www.elsevier.com/locate/jhep

Journal of Hepatology 51 (2009) 682–689

Thrombotic risk factors in patients with liver cirrhosis: Correlationwith MELD scoring system and portal vein thrombosis developmentq

Maria Assunta Zocco1,*, Enrico Di Stasio2, Raimondo De Cristofaro1, Marialuisa Novi1,Maria Elena Ainora1, Francesca Ponziani1, Laura Riccardi1, Stefano Lancellotti1,

Angelo Santoliquido1, Roberto Flore1, Maurizio Pompili1, Gian Lodovico Rapaccini1,Paolo Tondi1, Giovanni Battista Gasbarrini1, Raffaele Landolfi1, Antonio Gasbarrini1

1Department of Internal Medicine, Catholic University of Rome, Gemelli Hospital, Largo A. Gemelli 8, 00168 Rome, Italy2Department of Biochemistry and Clinical Biochemistry, Catholic University of Rome, Rome, Italy

See Editorial, pages 632–634

Background/Aims: Prognostic scores currently used in cirrhotic patients do not include thrombotic risk factors (TRFs).

Predicting factors of portal vein thrombosis (PVT) development are still unknown. We wanted to describe TRFs as a func-

tion of liver disease severity using the MELD score and assess the role of local and systemic TRFs as predictors of PVT

development in cirrhotic patients.Methods: One hundred consecutive patients with liver cirrhosis were included in the study. TRFs, D-dimers, MELD

score, portal vein patency and flow velocity were evaluated in all subjects at baseline and every 6 months thereafter. Vari-

ables able to predict PVT development within 1 year were identified by means of multiple logistic regression.

Results: The plasma levels of protein C and antithrombin were lower and the concentration of D-dimers was higher in

patients with advanced disease. Plasma levels of antithrombin, protein C and protein S resulted significantly lower in PVT

group at univariate analysis, but reduced portal vein flow velocity was the only variable independently associated with

PVT development.

Conclusions: Lower concentrations of natural coagulation inhibitors are frequently detected in patients with liver cirrho-sis. A reduced portal flow velocity seems to be the most important predictive variable for PVT development in patients with

cirrhosis.

� 2009 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

Keywords: Thrombotic risk factors; Liver cirrhosis; MELD score; Portal vein thrombosis; Portal flow velocity

0168-8278/$36.00 � 2009 European Association for the Study of the Liver.

doi:10.1016/j.jhep.2009.03.013

Received 9 October 2008; received in revised form 19 February 2009;

accepted 17 March 2009; available online 23 April 2009

Associate Editor: J. Boschq The authors who have taken part in this study declared that they do

not have anything to disclose regarding funding from industry orconflict of interest with respect to this manuscript.

* Corresponding author. Tel.: +39 347 0597805; fax: +39 0635502775.

E-mail address: mariaazocco@hotmail.com (M.A. Zocco).Abbreviations: TRFs, thrombotic risk factors; PVT, portal vein

thrombosis; AT, antithrombin; APA, antiphospholipid antibodies;MELD, Model for End-stage Liver Disease; LAC, lupus anticoagu-lant; INR, international normalized ratio; APTT, activated partialthromboplastin time; ELISA, enzyme-linked immunosorbent assay;PT, prothrombin time.

1. Introduction

The liver has many haemostatic functions, includingthe synthesis of most coagulation factors and inhibitorsas well as fibrinolytic factors [1,2]. The balance betweenprocoagulant and anticoagulant factors is essential toavoid excessive thrombin generation under physiologi-cal conditions [3]. Therefore, it is not surprising thatadvanced liver disease results in a complex pattern ofdefects in haemostatic functions in the form of reducedsynthesis of coagulation factors, inhibitors, and abnor-mal clotting factors, abnormalities of fibrinolytic

Published by Elsevier B.V. All rights reserved.

M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689 683

activity, disseminated intravascular coagulation andplatelet function defects [4–7].

Moreover, thrombosis of the intrahepatic veins is fre-quently observed in cirrhosis and has been associatedwith its progression [8], while occlusion of small sizedintrahepatic veins and sinusoids has been considered apotential triggering factor of liver tissue remodelling [9].

Recently, the prevalence of several genetic andacquired thrombotic risk factors in patients with chronichepatitis B or C as well as their possible association withnecroinflammatory activity and extent of fibrosis hasbeen evaluated [10,11]. It is well known that protein C,protein S and antithrombin (AT) serum levels aredecreased in patients with liver disease [12,13] but, todate, few reports are available about the associationbetween the above mentioned and other natural antico-agulant factors produced by the liver and different stagesof liver cirrhosis.

Portal vein thrombosis (PVT) is an important compli-cation of cirrhosis, and is mostly associated with theoccurrence of hepatocellular carcinoma [14]. The inci-dence of non-neoplastic PVT in cirrhotic patients isunknown, while its prevalence ranges from 0.6% to16% [15–17]. Since PVT can be an important source ofmorbidity and mortality, early detection and treatmentfor de novo thrombosis is an important issue, especiallyin patients on the waiting list for liver transplantation[18]. Over the last few years the etiology of PVT hasbeen better defined and large case series have improvedour understanding of the natural history of thiscondition [19]. Male sex, previous abdominal surgeryincluding splenectomy and portocaval shunts, encepha-lopathy, ascites, past history of bleeding varices, lowplatelet count, and Child–Pugh class C have been con-sidered predisposing factors to PVT in liver cirrhosis[18,20,21]. Previous sclerotherapy of varices has beenshown to be a risk factor in some studies [22,23]; otherauthors did not confirm this datum [16,24] or haveshown that sclerotherapy may only be a trigger factorfor PVT in patients with genetic thrombophilia [25,26].Two studies found antiphospholipid antibodies (APA)in more than half of cirrhotic patients with PVT[27,28] but no relationship was found between anticardi-olipin antibodies and PVT in another study [29]. More-over, inherited (such as the factor V Leiden 1691 G–Amutation and the prothrombin 20210 G–A mutation),or acquired (such as the reduced levels of naturalinhibitors of coagulation like protein C, protein S andantithrombin) coagulative defects leading to a hyperco-agulable state, have been found in cirrhotic patients withPVT [16,30–34].

On the whole, the occurrence of PVT in liver cirrhosisappears to be a multifactorial complication, in which theinvolvement of inherited and acquired thrombotic riskfactors as well as local anatomical and hemodynamicfactors may be involved.

This study was designed to determine whetheracquired systemic thrombogenic factors and local fac-tors related to hepatic haemodynamics and portalhypertension might be related to liver disease progres-sion and to the development of PVT.

2. Patients and methods

2.1. Patient population and study design

One hundred consecutive adult patients with cirrhosis (76 men, 24women) with a median age of 60 years (range 31–81 yrs) were enrolledin the study, which conformed to the ethical standards of the Declara-tion of Helsinki and was approved by the local ethics committee. Livercirrhosis was diagnosed by histological examination of liver biopsy orby unequivocal hematochemical, ultrasound (US) or endoscopic find-ings suggesting advanced liver disease with portal hypertension. Theseverity of liver disease was estimated according to Model for End-stage Liver Disease (MELD) scoring system [35].

Patients with patent paraumbilical vein, reversed portal blood flow,known hepatocellular carcinoma or any other malignancy, knownhemostatic disorders other than liver disease, bacterial infection, a clin-ical history of peripheral venous thrombosis, or those who hadreceived any form of antiviral and/or immunomodulatory therapywithin the last 6 months were excluded from the study. No patientwas taking oral contraceptives, anticoagulation or anti-platelet drugs.

Exclusion criteria were also the presence of the most commoninherited coagulation abnormalities (Factor V Leiden and prothrom-bin (G20210A) genes mutation), as well as, based on familiar screen-ing, antithrombotic protein deficiency (protein C, protein S,antithrombin (AT)). Only 2 patients of the initial population evaluatedwere affected by Factor V Leiden (heterozygote state) or prothrombin(G20210A – heterozygote state) mutation. Based on this finding, wedecided to exclude these patients from further analysis.

The demographic, clinical and laboratory data of the patientsenrolled in the study are summarized in Table 1. The following param-eters were considered as thrombotic risk factors (TRFs): deficiency inAT, protein S, protein C, presence of lupus anticoagulant (LAC) anti-bodies, elevated homocysteine and, cryoglobulinemia. TRFs and D-dimer levels were evaluated as functions of a stratified MELD score;an arbitrary cut-of value was set at 13 (usually applied in our institu-tion to select patients for liver transplantation).

The patients were followed up for one year and were evaluated atbaseline and every 6 months by liver function tests, D-dimer, TRFsand abdominal Doppler US.

Concerning the development of PVT during follow-up, we evalu-ated, for their prognostic significance, the clinical and demographiccharacteristics of patients at baseline. In particular previouslydescribed TRFs were analyzed in addition to age, gender, platelet cellcount, international normalized ratio (INR), activated partial throm-boplastin time (APTT), D-dimer levels, grading of oesophageal varicesand MELD score.

PVT was suspected by the occurrence of endoluminal material inthe main trunk of portal vein and/or its branches at grey-scale ultraso-nography or by the presence of a filling defect at color or power Dopp-ler ultrasonography. In all cases the diagnosis was confirmed bydynamic abdominal computed tomography that also allowed a moreprecise distinction between partial and complete obstructive thrombo-sis. Portal cavernoma was defined by the presence of multiple smallchannels that replaced the trombosed portal trunk.

The presence and staging of oesophageal varices was evaluated bymeans of upper endoscopy.

2.2. Blood collection and processing

After informed consent, blood was drawn without stasis by cleanvenipuncture and collected in vacuum tubes containing 105 mmol/Ltrisodium citrate as an anticoagulant (Vacutainer; Becton–Dickinson,

Table 1

Clinical, biochemical and demographic characteristics of enrolled

patients at baseline.

Characteristic [normal values] Patients (n = 100)

Age (yr) 59.5 (11.5)Male 76

EthiologyViral 56Alcoholic 28Othersa 15

Child–Pugh classA (n) 50B (n) 25C (n) 25MELD score 15.1 (7.1)Bilirubin (g/dl)[0.3–1.2] 3.7 (5.6)Albumin (g/l)[3.4–4.8] 3.1 (0.5)Creatinine (mg/dl)[0.7–1.2] 1.3 (1.1)Platelet count (�103/l)[140–450] 96.1 (49.9)INR[0.8–1.2] 1.37 (0.35)Partial thromboplastin time (s)[20–36] 36.4 (8.9)Protein C (%)[70–140] 42.1 (19.4)Protein S (%)[64–140] 73.1 (20.8)Antithrombin (%)[70–120] 52.4 (19.5)Homocysteine (lmol/l)[5–15] 10.8 (7.0)D-dimer (ng/ml)[<278] 1326 (1531)LAC antibodies positive 10Anti-b2 glycoprotein 1 (U/ml)[<20] 35 (8) (for samples

with anti-b2GpI P20 U/ml, n = 13)Cryoglobulinemia 12

Quantitative values are expressed as mean (SD). Categorial variablesare expressed as frequencies.

a Other: autoimmune or criptogenic cirrhosis.

684 M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689

Meylan, France) at a blood anticoagulant ratio of 9:1. Full bloodcount, platelets count, prothrombin time (PT) or INR, APTT, clinicalchemistry parameters were measured using commercially availableassays. Coagulation tests were performed using an automatic coagulo-meter (TOP, Instrumentation Laboratory Milano, Italy) and commer-cially available reagents from the same company. Using an auto-analyzer STAGO DIAGNOSTIC (Boehringer Mannheim, Mannheim,Germany) and following the manufacturer’s instructions, AT (Sta-chrom AT), protein C (Asserachrom PC), and protein S (AsserachromPS) were determined by enzyme-linked immunosorbent assay(ELISA). Results appear as a percentage of protein activity. Deficiencyin AT III, protein S, and protein C were diagnosed when the proteinlevel was below 70%, 64% and 70%, respectively. The presence ofLAC antibodies was detected by diluted Russell Viper Venom ClottingTime (dRVVCT – Instrumentation Laboratory Milano, Italy – screen-ing and confirm) [36].

Anti-b2 glycoprotein levels were measured using commercialenzyme linked immunosorbent assay (ELISA) kits (Imuclone Anti-Beta2GP1 IgG and IgM ELISA, Instrumentation Laboratory, Milano,

Italy) and were expressed in units, according to the manufacturer’sinstructions. The dRVVCT assay was performed by a commercial kit(‘‘dRVV test” and ‘‘dRVV confirm”) using an automatic coagulometer(‘‘Top”, Instrumentation Laboratory, Milano, Italy). This test wasconsidered positive if the dRVVCT was >38” and the value of the ratiodRVVCT/dRVVCT confirm was P1.25.

2.3. Doppler measurements

Doppler US examinations were performed using a Color-PowerDoppler US scanner Technos (Esaote, Genova, Italy) with a 3.5–5 MHz convex probe.

All subjects were examined after overnight fasting in a supine posi-tion after a rest of 15 min in order to avoid any influence of food, pos-ture and exercise.

The portal vein was examined following current guidelines whichallow a reduction of interobserver variability to non-significant levels[37]. In particular, care was taken to ensure that the angle of inson-ation between the Doppler beam and the portal trunk was between30 and 60�. The doppler study was performed by positioning the sam-ple volume at about the middle of the portal trunk (between the liverhilum and the spleno-mesenteric junction) by approaching from theepigastrium with the probe placed slightly obliquely.

Portal flow velocity was calculated automatically by the instrumentas time averaged maximum velocity, which is the parameter providingthe most reliable and repeatable results [37]. Doppler examinationswere performed by two experienced sonographers (MP, LR) withoutprior knowledge of clinical and biochemical status of the studypopulation. Three blood flow tracings lasting no less than 4 s wererecorded for each vessel and the mean of three was noted as the finalreading.

2.4. Statistical analysis

All data were analysed using the statistical package SPSS 15.0 ver-sion (SPSS Inc., Chicago, IL). Continuous variables are expressed asmeans ± SD, categorical variables are displayed as frequencies. Statis-tical analysis was performed using the appropriate parametric or notparametric test (t test or the Mann–Whitney test for comparisons ofcontinuous variables between groups, corrected v2 or Fisher’s exacttest for categorical data). Multivariate binary logistic analysis was per-formed to evaluate the relationship between the development of portalvein thrombosis and TRFs. The receiver operating characteristic curvewas used to identify the best discriminating value of portal vein flowvelocity at baseline associated with the development of portal throm-bosis. The coefficients obtained from the logistic regression were alsoexpressed in terms of odds ratio with 95% confidence intervals. Atwo tailed p value less than 0.05 was considered significant.

3. Results

3.1. TRFs and liver dysfunction

Levels of the natural anticoagulant proteins, AT andprotein C were low in the entire group of cirrhoticpatients (mean values 52.4% and 42.1%, respectively;normal values: AT, 70–120% and protein C, 70–140%)(Table 1).

Associations of high MELD (defined as P13) or lowMELD (<13) with levels or presence of TRFs are pre-sented in Table 2. Patients with advanced staging hadlower protein C and AT levels (32.1% vs. 54.6%,p < 0.001 and 42.6% vs. 63.4%, p < 0.001, respectively)and higher D-dimers levels (2008 ng/ml vs. 520 ng/ml,p < 0.001) (Fig. 1).

Table 2

Thrombotic risk factors stratified according to disease staging by categorized MELD score.

Factor MELD < 13 (n = 44) MELD P 13 (n = 56) p Value

Protein C (%) (normal 70–130) 54.6 (18.8) 32.1 (13.2) <0.001

Protein S (%) (normal 64–140) 75.5 (16.4) 71.1 (23.7) 0.288Anti-thrombin III (%) (normal 70–120) 63.4 (17.4) 42.6 (15.9) <0.001

D-dimer (ng/ml) (normal <278) 520 (384) 2008 (1791) <0.001

Homocysteine (lmol/l) (normal 5–15) 9.8 (3.3) 11.6 (9.0) 0.186Cryoglobulinemia (n) (%) 5 (11.0) 7 (12.6) 0.464LAC antibodies positive (n) (%) 5 (11.0) 5 (8.9) 0.468

Quantitative values are expressed as means (SD) and categorial variables are displayed as frequencies (%). Significant values are expressed in bold.

0

10

20

30

40

50

60

70

80

MELD <13 MELD 13

Protein C (%)Antithrombin (%)D-dimer x 100 (ng/ml)

≥

p<0.001P<0.001

Fig. 1. Protein C, AT and D-dimer, stratified according to categorized

MELD score.

Table 3

Individual portal flow rate and thrombosis characteristics of patients who deve

Patients ID Baseline portalflow rate (cm/s)

6 months portalflow rate (cm/s)

6 months PVT 6 monthromextens

1 19 15 No –2 11 nd Complete Portal9 10 10 No

19 12 9 Partial Portal20 12 13 No –

28 14 11 No –40 11 9 Partial Portal

42 10 nd Complete Portalright pbranchportal

49 10 10 No –64 9 11 No –67 12 10 Partial Portal69 11 10 No –

nd = not detectable flow rate.

M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689 685

3.2. Role of TRFs on the development of portal vein

thrombosis

Eighty-one out of the 100 patients enrolled completedthe study, 6 patients underwent ortothopic liver transplan-tation and 13 died during follow-up. Eight subjectspresented PVT at baseline and were excluded fromprognostic evaluation on PVT development. Among thepatients evaluated during follow-up (n = 73), 12 developedde novo thrombosis within 1 year. Individual portal flowvelocity and thrombus characteristics of these patientsare described in Table 3, whereas flow velocity timingchanges in patients without PVT are represented in Fig. 2.

The association of TRFs, clinical and demographiccharacteristics at baseline with PVT development duringfollow-up is shown in Table 4. As expected, a significantassociation between PVT development and high MELDscore (p = 0.012) was observed. Serum levels of AT,protein C and protein S at baseline resulted significantlylower in patients who developed PVT (p = 0.011,p < 0.001 and p = 0.028, respectively). Similarly,

loped PVT within 1 year of observation.

thsbusion

1 year portalflow rate (cm/s)

1 year PVT 1 year thrombusextension

11 Partial Left portal branchtrunk nd Complete Portal cavernoma

9 Partial Right portal branchtrunk 9 Partial Portal trunk

9 Partial Portal trunkRight portal branch

10 Partial Portal trunktrunk 9 Partial Portal trunk

Right portal branchtrunkortalleft

branch

nd Portal cavernoma

nd Complete Portal trunknd Complete Portal trunk

trunk 10 Partial Portal trunk10 Partial Portal trunk

baseline 6 months 1 year0

10

20

30

40

Timing

Port

al F

low

Rat

e (c

m/s

ec)

Fig. 2. Flow rate timing changes in patients without development of

PVT.

686 M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689

lowered platelet count was observed in PVT group(p < 0.001).

Concerning haemodynamic parameter, patients withreduced portal vein flow velocity at baseline (<15 cm/s) showed a significantly higher occurrence of PVTdevelopment (91.7% vs. 19.7%, p < 0.001).

The patients who developed portal thrombosis duringfollow-up had a mean basal value of portal flow velocity(11.8 ± 2.6 cm/s) significantly lower than that of thepatients without portal thrombosis (19.6 ± 5.7 cm/s)(p < 0.001). Using the receiver operating characteristic

Table 4

Demographic, clinical and biochemical data at baseline in patients who deve

observation.

Factor No PVT (n = 61

Age (yr) 59.2 (11.3)Male 44 (72.1%)

EthiologyViral 38 (62.3%)Alcoholic 14 (23.0%)Others 9 (14.7%)MELD score >13 26 (42.6%)INR 1.35 (0.36)Platelet cell count (�103/l) 103.3 (52.1)ATIII 54.7 (19.4)Protein C (%) 45.9 (18.9)Protein S (%) 77.9 (20.0)D-dimer (ng/ml) 1038 (1209)APTTa 1.21 (0.33)LAC positive 8 (13.1%)Anti-b2 glycoprotein 1 (U/ml) [<20] 34 (9)b

Cryoglobulins positive 10 (16.4%)Homocysteine (lmol/l) 10.1 (4.4)Portal flow rate <15 cm/s 12 (19.7%)

Oesophageal varices0/F1 52 (85.2%)F2 6 (9.8%)F3 3 (4.9%)

Quantitative values are expressed as means (SD). Categorial variables are da Results are expressed as a ratio of test to reference coagulation times, usb Values refer to samples with anti-b2GpI P 20 U/ml (n = 10 and n = 3, r

curve, 15 cm/s was identified as the best discriminatingvalue in the prediction of portal thrombosis develop-ment (sensitivity 85.7%, specificity 78.0%). Patients withportal vein flow velocity at baseline <15 cm/s showed asignificantly higher occurrence of PVT development(47.8% vs. 2.0%, p < 0.001).

A multivariate logistic backward regression modelwas built utilizing previously described variables at base-line associated with the presence of PVT within 1 year. Inthe final model, reduced flow rate was the only variableindependently associated with the development of portalvein thrombosis (OR 44.9, 95% CI 5.3–382.7; p < 0.001).

4. Discussion

The concern that liver function failure is related tothe development of severe coagulopathy is firmly estab-lished among hepatologists. This condition is mainlyreflected by the prothrombin time (PT) and by APTTprolongation. PT is an excellent marker of liver failureand a strong independent prognostic factor of survivalin patients with chronic liver disease. For this reason,new and old models of prognosis for patients withadvanced liver disease include PT (or INR) as one ofits components [38]. However, recent studies haveshown that cirrhotic patients, in addition to diminished

loped (PVT) or not (no PVT) portal win thrombosis within 1 year of

) PVT (n = 12) p Value

55.2 (10.9) 0.26510 (83.3%) 0.720

0.8246 (50.0%)4 (33.3%)2 (16.7%)10 (83.3%) 0.0121.34 (0.34) 0.43458.3 (20.9) <0.001

41.2 (13.7) 0.011

27.8 (7.5) <0.001

60.7 (22.8) 0.028

1660 (2405) 0.3991.20 (0.20) 0.9212 (16.7%) 0.46437 (5)b 0.6452 (16.7%) 0.42716.4 (16.0) 0.19511 (91.7%) <0.001

0.6829 (75.0%)2 (16.7%)1 (8.3)

isplayed as frequencies (%). Significant values are expressed in bold.ing as reference a normal plasma tested in parallel with test plasmas.espectively).

M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689 687

hepatic synthesis of clotting factors, also have a pro-found deficit of natural anticoagulants, such as proteinC and AT, which may counterbalance the bleeding ten-dency caused by the deficiency in procoagulants [3,39].

In a previous study, TRFs were found to be indepen-dently associated with the extent of fibrosis in patientswith chronic hepatitis [10].

The results obtained in the present study confirm sev-eral earlier reports showing that haemostatic tests couldbe of great practical value in the assessment of hepato-cyte function in liver disease [6,13,27,40,41]. Not surpris-ingly, our study showed that the reduction in the level ofantithrombotic proteins is strongly related to the sever-ity of liver cirrhosis according to the MELD scoring sys-tem. Moreover, markers of hemostasis activation suchas D-dimer were increased in the majority of ourpatients. D-dimer was significantly higher in patientswith MELD P 13, suggesting that the extent of hemos-tasis activation is related to severity of cirrhosis. The sig-nificant negative correlation between levels of D-dimerand both AT and protein C provides further evidencethat a hypercoagulation state is due to the loss of thenatural anticoagulant proteins.

One possible consequence of the hemostasis activa-tion could be the development of PVT, which is a nega-tive prognostic factor in patients with advanced liverdisease [42,43]. Understanding the factors predisposingto PVT would be important in identifying the subgroupof patients at higher risk and may ultimately aid deci-sions regarding the use and duration of anticoagulationtherapy.

As already mentioned, PVT in patients with liver dis-ease is the result of concomitant local, acquired andinherited thrombophilic factors [16,44]. Our study, eval-uated prospectively for the first time the role of acquiredTRFs and local hemodynamic factors as predictors ofPVT development. We found that plasma levels of anti-coagulant proteins, such as protein C, protein S and AT,were lower in cirrhotic patients who developed PVT dur-ing follow-up than in those without PVT. The associa-tion between grading of oesophageal varices andthrombotic events in cirrhotic patients reported by oth-ers authors [23] could not be supported by our data.Moreover, no association was found between PVTdevelopment and shorted aPTT, which has been previ-ously associated with venous thromboembolism [45].Conversely, other factors resulted to be related to thedevelopment of PVT at least at univariate analysis.The association of low platelet count to PVT has beenalready found by Francoz et al. [18] and we agree withthese authors that in cirrhotic patients the impact ofportal hypertension reflected by hypersplenism over-comes the proper effect of low platelet count in prevent-ing PVT occurrence. Furthermore, an important rolewas played by an advanced disease staging as well asby the reduction of portal flow velocity. These data

are not surprising if we consider that PVT is usually alate event in the natural history of liver cirrhosis occur-ring more frequently in patients with uncompensateddisease [21]. This is further confirmed by the epidemio-logical data showing that the PVT prevalence is about15% in uncompensated patients submitted to liver trans-plantation [18,20,46] and is higher than that reported ina series of consecutive outpatients or in hospitalized cir-rhotics (7.5–11.2%) who show a minor degree of liverfunction failure [16,29,47].

Moreover, it is interesting to note that the only factorconfirmed at multivariate analysis as predictors of PVTdevelopment was the reduced portal flow velocity, thussuggesting that local haemodynamic factors could bemore important than systemic anticoagulant deficiencyin the pathogenesis of PVT.

It is well known that portal flow velocity is inverselyrelated to Child–Pugh score being lower in patients inclass C compared to patients in classes B or A [48]and that when lower than 10 cm/s it is an independentrisk factor for patients’ survival [49]. However, we havedemonstrated for the first time that the decreased veloc-ity and stagnation of portal blood flow may act as themost important prognostic factor for PVT development.One possible explanation of these results is that theamount of active thrombin generated in the portal circu-lation, although reduced if compared to a normal condi-tion, is not washed away by an adequate flow. It isreasonable to argue that this effect, together with thepresence of reduced levels of protease inhibitors, asshown in the present study, could favour triggering ofcoagulation and local thrombosis.

Further studies are needed in order to assess the rel-evance of these parameters in a larger population sam-ple, to identify additional risk factors and to assess theprognostic effect of anticoagulation therapy and porto-systemic shunting in patients with established PVT.

Acknowledgements

This work has been supported by an unrestrictedgrant provided by Fondazione Ricerca in Medicina,Bologna, Italy.

References

[1] O’Grady JG, Langley PG, Isola LM, Aledort LM, Williams R.Coagulopathy of fulminant hepatic failure. Semin Liver Dis1986;6:159–163.

[2] Mammen EF. Coagulation abnormalities in liver disease. Hae-motol Oncol Clin North Am 1992;6:1247–1257.

[3] Tripodi A, Salerno F, Chantarangkul V, Clerici M, Cazzaniga M,Primignani M, et al. Evidence of normal thrombin generation incirrhosis despite abnormal conventional coagulation tests. Hepa-tology 2005;41:553–558.

[4] Kelly DA, Summerfield JA. Hemostasis in liver disease. SeminLiver Dis 1987;7:182–191.

688 M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689

[5] Mammen EF. Coagulation defects in liver disease. Med ClinNorth Am 1994;78:545–554.

[6] Al Ghumlas AK, Abdel Gader AG, Al Faleh FZ. Haemostaticabnormalities in liver disease: could some haemostatic tests beuseful as liver function tests? Blood Coagul Fibrinolysis2005;16:329–335.

[7] Carrr JM. Disseminated intravascular coagulation in cirrhosis.Hepatology 1989;10:103–110.

[8] Wanless IR, Wong F, Blendis LM, Greig P, Heathcote EJ, LevyG. Hepatic and portal vein thrombosis in cirrhosis: possible role indevelopment of parenchymal extinction and portal hypertension.Hepatology 1995;21:1238–1247.

[9] Wanless IR, Liu JJ, Butany J. Role of thrombosis in thepathogenesis of congestive hepatic fibrosis (cardiac cirrhosis).Hepatology 1995;21:1232–1237.

[10] Papatheodoridis GV, Papakonstantinou E, Andrioti E, Cholang-itas E, Petraki K, Kontopoulou I, et al. Thrombotic risk factorsand extent of liver fibrosis in chronic viral hepatitis. Gut2003;52:404–409.

[11] Assy N, Bekirov I, Mejritsky Y, Solomon L, Szvalb S, Hussein O.Association between thrombotic risk factors and extent of fibrosisin patients with non alcoholic fatty liver diseases. World JGastroenterol 2005;11:5834–5839.

[12] He XF, Wen ZB, Liu MJ, Zhang H, Li Q, He SL. Levels of plasmades-gamma-carboxy protein C and prothrombin in patients withliver diseases. World J Gastroenterol 2004;10:3073–3075.

[13] Vukovich TH, Teufelsbauer H, Fritzer M, Kreuzer S, Knoflach P.Hemostasis activation in patients with liver cirrhosis. Thromb Res1995;77:271–278.

[14] Webster GJM, Burroughs AK, Riordan SM. Review article:portal vein thrombosis-new insight into aetiology and manage-ment. Aliment Pharmacol Ther 2005;21:1–9.

[15] Okuda K, Ohnishi K, Kimura K, Matsutani S, Sumida M, GotoN, et al. Incidence of portal vein thrombosis in liver cirrhosis. Anangiographic study in 708 patients. Gastroenterology1985;89:279–286.

[16] Amitrano L, Guardascione MA, Brancaccio V, Margaglione M,Manguso F, Iannaccone L, et al. Risk factors and clinicalpresentation of portal vein thrombosis in patients with livercirrhosis. J Hepatol 2004;40:736–741.

[17] Belli L, Romani F, Sansalone CV, Aseni P, Rondinara G. Portalthrombosis in cirrhotics. A retrospective analysis. Ann Surg1986;203:286–291.

[18] Francoz C, Belghiti J, Vilgrain V, Sommacale D, Paradis V,Condat B, et al. Splanchnic vein thrombosis in candidates for livertransplantation: usefulness of screening and anticoagulation. Gut2005;54:691–697.

[19] Webster GJM, Burroughs AK, Riordan SM. Review article:portal vein thrombosis-new insight into aetiology and manage-ment. Aliment Pharmacol Ther 2005;21:1–9.

[20] Nonami T, Yokoyama I, Iwatsuki S, Starzl TE. The incidence ofportal vein thrombosis at liver transplantation. Hepatology1992;16:1195–1198.

[21] Yerdel MA, Gunson B, Mirza D, Karayalc�in K, Olliff S, BuckelsJ, et al. Portal vein thrombosis in adults undergoing livertransplantation: risk factors, screening, management, and out-come. Transplantation 2000;69:1873–1881.

[22] Politoske D, Ralls P, Korula J. Portal vein thrombosis followingendoscopic variceal sclerotherapy. Prospective controlled com-parison in patients with cirrhosis. Dig Dis Sci 1996;41:185–190.

[23] Mangia A, Villani MR, Cappucci G, Santoro R, Ricciardi R,Facciorusso D, et al. Causes of portal venous thrombosis incirrhotic patients: the role of genetic and biologic factors. Eur JGastroenterol Hepatol 2005;17:745–751.

[24] Kawasaki S, Henderson JM, Riepe SP, Brooks WS, Hertzler G.Endoscopic variceal sclerosis does not increase the risk of portalvenous thrombosis. Gastroenterology 1992;102:206–215.

[25] Amitrano L, Brancaccio V, Guardascione MA, Margaglione M,Sacco M, Martino R, et al. Portal vein thrombosis after varicealendoscopic sclerotherapy in cirrhotic patients: role of geneticthrombophilia. Endoscopy 2002;34:535–538.

[26] Yonemitsu Y, Yanaga K, Matsumata T, Sugimachi K. Portal veinthrombosis due to antithrombin III deficiency. A case report.Angiology 1995;46:1043–1047.

[27] Violi F, Ferro D, Basili S, D’Angelo A, Mazzola G, Quintarelli C,et al. Relation between lupus anticoagulant and splanchnicvenous thrombosis in cirrhosis of the liver. Br Med J1994;309:239–240.

[28] Prieto J, Yuste JR, Beloqui O, Civeira MP, Riezu JI, Aguirre B,et al. Anticardiolipin antibodies in chronic hepatitis C: implica-tions of hepatitis C virus as the cause of the antiphospholipidsyndrome. Hepatology 1996;23:199–204.

[29] Mangia A, Margaglione M, Cascavilla I, Gentile R, Cappucci G,Facciorusso D, et al. Anticardiolipin antibodies in patients withliver disease. Am J Gastroenterol 1999;94:2983–2987.

[30] Amitrano L, Brancaccio V, Guardascione MA, Margaglione M,Iannacone L, D’Andrea G, et al. Inherited coagulation disordersin cirrhotic patients with portal vein thrombosis. Hepatology2000;31:345–348.

[31] Majluf-Cruz A, Hurtado-Monroy R, Sansores Garcia L, Labar-dini-Mendez J. The incidence of protein C deficiency in throm-bosis-related portal hypertension. Am J Gastroenterol1996;91:976–980.

[32] Romero Gomez R, Suarez Garzia E, Lopez Lacomba D, GrandeL, Marchsnte I, Castro Fernandez M. Antiphopholipids antibod-ies: relationship to portal vein thrombosis in cirrhotic liverpatients. J Clin Gastroenterol 2000;31:237–240.

[33] Erkan O, Bozdayi AM, Disibeyaz S, Oguz D, Ozcan M, Bahar K,et al. Thrombophilic gene mutations in cirrhotic patients withportal vein thrombosis. Eur J Gastroenterol Hepatol2005;17:339–343.

[34] Amitrano L, Guardascione MA, Ames PR, Margaglione M,Iannaccone L, Brancaccio V, et al. Increased plasma prothrombinconcentration in cirrhotic patients with portal vein thrombosisand prothrombin G20210A mutation. Thromb Haemost2006;95:221–223.

[35] Kamath PS, Wiesner RH, Malinchoc M, Kremers W, TherneauTM, Kosberg CL, et al. A model for end-stage liver disease.Hepatology 2001;33:464–470.

[36] Huemer RP, Lee KD. Automated Lowry method for microgramprotein determination. Analyt Biochem 1970;37:149–153.

[37] Sabba C, Merkel C, Zoli M, Ferraioli G, Gaiani S, Sacerdoti D,et al. Interobserver and interequipment variability of echo-Doppler examination of the portal vein: effect of cooperativetraining program. Hepatology 1995;21:428–433.

[38] Kamath PS, Wiesner RH, Malinchoc M, Kremers W, TherneauTM, Kosberg CL, et al. A model to predict survival in patientswith end-stage liver disease. Hepatology 2001;33:464–470.

[39] Bosch J, Reverter JC. The coagulopathy of cirrhosis: myth orreality? Hepatology 2005;41:434–435.

[40] De Caterina M, Tarantino G, Farina C, Arena A, di Maro G,Esposito P, et al. Haemostasis unbalance in Pugh-scored livercirrhosis: characteristic changes of plasma levels of protein Cversus protein S. Haemostasis 1993;23:229–235.

[41] Raya-Sanchez JM, Gonzales-Reimers E, Rodriguez-Martin JM,Santolaria- Fernandez F, Molina-Perez M, Rodriguez-Moreno F,et al. Coagulation inhibitors in alcoholic liver cirrhosis. Alcohol1998;15:19–23.

[42] Janssen HL, Wijnhoud A, Haagsma EB, van Uum SH, vanNieuwkerk CM, Adang RP, et al. Extrahepatic portal veinthrombosis: aetiology and determinants of survival. Gut2001;49:720–724.

[43] Gayowski TJ, Marino IR, Doyle HR, Echeverri L, Mieles L,Todo S, et al. A high incidence of native portal vein thrombosis in

M.A. Zocco et al. / Journal of Hepatology 51 (2009) 682–689 689

veterans undergoing liver transplantation. J Surg Res1996;60:333–338.

[44] Sobhonslidsuk A, Reddy KR. Portal vein thrombosis: a concisereview. Am J Gastroenterol. 2002;97:535–541.

[45] Tripodi A, Chantarangkul V, Martinelli I, Bucciarelli P, Mann-ucci PM. A shortened activated partial thromboplastin time isassociated with the risk of venous thromboembolism. Blood2004;104:3631–3634.

[46] Manzanet G, Sanjuan F, Orbis P, Lopez R, Moya A, Juan M,et al. Liver transplantation in patients with portal vein thrombo-sis. Liver Transpl 2001;7:125–131.

[47] Gaiani S, Bolondi L, Li Bassi S, Zironi G, Siringo S, Barbara L.Prevalence of spontaneous hepatofugal portal flow in livercirrhosis. Clinical and endoscopic correlation in 228 patients.Gastroenterology 1991;100:160–167.

[48] Zironi G, Gaiani S, Fenyves D, Rigamonti A, Bolondi L, BarbaraL. Value of measurement of mean portal flow velocity by Dopplerflowmetry in the diagnosis of portal hypertension. J Hepatol1992;16:298–303.

[49] Zoli M, Iervese T, Merkel C, Bianchi G, Magalotti D, MarchesiniG, et al. Prognostic significance of portal hemodynamics inpatients with compensated cirrhosis. J Hepatol 1993;17:56–61.