Mechanism of the anticoagulant effect of warfarin as evaluated in rabbits by selective depression of individual procoagulant vitamin K dependent clotting factors

Mechanism of the anticoagulant effect of

warfarin as evaluated in rabbits by selective

depression of individual procoagulant vitamin

K-dependent clotting factors.

A Zivelin, … , L V Rao, S I Rapaport

J Clin Invest. 1993;92(5):2131-2140. https://doi.org/10.1172/JCI116814.

We have evaluated the contribution of depression of individual procoagulant vitamin K-dependent clotting factors to the ability of warfarin to protect rabbits against tissue factor-induced coagulation. Mean activities of individual procoagulant factors were determined, in assays with rabbit substrates, for a group of rabbits achieving a protective degree of

anticoagulation with warfarin. Values were: factor VII, 12%; factor IX, 7%; factor X, 14%, and prothrombin, 13%. The effect upon tissue factor-induced coagulation of selective

immunodepletion of each factor to a comparable level was then evaluated.

Immunodepletion of plasma factor X or prothrombin, but not of factor VII or factor IX, protected otherwise normal rabbits against tissue factor-induced coagulation. Next, we determined the effect upon the protection in warfarin-treated rabbits of selectively restoring factor X or prothrombin before infusing tissue factor. When either factor was selectively restored, warfarin's protective effect was abolished. Moreover, selective restoration of

prothrombin sensitized warfarin-treated rabbits to coagulation more severe than observed in nontreated control rabbits. One may extrapolate from these data that depression of both factor X and prothrombin are required for warfarin's clinical antithrombotic efficacy and that depression of plasma prothrombin is particularly important.

Research Article

Find the latest version:

Mechanism of the

Anticoagulant

Effect of Warfarin

as

Evaluated

in Rabbits

by Selective Depression of Individual

Procoagulant Vitamin K-dependent Clotting Factors

Ariella Zivelin, L. Vijaya Mohan Rao, and Samuel 1. Rapaport

DepartmentsofMedicine and Pathology, University of California, SanDiego, California92093

Abstract

We have evaluatedthecontribution of depression of individual procoagulant vitamin K-dependent clotting factors to the abil-ity of warfarin to protect rabbits against tissue factor-induced coagulation. Mean activities of individual procoagulant factors were determined, in assays with rabbit substrates, for a group ofrabbitsachievingaprotective degree of anticoagulation with warfarin. Values were: factor VII, 12%; factor IX, 7%; factor X, 14%, and prothrombin, 13%. The effect upon tissue factor-in-ducedcoagulation of selective immunodepletion of each factor toacomparable level wasthenevaluated. Immunodepletion of plasmafactor X or prothrombin, but not of factor VII or factor IX, protected otherwise normal rabbits against tissue factor-induced coagulation. Next, we determined the effect upon the protection in warfarin-treated rabbits of selectively restoring factor X or prothrombinbefore infusing tissue factor. When either factor was selectively restored, warfarin's protective ef-fect was abolished. Moreover, selective restoration of prothrombin sensitized warfarin-treated rabbits to coagulation more severe than observed in nontreated control rabbits. One may extrapolate from these datathatdepressionofboth factor

X and prothrombin are required for warfarin's clinical anti-thromboticefficacyandthatdepressionof plasma prothrombin is particularly important. (J. Clin. Invest. 1993. 92:2131-2140.) Key words: tissue factor * intravascular coagulation . factorVII* factorX*prothrombin

Introduction

Warfarinand relatedvitamin K antagonists were used clini-cally for many years before it wasrecognized that they im-pairednotonlythesynthesis ofprocoagulant factors, factors VII, IX,X, andprothrombin, but alsothesynthesis ofanticoag-ulantfactors,protein C and protein S. Although warfarin's anti-thrombotic efficacy is established in a variety of clinical cir-cumstances(1),it is not clear why in these circumstances the effect of depressingprocoagulantfactorsoutweighs the effect of depressing anticoagulantfactors.

Addresscorrespondence to Dr. Samuel I. Rapaport, UCSD Medical

Center,MailCode8423,200 West Arbor Drive, San Diego, CA 92103. Ariella Zivelin's present address is Institute ofThrombosis and He-mostasis, Chaim Sheba Medical Center, Tel Hashomer, 52621, Israel;

L.VijayaMohan Rao's present address is Department of Biochemis-try, TheUniversityof Texas Health Center at Tyler, Tyler, TX 75710.

Receivedforpublication 24 March 1993 and in revisedform 3 June 1993.

In 1958, Siseetal. (2) observed that both hemorrhage and thrombosis in patients receivingavitamin K antagonistwere

moreclosely related to specific plasma prothrombin activity thantothe"prothrombin time,"atestparticularly sensitiveto

the effect of vitamin K antagonists upon plasma factor VII activity. Reports ofvenousthrombo-embolism in patients with hereditary factor VII deficiency (3, 4) also raised uncertaintyas

totheimportance ofdepressed factor VII activity for warfarin's antithrombotic efficacy. Recently, Furie et al. (5) reported that theplasma native prothrombin antigen level correlatedclosely with the presence of bleedingor thromboembolic complica-tionsin patients given warfarin. Moreover, Xietal.(6) have shown that adding purified prothrombin to pooled plasma frompatients givenavitamin K antagonist increased thrombin

generation whereas adding purified factors VII, IX, or X

didnot.

Only a few animal studies have been carried out on the mechanism ofthe antithrombotic effect of warfarin. Izakand Galewsky (7)reported in 1966 thattreatmentwith warfarin

failed toprevent intravascular coagulation in rabbits infused withacrude tissue factor (TF)1 preparation. Wessler and col-leagues (8-10), using a rabbit stasis thrombosis model,

re-ported that warfarin protected against thrombosisnotonly by depressing factor VII initially and other vitamin K-dependent procoagulant factors after 2 d but also byenhancing,after 6 d,

aninhibitory activity in plasma against factor Xa. Thereareno confirmatory studies in the literature of this last observation.

We report here studies inarabbit model system of the

con-tribution ofthe depression ofeach procoagulant vitamin K-de-pendentclotting factortowarfarin's functionas an anticoagu-lant agent. The data provide evidence for the importanceof depression of both plasma prothrombin activity and plasma factor X activity for the ability of warfarintoprotect rabbits against TF-induced intravascular coagulation.

Methods

Reagents

General. Sodium warfarin for intravenous administrationwas agift

from DuPontPharmaceuticals(Willimgton, DE).Rabbit anti-human

prothrombin antiserumand rabbitanti-human factor X antiserum werefromAmericanBioproducts (Parsippany,NJ). Rabbitplasmafor

protein purification was purchased from Pel-Freez Biologicals (Rogers,AR).

Reagentsforclottingassays. Rabbit brainthromboplastin (TF)

wasfrom Sigma Chemical Co. (St. Louis, MO). AutomatedAPTT was

from GeneralDiagnostics(MorrisPlains, NJ).KaolincephalinAPTT Reagent was from Helena Laboratories (Beaumont, TX). Bovine thrombin and humanfibrinogenreferenceplasmawerefrom Baxter Diagnostics (Deerfield, IL). Human substrate plasmas deficient ina 1.Abbreviationused in this paper: TF, tissue factor.

J.Clin. Invest.

© The AmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/93/11/2131/10 $2.00

specificclotting factor werefrom patientsseen attheUniversity of

California, San Diego (UCSD) MedicalCenter, fromGeorge King Bio-Medical (Overland Park, KS), or were prepared by

immu-noadsorbing normal human plasma.

Purified coagulation proteins

RabbitTF.Purified rabbit brainTFapoprotein( 11 )wasreconstituted

intophospholipidvesiclescontaining phosphatidylserineand

phospha-tidylcholinein aratio of4:6(12)and stored at-750C.

Rabbit factors VII, IX, X, andprothrombin. The initial steps of purificationofrabbit vitaminK-dependentclottingfactorswere

essen-tially the same asfor the humanproteins (13, 14),buttheelution profile fromaDEAE-Sephadex columndiffered. Rabbit factorIX was elutedseparately from theotherclotting factorsand waspurified to

homogeneityasdescribed earlier ( 15). Prothrombin, factor VII, and factor X, which overlappedonelution fromaDEAE-Sephadex

col-umn, wereappliedto a5X38-cmheparinagarose column(14). The columnwasdeveloped withaninitial buffer (0.02Msodium citrate,1 mMbenzamidine,pH7.5)toyield partially purified factorVII.Then, boundproteinswereelutedwithalinear saltgradient of0-0.6MNaCl

in0.02Mcitrate, 1 mMbenzamidine,pH 7.5(totalgradient volume, 2,000 ml). Thisseparatedprothrombin from factor X, yielding appar-entlyhomogeneousprothrombinandpartially purified factorX.

HumanprothrombinandfactorX.Humanprothrombinandfactor

X werepurifiedtoapparenthomogeneityasdescribed earlier ( 14).As measured inchromogenicassays, theprothrombin preparation

con-tained <0.01% thrombin and the factor X preparation contained

<0.01%factorXa(wt/wt).

Antibodies

to

rabbit

clotting

factors

Anti-factorVIIIgG. Partially purified rabbit factorVIIwas subjected

toSDS-PAGEona 10% polyacrylamideslabgel.Themajor stained protein band, which corresponded to the molecularweight of factor

VII,was electroeluted and used as theantigentoimmunizeagoat as

described earlier ( 15).Theantiserumwasheatinactivatedat560Cfor

30 min and the IgG fractionwasseparatedbyprecipitation at40%

ammoniumsulfate saturation followed by DEAE-Affi-Gelblue

chro-matography.TheIgGhad aspecificactivity of10-50inhibitor U/mg protein, bound stronglytorabbit factorVII on animmunoblot,and

hadnoinhibitory activity against rabbit factors IX,X, orprothrombin.

Forthisandsubsequentantibodies,oneinhibitor unitwasdefinedas theconcentrationof IgGin anequalpartmixture of IgGandrabbit plasmathatinhibited50%oftheclotting factor activityin themixture in 30-60min at room temperature.

Anti-factorIXIgG.AmAbagainst rabbit factorIX(mAb 170.1)

waspreparedby standardtechniquesandpurified byuseofamAb

purificationkit(Bio-RadLaboratories, Richmond, CA).mAb 170.1 wasofsubclassIgGl,had aninhibitory activity of200U/mg IgG,and boundonlytorabbit factorIX onan immunoblot of rabbit plasma. Monospecific polyclonal antibodies against rabbit factor IX were

raisedin a goat asdescribed earlier (15).

Anti-factorXIgG. mAbsagainst rabbit factorX were prepared usingstandardtechniques withpartially purifiedrabbitfactorX asan

antigenand the IgG waspurifiedasdescribed above. mAb 409 was used to prepare rabbitfactor X-deficientsubstrate plasma.Antibody

mAb 541,whichwasofsubclassIgG3 and had a titer of 75 U/mg IgG, wasinjectedintorabbits to immunodeplete plasmafactorXactivity.

Antiprothrombin IgG. Our purified rabbit prothrombin

prepara-tion lost most of itsactivityonovernightdialysisat4°Cinto Hepes buffer. SDS-PAGE analysis of thedialyzedsample revealed amajor proteinbandcorrespondingtoprethrombin 1. This band was electro-eluted from a Coomassie blue-stained SDS-PAGE slab gel and used as

antigentoraise amonospecificantiserumin a goat. The antiserum was

affinity purifiedon aprethrombin1 column that was prepared by cou-pling 300 mg of rabbit prethrombin 1 to 25 mlAffi-Gel15 as described in themanufacturer's (Bio-Rad Laboratories)technicalbulletin. About

100mlofantiserumwasprecipitatedat40% ammonium sulfate

satura-tion. Theprecipitatewasdissolved anddialyzed extensivelyinTBS,

and mixed from 4 htoovernightwith theimmobilized prethrombin1

at4VC. Unbound orweaklyboundproteinswerewashed with TBS

followedbyTBScontaining1 MNaCl.TheboundIgGwaseluted with

0.1Mglycine,pH2.4, intotesttubescontaining 0.1vol of 1.0 M Tris. TheelutedIgGwasconcentratedby precipitation with 80% ammo-nium sulfatesaturation, dissolved,anddialyzed extensivelyinHepes buffer.Theaffinity-purified IgGhadaspecific activityof 3inhibitory U/mgofprotein.

ControlIgGpreparations.Ananti-humanthyroglobulinIgGthat wasunreactiveagainst clotting factorswaspurified from ascitic fluid kindly providedby Dr. M. Zanetti (Department ofMedicine, UCSD). NonimmunegoatIgGwaspurifiedasdescribedabove from

commer-cial goatserum(SigmaChemicalCo.).

Rabbit-specific

deficiency substrate

plasmas

Rabbit substrateplasmasdeficientin factorVII, IX, or X were

pre-pared byimmunoaffinitychromatography againstgoatanti-factorVII IgG,againstamixture of anti-factorIXmAb 170.1and goat

anti-fac-torIXIgG,andagainstanti-factor X mAb409.In preparing the im-munoadsorbant columns, the IgGs were coupled to Affi-Gel 15, - 5

mgof IgG/ml of beads, accordingtothe manufacturer's instructions. Platelet-poorcitrated rabbit plasma( 15-25ml)wasadded to a 5-10-mlspecific antibodycolumn and rotatedon amixer for 4-6 h at4VC.

Factors VII and Xsubstrateplasmashad< 1%activityof thespecific factor. Factor IX substrateplasmahad - 3% residual factor IX

activ-ity. These substrate plasmas were stored at -750C.

A serum reagentfor usein specific rabbit and human prothrombin

(factor II)assayswaspreparedby clottingeither rabbit or human blood in a glasstube containing a 1:100-200 vol of rabbit brain TF (Sigma

ChemicalCo.).After2-4 h the serumwasseparatedfrom the clotby

twosuccessive centrifugations for 10 min at 10,000g, followed by dialysisof the serum into TBS. Serum so prepared contained < 1%

prothrombinactivityand>60%activityoffactorsVIIand X.

Barium

sulfate-adsorbed rabbit plasma

Barium-adsorbed rabbit plasma, prepared as described earlier ( 16), was added in assaysfor factorsVII, IX, and X to ensure that variation in levels of factors V and VIII in test samples did notinfluencethe

clotting times ofthese assays.Since the serum used in the prothrombin assay lackedfibrinogen, a specialbarium-adsorbedplasma reagent with ahigherfibrinogen level was prepared forthis assay from blood

ob-tained fromarabbit given anintravenousinjectionof 100,ug/kgof

Escherichia coli endotoxin (0111 :B4; Sigma Chemical Co.) 24 h

earlier.

Animal

experimental procedures

Animalprotocols. Female NewZealandrabbits (1.5-2.4 kg) wereused inprotocols approved by theAnimal Subjects CommitteeofUCSD. Rectaltemperatures, takenbeforeandafteranexperimenttomonitor againstcontamination of infused material with endotoxin, were un-changed.

Bloodsampling, injections, and infusions. Blood samples of 1.5-2

ml wereobtained by inserting a 23-gauge needle into amarginal ear vein that had been dilated by rubbing the ear with xylene. Nine parts of blooddripping freely from the hub were collected into an Eppendorf tube containing one part of a balanced citrate anticoagulant (0.06 M sodium citrate, 0.04 M citric acid). Platelet-poorplasma wasobtained bytwo successivecentrifugationsin an Eppendorfcentrifugefor 6 min at 15,000 g, at 4°C. The plasma was stored in - 0.2-mlaliquots in

plastic vials at -75°C. Blood for cell counts was collectedinto EDTA-coated microtainers (Fisher Scientific, Pittsburgh, PA). Larger amounts of blood for making rabbit reagents were obtained from a central ear artery.

TF was infused through a marginal ear vein as described earlier (12). Warfarin and IgG were injected either directly into a marginal ear vein through a 23-gaugebutterflyset orinto thetubingofarunning

infusion.In someexperimentsIgG was also added to thecontentsofan

Administration of warfarin. Rabbitsweregivenaspecialdiet de-pleted of vitamin K (ICNBiochemicals, Cleveland, OH), beginning

3-4 d before warfarin inearlyexperimentsandbeginningonday 1 of warfarin in laterexperiments.Therabbits were injectedintravenously

daily for a consecutive 5 d with 0.5mg/kgof al-mg/mlstock solution of sodium warfarin in isotonic sterile saline. Bloodsampleswere

ob-tained dailytomonitor warfarintreatmentbyamodified version ofthe

ProthrombinandProconvertin (P&P)testof Owrenand Aas (17). TF-induced intravascular coagulation. Rabbitswereinfusedover

- _{4 hwith100mlofisotonic saline containing} - 20ng/ml of

recon-stitutedTF.The total TFinfusedwas 1Mg/kg.At anadjusted flowrate of 25ml/h, TF,which had beeninjected intotheinfusion bag, beganto enterthe animal - 30minafterthe startoftheinfusion, i.e., after

materialfromtheinfusionbag hadreplaced thesalinein theinfusion tubing.Intravascular coagulationwasmonitored by serial measure-mentsoffibrinogen, factorV, andfactorVIII. Unlessotherwise stated,

testresults were expressed as percentof initialvaluesobtained froma

sampledrawnbeforeaninfusionwasstarted.

Five batchesofTF were used. Afirstgroupof15controlrabbits

were infusedwithbatch 1 or 2.Fibrinogen, factorV,andfactorVIII

levelswerereduced to - 50%bythe endoftheinfusion,and onerabbit

died shortly afterthe TFinfusionwasstarted.Asecond groupofsix control rabbitswereinfused withTFfrombatches 3, 4,or5, which

turned out to be more potent. Fibrinogen, factor V,andfactorVIII levelsfellby 75%,and threeof the six animals diedduringthecourse oftheinfusion. Data fromanexperimental TF-infused group were

always compared withdatafromthe control groupinfusedwith TFof equivalentpotency.

Clottingfactor

assays

Automatedinstruments yieldingaphoto-opticalendpointwereused.

Sampleswere testedinduplicate,and the mean was taken as the value

forthespecimen.Testplasmaswerediluted forassay in coldTBS/BSA unless otherwisenoted.Pooledcitrated platelet-poorplasmafrom six

toeight rabbitswasused as arabbit plasmareferencestandardwithan

assigned value of 100% forallfactorsexceptfibrinogen (seebelow).

Whenplasma sampleswere assayedfor activity after rabbitsweregiven purified human factor X orprothrombin, clotting times were con-verted to percent activity ofthefactor from both astandard curve preparedwithrabbitreferenceplasma and astandardcurveprepared withahumanreferenceplasma. Substrate plasmas for specificfactor

assays ofrabbitplasma were: for factorsVII, X, andprothrombin,of rabbitorigin;for factor IX, ofhuman or rabbit origin; forfactorsVand VIII,ofhuman origin.

TheP&P test. In the test asmodifiedforthisstudy, 100 Mlofrabbit

plasmadiluted 1:5 with barium sulfate-adsorbed rabbitplasma was

incubated for3 min at370Cwith100 Ml rabbitbrain TF(Sigma Chemi-calCo.).Clottingwastriggered bytheadditionof100

_Al

of25mM

CaCl2, which inthis and other tests citedbelow had been warmed to37°C.

Measurementoffibrinogen.Themethodof Clauss ( 18)wasused. 100Mlof 80 U/mlbovinethrombin was addedto200_{AL rabbit plasma}

dilutedin Veronalbuffer (28.4mMsodiumbarbital, 125 mMNaCl,

pH7.35).Clotting timeswereconverted tofibrinogenconcentration

fromastandard curve preparedwithahumanfibrinogen.

Factor V was measured in aone-stage assay in which 100

M1

of human immunodepleted factor V-deficient plasma was incubated

with 100Mlof a 1:50-200 dilution of the test sample and 100 Ml rabbit brainTF(SigmaChemical Co.) for 3 min, and clotting wastriggeredby theadditionof 100Ml 25 mMCaCl2.

Factor VIII was measured in a one-stage assay in which 100

M1

of

hereditaryhumanfactor VIII-deficient plasma,100 Mlof a1:10or1:20

dilutionof test sample, and100AlofAPTTreagent were incubated for 5 min, andclottingwastriggered bytheaddition of 100 Ml of 30 mM

CaCl2.

FactorVIIwasmeasured inaone-stage assay in which100Alofan

equal part mixture of rabbit immunodepleted factor VII-deficient

plasmaandbarium-adsorbedrabbitplasma, 100Mlofa1:50-200

dilu-tion oftestplasma,and100 Ml rabbit brainTF(SigmaChemicalCo.) wereincubated together for3 min,andclottingwastriggeredbythe

addition of 100Mlof 25mMCaCI2.

FactorX wasmeasuredessentiallyas forfactorVII except that

rabbitimmunodepleted factorX-deficientplasmawasusedasthe

sub-strate.

Prothrombinwasmeasured inaone-stage assay inwhich 100MA of

anequalpartmixture of rabbitserumreagent andspecial

barium-ad-sorbedrabbitplasma,100 Ml ofa1:10-40 dilutionoftestplasma,and 100Ml of rabbit brainTF(SigmaChemicalCo.)wereincubated to-getherfor3min,andclottingwastriggered bythe addition of100Ml of 25 mMCaCl2.

Factor IXwasmeasured inaone-stage assay in which 100M1ofan equal part mixture ofrabbit immunodepleted factor IX-deficient plasmaandbariumadsorbedoxplasmaorofhumanfactor

IX-defi-cient plasmaandbarium-adsorbedrabbitplasma, 100Mlofa1:5-10

dilution oftestplasma,and100Mlof APTTreagentwereincubated for 5min,andclottingwasthentriggered bythe addition of100 MAlof 35 mMCaCl2.Verysimilarvalues wereobtained witheither substrate for

rabbit plasmatestsamples whenplasma factorIXactivitywas <25%, i.e.,withinthe range criticalforthe presentexperiments.

Electroimmunoassays

Humanprothrombinantigenand human factor Xantigenwere

mea-sured by electroimmunoassay (EIA)with specific antibodies in 1%

Seakem agarose ME inaTris-Tricinebuffer, pH8.6(81mMTris-HCl

and 24 mMtricine). Antibodyconcentrations were0.5% for rabbit

anti-humanprothrombinantiserumand 0.4% for rabbit anti-human

factorXantiserum.

Hematological

tests

Hematocrit and platelet counts were determined with a Coulter ST counter(Coulter Electronics Inc.,Hialeah, FL) inthehematology labo-ratoryofthe UCSDMedical Center.

Electrophoresis

Reduced and nonreducedSDS-PAGEon8-10%gelswereperformed

accordingtothe methodofLaemmli ( 19).Proteinsweretransblotted ontoImmobilonPmembrane(Millipore, Bedford, MA) accordingto

the methodof Towbinet al.(20).Electroelution ofproteinbandsfrom

astained SDS-PAGE wascarried out with anelectroelution system

(Bio-Rad Laboratories) accordingtothemanufacturer'sinstructions.

Statistical methods

Allvalues reported in the tables aremean±SDand all values reported in the graphs are means±SEM. When datafromtwo groups were

com-pared,atwo-tailedstudent's groupttest wasused.

Results

The

effect

of warfarin upon rabbit vitamin K-dependent proco-agulantfactors. In aninitial experiment with 13 rabbits given 0.5 mg/kg warfarin daily, animals werefoundtofall intotwo

nonoverlapping groups. One group, referred to hereafter as

high responders, hada meanP&Pactivityonday 4 of warfarin

of 17±8% (SEM). Asecondgroup,referredtohereafteraslow responders, had a meanP&Pactivityon day 4of warfarin of

48±3%(SEM).

Activity levelsof factors II, VII, IX, and Xweremeasured

daily for4din 23 high responder rabbits. FactorVIIactivity

hadfallentoits fullextentby24 h, which was the time the first

postwarfarin samplewasdrawn.FactorsIX, X, and prothrom-binwerealso reduced at24 h,but 2 to 3 dwere required for factors IX and X to fall to their full extent, and 4 d were re-quiredforthe fall in prothrombin activityto level off. Mean

K-depen-dent procoagulantfactors from a large group ofrabbits after 4 d ofwarfarin arelisted in Table I. It isemphasizedthat thesedata were obtained in assays with rabbit substrates.

The

effect

ofwarfarin upon coagulation induced by the

infu-sionofTF.Ninecontrolnon-warfarin-treated rabbits infused

over4 hwith 1ag/kg ofTF(batch 1) had clear-cut evidence of intravascular coagulation (Fig. 1). By the end of the infusion mean values were 50% for fibrinogen, 35% for factor V, and 42%for factor VIII. In six high responder warfarin-treated rab-bits(meanP&Ptestactivity of17±8%)infused withthesame batchofTF,intravascularcoagulationwasattenuated.By the endoftheTFinfusion mean values were 88% forfibrinogen, 93%for factor V, and80% for factorVIII(Fig.1). These mean valuesdiffered from the mean values for the control rabbits (P

<0.05forfibrinogen,andP <0.001 for factorsVandVIII). In five low responder warfarin-treated rabbits (mean P&P test activity of48±3%)infused withthe TF,intravascular coagula-tion wasonlypartiallyattenuated. By the end of theinfusion mean values were 75%for fibrinogen, 50% for factor V, and 43% for factorVIII(Fig. 1). Therefore, only highresponder rabbits, witha P&P test ofbetween 5 and 35% after4 d of warfarin,were used infurther studies.

Evaluation of the

effect

upon

TF-induced intravascular

coagulation ofselective

depression offactors

VII,

IX, X,

and prothrombin

Antibodieswereadministered to otherwise normal rabbitsto

depressselectively plasma activitiesof the vitamin K-depen-dentprocoagulant factorstolevelssimilartothoseobtainedin high responderrabbits after4dof warfarin(Fig. 2).The rab-bitswere theninfused withTFfor4hand thedegree ofintravas-cular coagulation was compared to that observed in appro-priatecontrol groups.

The

effect of

different

degrees of

selective reduction

of

plasmafactor

VII

activity upon

TF-induced intravascular

coag-ulation. 12 rabbitswere

injected

with 250

inhibitory

U of purifiedgoatanti-rabbit factorVIIIgG,andanadditional250 Uof IgGwereaddedtotheinfusion bag.Theanimals could be divided intotwodistinctgroups. Inonegroupof six

animals,

with a mean valuefor factorVIIactivity before immunodeple-tion of95%,thefactorVIIlevelfellto 10-15%

(Fig. 2).

In a

second groupof six rabbits, witha meanvaluefor factorVII

activity beforeimmunodepletion of 56%,themeanfactorVII

levelfellto< 1%.

Immunodepression of plasma factor VII activity to 10-15%, i.e.,to alevelcomparabletothemeanlevelobtained in rabbits

given

warfarin (Table I), failedtopreventTF-induced

TableL Mean Valuesforthe P&PTestand Individual Vitamin

K-dependent ProcoagulantFactors

inHigh Responder Rabbitsafter4dof Warfarin

Assay n Activity±SD Range

P&P 38 16±8 5-35 Factor II 38 13±10 3-54 FactorVII 38 12±11 0.5-50 Factor IX 23 7±4 2-25 Factor X 38 14±7 4-36

100 75

z

F--I:

z

LLI

CY

C)

FIBRINOGE

o\!

501 25

100 75 50 25

125 100 75 50 25

0 1 2 3 4

TIME (Houi

N A

-O

Figure1.Theeffect of

treatment with warfarin

upontheextentof TF-induced intravascular coagulationas mea-5 6 suredbythe decrease in

plasmalevelsof fibrino-v

_B

gen

(A),

factorV

(B),

and factor VIII(C).

-__

The...infusionperiod is

indicatedbyabar. (n) High respondersto

war-...farintreatment(n 6);

(A) low respondersto

warfarintreatment(n

5 6 7 =5); (V) control rabbits nottreated with

war-III

C

farin(n=

9).

Values

aremean±SEM of the

-Q meanpercentbefore in-fusion. Mean valuesat

the end of the infusion forhigh responders and control animals dif-fered: P<0.05 for

fi-5 6 7 _brinogenand P<0.001

rs) for factorVand VIII.

intravascular coagulation. By the end ofa 4-h TF infusion

mean valueswere40% forfibrinogen, 33% for factorV,and 31% for factorVIII. These values differed from the values ob-served in control warfarin-treated animalsinfused with TF but

>1

LLI C) (x

C)

a

i'-. 12

10

C)

<7

H2

z 2

LLJ I L)

a-.5

.50

FACTOR VII

0 1 2 3 4 5

TIME (Hours)

FACTOR X

0 1 2 3 4 5

TIME (Hours)

H-FACTOR IX

>125

j-100

CI

< 75

Z 20

cLI

15.

T-!_ t '

LU 0 2 3 4 5 6 7 8

a _TIME

_(Hours)

>- PROTHROMBIN

H_1q1L

!5

'5

0 1 2 3 4, 5 6

TIME (Hours)

6

>12 10

C)

< 7 H2

_ J2 U1

L)

6UW

a-Figure 2. Specific immunodepletionofplasmalevels of factorsVII, IX, X, andprothrombintolevelscomparabletothose obtainedin

high responder rabbits given warfarin. Specificantibodiesweregiven (I ) afterthe initial bloodsamplewasdrawn. The animalswerethen

infusedwith TFovertheperiodindicatedbythe bar. Valuesaremean

percentactivity±SEM.Detailsaregivenin thetext. 2134 A. Zivelin,L. V M.Rao,and S. I.Rapaport

0 1 2 3 4

FACTOR \

0 1 2 3 4

FACTOR VI

15 ,

_{-_}

..I .

?01- U

5

-10 _'-* --. /

D 7 5

2

1 1 )0-_f 75

501

15

, I 2-!

.--I

' 2

5.

ul ul

125

FACTOR V

B

100 U

75- 50-25

0 1 2 3 4 5

125 -,- .IT T

100

75

50 25

0 1 2 3 4

TIME

(Hours)

Figure 3. Evidence that selectivedepression of factorVII to 10-15%of

normalplasma activity failstoprevent

TF-in-duced intravascular

co-agulationasmonitored byconsumptionof plasmafibrinogen(A), factorV(B), and factor VIII (C). Values are mean(±SEM) percent of the mean initial value.(.)Anti-factor

VII-treated rabbits (n

=6 );(.) controlrabbits (n=3);(n) warfarin-treatedrabbits (n =7).

Allrabbitswereinfused

with TF over the period indicated by the bar.

z

LUL I.J

H

z

1-LLJ

C

z

LU C-) LUI

a-FIBRINOGEN

A

100- ₀

75 !

50 Figure4. Evidence that

S!_~-~' immunodepression of

25 factor VII to< 1%of

0 , . , normalplasmaactivity

0 1 2 3 4 5 6 protectsagainst TF-in-duced intravascular

co-125

ATO agulationasmonitored

FACTOR V B by consumption of

fi-100 - *yibrinogen(A), factor V

75 _Q (B),andfactor

VIII

-(C). Datafrom rabbits 50 N with selective

depres-25 sion of factorVIIto

10-15%of normal

0 , plasmaactivityarealso

0 ₁ 2 3 4 5 6

plotted forcomparison.

Values are mean

125 (±SEM)percent ofthe

FACTOR VIII C meanvaluesbefore the

100 e injection of the IgG.

_(o)

75 Rabbits with< 1%

plasmafactorVII

activ-50 _{ity (n}=_6);

(e)

rabbits

25 with 10-15%plasma factorVIIactivity(n

0 =6).The TFinfusion

0 1 2 3 4 5 6 _{period isindicatedby}

TIME

(Hours)

the

bar.

not fromthe values observed in control nontreated animals

infused with TF(Fig. 3).

In rabbitsimmunodepleted toplasma factor VII levels of

<1%, a 4-h infusion ofTF resulted in the following mean

values:85% for fibrinogen,72%for factorV, and 75% for factor VIII.Immunodepletion of plasma factorVIIto < 1%,in con-trast toimmunodepletion of plasma factorVII to 10-1 5%,

pro-tected rabbits against TF-induced intravascular coagulation

(Fig.

4).

Theinability ofselective reduction ofplasmafactorIXactiv-itytoprotect against TF-induced intravascular coagulation. 12 rabbitswereinjected intravenously with500 U/kgof

monoclo-nalanti-factorIXIgG. 1hlater eight rabbitswereinfused with

TFandfour rabbitswereinfusedwithsaline.Nine rabbits were

injected withacontrol monoclonal anti-humanthyroglobulin

IgG (2.5 mg/kg). 1 hlaterfive oftheseanimalswere infused

with TF and the remaining four animals were infused with

saline.

At 1 h after the anti-factor IX IgG injection the mean

plasmafactor IXactivityof the animalsgivenTFwas13%,and

by2hafter theinjection it had fallento 7%(Fig. 2).The latter was the same mean plasma activity as that found in a large group of rabbitsgivenwarfarin(Table I). Bythe endof theTF infusion meanvalues were 46% for fibrinogen, 40% for factor V, and 42% for factor VIII (Fig. 5).Thesevalues did not differ

fromthe values in animalsinfused withTFafterreceiving the controlmonoclonal IgG. Thus, selective depletion of plasma

factor IX activityto a level comparable to that measured in

rabbitsgiven warfarin failedto protect rabbits against

TF-in-ducedcoagulation.

Theprotectiveeffectofselective reduction offactorXactivity upon TF-induced intravascular coagulation. 11 rabbits were injected with - 300 inhibitory U of monoclonal anti-factor X

lo0 75S

z

LJ

2:

I

z

l I

7F

LL

C

z

LU C-)

LLU

c0.

50 25

125 FACTOR V

B

100 75

50

25 T

0₀ 1_{1, 2}

3.

₃ 4_{4 5}5 6_{6 7}

1P9 100 75 50

25

0 1 2 3 4 5 6

TIME

(Hours)

Figure5. Evidence that

selective

immunode-pression of plasma fac-torIXactivityfailsto

prevent TF-induced in-travascularcoagulation

asmonitored by

con-sumption offibrinogen

(A), factorV(B),and

factorVIII(C).Values aremean(±SEM) per-centofthemeaninitial values.

(e

) Rabbits

re-ceivinganti-IXIgGand infused with TF (n

=8); (m) rabbits receiv-ing controlIgGand in-fused withTF(n= 5);

(o) rabbitsreceiving ei-ther controlIgG (n=4)

oranti-IXIgG (n=4)

and infused with saline. The infusionperiodis indicatedby thebar.

z

LUJ

~-jII

l-z

LL CD C-)

LJ

0n

FACTOR VIII

~.~-0

I

I -_

*

-I~

FIBRINOGEN

A

Moo-O °5

11

-Do

I

0 1 2 3 4 5 6

FACTOR VIII

C

-to_ O5 0---_

IgG, and an additional 125-150 U was added to the infusion bag. Within 30 min of the initial injection factor X activity had fallen to 10-20% (Fig. 2), which was comparable to the mean 14%activity ofhigh responder rabbits given warfarin (Table I). Sevenoftherabbitswereinfusedwith TF, and the four remain-ing rabbits were infused with saline as controls. In the latter the anti-factor X IgG had no effect upon fibrinogen, factor V, or platelets, butdid alter the measurement of factor VIII as de-scribedbelow.

Selective

immunodepletion

of

plasma

factor X

activity

substantiallyprotected therabbits against TF-induced intravas-cular coagulation. At the end of the TF infusion the mean fibrinogen level was 81% (Fig. 6 A), which differed from the 19% mean fibrinogen level of control rabbits infused with TF and was comparable to the 85% meanfibrinogen level of war-farin-treated rabbits infused with TF. At the end of the TF infusion the mean factorVlevel was 68% (Fig. 6 B), which differed from the mean factorVlevel ofcontrol nontreated rabbitsinfused withTF and was comparable to the 78% mean factorVactivity of warfarin-treated rabbits infused withTF.

A

problem

became apparent in

measuring plasma

factor VIII levels in factorX-depleted rabbit plasma. In contrast to thefactorVassay, in which rabbit plasmatestsamples were assayed at a 1:50-200 dilution, testsamples were assayedfor factorVIIIata1:20dilution.MeanfactorVIIIlevels in thefirst sample measured after the injection of anti-factor X IgG, which was 30 min after beginning an infusion, were reduced

125 FACTOR V B

100

m-75

-I-50

-25

0 1 2 3 4 5 6 7 12

0 1 2 3 4 5 6

TIME

(Hours)

Figure 6. Evidence that

selective

immunode-pression of plasma fac-torXactivitytolevels

comparabletothose

ob-tained inhigh

re-spondersgiven warfarin

protectsagainst TF-in-ducedintravascular

co-agulationasmonitored byconsumption of

fi-brinogen (A), factorV

(B),andfactorVIII

(C). The TFinfusion

periodisindicated by

the bar. (v) Nontreated

controlrabbits(n=3);

(o) warfarin-treated rabbits(n= 7);

(.)

rabbitstreatedwith anti-factorXIgG(n

=7). Values are mean percent(±SEM)of meaninitial values for

fibrinogenandfactor Vand mean percent of thesample 30 min after

injectionof IgG for fac-tor VIII.Atthe endof the TFinfusion the P valuesfor differencesin means for the anti-X/TF animals and the

warfarin/TFanimals were: for fibrinogen, P=0.46; for factor V, P

=0.22;forfactorVIII, P=0.73. Thecorresponding P values for

dif-ferencesin means for theanti-X/TFanimals and the control/TF

animalswere:for fibrinogen,P<0.0001;forfactor V,P=0.001; for factorVIII,P=0.0 12.

both for animals infused withsaline (56±6%) and for animals infusedwith TF (66±6%). Since the anti-rabbit factor X IgG did not inhibit human factor X in the factor VIII-deficient substrate plasma, it appeared that a markedly reduced factor X level in a rabbitplasma test sample relative to the factor X level of the rabbitplasma reference standard influenced the test re-sult of our factor VIII assay.

Therefore, in the analysis of the factor VIII data, the factor VIII level 30min after beginning an infusion ofTF, rather than thefactorVIIIlevel beforeinjecting the anti-factor X IgG, was assignedavalue of100%. When the data were plotted as per-centof this initial value (Fig. 6 C), the mean plasma factor VIII level at the end of the TF infusion differed from the mean factor VIIIlevel ofcontrol rabbits infused with TF and was comparable to the meanfactorVIIIlevel ofwarfarin-treated rabbitsinfused with TF. Thus, all data were consistent in estab-lishing that selective depression of factor X to a level compara-bleto that obtained in high responder rabbits given warfarin protectedagainstTF-induced intravascularcoagulation in our rabbit model.

The protective effect of selective reduction of plasma prothrombin activity upon TF-induced intravascular coagula-tion. 10 rabbitswere

injected

with 150

inhibitory

Uof affin-itypurifiedgoat anti-rabbit prothrombin IgG, and an addi-tional75Uof IgGwasaddedtotheinfusionbag.Insixrabbits the meanprothrombin activityfell to 10%(Fig.2),whichwas comparabletothe 13%activityofhigh responder rabbits given warfarin (Table I). Theserabbitswereinfusedwith TF. Three rabbits, inwhich theplasmaprothrombin activity fellto be-tween 2 and 3%,andafourthrabbit withaplasma prothrom-binactivityof 1 1%wereinfused with salineas acontrol for the effectof theinjection of theIgGupon hemostatic factors and theirmeasurement.

Mean valuesforfibrinogenatthe endofa4-hinfusionwere

93% for control prothrombin-depleted animals infused with saline, 19% for controlnonprothrombin-depletedanimals in-fusedwith TF, and 75% forprothrombin-depleted animals in-fusedwithTF.This last valuewascomparabletothe85%mean

fibrinogen level ofwarfarin-treated animals infused with TF

(Fig.7A). Mean values for factorV(Fig.7B)atthe endofa

4-hinfusionwere79% for controlprothrombin-depleted ani-malsinfused withsaline, 17%forcontrol

non-prothrombin-depleted rabbits infused with TF, 61% for

prothrombin-de-pleted animalsinfused with TF,and78%for warfarin-treated animalsinfused withTF.Althoughthemeanvaluefor factorV

ofthe

prothrombin-depleted

rabbits

given

TFwaslowerthan the value for factor V of the warfarin-treated animals (P

=0.012), itwas

impressively higher

than the value for the

con-trol TF-infused rabbits(P<0.0005).

Thedata on factorVIIIlevelswereless clearcut(Fig.7C).

Mean factor VIII levels fell 30 min after

injection

of anti-prothrombinIgGto70-75%ofmean valuesbeforethe injec-tion in both animals infused withTFand inanimals infused withsaline. Atthe end of infusionthe meanfactorVIIIlevel was lower for the animals infused withTFthan for the animals infused with saline(P=_0.004).It wasalso lower than for the warfarin-treated animals infused withTF(P= 0.008).

How-ever,the mean factor VIII level for theprothrombin-depleted rabbitsinfused withTF washigherthanfor the control rabbits infused withTF(P=_0.003).

Incontrast to theinjectionof the otherIgG

preparations,

theinjectionoftheanti-rabbitprothrombin IgGcauseda50%

FIBRINOGEN

A

100

75 N"V 50

25

0 1 2 3 4 5 6 7

z

-lJ

H

z

U-Ld M:

0 z

C-)

LLJ

l0J

FACTOR VIII C

* U

EA\`1

. _.m 100

75 50 25

z

LII

M-I

z

LlL

0

H-z

C-)

25

CI

125 FACTOR V B

100

II

75 _.~

50-25

0 1 2 3 4 5 6 7

125, _ __ _

.__W-10

7'

5

2

-o

FACTOR VIII

C

'0 =

)5 | _

0I

0 1 2 3 4 5 6 7

TIME

(Hours)

Figure 7. Evidence that

selective immunode-pression ofplasma prothrombin activity to levels comparable to thoseobtainedinhigh

responders given

war-farinprotectsagainst

TF-induced intravascu-larcoagulationas

moni-toredby consumption offibrinogen (A), factor

V(B),and factor VIII

(C).(.) Rabbits given antiprothrombin IgG

andinfused withTF (n

=6); (o)

warfarin-treatedrabbits infused

with TF (n =7); (v)

control nontreated rab-bitsinfused withTF(n

=_{3); (o)}control

rab-bitsgiven

anti-prothrombinIgG and infused withsaline(n

=4). Values are mean percent (±SEM) of

meaninitial values be-fore theIgGinjection

(4),orbefore the

infu-sion. Theinfusion pe-riod isindicatedby the bar.Atthe end of the TFinfusionthe P valuesfor differencesbetween meansforanti-II /TF andwarfarin/TF animalswere:forfibrinogen,P=0.2; for factorV, P=0.0 12; for factorVIII, P=0.008. ThecorrespondingPvaluesfor differencesbetweenmeansof anti-II/TFanimals andcontrol/TF

animalswere:forfibrinogen,P=_{0.002; for factor}V, P<0.0005;for

factorVIII,P=0.003.

fall in the mean platelet count of both rabbits infused with saline andrabbits infused withTF(Table II). Because plasma prothrombinconcentration exceeds by many-fold the concen-trations of factors VII, IX, or X, the platelet fall after anti-prothrombinIgG is presumed to stem from the binding oflarge enough amounts of prothrombin-antiprothrombin immune complexes to platelets to cause platelet sequestration. After 2 h, platelet counts began to rise both in rabbits infused with saline and inrabbits infusedwith TF, which indicated that sufficient thrombin had notbeengenerated in the rabbits infused with

TFtodepress the platelet count further.

Viewed as a whole the above data on fibrinogen, factor V, factor VIII, and platelets provide convincing evidence that se-lectiveimmunodepletion of plasma prothrombin activity sub-stantially protected rabbits against TF-induced intravascular coagulation.

Selective restoration offactor X in warfarin-treated rabbits infusedwithTF.Itwasclear from the data above that immuno-depletionof either plasma factor X or prothrombin to activities comparable to those obtained in warfarin-treated rabbits suf-ficed to protect otherwise normal rabbits against TF-induced intravascularcoagulation. Our next step was to determine the effect of selective restoration of factor X upon the protection warfarin afforded against TF-induced intravascular

coagula-tion.

TableII.Mean Platelet Count ofRabbits Given

AntiprothrombinIgGandInfusedwith Either TForSaline Mean platelet count

Time TFinfusion (n=6) Saline infusion (n=₄₎

xI09/Iiter±SD

0 h* 520±125 586±151

30 min 263±70 270±144 2 h 374±80 363±122

4h 393±84 381±126

*Before IgG and thebeginningofaninfusion.

Since itwasnecessaryto use

purified

human factorXfor thispurpose, ina

preliminary experiment,

1 U of human factor X was added in vitro to 1 ml of rabbit factor X-deficient plasma, and the resultantplasma factorX

activity

was

assayed

with rabbitfactor X-deficientsubstrate. Thefollowingvalues were obtained: 100%, when clottingtimes wereconvertedto

percent activity from ahuman plasma reference curve;

only

10-15%,

whenclottingtimeswereconvertedtopercent

activity

from arabbitplasmareferencecurve. Asevaluated ina

one-stagefactorXassay withrabbitreagents, humanfactorXhada

much loweractivityinrabbit plasmathandid rabbit factorX.

Becauseof thisobservation,threewarfarin-treated rabbits were injected with 400-500 Uof purifiedhumanfactorX be-foreinfusingTF.Thisraisedthemeaninitial

plasma

factorX

activity of 19%to57%asreadfromarabbit reference

plasma

curve, and to 600% asread from ahuman plasma reference

curve(Fig. 8).ThemeanplasmahumanfactorX

antigen

level after theinjectionwas475%.Bytheendofthe 4-h TFinfusion meanplasmafactorXactivityhad fallento25%asreadfroma

rabbit reference curve, and to 220% as read from a human referencecurve.

The 4-hinfusionofTF resulted ina fall ofmeanplasma fibrinogenlevelto19% andmeanplasmafactorVlevelto16%. These mean values wereidentical tothose observed in three TF-infused control rabbits not given warfarin. The mean

plasmafactorVIIIactivity fellto anintermediatevalueof 37%, whichwashigher(P=0.011_{)than the mean}factorVIIIlevelof

16% in the control rabbits not given warfarin but lower (P =0.010)thanthemeanfactorVIIIlevelof64% inseven

war-farin-treated rabbitsnot

injected

withfactorX.

X Figure 8. Mean plasma

600 factor X levels in three 450 [ warfarin-treated rabbits < 300 [| °>*injectedwith human

150 ~0 y factorXand infused withTF.Values are LLJ

750

_{-percent±SEM.}

(.)

Fac-Cy _{25 A/ _ j} j tor X activity measured a 0 _withrabbit substrate

0 1 2 3 4 5 6 7 _{and read from a rabbit}

TIME (Hours) _{plasma reference curve;} (in)factorXactivity

measuredwith rabbit substrate and read fromahumanplasma

These data provided evidence that restoring plasma factor Xactivity in warfarin-treated rabbits could abolish the protec-tion warfarin affordedagainst TF-induced intravascular coagu-lation. Despite the low concentration of prothrombin activity in the warfarin-treated animals (range, 6-25%), sufficient thrombin was generated to induce intravascular coagulation as extensiveasthat observed in control non-warfarin-treated rab-bits.

Selective restoration of prothrombin in warfarin-treated rabbits infused with TF. Eight warfarin-treated rabbits were injected with - 75 U/kg ofhuman prothrombin. Three

rab-bitswereinfused with salineas controls and five rabbits were infused with TF, but, as described below, data from one were excluded. In the remaining four rabbits the mean plasma prothrombin activity 1 h after the injection of prothrombin was

_134+21%,

and themean

plasma

human

prothrombin

anti-gen was 124±21%. Mean activities of the other procoagulant vitaminK-dependent factors in these four rabbits on the day of theexperimentwere:factorVII, 14%; factor IX, 6%; and factor X, 15%.

1 haftertheinjection ofprothrombin, which was just be-forebeginninganinfusionofTForsaline, the platelet count andfibrinogen levels were unchanged but mean values in per-centofinitial values wereslightly elevatedfor bothfactor V (133±29%) and factor VIII (125±34%). Since our purified prothrombin contained a trace of thrombin (< 0.01%), it is possible that there were traces ofcirculating thrombin-acti-vatedfactors V and VIIIinthese animals.

Restoring plasma prothrombin activity notonly reversed theprotectionbywarfarinbutstrikingly sensitizedtherabbits to TF-induced intravascularcoagulation. Plasma fibrinogen, factorV, andfactorVIIIfelldramatically duringtheTF infu-sion(Fig. 9).The first rabbit died ofuncontrollable bleeding secondarytodefibrination aftera1-h samplewastaken,andits datawereexcluded.Asecond rabbit died afterasampleat2 h was taken, and the data fromthisanimal were included. In the remainingthreeanimalstheTFinfusionwasstoppedwithin2

to3 h becauseofpersisting oozing ofblood from-thesampling sites. Ascan be seen in Fig. 9, plasma

fibrinogen

levels fell within 2 h to unmeasurable levels (< 25 mg/dl), andfactorV

and VIII mean activities were <5%. In addition, the mean

platelet countfell from aninitialvalue of759 X 109/literto 100 x 109/ liter after2 hofTFinfusion.

Hematologic parameters. The meanhematocritvalue for the 64 rabbits fell as a consequence ofinfusions andblood sampling fromaninitialvalueof36to33% atthe endofthe infusion period. Thefallinananimal's hematocritwas inde-pendent ofwhether theanimal

experienced

intravascular coag-ulation.In 22rabbits with intravascularcoagulation,themean

platelet count decreased from an initial value of 634±127

x 109/literto a value of 388±151 X 109/literat the end of a

4-h TFinfusion (P<0.001).In32 animals without intravascu-larcoagulation, mean platelet levels were unchanged after a4-h

infusion. The data from 10 animals inwhich,as mentioned earlier, plateletcountsfellafter theinjectionof antiprothrom-bin IgGwereexcluded incalculatingthesemeans(TableII).

Discussion

Theexperiments reported here provide data on how warfan functionsin vivo as an anticoagulant when TFwas usedto

initiate bloodcoagulation (21).In apreliminary experiment z

I 1-J

Hl-z

LL

CY

H-llJ

0I

.50 _FIBRINOGEN A Figure 9. Evidence that

.25 F Nselective restoration of

prothrombin

in

war-LOO- *7° °

_{farin-treated}

_rabbits 75 \ _ sensitizesthe rabbits to 50 ! extensiveTF-induced

25 intravascular

coagula-tionasmonitoredby

0 1 2 3 4 5 consumptionof

fibrino-50 gen(A), factor V (B),

.2525FACTOR V

B

andfactor VIII

(C).

Values aremean

Lo _{(±SEM)percent of the}

75 meanvaluesbefore the

50 injectionofhuman

prothrombinorbefore

25 thebeginningof the

in-o 1 2 fusion in the other

0 1 2 3 4 5 _{groups. All rabbits were}

SO FACTOR VIII C infused withTF. (C)

.25 Warfarin-treatedrabbits

LOO

BOwith

restored

prothrom-75

13 _bin(n=_{4); (m)control}

nontreatedrabbits (n

50 =3); (o)

warfarin-25 treated rabbits not given

,- \, 1 ._ W prothrombin(n=

7).

1 2 3 4 ' The infusionperiodfor

TIME (Hours) the controlgroupsis

in-dicated bythe bar.The

arrowdenotes the endofthe meaninfusion period forthe experi-mental group. Theasteriskdenotes an unmeasurablefibrinogenlevel

of<25mg/dl.

the responseof rabbitsto0.5mg/kgof sodium warfaringiven intravenously daily for4dwasdeterminedasmonitoredbya

sensitive

prothrombin

timetechnique (a modification of the P&P test

[17]).

The rabbits fell into two groups: high

re-sponders,inwhich themeanplasmaP&Pactivitydecreasedto

17%, and lowresponders,inwhichthemeanplasmaP&P activ-ity decreasedtoonly48%. Thereasonisunknown,buta

differ-ence in responsiveness to warfarin of rabbits has also been noted by others (22). Contraryto a

single early

report that warfarin treatmentfailedtoprevent intravascularcoagulation inrabbits infused withcrude TF

(7),

underour

experimental

conditions

high

responderrabbitswereprotected against TF-induced intravascularcoagulation (Fig. 1).

The P&P activity range of 5-35% of protected high

re-sponder rabbits is similar to the 10-30% range first

recom-mendedasthetherapeuticrangefor patients monitoredwith theP&P test(23). Moreover, themean P&Pactivity ofour

high responder rabbitsof 16% iscomparabletothe 15%P&P activity later recommended foranticoagulanttherapy in pa-tients with arterial disease(24).In our

opinion,

this reinforces thevalidityofextrapolatingdatafromourrabbit modeltoan

understandingofthemechanism ofthe clinical antithrombotic efficacyof warfarin.

Aninitialapproachwas todeterminetheeffect upon

TF-induced coagulationofselectiveimmunodepletion of factors VII, IX, X, andprothrombin to mean levels comparableto

those obtainedinhigh responderrabbitsgivenwarfarin(Table I). It wasthought important to measure the activity of the procoagulant vitaminK-dependentfactorsastheyfunctioned in rabbit plasma. Specific rabbit substrate plasmas for each factor wereprepared and assays using these substrates were

2138 A.Zivelin,L. V. M.Rao, andS. I.Rapaport

standardized. As documented elsewhere (25) this proved es-sential for accurate determination of the plasma activity of fac-tors VII, X, andprothrombin.

Selective immunodepletion of plasma factor VII activity to a mean level ( 1%) comparable to that obtained in high re-sponderrabbitsgivenwarfarin ( 12%) failed to protectagainst TF-induced intravascularcoagulation (Fig.3). We inferfrom thisexperimentalresult that the fall inplasma factor VII activ-ity in patients treated with warfarin, e.g., to the 20% level ob-served inagroup ofpatients stablyanticoagulated witha vita-min Kantagonist (26), contributes little if any to warfarin's antithrombotic efficacy.

Additional observations supportthisconclusion. As men-tioned earlier,venousthromboembolism has beenreported in patients withhereditary factor VIIdeficiency (3, 4). Adding purifiedfactorVII to aplasma level of 5% generated maximal prothrombinaseactivity when the coagulation reactionsof he-reditary factor VII-deficientplasma wereinitiatedin vitrowith diluteTF(6). Finally,aplasmafactorVIIlevelof between 5 and10%hasbeenreported to be adequate forsurgical hemosta-sis inpatients with hereditary factorVIIdeficiency (27, 28).

Asexpected,sinceTFinitiates coagulationthrough the for-mation of factor VII(a)/TFcomplexes,selective immunode-pletion of plasma factor VII activity to < 1% protected the rabbits againstTF-inducedintravascularcoagulation(Fig. 4). However, the intensity of warfarintherapy needed toachieve

anantithrombotic effect fromsuch severedepletion ofplasma factorVIIactivitywould have an unacceptablerisk ofclinical bleeding.

Selectiveimmunodepressionof plasma factorIX tolevels comparable to the 7% meanlevel ofhigh responder rabbits givenwarfarin also failedtoprotect rabbitsagainstTF-induced intravascular coagulation (Fig. 5). Therefore, in at least some clinical circumstances depression ofplasma factorIXactivity may notcontribute significantlytotheantithrombotic efficacy ofwarfarin.However,it ispremature to conclude that depres-sionof factorIXactivity failstocontributetothe antithrom-boticefficacy of warfarin when coagulation is initiated by a lowerconcentration ofTF or byfactor XIa. Further data are needed but may not beobtainablewith the present model sys-tem, which isinsensitivetoinfusion of concentrations ofTF

much lower than 1ytg/kg(29).

In contrastto the above, selective immunodepression of either factorX(Fig.6)orprothrombin (Fig. 7)tomeanlevels comparabletothose observed inhigh responderrabbitsgiven warfarin(TableI)protected rabbits against TF-induced intra-vascularcoagulation. Theseobservationsinotherwise normal rabbits prompted us to carry out additional experiments to determinewhetherdepression of factorXaloneor prothrom-bin alone would also protectagainst TF-induced coagulationin

warfarin-treated

animals,inwhichprotein C and S levels are depressed. Therefore, either plasma factorX orprothrombin was restored in high responder rabbits given warfarin before theywereinfused withTF.Becausehighlypurified rabbit

fac-tor Xandrabbitprothrombinwerenotavailable insufficient quantity for these experiments, purified human proteins hadto be used.

Theinjection of 400-500Uof human factorXinto a2-kg rabbit, which, as discussed earlier, raised theanimal'splasma factorXlevelfrom- _{20% toonly}- _{60% asdetermined from}

a rabbit plasma reference curve,reversedtheprotection

war-farin afforded against TF-induced intravascular coagulation.

Despite a mean plasma prothrombin activity of only 15%, manifestations of intravascular coagulation in theseanimals were virtuallyindistinguishable fromthose observed when TF was infused into control animals with a normal plasma prothrombin level.

However, theeffect of restoringfactor X in these experi-ments must be evaluatedwiththecaveatthatthetestresult in our one-stage rabbit factor X assay isdetermined bythe initial rates, over secondsafterrecalcifyingthe reactionmixture, of the generation and function ofhuman factor Xa in rabbit plasma. Because theseinitialrates areslower than ifthesample containedanequivalent concentration ofrabbitfactor X, clot-ting timesconvertedtopercentactivityfromarabbit

plasma

referencecurve willyieldamuch lower valueforfactorX activ-ity than clotting times converted to percent activity from a

human plasma reference curve. It is possible that the large amounts ofhuman factor Xthatwere injected into

rabbits,

which raised plasma levels to- _{600%as measured}witha hu-man plasmareferencestandard(Fig. 8),could have had much moreofaprocoagulanteffectover the4hofaTFinfusionthan onemightassume from the - 40% increase inplasma factorX

concentrationthat wasmeasuredwith arabbit plasma refer-ence standard.

Restoring prothrombin in high responder rabbits given warfarin not only reversed warfarin's protection but sensitized the animals toTF-induced intravascularcoagulation (Fig. 9). Incontrast to the protection seen in rabbits immunodepleted of plasmafactor X, the comparably low factor X activity of warfarin-treated animals with restored prothrombin failed to prevent massive TF-induced intravascular coagulation with un-controllablebleedingdue to defibrination. It is assumed that thisreflectedaninabilityof depressedproteinC and S levels to regulate the TF-induced coagulation. It is also believed that the differencebetweenreversingprotectionby restoringfactorX

and enhancing intravascular coagulation by restoring prothrombin reflects prothrombin's role as the substrate for thrombin.

In the present experiments the prevention of a fall in plasma fibrinogen, factorV, and factor VIII levels has been taken asevidence forreduced TF-induced intravascular coagu-lation.Althoughunquestionablytrue forfibrinogen,onecould questionwhetherthemaintenance of factorVandVIIIlevels in warfarin-treated animals infused with TF could have re-sulted atleast in part from reduced plasma protein C and S levels. This seems most unlikely to us in view of the striking depletion ofplasma factorVand VIII observed after TF infu-sion inwarfarin-treatedanimals with restored prothrombin.

Theprothrombintime test is sensitive to the combined ac-tivity of factor VII,factor X, and prothrombin. However, when warfarin is given to a patient the initial prolongation of the prothrombin timestems primarily from a reduced plasma

fac-tor VIIactivity,since factor VII has an intravascular

half-clear-ancetimeofonly 5h(30). Although an initial fall in plasma factorVIIto levelsof 10- 15% within 1-2d of starting warfarin therapy may yield a prothrombin time in the accepted thera-peutic range, yet, as strongly supported by the data here, a patientwill notbe protected against thrombosis. Indeed,

be-cause proteinC also hasa similarly short intravascular

half-clearancetime (31 ), the prolonged prothrombin time during

the first dayortwo oftreatment with warfarin is thought to

mask aninitial

period

ofpotential hypercoagulability.

efficacy

of warfarinin

patients

willnotbeobtaineduntil both

plasma

factorX

activity

andplasma prothrombin activityare

substantially depressed.

In

particular,

ourdata

provide

impres-sive in vivo evidenceof the

key

importance

of

depression

of

prothrombin activity

forwarfarin's

anticoagulant

effect.It is

increasingly recognized (32, 33)

that when

patients

aregiven both

heparin

and warfarin fortreatment ofanacute throm-botic

episode,

the

heparin

shouldbecontinued for several days after

starting

thewarfarin

regardless

of the

prothrombin

time. Inview of theintravascular half-clearance time in humans for factorXof- 40_h

₍₃₄₎

_andfor

_prothrombin

_of- _60h

_(34),

_it

seems

appropriate

to us toextend the

period

of

overlap

to6 d in

patients

withevidence of extensivevenousthrombosis.A clini-cal

study

inwhich

prothrombin

levelsare

measured,

eitheras

specific prothrombin activity

orasnative

prothrombin antigen

(5,

35),

before

heparin

is discontinuedshould

provide

the in-formationneededtovalidatethe aboverecommendation. Acknowlednments

Weacknowledge theexperttechnical assistance ofMr. AnD.Hoang.

This workwassupported by NationalHeartLung and Blood

Insti-tute(NHLBI)grantsHL-27234 (to S. I. Rapaport)andHL-42813(to

L. V.M.Rao).L. V.M. RaoisarecipientofaResearchCareer Devel-opment AwardfromNHLBI.

References

1.Hirsh,J.1991.Oralanticoagulant drugs.N.Engl.J.Med. 324:1865-1875.

2.Sise,H.S., S.M.Lavelle, D.Adamis,and R.Becker. 1958. Relationof

hemorrhagicandthrombosistoprothrombinduringtreatmentwith

coumarin-typeanticoagulants.N.Engl.J.Med. 259:266-271.

3.Godal,H.C.,K.Madsen,andR.Nissen-Meyer.1962.Thrombo-embolism inpatientswith totalproconvertin(factorVII)deficiency.Areportontwocases.

ActaMed.Scand. 171:325-327.

4.Hall, C.A.,S. I.Rapaport, S.B.Ames, and J.A.Degroot.1964.Aclinical andfamily studyofhereditaryproconvertin(factor VII) deficiency.Am.J. Med. 37:172-181.

5.Furie,B., C. F.Diuguid,M.Jacobs,D. L.Diuguid,andB.C.Furie. 1990. Randomizedprospectivetrial comparingthenativeprothrombin antigenwith

theprothrombintimeformonitoringoralanticoagulant therapy.Blood. 75:344-349.

6.Xi,M., S.Beguin,andH.C. Hemker. 1989. Therelativeimportanceof the

factorsII,VII,and Xfortheprothrombinaseactivityinplasmaoforally

anticoag-ulatedpatients.Thromb.Haemostasis.62:788-791.

7.Izak,G.,and K.Galewsky.1966.Studiesonexperimentallyinduced

hyper-coagulablestateinrabbits. Thrombos.Diath. Haemorrh. 16:228-242.

8.Wessler,S.,S.N.Gitel,H.Bank,U.Martinowitz,andR.C.Stephenson.

1979.An assayof the antithrombotic action of warfarin:Itscorrelation with the inhibition of stasis thrombosis in rabbits. Thromb.Haemostasis.40:486-498.

9.Gitel,S.N.,andS. Wessler. 1979.Theantithrombotic effects of warfarin andheparinfollowinginfusions of tissuethromboplastininrabbits: clinical

im-plications.J.Lab. Clin. Med.94:481-488.

10.Gitel,S.N.,andS. Wessler. 1983.Dose-dependentantithrombotic effect of warfarin in rabbits. Blood. 61:435-438.

11.Rao,L. V.M.,and A. D.Hoang.1989.Purificationandcharacterization

of rabbit brain tissue factor.Thromb.Res.56:109-118.

12.Warr,T.A.,L. V. M.Rao,and S. I.Rapaport.1990.Disseminated intra-vascularcoagulationinrabbits induced by administration of endotoxinortissue

factoneffect ofanti-tissue factor antibodiesandmeasurementofplasmaextrinsic

pathwayinhibitoractivity. Blood.75:1481-1489.

13.Bajaj,S. P., S. 1. Rapaport,andS.F. Brown.1981.Isolationand character-izationof human factorVII.J.Biol. Chem. 256:253-259.

14.Bajaj,S. P.,S.1. Rapaport, and C. Prodanos. 1981.Asimplifiedprocedure forpurificationofhumanprothrombin,factorIXandfactorX.Prep.Biochem.

11:397-412.

15.Rao,L. V.M.,and A. D.Hoang.1991.Purification and characterization of rabbit factorIXandits existenceas atwo-chain factor IXa incirculating

plasma.Thromb.Res.64:57-68.

16.Warr,T.A.,L. V. M.Rao,andS.I.Rapaport.1990.Asensitive,accurate

assayfor extrinsicpathwayinhibitor(EPI)activityinrabbitplasma:paradoxical

effect ofexcessexogenousfactorX.Thromb. Res. 59:773-782.

17.Owren,P.A., andK. Aas.1951.The controlofdicumaroltherapyand the

quantitativedetermination ofprothrombinandproconvertin. Scand.J. Clin.

Lab. Invest. 3:201-208.

18.Clauss,A.1957.Gerinnungs physiologisheschnell methodezue

bestim-mugdesfibrinogens.ActaHaematol. 17:237-246.

19.Laemmli,U. K.1970.Cleavageof structuralproteins duringtheassembly

oftheheadofbacteriophageT4.Nature(Lond.).227:680-685.

20.Towbin, H.,T.Staehelin,andJ.Gordon. 1979.Electrophoretictransferof proteinsfrompolyacrylamide gelstonitrocellulosesheets:proceduresandsome

applications.Proc.Natl. Acad.Sci. USA. 76:4350-4354.

21.Rapaport,S.I.,andL. V.M. Rao.1992.Initiation andregulationoftissue

factor-dependentbloodcoagulation.Arterioscler.Thromb. 12:1111-1121. 22.Edwards,R.L.,E.Schreiber,andW. Brande. 1986. Theeffectof sodium warfarinonrabbitmonocytetissuefactorexpression.Thromb. Res. 42:125-137. 23.Owren, P. A. 1955. Long-termdicumaroltreatmentincardiovascular disease:techniqueand results.InInternationalConferenceonThrombosis and

Embolism, 1st.Basel,1954. Schwabe & Co.AG,Basel. 1085-1094.

24.Borchgrevink,C. F. 1960. Long-termanticoagulant therapyinangina pectorisand myocardialinfarction: clinical trial of intensiveversusmoderate

treatment.ActaMed. Scand. 168(Suppl. 359):1-52.

25.Zivelin,A.1993.Individualimportanceof factors II,VII,IXand Xfor the antithromboticeffect of the vitaminKantagonistwarfarin.Ph.Dthesis.

Univer-sityofCalifornia,SanDiego. 151 pp.

26.Loeliger,E.A.,B.vanderEsch,M. J.Mattern,and A.S.A.den

Bra-bander. 1963. Behaviour of factorsII,VII, IX,and Xduring long-termtreatment

withcoumarin.Thrombos.Diath. Haemorrh. 9:74-89.

27.Strauss,H. S. 1965. SurgeryinpatientswithcongenitalfactorVII

defi-ciency(congenitalhypoproconvertinemia). Experiencewithone caseandreview of theliterature.Blood.25:325-334.

28.Briet,E.,andG. Onvlee. 1987.Hip Surgeryinapatientwithseverefactor

VIIdeficiency.Haemostasis. 17:273-277.

29.Sandset,P.M.,B. J.Warn-Cramer,L.V. M.Rao,S. L.Maki,andS.I.

Rapaport. 1991.Depletionof extrinsicpathwayinhibitor(EPI)sensitizes rabbits

todisseminated intravascularcoagulationinducedwith tissue factor: evidence

supportingaphysiologicroleforEPIasanaturalanticoagulant.Proc.Nati.Acad.

Sci. USA.88:708-712.

30.Loeliger,E.A.,B.vanderEsch,C.C.terHaarRomney-Watchter,and

H. L.Booji. 1960.Factor VII:itsturnover rateand itspossiblerole in

thrombo-genesis. Thrombos. Diath. Haemorrh. 4:146-200.

31.Weiss, P., G.A. Soff,H. Halkin,and U. Seligsohn. 1987. Decline of

proteinsCandSandfactorsII,VII,IXandXduringtheinitiationofwarfarin therapy.Thromb.Res.45:783-790.

32.Wessler, S.,andS.N.Gitel. 1984. Warfarin: from bedsidetobench.N.

Engl.J. Med. 311:645-652.

33.Hyers,T.H.,R. D.Hull,and J.G.Weg.1992. Antithrombotictherapyfor

venousthromboembolic disease. Chest. 102(Suppl.):409S-425S.

34.Williams,W.J.1983. Lifespanofplasmacoagulationfactors.In

Hematol-ogy,3rded. W. J.Williams,E.Beutler,A.J.Erslev,andM. A.Lichtman,editors. McGraw-HillInc.,NewYork.1232-1237.

35. Braun, P. J., and K. M.Szewczyk. 1992.Relationshipbetween total

prothrombin,nativeprothrombinandtheinternational normalized ratio(INR).

Thromb.Haemostasis.68:160-164.