Pharmacokinetic/pharmacodynamic modelling of NSAIDs in a model of reversible inflammation in the cat

Transcription

1 British Journl of Phrmcology (25) 146, & 25 Nture Pulishing Group All rights reserved /5 $3. Phrmcokinetic/phrmcodynmic modelling of NSAIDs in model of reversile inflmmtion in the ct 1,3 Jerome M. Girudel, 1 Armelle Diquelou, 1 Vlerie Lroute, 2 Peter Lees & *,1 Pierre-Louis Toutin 1 UMR 181 de Physiopthologie et Toxicologie Expérimentles INRA/ENVT,Ecole Ntionle Ve te rinire de Toulouse,23 chemin des Cpelles,BP 87614,3176 Toulouse Cedex 3,Frnce nd 2 Deprtment of Veterinry Bsic Sciences,Royl Veterinry College, Hwkshed Cmpus,North Mymms,Htfield,Hertfordshire AL9 7TA Keywords: Arevitions: 1 Dt on the reltionships etween plsm concentrtion nd nlgesic nd nti-inflmmtory effects of NSAIDs re limited ecuse most inflmmtion models do not permit phrmcokinetic/ phrmcodynmic (PK/PD) modelling to e redily performed. 2 In this study, kolin-induced inflmmtion model in the ct ws evluted for pre-clinicl chrcteriztion of the phrmcodynmic profiles of NSAIDs (determintion of efficcy,potency, sensitivity (tht is the slope of the concentrtion effect reltionship) nd durtion of drug response), using meloxicm s proe rticle. 3 Indirect response PK/PD models descried the time course nd mgnitude of responses produced y.3 mg kg 1 meloxicm dministered sucutneously. For endpoints for which spontneous recovery from inflmmtion ws superimposed on drug response, PK/PD model with timedependent K in ws used to llow for the spontneous chnges of the inflmmtion with time. 4 The selected endpoints were suitle for studying simultneously the nlgesic,nti-inflmmtory nd ntipyretic effects of meloxicm,llowing comprison of reltive potencies for these effects. Men7s.d. IC 5 or EC 5 vlues (ng ml 1 ) were (ody temperture), (locomotion vrile), (pin score), (lmeness score) nd (skin temperture difference). Corresponding men times7s.d. of pek responses (h) were , ,5.275., nd ,respectively. 5 As the phrmcokinetic profiles of meloxicm in cts nd humns re similr,simultions of severl dosge regimens in the ct provided pre-clinicl sis,illustrting the vlue of the ct model for predicting clinicl dose regimen for evlution in mn. The predicted loding doses (mg kg 1 )of meloxicm in the ct producing 7% of the mximum ttinle responses were.29 (ody temperture),.32 (lmeness score),.33 (overll locomotion vrile),.36 (pin score) nd.5 (skin temperture difference). The vlues re similr to or somewht greter thn the cliniclly recommended doses oth in cts (.3 mg kg 1 ) nd humns ( mg,tht is,etween.1 nd.3 mg kg 1 ). 6 These findings indicte the potentil vlue of the ct s lortory model,nd of PK/PD modelling pproch in ssisting NSAID development progrms in nimls nd humns. British Journl of Phrmcology (25) 146, doi:1.138/sj.jp.76372; pulished online 22 August 25 PK/PD modelling; NSAID; ct; reversile inflmmtion; meloxicm; kolin COX,cyclooxygense; COX-2,cyclooxygense-2 isoform; HPLC,high-performnce liquid chromtogrphy; i.v., intrvenous; NSAID,nonsteroidl nti-inflmmtory drug; PK/PD,phrmcokinetic/phrmcodynmic; s.c., sucutneous; UV,ultrviolet Introduction In vivo phrmcokinetic/phrmcodynmic (PK/PD) modelling is incresingly ccepted s powerful pproch for determining phrmcodynmic prmeters,nd thus for selecting effective nd sfe dosge regimens for clinicl use. However,despite lrge ody of scientific literture on nonsteroidl nti-inflmmtory drug (NSAID) phrmcokinetics nd phrmcodynmics,reltively few pre-clinicl studies hve ttempted to model lood or plsm concentrtion time profiles with the time course of NSAID *Author for correspondence; E-mil: pl.toutin@envt.fr 3 Current ddress: Novrtis Snte Animle S.A.S.,14 Bd Richelieu BP 43,F Rueil Mlmison cedex,frnce. effects (Toutin et l.,1994; Grndos-Soto et l.,1995; Lndoni & Lees,1995; Lndoni et l.,1995; Torres-Lopez et l.,1997; Flores-Murriet et l.,1998; Jos et l.,21; Toutin et l.,21; Lees,23). PK/PD modelling pproches permit the in vivo determintion of numericl vlues for the three pivotl phrmcodynmic prmeters of drug,nmely efficcy,potency nd sensitivity (tht is the slope of the concentrtion effect reltionship). These prmeters llow prediction of the mgnitude nd time course of drug effect for ny formultion,route of dministrtion or dosge regimen,provided corresponding phrmcokinetic dt re ville (Toutin et l.,21; Toutin,22). In ddition,if drug plsm concentrtions required to produce given degree of phrmcologicl effect re similr in nimls

2 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 643 nd mn (Levy,1993),PK/PD modelling in non-humn species offers vlule pproch to dosge regimen prediction for humn use. Currently,the ppliction of PK/PD principles to determintion of the pre-clinicl profile of NSAIDs is limited y the vilility of vlidted niml models nd the inility of existing models to provide cliniclly relevnt endpoints of drug response. Most of the inflmmtion models tht hve een developed in rodents nd dogs (e.g. crrgeenn-induced pw oedem,uric cid-induced rthritis) re too short-lsting to permit n ccurte determintion of the complete drug concentrtion effect reltionship. This prtilly ccounts for the fct tht dose-titrtion pproches,sed on pre-chllenge NSAID dministrtion,hve een preferred. A further drwck of rodent models is tht they do not redily permit mesurement of cliniclly relevnt endpoints (e.g. lmeness scoring). On the other hnd,irritnts such s Freund s complete djuvnt result in sustined nd reltively stle inflmmtory response in the dog (Botrel et l.,1994; Toutin et l.,1994; 21),llowing mesurement of relevnt nd sensitive endpoints (e.g. verticl force pplied y the hind lim, lmeness score,etc.). However,for ethicl resons,the use of this irreversile inflmmtion model my e questioned. Consequently,n cceptle inflmmtion model should e neither too severe nor irreversile,ut should e sufficiently long lsting to llow n ccurte evlution of the three min responses to NSAIDs,nmely their nti-inflmmtory,ntipyretic nd nlgesic effects. The model should lso e developed in medium-sized species (e.g. dog,ct) to enle oth collection of cliniclly relevnt endpoints nd sequentil lood smpling for determintion of drug concentrtion in lood/plsm time profiles. Recent studies y our group hve shown tht sucutneous (s.c.) injection of kolin in ct s pw produces well-defined, reproducile nd reversile inflmmtory response (Girudel et l.,25). The signs of inflmmtion induced with 5 mg kolin were reltively constnt etween 2 nd 4 dys fter kolin injection,llowing dministrtion of the NSAID on dy 2. This model lso incorported () quntittive mesurement of ojective endpoints relevnt to therpeutic efficcy nd () the possiility of sequentil lood smpling. Bsed on sttisticl (reproduciility nd ccurcy of the mesurement) nd iologicl significnce,it ws nticipted tht ody temperture,git scoring,times for performnce of locomotion tests nd possily pw volume nd skin temperture might e suitle for PK/PD studies (Girudel et l.,25). Moreover,the model ws found to e suitle for studying simultneously the nlgesic,nti-inflmmtory nd ntipyretic effects of NSAIDs. This is of considerle interest,ecuse it is possile nd even likely tht concentrtion effect reltionships will differ for ech of these effects. Wheres nlgesi seems to occur in concentrtion-dependent mnner,ntiinflmmtory effect is more closely correlted with the time of exposure to given drug (Wlker,1995). Such differences nd the chrcteristics of the dose effect reltionship (efficcy, potency nd slope,for exmple, ll or nothing response compred with progressive grded response) cn e redily determined using pre-clinicl PK/PD modelling pproch. The im of the current study ws to ssess the vlue of the feline pw inflmmtion model for pre-clinicl chrcteriztion of the full phrmcologicl profile of NSAID nd then to pply it to prediction of dosge regimens for evlution in other niml species nd in mn. ws selected s reference NSAID,ecuse it hs een used extensively s therpeutic gent in oth nimls nd humns,nd its dosge regimen is now well estlished oth in mn nd the ct (Engelhrdt,1996; Turck et l.,1996; Busch et l.,1998; Slingsy & Wtermn-Person,2; Lscelles et l.,21). Methods Animls The study ws performed in six helthy Europen short-hired cts of oth sexes (three mles,three femles),mintined in temperture-controlled environment (2721C) nd either loose-housed in colony (etween experiments) or kept in individul stinless steel cges (during experimentl phses). Weights nd ges rnged from 3.2 to 4.6 kg nd 1.4 to 1.6 yers,respectively. They were fed ech evening fter the lst mesurements with commercil dry food. The housing nd experimentl fcilities t the Ntionl Veterinry School of Toulouse were pproved y the French Ministry of Agriculture,nd niml cre nd conduct of the study were performed in ccordnce with the Guide for the Cre nd Use of Lortory Animls (Institute of Lortory Animl Resources,Commission on Life Sciences,Ntionl Reserch Council,1996). Drugs nd chemicls Kolin (hydrted luminium silicte) ws purchsed from Sigm-Aldrich (Sint Quentin Fllvier,Frnce). The solution for injection of meloxicm (5 mg ml 1 ) ws otined from Boehringer Ingelheim Vetmedic GmH (Ingelheim/Rhein, Germny). Animl preprtion nd inflmmtion induction Animl nesthesi nd preprtion s well s inflmmtion induction were performed using stndrdized procedures (Girudel et l.,25). At 4 dys prior to inflmmtion induction,ech ct ws nesthetized to insert nd secure centrl venous ctheter in jugulr vein. Both hind pws were shved from the toes up to the hock joint nd mrked for skin temperture,withdrwl time nd pw volume mesurements (for detils,see Girudel et l.,25). On the dy of inflmmtion induction (dy ),ech ct ws re-nesthetized to fcilitte the s.c. injection of 1.75 ml sterile suspension of out 5 mg kolin. Endpoint mesurements All nimls were ccustomed to experimentl conditions nd trined for recording of the endpoints during 4 weeks efore the onset of the tril. During the experiment,the endpoints were recorded y the sme two trined opertors using stndrdized procedures (Girudel et l.,25). Briefly,the git ws scored with numericl rting scle (NRS) nd the corresponding lmeness score rnged from (no lmeness) to 5 (voidnce of ny contct of the ffected pw with the ground). Pin ws evluted s the time required y the ct to withdrw its pw fter stimultion with the rdint het

3 644 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model emitted from n nlgesi meter (Model 39,IITC Inc./Life Science,Woodlnd Hills,CA,U.S.A.). The niml ws confined for few min in Plexigls chmer dpted to the size of ct,nd plced on top of the glss pnel of the device. This method voids niml mnul restrint nd delegtes to the ct control of the durtion of the stimulus. Rdint het ws delivered s em of focussed light of fixed intensity (2% of mximl intensity). The stimultion ws stopped s soon s the niml strted to withdrw its pw. Only cler withdrwl or lterl trnsltory movement of the pw ws ccepted s n ccurte cutoff point. The withdrwl time ws trnsformed into pin score (%) ccording to the following eqution: R ¼ 1 MT BT CT BT where MT (s) is the withdrwl time fter meloxicm dministrtion,bt (s) is the withdrwl time shortly efore injection of meloxicm nd CT (s) is the men withdrwl time recorded for the sme pw efore kolin injection. Locomotion ws quntified s the time required y the ct to perform series of tests (niml scending or descending wooden stircse nd niml creeping under grid). Using n equivlent of eqution (1),the three locomotion times were expressed s percentges,therey tking ccount of oth seline nd control (pretretment) vlues. The resulting locomotion vriles were then nlysed collectively y summing the corresponding percentges to otin n overll locomotion vrile. Hind pw skin temperture ws mesured with n infrred thermometer (Rynger s MX4 t Rytek s,fisher Biolock Scientific,Illkirch,Frnce); the difference etween the temperture of the injected nd the control pw ws determined. Finlly,the volume of the inflmed hind pw ws mesured with wter displcement technique nd ody (rectl) temperture ws recorded with n electronic thermometer. Experimentl design The six cts were rndomly llocted to two groups of nimls. In the first period of the study,the right pw of the cts ws injected with kolin. A jugulr ctheter ws lso plced in the cts dministered with meloxicm on dy 2 (group 1); the other cts (group 2) were shm-prepred (sme nesthesi nd ndge s treted cts,ut no ctheter insertion in control cts) nd did not receive meloxicm. After wshout intervl of t lest 5 weeks,the left pws were injected with kolin (second period) nd the tretments were crossed over. ws injected s.c. 47 h fter kolin injection t dose rte of.3 mg kg 1 ody weight,which is the mnufcturer s recommended dose for cts. Blood smples (1 ml) were tken efore nd 5,2,4,6 nd 9 min nd 2,3, 6,9,12,24,36,48,72 nd 96 h fter meloxicm dministrtion. Smples were collected in heprinized tues nd centrifuged within 15 min t 41C,3 g for 1 min. Plsm smples were frozen t 21C until nlysed for meloxicm concentrtion y high-performnce liquid chromtogrphy (HPLC). Endpoints were mesured efore (dys 3 nd 2) nd up to 5 dys fter kolin injection (dys 1 5). On dy 2, mesurements for the treted ct were performed 1.5 h efore nd.75,1.5,3,5,8 nd 12 h fter meloxicm dministrtion. Mesurements were lso tken 1,3,4 nd 5 dys fter kolin ð1þ injection nd on dy 7 for control cts nd dy 8 for treted cts to ssess recovery from the induced inflmmtion. At the end of the study,no niml exhiited ny persisting clinicl sequele nd ll were re-homed. Anlysis of meloxicm in plsm Plsm smples were nlyzed y vlidted HPLC procedure using ultrviolet (UV) detection. Briefly,internl stndrd (piroxicm) nd meloxicm were extrcted from plsm y solid-phse extrction. The HPLC pprtus comprised pump system equipped with n utomtic injector nd UV detector (36 nm). Seprtion ws chieved y reverse-phse column (Hypersil BDS C 18,3mm,15 2. mm),using gurd column (Hypersil BDS C 18,1 2. mm). The moile phse ws 4 : 6 mixture of 1% cetic cid : methnol t flow rte of.12 ml min 1. Under these conditions,meloxicm nd piroxicm were eluted t retention times of 9.9 nd 7.4 min, respectively. The method ws liner over the clirtion rnge of 1 15 ng ml 1,using weighted liner regression model. Within-dy nd dy-to-dy coefficients of vrition were less thn 9% nd the ccurcy rnged from 96 to 99%,indicting n pproprite precision nd ccurcy for the nlyticl method. The lck of interference from endogenous compounds ws verified on lnk plsm from untreted cts,estlishing the specificity of the method. The vlidted limit of quntifiction ws 1 ng ml 1. Dt nlysis Phrmcokinetic nd PK/PD modelling were performed y lest-squres regression nlysis using WinNonlin Professionl softwre (WinNonlin s,version 4..1,Phrsight Corportion, Mountin View,CA,U.S.A.). plsm concentrtions were fitted for ech ct using n eqution corresponding to two-comprtmentl model with first-order sorption nd lg time: CðtÞ¼ ðy 1 þ Y 2 Þ expð kðt lgþþ ð2þ þ Y 1 expð l 1 ðt lgþþ þ Y 2 expð l 2 ðt lgþþ where C(t) (ng ml 1 ) is the meloxicm plsm concentrtion t time t, Y 1, Y 2 (ng ml 1 ) re the coefficients of the exponentil terms, l 1, l 2 (h 1 ) re the exponents of the exponentil terms, k (h 1 ) is the first-order rte constnt of sorption nd lg (h) is the lg time for sorption. The dt were weighted y the inverse of the squred-fitted vlue nd goodness of fit ws determined using the Akike Informtion Criterion (AIC) (Ymok et l.,1978) nd y visul inspection of the fittings nd the residuls. Individul niml phrmcokinetic prmeters of meloxicm were then used s constnts in the integrted PK/PD model. The PK/PD reltionships were descried using indirect phrmcodynmic response models (Dynek et l.,1993). In these models,the mesured response (R) is ssumed to result from fctors controlling either the input or the dissiption of the mesured response: dr=dt ¼ K in K out R ð3þ where dr/dt is the rte of chnge of the response over time, K in represents the zero-order rte constnt for production of the response nd K out the first-order rte constnt for loss of the response.

4 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 645 For the skin temperture difference (mesured s the difference etween the temperture of the inflmed pw nd the control pw),dt were descried with the following model: " dr=dt ¼ K in 1 I # mxcðtþ n IC n 5 þ K out R ð4þ Cn ðtþ where K in (1Ch 1 ) represents the zero-order rte constnt for production of the therml inflmmtory response nd K out (h 1 ) is the first-order rte constnt for dissiption of the therml component of inflmmtion. The drug effect ws descried with Hill eqution in which C (t) (ng ml 1 ) is the meloxicm plsm concentrtion t time t,ic 5 (ng ml 1 ) is the meloxicm plsm concentrtion producing hlf the mximum drug effect (i.e.,hlf I mx ), I mx is the mximum possile inhiition nd n is the exponent expressing the sigmoidicity of the meloxicm concentrtion effect reltionship. The control vlue for the response (R, 1C) corresponds to the stedy skin temperture plteu chieved 2 dys fter kolin injection; it is determined y oth K in nd K out from the eqution: R ¼ K in ð5þ K out The time course of the nlgesic effect ws est descried using the model represented y eqution (4). The derived IC 5 ws therefore the meloxicm plsm concentrtion,producing hlf of the mximum inhiition of the fctors controlling pin production. The ntipyretic effect ws descried s the consequence of stimultion of the fctors controlling het loss (Dynek et l.,1993; Toutin et l.,21) y the eqution: " dr=dt ¼ K out R K out 1 þ ð # R R mx 1ÞCðtÞ n EC n 5 þ R ð6þ Cn ðtþ where K out (h 1 ) represents the first-order rte constnt for het loss, R (1C) is the stedy ody temperture efore meloxicm dministrtion on dy 2, R mx (1C) is the mximum response ttriuted to the drug (i.e.,the minimum ody temperture predicted fter meloxicm dministrtion) nd EC 5 (ng ml 1 ) is the meloxicm plsm concentrtion producing hlf the mximum stimultion of het loss. C (t) nd n re s descried in eqution (4). For those endpoints for which spontneous recovery from inflmmtion (ssessed from the dt otined in the sence of meloxicm) ws superimposed on drug response (lmeness score nd locomotion vriles), K in ws no longer considered s prmeter,ut s time-dependent vrile. A gmm function (Wise,1985) ws used to descrie the time course of K in nd eqution (3) ws therefore trnsformed into n eqution tht tkes into ccount the spontneous time chnges in inflmmtion intensity: dr=dt ¼ At expð tþ K out R ð7þ where A (response unit h 2 ), (no unit) nd (h 1 ) re the prmeters of the gmm function descriing the time course nd intensity of the inflmmtion production rte (s ssessed y the lmeness score nd the overll locomotion vrile), K out (h 1 ) is the first-order rte constnt for loss of the response nd t is the time fter kolin injection. Estimtes of A,, nd K out were otined y fitting this model to the dt otined during the control period (tht is when the cts were injected with kolin ut did not receive meloxicm). These vlues were then used s initil estimtes for fitting the dt collected in the period during which meloxicm ws dministered on dy 2 (referred to s the meloxicm period). For these dt, comined model incorporting drug effect ws used to ccount for the meloxicm effect s well s the confounding spontneous chnges in the inflmmtory condition. In this model, production of the inflmmtory response is governed y the time-dependent gmm function,which is multiplied y n inhiition function fter meloxicm dministrtion: " # dr=dt ¼ At expð tþ 1 I mxc n ðt delyþ IC n 5 þ Cn ðt delyþ K out R where dely is the time etween kolin injection nd meloxicm dministrtion, C (t dely) is given y eqution (2) if t4dely or is equl to if tpdely,nd other prmeters re s indicted in eqution (4). Using men phrmcokinetic nd phrmcodynmic prmeters,simultions were performed to predict the time course of meloxicm response for ech endpoint for doses rnging from.1 to 1 mg kg 1. Becuse the distriution of individul phrmcokinetic prmeters ws close to log-normlity,the geometric men ws clculted. Dt from ll individul nimls were used to compute men phrmcokinetic prmeters ut,only those cts with cceptle phrmcodynmic fittings (five nimls for most PD prmeters) were tken into ccount for the clcultion of verge (rithmetic men) phrmcodynmic prmeters. For ech simultion, summry prmeter,the verge drug response,ws used to chrcterize the time course of ech endpoint fter meloxicm dministrtion. All drug effects were considered to hve wned 24 h fter meloxicm dministrtion nd ll clcultions were therefore performed y considering the 24-h period following s.c. injection of the drug. Simulted verge drug responses were clculted for ll endpoints s follows: E ¼ AUR AUR X 24 where E (1C,% or without unit for the lmeness score) is the verge drug response over the first 24 h fter meloxicm dministrtion,aur (1C h,% h or h) is the re under the time response profile for the sme time period ut without meloxicm dministrtion nd AUR X (1C h,% h or h) is the sme re ut fter dministrtion of X mg kg 1 meloxicm. The verge drug response ws then expressed s percentge of the mximum possile verge drug response, tht is the response otinle for very high s.c. dose of meloxicm (1 mg kg 1 ) (Figure 1). The 5,7 nd 9% effective doses were lso clculted for ech endpoint. Sttisticl nlysis Results re presented s men dt nd stndrd devition (s.d.). For hlf-lives,the hrmonic men nd corresponding 95% confidence intervl were clculted. ð8þ ð9þ

5 646 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model Lmeness score Results AAR 1 mg/kg 24 hours AAR.3 mg/kg mg/kg.3 mg/kg 1 mg/kg Figure 1 Method for computtion of the drug response s percentge of the mximum possile drug response. Time course of the simulted men lmeness score (from to 5,with indicting no lmeness nd 5 mximum lmeness) fter injection of 5 mg kolin into the hind pw of six cts under control conditions ( mg kg 1 ), fter.3 mg kg 1 nd 1 mg kg 1 meloxicm dministrtion. 1 mg kg 1 ws the dose ssumed to give the mximum possile response. AAR :3 mgkg 1 (h) (drk grey re) nd AAR 1 mg kg 1 (h) (drk nd light grey re) re the res ove the time response profiles during the 24 h following the dministrtion of.3 nd 1 mg kg 1 meloxicm,respectively. AAR :3 mgkg 1 (h) represents pproximtely 64% of AAR 1 mg kg 1 (h); therefore single s.c. dministrtion of.3 mg kg 1 meloxicm is ssocited with drug response tht corresponds to 64% of the mximum drug response tht cn e chieved with s.c. dministrtion of meloxicm. The iexponentil decline in meloxicm plsm concentrtion fter s.c. dministrtion of.3 mg kg 1 in ech of six cts ws interpreted s two-comprtment open model with first-order sorption nd short lg phse (Figure 2). The pprent totl plsm clernce (Cl F 1,tht is clernce scled y iovilility) ws low ( ml kg 1 h 1 ) nd the pprent stedy-stte volume of distriution (Vss F 1,volume of distriution scled y iovilility) ws reltively smll ( ml kg 1 ). Pek meloxicm plsm concentrtion (C mx ¼ ng ml 1 ) ws chieved reltively rpidly (T mx ¼ h) nd the lg time for sorption ws very smll (lg ¼ h). The hrmonic men nd 95% confidence intervls for the terminl hlf-life (h) nd the pprent hlf-life of sorption (h) were 37. (27.1; 58.3) nd.42 (.25; 1.18),respectively. Both mgnitude nd time course of the inflmmtory response were mesured s chnges in the vlues of the selected endpoints (Figure 3). Figure 3 demonstrtes tht the inflmmtory response developed rpidly fter kolin injection. Within 1 2 dys,vlues of these endpoints (with the exception of pw volume) reched mximum,efore returning to sl levels (pre-inflmmtion vlues) pproximtely 1 week fter kolin dministrtion. For ody temperture,lmeness score nd the overll locomotion vrile,vlues reched mximum on the morning of dy 2 nd n pprent stedy-stte inflmmtory condition ws mintined for pproximtely 2 further dys (Figure 3c e). Skin temperture of the inflmed pw followed the sme time course s the lmeness score nd the overll locomotion vrile. Circdin chnges in ech mesurement were oserved: for skin temperture this ws noted for oth pws, plsm concentrtion (ng/ml) Oserved vlues Fitted vlues Time (h) Figure 2 Oserved (men7s.d.) nd fitted meloxicm plsm concentrtion (ng ml 1 ) vs time (h) fter s.c. dministrtion t nominl dose of.3 mg kg 1 in six cts. ut ws more pronounced for the control thn the inflmed pw (Figure 3). Figure 4 illustrtes the effect on the time course of the inflmmtory response of.3 mg kg 1 dose of meloxicm dministered 2 dys fter the inflmmtion induction. Inflmed pw volumes for the control nd the meloxicm periods were very similr on dys 1 3,with slightly,ut not significntly,lower volumes for the treted nimls therefter. For ll remining endpoints,on the other hnd,significnt inhiitory effects were consistently produced y meloxicm. After meloxicm dministrtion,the men time of mximum decrese in ody temperture occurred t 5.6 h (rnge 5 8 h) (Figure 4d nd Tle 1) nd the men time of pek response for the lmeness score ws lso 5.6 h (rnge 3 12 h) (Figure 4c nd Tle 2). Skin temperture differences followed the sme time course,with mximum decrese (rnge 2.4 to 7.31C) otined pproximtely 4.5 h fter meloxicm dministrtion. The time course of the pin score correlted well with other surrogte clinicl responses,notly skin nd ody temperture (Figure 4,d nd f),displying n verge decrese of 228% compred to pretretment vlues. The mximl decrese in pin score occurred t 5.2 h (rnge h) fter meloxicm dministrtion (Tle 1 nd Figure 4f). Becuse the onset of the meloxicm response occurred reltively rpidly,the totl durtion of the response ws considered to correspond to the time etween drug dministrtion nd disppernce of drug response. This rnged from 12 to 32 h,with slightly smller vlues for skin temperture differences nd n verge durtion of drug response for ll the other endpoints of pproximtely 24 h. For ll endpoints,the response time profile nd the meloxicm concentrtion time profile were not in phse. Figure 5 illustrtes the dely etween the mximum decrese in the overll locomotion vrile (occurring pproximtely 12 h fter meloxicm dministrtion) nd the mximum meloxicm plsm concentrtion (t 2 h fter drug injection). The ppliction of indirect response models together with sigmoid E mx model for drug effect delt dequtely with this temporl dely etween phrmcokinetic nd phrmcodynmic dt. Figure 6 illustrtes for representtive ct the fittings for three endpoints nlyzed with the clssicl indirect response models,nd Tle 1 gives the corresponding men phrmcodynmic prmeters for skin temperture difference,

6 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model Inflmed pw Control pw 4 Inflmed pw Control pw Pw volume (ml) Skin temperture ( C) c d Lmeness score Body temperture ( C) e Overll locomotion vrile (%) f Pin score (%) Figure 3 ( f) Time course of oserved vlues (men7s.d.) of six endpoints following injection of 5 mg kolin in the hind pw of six cts. Corresponding vlues (open circles) for the control (noninflmed) pw re given for pw volume () nd skin temperture (). For lmeness score (c), score of indictes no lmeness nd score of 5 mximum lmeness. For ody temperture (d) return to pre-tretment vlues ws otined 5 dys fter meloxicm dministrtion. The overll locomotion vrile (e) corresponds to composite vrile otined from climing,descending nd creeping tests. For pin score (f),vlues for the contrlterl pw re not given ecuse these were influenced y the pin experienced on the inflmed pw. pin score nd ody temperture. For the lmeness score nd the overll locomotion vrile,the time course of the response ws fitted with PK/PD model,tking the spontneous development of the inflmmtory response into ccount (Figure 7 nd ). For these endpoints,mximl inhiition (I mx or efficcy) rnged from 64 to 75% (Tle 2). On the other hnd,for ody temperture R mx ws similr to norml temperture,indicting tht complete suppression of hyperthermi cn e chieved with meloxicm. For nlgesi, I mx ws greter thn % ecuse pin thresholds for cts

7 648 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 45 Control 4 Control Pw volume (ml) Skin temperture ( C) c 5 Control d 41 Control Lmeness score Body temperture ( C) e Overll locomotion vrile (%) Control f Pin score (%) Control Figure 4 ( f) Time course of oserved vlues (men7s.d.) for six endpoints (pw volume (),skin temperture (),lmeness score (c),ody temperture (d),overll locomotion vrile (e) nd pin score (f)) following injection of 5 mg kolin into the hind pw of six cts nd dministrtion of.3 mg kg 1 meloxicm on dy 2 (rrow). Men oserved vlues for the control period (open circle) re lso presented s comprtive curve. Considering the 24 h following meloxicm dministrtion (grey re),cler-cut effects were oserved for ll endpoints except pw volume (). treted with meloxicm were consistently higher thn those recorded on the sme cts efore inflmmtion induction (twofold increse on verge) (Tle 1). sensitivity (tht is,the slope of the concentrtion effect reltionship in the Hill eqution) ws reltively high for ll mesured endpoints, rnging from 6.1 to 1.,thus illustrting tht there is n lmost ll or nothing effect on the input or output processes of the indirect response models. Another feture of the dt

8 Tle 1 Oserved nd fitted phrmcodynmic prmeters descriing meloxicm nti-inflmmtory (skin temperture difference),nlgesic (pin score) nd ntipyretic (ody temperture) effects fter single s.c. dministrtion of nominl dose of.3 mg kg 1 meloxicm in five cts End point T min (h) R min (1C or%) J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 649 K in (1Ch 1 or % h 1 ) R (1C) R mx (1C) K out (h 1 ) I mx (%) IC 5 or EC 5 (ng ml 1 ) n (no unit) Skin temperture NA NA difference (1C) Pin score (%) NA NA Body temperture (1C) NA NA Vlues re men7s.d. NA: not pplicle in the model. T min (h) nd R min (sme unit s endpoint,i.e.,1c or %) re oserved vlues for the time of occurrence of the pek response nd the mximum meloxicm response expressed s decrese in the endpoint vlue,respectively. Dt were fitted using n indirect response model in which meloxicm produces its phrmcodynmic effect y inhiiting the fctors controlling K in. The rte of chnge of the response over time nd the significnces of K in, K out,ic 5,EC 5, I mx nd n (upper limit fixed t 1) re given y eqution (4). For ody temperture,nother indirect response model ws used descriing drug effect s stimulting het loss (see eqution (6)). Tle 2 Oserved nd fitted phrmcodynmic prmeters descriing improvement of locomotion fter single dministrtion of nominl dose of.3 mg kg 1 meloxicm in five cts End point T min (h) R min (no unit or %) A (h 2 or % h 2 ) Alph (no unit) (h 1 ) K out (h 1 ) I mx (%) IC 5 (ng ml 1 ) n (no unit) Lmeness score (no unit) Overll locomotion vrile (%) Vlues re men7s.d. T min (h) nd R min (sme unit s endpoint,i.e.,no unit (lmeness score) or % (overll locomotion vrile)) re oserved vlues for the time of occurrence of the pek response nd the mximum meloxicm response expressed s decrese in the endpoint vlue,respectively. Locomotion times were trnsformed to percentges (using n equivlent of eqution (1)) nd dded to otin n overll locomotion vrile. Dt were fitted with n integrted PK/PD model,tking into ccount the spontneous evolution of the inflmmtion (see eqution (8)). The rte of chnge of the response over time s well s the mening of the prmeters of the gmm function (A, nd ) nd K out,ic 5, I mx nd n re given y eqution (8). Overll locomotion vrile (%) Locomotion vrile concentrtion Time fter meloxicm dministrtion (h) Figure 5 Time courses of oserved meloxicm plsm concentrtion (ng ml 1 ) nd the overll locomotion vrile (%) in representtive ct fter dministrtion of.3 mg kg 1 meloxicm. The locomotion times corresponding to the climing,descending nd creeping tests were trnsformed into percentges nd summed to compute n overll locomotion vrile descriing the ct s locomotion. ws the reltively high inter-niml vriility for most phrmcodynmic prmeters (Tles 1 nd 2). However, coefficients of vrition for IC 5 nd EC 5 were reltively low,rnging from 16% (ody temperture) to 35% (skin temperture difference). Men vlues of IC 5 nd EC concentrtion (ng/ml) rnged from 777 (ody temperture) to 1298 ng ml 1 (skin temperture difference). Men phrmcokinetic nd phrmcodynmic prmeters were used to simulte dosge regimens rnging from.1 to 1 mg kg 1 meloxicm. Simultions for ody temperture nd pin score re illustrted in Figure 8. As consequence of the steepness of the concentrtion effect reltionship,it is predicted tht the lowest dose (.1 mg kg 1 ) would exert no nlgesic effect,.2 mg kg 1 would provide only wek nd trnsient effect,while.5 mg kg 1 would provide good efficcy for lmost 24 h. Tle 3 presents the simulted verge drug responses for rnge of dosge regimens (single-dose dministrtions of.1 1 mg kg 1 meloxicm) nd the doses producing 5,7 nd 9% of the mximum possile verge drug response tht cn e otined with single s.c. dministrtion of meloxicm. Discussion The im of this study ws to investigte the vlue of new inflmmtion model for pre-clinicl chrcteriztion of NSAIDs using PK/PD modelling pproch,with meloxicm s test drug. Informtion on the reltionship etween plsm concentrtion nd nlgesic,nti-inflmmtory nd ntipyretic effects re very limited for this clss of compound,despite oth mechnistic nd clinicl relevnce (Lndoni & Lees,

9 65 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model Body temperture ( C) Pin score (%) Skin temperture difference ( C) c Skin temperture concentrtion Time fter meloxicm dministrtion (h) Body temperture 4 concentrtion Time fter meloxicm dministrtion (h) Pin score concentrtion Time fter meloxicm dministrtion (h) Figure 6 ( c) Time courses of oserved nd fitted meloxicm plsm concentrtion (ng ml 1 ) nd () skin temperture difference (1C),() ody temperture (1C) nd (c) pin score (%) in representtive niml fter dministrtion of.3 mg kg 1 meloxicm. An indirect response model descriing drug effect s inhiiting inflmmtion production ws used to fit the time course of skin temperture difference. 1995; Lndoni et l.,1995; Brown et l.,1998; Flores-Murriet et l.,1998). In the present feline model,oth of these spects of PK/PD modelling were successfully documented y using two types of endpoints: (i) those exploring specific component of the inflmmtory response nd hving mechnistic interest,such s centrl nd locl hyperthermi (ody nd skin temperture),hyperlgesi (pin score) nd oedem (pw volume) nd (ii) hyrid endpoints hving direct clinicl relevnce nd reflecting oth the pin experienced y the ct nd functionl impirment due to oedem (lmeness score nd overll locomotion vrile). Idelly,the chnge in the vlue of n endpoint produced y drug should e relted solely to its phrmcodynmic properties. However,the time course of the phrmcologicl response my lso e influenced y inflmmtion induction, progression nd then recovery. This confounding fctor concentrtion (ng/ml) concentrtion (ng/ml) concentrtion (ng/ml) Overll locomotion vrile (%) Lmeness score Time fter kolin dministrtion (dys) Time fter kolin dministrtion (dys) Figure 7 (,) Time courses of oserved nd fitted () overll locomotion vrile nd () lmeness score in the sme niml s in Figure 6. ws dministered t dy 2 (rrows). An indirect response model descriing drug effect s inhiiting inflmmtion production comined with model of inflmmtion progression ws used to fit the time course of the locomotion vrile. deserves specil ttention for drugs such s meloxicm,with reltively long durtion of effect. In the present study,for exmple,mesurements of the overll locomotion vrile nd lmeness score,otined fter the drug response hd wned, did not give results similr to those otined efore drug dministrtion. Therefore, mthemticl model descriing the inflmmtory response to kolin ws used to tke progression of the inflmmtory process into ccount. This model showed good ility to descrie the time course of the endpoints during the control period (no meloxicm dministrtion). Moreover,comining this model with the PK/PD model predicted ccurtely oth time course nd extent of drug responses,nd llowed estimtion of individul phrmcodynmic prmeters of efficcy,potency nd sensitivity. Body temperture,pin nd pw volume re routinely used erly in the course of new drug development for mechnistic purposes to chrcterize the phrmcologicl profile of NSAIDs (Riendeu et l.,21). In this study,pw volume

10 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 651 ws the only endpoint for which the test drug hd no evident effect on the experimentlly induced inflmmtion. A similr finding hs een oserved in other studies in which the NSAID ws dministered fter inflmmtion induction (Holspple & Yim,1984; Engelhrdt et l.,1995; Zhng et l.,1997; Girudel et l.,25). Possile explntions re tht oedem response involves mny other meditors thn prostglndins nd tht the time required for the clernce of fluid lredy Body temperture ( C) Pin score (%) mg/kg.2 mg/kg.3 mg/kg.4 mg/kg.5 mg/kg 1 mg/kg Time fter meloxicm dministrtion (h) mg/kg.2 mg/kg.3 mg/kg.4 mg/kg.5 mg/kg 1 mg/kg Time fter meloxicm dministrtion (h) Figure 8 (,) Simulted time profiles of () ody temperture nd () pin score for single-dose dministrtions of.1,.2,.3,.4,.5 nd 1 mg kg 1 meloxicm. ccumulted in tissues is considerle (Lees,23; Girudel et l.,25). One of the key requirements of n inflmmtion model is sustined hyperlgesic response tht cn e reversed y NSAID dministrtion (Zhng et l.,1997; Dirig et l.,1998). This requirement ws met in the present study, complete reversl of hyperlgesi eing chieved pproximtely 1 h fter meloxicm dministrtion. In ddition,the threshold for pin senstion t lter time points significntly exceeded the levels oserved efore inflmmtion induction. This hypolgesic stte,in which the niml withstnds longer exposure to het stimulus,hs recently een descried y others (Frncischi et l.,22). The fct tht ntinociception nd nlgesi fter NSAID dministrtion my involve mechnisms of ction other thn COX inhiition (Sndrini et l.,22; Pinrdi et l., 23; Koppert et l.,24) my provide n explntion for the hypolgesic stte oserved in the present study nd y Frncischi et l. (22). As in previous studies (Shirot et l.,1984),skin temperture ws shown to e good mrker of inflmmtion development nd suppression. The finding in the present study of circdin rhythm of skin temperture chnge tht ws more pronounced in control pws extends erlier oservtions otined in the rt (Kessler et l.,1983). After meloxicm dministrtion, smll decrese in skin temperture of the control pw ws oserved in some cts,nd this ws proly relted to the decrese in ody temperture (Girudel et l.,25). Skin temperture difference,on the other hnd,ws not confounded y the decrese in ody temperture nd this derived vrile could therefore e conveniently used for PK/PD modelling. The time courses of ll endpoints nlysed using PK/PD modelling were well descried y the selected indirect response models nd permitted evlution of two importnt drug phrmcodynmic properties,nmely sensitivity nd potency. Sensitivity,which gives informtion on the rnge of efficcious concentrtions,ws high in this study,indicting tht there is proly threshold concentrtion close to the IC 5 elow which no drug effect occurs. This explins why no drug effect persisted eyond 3 h,despite reltively high meloxicm plsm concentrtions (2 7 ng ml 1 ) up to 72 h fter dosing. Clssiclly,potency for nlgesic nd ntipyretic effects of NSAIDs is higher thn potency for their ntiinflmmtory effect (Toutin et l.,21; Lees,23). In the present study,however,the IC 5 computed for the pin score ws of mgnitude similr to potencies otined for other endpoints. This might e explined y the fct tht the potency computed for the pin score reflects not only the ntihyper- Tle 3 Simulted verge drug responses (expressed s percentges of the mximum possile verge drug response, i.e.,the response otined with 1 mg kg 1 meloxicm) for different end points nd doses rnging from.1 to 1 mg kg 1 Averge response (from to %) dose (mg kg 1 ) End point E.1 E.2 E.3 E.4 E.5 E 1 ED 5 ED 7 ED 9 Skin temperture difference Pin score Body temperture Lmeness score Overll locomotion vrile E X (%) represents the predicted verge drug response for single s.c. dose of X mg kg 1 meloxicm. ED Y (mg kg 1 ) is the meloxicm dose producing Y% of the mximum possile verge drug response tht cn e otined with single s.c. dministrtion of meloxicm.

11 652 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model lgesic effect of meloxicm (possily due to inhiition of prostglndin synthesis) ut lso its hypolgesic effect. As the in vivo determined potencies clculted in the present study re likely to reflect meloxicm effects on prostglndin synthesis,it is relevnt to compre these IC 5 s with the potency for COX-2 inhiition otined for meloxicm in feline whole-lood ssys (Girudel et l.,25). These potencies were of the sme mgnitude (71 ng ml 1 for the in vitro determined potency for COX-2 inhiition nd 883 nd 911 ng ml 1 for the inhiition of the pin nd lmeness production,respectively),which is consistent with COX-2 inhiition eing the mjor mechnism of ction of meloxicm. Regrding clinicl relevnce,the present study showed tht endpoints,such s ody nd skin temperture nd pin score, could e lso used s surrogte endpoints of NSAID efficcy. These endpoints demonstrted good metrologicl performnces,ut,eqully importnt,they displyed time course similr to the more cliniclly relted endpoints (lmeness score nd overll locomotion vrile),resulting in durtion of meloxicm response tht ws very consistent for ll endpoints. In ddition,it ws shown tht the clinicl remission oserved on dys 4 nd 5 ws generlly greter fter meloxicm compred to control. Interestingly,this lso occurred when return to pre-dministrtion vlues ws otined 24 h fter meloxicm dministrtion,suggesting tht the susequent fster clinicl remission when nimls were treted with meloxicm might e due to some dditionl effect of the drug. This finding is potentilly of considerle clinicl significnce nd deserves further investigtion. Another feture of clinicl interest ws the high inter-niml vriility oserved for most phrmcodynmic nd disese prmeters. Similrly,ntiinflmmtory therpy in humns is lso chrcterized y mrked individul vrition in ptient responses (Levy,1998). Drug potencies nd sensitivities computed with indirect response models re not of direct clinicl ppliction,ut they re essentil when using preclinicl dt to predict effective dosge regimens. In the present study,these prmeters reflect drug effect on physiopthologicl mechnism (likely to e prostglndin synthesis for NSAIDs) nd re not genuine prmeters descriing drug response on the relevnt endpoints (e.g. lmeness). Therefore,other prmeters (ED 5,ED 7, ED 9 ) were clculted y simulting different dosge regimens nd this llowed chrcteriztion of the dose response reltionship of meloxicm for ech endpoint. Such simultions my e very helpful in ssisting clinicins to select dosge regimen not only in the species investigted ut lso in mn. This my e especilly relevnt for meloxicm,ecuse the phrmcokinetic profile in cts closely resemles tht otined in humns (Turck et l.,1996). In the only pulished phrmcokinetic study in cts (Scientific Discussion on Metcm 5 mg kg 1 Solution for Injection for Dogs nd Cts. EMEA CVMP/263/-Rev.3),it ws shown tht the iovilility from the s.c. route of dministrtion ws %. The clernce determined in this study (Cl F 1, 6. ml kg 1 h 1 ) cn therefore e compred to the i.v. clernce determined in mn (6.1 ml kg 1 h 1 for ody weight of 7 kg) (Turck et l.,1996). As these vlues re virtully identicl nd s drug potencies re commonly similr etween species (Levy, 1993; Busch et l.,1998),it is suggested tht the present results could lso e relevnt for predicting dosge regimen in mn. For NSAIDs with different phrmcokinetic profiles in cts nd humns,dose prediction must tke into ccount differences in drug clernce. In this study,it cn e concluded tht dose of.25.3 mg kg 1 (i.e., dose close to the ED 5 for ll the endpoints) might e n effective loding dose in the ct. This dose is very similr to the cliniclly recommended dose not only in cts (.2 or.3 mg kg 1 ) ut lso in humns ( mg,i.e., etween.1 nd.3 mg kg 1 for ody weight of 5 7 kg). This comprison demonstrtes the potentil usefulness of this preclinicl PK/PD modelling pproch for predicting dosge regimen,not only in the trget species ut lso in humns. In conclusion,the present investigtion provides evidence tht reversile nd ethicl model of inflmmtion in medium-sized species llowed mesurement of rnge of endpoints chrcterizing the nti-inflmmtory,nlgesic nd ntipyretic effect of NSAID. For these drug properties,the min phrmcodynmic prmeters (efficcy,potency nd sensitivity) nd time course of drug response were determined. This fcilitted estlishment of drug concentrtions nd dosge regimens tht my e used to pln dose-rnging studies in humn drug development. We thnk Joseph Mligoy nd Ndine Gutier for skilled technicl ssistnce. This study ws supported y Novrtis Animl Helth. References BOTREL, M.A., HAAK, T., LEGRAND, C., CONCORDET, D., CHEVALIER, R. & TOUTAIN, P.L. (1994). Quntittive evlution of n experimentl inflmmtion induced with Freund s complete djuvnt in dogs. J. Phrmcol. Toxicol. Methods, 32, BROWN, R.D., KEARNS, G.L. & WILSON, J.T. (1998). Integrted phrmcokinetic phrmcodynmic model for cetminophen, iuprofen,nd plceo ntipyresis in children. J. Phrmcol. Biophrm., 26, BUSCH, U., SCHMID, J., HEINZEL, G., SCHMAUS, H., BAIERL, J., HUBER, C. & ROTH, W. (1998). Phrmcokinetics of meloxicm in nimls nd the relevnce to humns. Drug Met. Dispos., 26, DAYNEKA, N.L., GARG, V. & JUSKO, W.J. (1993). Comprison of four sic models of indirect phrmcodynmic responses. J. Phrmcol. Kinet. Biophrm., 21, DIRIG, D.M., ISAKSON, P.C. & YAKSH, T.L. (1998). Effect of COX-1 nd COX-2 inhiition on induction nd mintennce of crrgeenn-evoked therml hyperlgesi in rts. J. Phrmcol. Exp. Ther., 285, ENGELHARDT, G. (1996). Phrmcology of meloxicm, new nonsteroidl nti-inflmmtory drug with n improved sfety profile through preferentil inhiition of COX-2. Br. J. Rheumtol., 35 (Suppl 1),4 12. ENGELHARDT, G., HOMMA, D., SCHLEGEL, K., UTZMANN, R. & SCHNITZLER, C. (1995). Anti-inflmmtory,nlgesic,ntipyretic nd relted properties of meloxicm, new non-steroidl ntiinflmmtory gent with fvourle gstrointestinl tolernce. Inflmm. Res., 44, FLORES-MURRIETA, F.J., KO, H.C., FLORES-ACEVEDO, D.M., LOPEZ-MUNOZ, F.J., JUSKO, W.J., SALE, M.E. & CASTANEDA- HERNANDEZ, G. (1998). Phrmcokinetic phrmcodynmic modelling of tolmetin ntinociceptive effect in the rt using n indirect response model: popultion pproch. J. Phrmcokinet. Biophrm., 26, FRANCISCHI, J.N., CHAVES, C.T., MOURA, A.C., LIMA, A.S., ROCHA, O.A., FERREIRA-ALVES, D.L. & BAKHLE, Y.S. (22). Selective inhiitors of cyclo-oxygense-2 (COX-2) induce hypolgesi in rt pw model of inflmmtion. Br.J.Phrmcol.,137,

12 J.M. Girudel et l PK/PD modelling of NSAIDs in ct model 653 GIRAUDEL, J.M., DIQUELOU, A., LEES, P. & TOUTAIN, P.L. (25). Development nd vlidtion of new model of inflmmtion in the ct nd selection of surrogte endpoints for testing nti-inflmmtory drugs. J. Vet. Phrmcol. Ther., 28, GIRAUDEL, J.M., TOUTAIN, P.L. & LEES, P. (25). Development of in vitro ssys for the evlution of cyclooxygense inhiitors nd ppliction for predicting the selectivity of NSAIDs in the ct. Am.J.Vet.Res.,66, GRANADOS-SOTO, V., LOPEZ-MUNOZ, F.J., HONG, E. & FLORES- MURRIETA, F.J. (1995). Reltionship etween phrmcokinetics nd the nlgesic effect of ketorolc in the rt. J. Phrmcol. Exp. Ther., 272, HOLSAPPLE, M.P. & YIM, G.K. (1984). Therpeutic reduction of ongoing crrgeenin-induced inflmmtion y lipoxygense,ut not cyclooxygense inhiitors. Inflmmtion, 8, JOSA, M., URIZAR, J.P., RAPADO, J., DIOS-VIEITEZ, C., CASTANEDA- HERNANDEZ, G., FLORES-MURRIETA, F., RENEDO, M.J. & TROCONIZ, I.F. (21). Phrmcokinetic/phrmcodynmic modelling of ntipyretic nd nti-inflmmtory effects of nproxen in the rt. J. Phrmcol. Exp. Ther., 297, KESSLER, F., SCHMIDT, K.L. & RUSCH, D. (1983). Thermometry in experimentl inflmmtion. I. Studies on the circdin rhythm of the pw temperture in helthy rts nd temperture ehvior in cute inflmmtion fter injection of kolin,crrgeenin nd dextrn. Z. Rheumtol., 42, KOPPERT, W., WEHRFRITZ, A., KORBER, N., SITTL, R., ALBRECHT, S., SCHUTTLER, J. & SCHMELZ, M. (24). The cyclooxygense isoenzyme inhiitors precoxi nd prcetmol reduce centrl hyperlgesi in humns. Pin, 18, LANDONI, M.F., CUNNINGHAM, F.M. & LEES, P. (1995). Phrmcokinetics nd phrmcodynmics of ketoprofen in clves pplying PK/PD modelling. J. Vet. Phrmcol. Ther., 18, LANDONI, M.F. & LEES, P. (1995). Comprison of the ntiinflmmtory ctions of flunixin nd ketoprofen in horses pplying PK/PD modelling. Equine Vet. J., 27, LASCELLES, B.D., HENDERSON, A.J. & HACKETT, I.J. (21). Evlution of the clinicl efficcy of meloxicm in cts with pinful locomotor disorders. J. Smll Anim. Prct., 42, LEES, P. (23). Phrmcology of drugs used to tret osteorthritis in veterinry prctice. Inflmmophrmcology, 11, LEVY, G. (1993). The cse for preclinicl phrmcodynmics. In: Integrtion of Phrmcokinetics, Phrmcodynmics nd Toxicokinetics in Rtionl Drug Development. ed. Ycoi,A.,Skelly,J.P., Shh,V.P. & Benet,L.Z. New York nd London: Plenum Press,pp LEVY, G. (1998). Predicting effective drug concentrtions for individul ptients. Determinnts of phrmcodynmic vriility. Clin. Phrmcokinet., 34, PINARDI, G., SIERRALTA, F. & MIRANDA, H.F. (23). Atropine reverses the ntinociception of nonsteroidl nti-inflmmtory drugs in the til-flick test of mice. Phrmcol. Biochem. Behv., 74, RIENDEAU, D., PERCIVAL, M.D., BRIDEAU, C., CHARLESON, S., DUBE, D., ETHIER, D., FALGUEYRET, J.P., FRIESEN, R.W., GORDON, R., GREIG, G., GUAY, J., MANCINI, J., OUELLET, M., WONG, E., XU, L., BOYCE, S., VISCO, D., GIRARD, Y., PRASIT, P., ZAMBONI, R., RODGER, I.W., GRESSER, M., FORD- HUTCHINSON, A.W., YOUNG, R.N. & CHAN, C.C. (21). Etoricoxi (MK-663): preclinicl profile nd comprison with other gents tht selectively inhiit cyclooxygense-2. J. Phrmcol. Exp. Ther., 296, SANDRINI, M., VITALE, G. & PINI, L.A. (22). Effect of rofecoxi on nociception nd the serotonin system in the rt rin. Inflmm. Res., 51, SHIROTA, H., KOBAYASHI, S., SHIOJIRI, H. & IGARASHI, T. (1984). Determintion of inflmed pw surfce temperture in rts. J. Phrmcol. Methods, 12, SLINGSBY, L.S. & WATERMAN-PEARSON, A.E. (2). Postopertive nlgesi in the ct fter ovriohysterectomy y use of crprofen, ketoprofen,meloxicm or tolfenmic cid. J. Smll Anim. Prct., 41, TORRES-LOPEZ, J.E., LOPEZ-MUNOZ, F.J., CASTANEDA-HERNANDEZ, G., FLORES-MURRIETA, F.J. & GRANADOS-SOTO, V. (1997). Phrmcokinetic phrmcodynmic modelling of the ntinociceptive effect of diclofenc in the rt. J. Phrmcol. Exp. Ther., 282, TOUTAIN, P.L. (22). Phrmcokinetic/phrmcodynmic integrtion in drug development nd dosge-regimen optimiztion for veterinry medicine. AAPS Phrm. Sci., 4,(rticle 38) ¼ ps443. TOUTAIN, P.L., AUTEFAGE, A., LEGRAND, C. & ALVINERIE, M. (1994). Plsm concentrtions nd therpeutic efficcy of phenylutzone nd flunixin meglumine in the horse: phrmcokinetic/ phrmcodynmic modelling. J. Vet. Phrmcol. Ther., 17, TOUTAIN, P.L., CESTER, C.C., HAAK, T. & LAROUTE, V. (21). A phrmcokinetic/phrmcodynmic pproch vs dose titrtion for the determintion of dosge regimen: the cse of nimesulide, Cox-2 selective nonsteroidl nti-inflmmtory drug in the dog. J. Vet. Phrmcol. Ther., 24, TURCK, D., ROTH, W. & BUSCH, U. (1996). A review of the clinicl phrmcokinetics of meloxicm. Br. J. Rheumtol., 35 (Suppl 1), WALKER, J.S. (1995). Phrmcokinetic phrmcodynmic correltions of nlgesics. In: Hndookof Phrmcokinetic/Phrmcodynmic Correltion. ed. Derendorf,H.,Hochhus,G. U.S.A.: CRC Press LLC,pp WISE, M.E. (1985). Negtive power functions of time in phrmcokinetics nd their implictions. J. Phrmcokinet. Biophrm., 13, YAMAOKA, K., NAKAGAWA, T. & UNO, T. (1978). Appliction of Akike s informtion criterion (AIC) in the evlution of liner phrmcokinetic equtions. J. Phrmcokinet. Biophrm., 6, ZHANG, Y., SHAFFER, A., PORTANOVA, J., SEIBERT, K. & ISAKSON, P.C. (1997). Inhiition of cyclooxygense-2 rpidly reverses inflmmtory hyperlgesi nd prostglndin E2 production. J. Phrmcol. Exp. Ther., 283, (Received My 13, 25 Revised July 13, 25 Accepted July 18, 25 Pulished online 22 August 25)