the species studied. Itwas 35%–49%, 89% and 32% in rats, dogs andmonkeys, respectively. Consistent with the dif-ference in the extent of absorption, the rate ofabsorption was also different across the species.The absorption rate constant ka was 0.18–0.28 and0.15 h�1 in rats and monkeys (Tables 4 and 5),respectively, whereas it was 1.3 h�1 in dogs (Table5). The species difference in the oral absorption ofBMS-378806 may result from differences inparacellular pathways in the species. He et al.


Z. YANG ET AL.398

[28] showed that polyethylene glycol and D-peptides, which were absorbed via the paracel-lular pathway, had significantly higher oralbioavailability in dogs than in rats, presumablydue to a larger pore size and greater frequency ofpores in dogs. In a series of papers published byChiou et al. [29–31], they demonstrated that bothrats and monkeys are better predictors of humanabsorption than dogs, with rats and monkeyshaving a nearly linear correlation to human oral

absorption. Therefore, it is reasonable to assumethat the extent and rate of absorption in rats andmonkeys, which are comparable between the twospecies, are good predictors of oral absorption ofBMS-378806 in humans. Furthermore, the aver-age percentage of dose absorbed in rats andmonkeys (39%) appeared to be in agreement withthe low end value predicted from Caco-2 Pc(40%). Consistently, both Caco-2 and in vivo datapinpoint that future lead optimizations shouldfocus on improving the oral absorption of thisseries of compounds and Caco-2 cells are asuitable screening model for this purpose.

To assess the liver first-pass metabolism inhumans, in vitro–in vivo correlation in clearancewas established first using liver microsomal data.A substrate concentration of 1 mm was selected forin vitro incubations. The selection of this concen-tration was justified as judged by a first-orderdecline in parent drug disappearance (Figure 2),which allows for determination of the in vitrointrinsic clearance, and by existence of in vitro–in vivo correlation in clearance in animal species(within 2-fold predictions, Table 2). The humanhepatic clearance predicted from liver micro-somes was 9.2 ml/min/kg, which is higher thanthe human clearance predicted from four differ-ent allometric scaling methods (1.0–4.7 ml/min/kg, Figure 7a). Therefore, human liver micro-somal data provided a conservative prediction ofthe percentage of dose escaping the liver first-pass extraction (fh) in humans, which was about54%. Consequently, the oral bioavailability ofBMS-378806 in humans, calculated as fa � fhð39% � 54%Þ, was predicted to be 20%.

In order to assess further the developability ofBMS-378806 and to guide the lead optimizationof additional candidates in the series, the com-plex Dedrick plot method was used to predict thehuman pharmacokinetic profiles of BMS-378806following i.v. and p.o. administration. Figure 7ashows the complex Dedrick plot of plasmaconcentration-time profiles following i.v. admin-istration to rats, dogs and monkeys. The datapoints in all three species were well fitted with atwo-compartment model, from which human i.v.pharmacokinetic profiles of BMS-378806 werederived (Figure 7b). The predicted human CLtot,Vss and T1/2 from the Dedrick plot method was2.5 ml/min/kg, 0.33 l/kg and 2.2 h, respectively.

0.8

1

1.2

1.4

1.6

1.8

-1 -0.50 0.51 1.52

Log CLtot = 1.25 + 0.556*Log BW, r = 0.993

Lo

g C

Lto

t (m

L/m

in)

Log BW (kg)

Rat

Monkey

Dog

Predicted human CLtot = 3.3 mL/min/kg

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1 1.5

Log Vss = -0.304 + 0.931*Log BW, r = 0.998

Lo

g V

ss (

L)

Log BW (kg)

Predicted human Vss = 0.37 L/kgRat

Monkey

Dog

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

-1 -0.50 0.51 1.5

Log t1/2 = 1.48 + 0.368*Log BW, r = 1.00

Lo

g t

1/2

(min

)

Log BW (kg)

Predicted human T1/2= 2.4 hr

Rat

Monkey

Dog

(a)

(b)

(c)

Figure 6. Allometric scaling of pharmacokinetic parametersof BMS-378806 from animal data to humans. (a) CLtot; (b) Vss;(c) T1/2



Because the Dedrick plot method is allometric innature, the pharmacokinetic parameters pre-dicted are similar to those obtained from directallometric scaling of pharmacokinetic parameters(Figure 6).

To predict human oral pharmacokinetic pro-files, the Dedrick plot method was not used toscale up animal oral data to humans directlybecause of species-dependent oral absorption.Instead, the human oral pharmacokinetic profileof BMS-378806 was simulated from the predictedhuman i.v. data, with an oral bioavailability of20% predicted from the mechanistic approachdiscussed above. The absorption rate constant ka

used in the simulation was 0.2 h�1, an average of

rat and monkey data. Figure 7b shows thepredicted human oral profile of BMS-378806; itis apparent that a daily dose of 2 g would beneeded to cover the protein binding adjustedEC50 value (230 nm) over a 24-h period. Clearly,the projected human dose for BMS-378806 is toolarge to be practical for once-daily dosing.

In comparison with other anti-HIV agents, thehuman Vss predicted for BMS-378806 is compar-able to or slightly less than that of indinavir,ritonavir, delavirdine and nucleoside/nucleotidereverse transcriptase inhibitors (0.4–0.7 l/kg)[32,33], but significantly lower than that of otherPIs and NNRTIs [16]. The predicted human T1/2

of BMS-378806 is significantly shorter than that ofPIs and NNRTIs except indinavir, ritonavir andnelfinavir [16]. From a pharmacokinetic point ofview, these results along with projected humandose demonstrate a significant need to improvethe half-life of future drug candidates in theseries by further reducing clearance and enlar-ging the volume of distribution.

For a drug–drug interaction point of view,BMS-378806 exhibited a number of advantages,which is of great importance for anti-HIV agents.In vitro metabolism studies with BMS-378806demonstrated that the compound was metabo-lized by 3 CYP enzymes to approximately thesame extent. This is in contrast to many PIs andNNRTIs where elimination in humans is primar-ily mediated by CYP3A4 [17,18]. In rats, the renalclearance also contributed to about 30% of thetotal body clearance of BMS-378806. If these datareadily translate to humans, it would suggestthat the disposition of the compound might notbe altered significantly by concomitant adminis-tration of CYP inhibitors. Moreover, unlike manyPIs and NNRTIs that are potent inhibitors ofCYP3A4 [17,18], BMS-378806 did not inhibitmajor drug-metabolizing enzymes to a signifi-cant extent in a recombinant system. Collectively,these results demonstrate that BMS-378806 haslow potential for metabolism-based drug–druginteractions.

In summary, comprehensive pharmacokineticand in vitro permeability and metabolism studieswere conducted on BMS-378806 to understand itsabsorption and disposition kinetics in the pre-clinical setting. In order to guide lead optimiza-tion efforts in the series, multiple approaches

1

10

100

1000

104

105

1

10

100

1000

104

105

0 0.5 1 1.5 2 2.5 3.53

RatDogMonkeyFitted

Pla

sma

Con

c/(D

ose/

BW

0.93

)nM

/(m

g/kg

0.93

)

Time/BW0.37, hr/kg0.37

0 5 10 15 20 25

IVPO

Pla

sma

Con

c (n

M)

Time (hr)

(a)

(b)

Figure 7. Complex Dedrick plot of animal i.v. data (a) andprediction of human i.v. and p.o. profiles of BMS-378806 at ani.v. dose of 100 mg and a p.o. dose of 2 g (b)


Z. YANG ET AL.400

were employed to predict the human pharmaco-kinetic parameters and profiles of BMS-378806and to benchmark the compound with marketedanti-HIV agents. BMS-378806 clearly exhibited anumber of desirable pharmacokinetic propertiesincluding a low extent of protein binding and alow potential for drug–drug interactions. How-ever, from a pharmacokinetic point of view,future efforts in lead optimization need to focuson improving oral absorption and achieving anadequate half-life for once-daily dosing. It is alsodesirable to achieve high brain penetration. Thepresent studies also demonstrate the value andapproaches of understanding pharmacokineticproperties in the early stage of drug discovery.

Acknowledgements

The authors would like to thank the reviewersand Drs Scott Grossman, Punit Marathe andDonald Tweedie for their valuable suggestions tothe manuscript.

References

1. Fauci AS. HIV and AIDS: 20 years of science. Nat Med2003; 9: 839–843.

2. Pomerantz RJ, Horn DL. Twenty years of therapy for HIV-1 infection. Nat Med 2003; 9: 867–873.

3. Powderly WG. Limitations of current HIV therapies:opportunities for improvement. J Acquir Immune DeficSyndr 2003; 33: S7–S16.

4. Lin PF, Blair W, Wang T, et al. A small molecule HIV-1inhibitor that targets the HIV-1 envelope and inhibits CD4receptor binding. Proc Natl Acad Sci USA 2003; 100:11013–11018.

5. Wang T, Zhang X, Wallace OB, et al. Discovery of 4-benzoyl-1-[(4-methoxy-1H-pyrrolo[2,3-b]pyridin-3-yl)ox-oacetyl]-2-(R)-methylpiperazine (BMS-378806): A novelHIV-1 attachment inhibitor that interferes with CD4-gp120 interactions. J Med Chem 2003; 46: 4236–4239.

6. Iwatsubo T, Hirota N, Ooie T, et al. Prediction of in vivodrug metabolism in the human liver from in vitrometabolism data. Pharmacol Ther 1997; 73: 147–171.

7. Obach RS, Baxter JG, Liston TE, et al. The prediction ofhuman pharmacokinetic parameters from preclinical andin vitro metabolism data. J Pharmacol Exp Ther 1997; 283:46–58.

8. Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, BelilesRP. Physiological parameter values for physiologicallybased pharmacokinetic models. Toxicol Ind Health 1997; 13:407–484.

9. Davies B, Morris T. Physiological parameters in laboratoryanimals and humans. Pharm Res 1993; 10: 1093–1095.

10. Gibaldi M, Perrier D. Pharmacokinetics, 2nd edn. Marcel-Dekker: New York, 1982, 409–417.

11. van Kempen GMP, van Vilet LJ. Mean and variance ofratio estimators used in fluorescence ratio imaging.Cytometry 2000; 39: 300–305.

12. Benet LZ, Oie S, Schwartz JB. Design and optimization ofdosage regimens: pharmacokinetic data. In Goodman andGilman’s The Pharmacological Basis of Therapeutics, Good-man-Gilman JG, Hardman JG, Limbard LE, Molinoff PB,Ruddon RW (eds). McGraw-Hill: New York, 1996, 1707–1792.

13. Mahmood I. Allometric issues in drug development.J Pharm Sci 1999; 88: 1101–1106.

14. Chiou WL, Robbie G, Chung SM, Wu TC, Ma C.Correlation of plasma clearance of 54 extensively meta-bolized drugs between humans and rats: mean allometriccoefficient of 0.66. Pharm Res 1998; 15: 1474–1479.

15. Boxenbaum H, Ronfeld R. Interspecies pharmacokineticscaling and the Dedrick plots. Am J Physiol 1983; 245:R768–R774.

16. Boffito M, Back DJ, Blaschke TF, et al. Protein binding inantiretroviral therapies. AIDS Res Hum Retroviruses 2003;19: 825–835.

17. Williams GC, Sinko PJ. Oral absorption of the HIVprotease inhibitors: a current update. Adv Drug DelivRev 1999; 39: 211–238.

18. Smith PF, DiCenzo R, Morse GD. Clinical pharmacoki-netics of non-nucleoside reverse transcriptase inhibitors.Clin Pharmcokinet 2001; 40: 893–905.

19. Kim RB. Drugs as P-glycoprotein substrates, inhibitors,and inducers. Drug Metab Rev 2002; 34: 47–54.

20. Mizuno N, Niwa T, Yotsumoto Y, Sugiyama Y. Impact ofdrug transporter studies on drug discovery and develop-ment. Pharmacol Rev 2003; 55: 425–461.

21. Prime-Chapman HM, Fearn RA, Cooper AE, Vanessa M,Hirst BH. Differential multidrug resistance-associatedprotein 1 through 6 isoform expression and function inhuman intestinal epithelial Caco-2 cells. J Pharmacol ExpTher 2004; 311: 476–484.

22. Litman T, Brangi M, Hudson E, et al. The multidrug-resistance phenotype associated with overexpression ofthe new ABC half-transporter, MXR (ABCG2). J Cell Sci2000; 113: 2011–2021.

23. Oxvegy C, Litman T, Szakaes G, et al. Functionalcharacterization of the human multidrug transporter,ABCG2, expressed in insect cells. Biochem Biophys ResCommun 2001; 285: 111–117.

24. Ukabam SO, Cooper BT. Small intestinal permeability tomannitol, lactulose, and polyethylene glycol 400 in celiacdisease. Dig Dis Sci 1984; 29: 809–816.

25. Terao T, Hisanaga E, Sai Y, Tamai I, Tsuji A. Activesecretion of drugs from the small intestinal epithelium inrats by P-glycoprotein functioning as an absorptionbarrier. J Pharm Pharmacol 1996; 48: 1083–1089.

26. Lentz KA, Polli JW, Wring SA, Humphreys JE, Polli JE.Influence of passive permeability on apparent P-glyco-protein kinetics. Pharm Res 2000; 17: 1456–1460.

27. Troutman MD, Thakker DR. Efflux ratio cannot assessP-glycoprotein-mediated attenuation of absorptive



transport: asymmetric effect of P-glycoprotein onabsorptive and secretory transport across Caco-2 cellmonolayers. Pharm Res 2003; 20: 1200–1209.

28. He YL, Murby S, Warhurst G, et al. Species differences insize discrimination in the paracellular pathway reflectedby oral bioavailability of poly(ethylene glycol) andD-peptides. J Pharm Sci 1998; 86: 626–633.

29. Chiou WL, Buehler PW. Comparison of oral absorptionand bioavailability of drugs between monkey and human.Pharm Res 2002; 19: 868–874.

30. Chiou WL, Barve A. Linear correlation of the fraction oforal dose absorbed of 64 drugs between humans and rats.Pharm Res 1998; 15: 1792–1795.

31. Chiou WL, Jeong HY, Chung SM, Wu TC. Evaluation ofusing dog as an animal model to study the fraction of oraldose absorbed of 43 drugs in humans. Pharm Res 2000; 17:135–140.

32. Tan X, Chu CK, Boudinot FD. Development and optimi-zation of anti-HIV nucleoside analogs and prodrugs: areview of their cellular pharmacology, structure activityrelationships and pharmacokinetics. Adv Drug Deliv Rev1999; 39: 117–151.

33. Fung HB, Stone EA, Piacenti FG. Tenofovir disoproxilfumarate: a nucleotide reverse transcriptase inhibitorfor the treatment of HIV infection. Clin Ther 2002; 24:1515–1548.


Z. YANG ET AL.402

Preclinical pharmacokinetics of a novel HIV-1 attachment inhibitor BMS-378806 and prediction of its...

Documents

Transcript of Preclinical pharmacokinetics of a novel HIV-1 attachment inhibitor BMS-378806 and prediction of its...