Article 1

see related editorial on page x

Clinical Impact and Safety of Anticoagulants

Liver
for Portal Vein Thrombosis in Cirrhosis
I. Pettinari, MD, PhD1, R. Vukotic, MD, PhD1, H. Stefanescu, MD, PhD2, A. Pecorelli, MD, PhD1, Mc Morelli, MD3, C. Grigoras2,
Z. Sparchez, MD, PhD2, P. Andreone, MD1 and F. Piscaglia, MD, PhD1 the VALDIG-BO (Vascular Liver Disease Interest study Group-Bologna)

Objectives: Portal vein thrombosis (PVT) is a frequent complication of cirrhosis. Benefit, safety, and duration of
anticoagulant treatment in this setting are controversial issues. The aim of this study was to analyze
the course of PVT in a large cohort of cirrhotic patients undergoing or not anticoagulation therapy.

Methods: The data of 182 patients who presented between January 2008 and March 2016 with cirrhosis
and PVT with at least 3 months of follow-up after the first PVT detection were analyzed. Eighty-one
patients received anticoagulants and 101 were untreated per physician discretion.

Results: The extension of the thrombosis decreased by >50% in 46 (56.8%, with complete recanalization in
31/46) patients under anticoagulation and in 26 (25.7%) untreated patients. Of the 46 patients who
underwent recanalization, 17 (36%) suffered recurrent thrombosis after stopping anticoagulation
therapy. Kaplan–Meier analysis showed a higher survival rate in the treated group (p = 0.010). At
multivariate analysis, anticoagulation was an independent factor associated with longer survival
(HR:0.30, CI:0.10–0.91, p = 0.014). The Child–Turcotte–Pugh classes B/C negatively influenced
survival (hazard ratio, (HR):3.09, confidence interval (CI):1.14–8.36, p = 0.027 for Child–
Turcotte–Pugh B and HR:9.27, CI:2.67–32.23, p < 0.001 for Child–Turcotte–Pugh C). Bleeding
complications occurred in 22 (21.8%) untreated and 16 (19.7%) treated patients, but in only
four cases was it judged to be related to the anticoagulant treatment. No death was reported as a
consequence of the bleeding events.

Conclusions: Anticoagulant treatment is a safe and effective treatment leading to partial or complete
recanalization of the portal venous system in 56.8% of cases, improving the survival of patients with
cirrhosis and PVT. Discontinuation of the therapy is associated with a high rate of PVT recurrence.
SUPPLEMENTARY MATERIAL accompanies this paper at https://doi.org/10.1038/s41395-018-0421-0

Am J Gastroenterol https://doi.org/10.1038/s41395-018-0421-0

Introduction Currently, the optimal management of PVT in cirrhosis remains

Portal vein thrombosis (PVT) is a frequent complication in
patients with cirrhosis, although it may often remain asympto-
matic, at least early after onset [1]. The prevalence of PVT ranges
from 0.6 to 26% in patients with cirrhosis [1–13]. The effects of
PVT on the course of cirrhosis and its overall prognostic signifi-
cance are still not clear. Some studies have suggested that PVT in
cirrhosis could increase the rate of decompensation and worsen
survival whereas others report an insignificant impact of PVT [14,
15]. However, PVT is usually associated with some major clinical
complications, including worsening of portal hypertension and
liver function, and an increased risk of suffering from varices and
gastrointestinal bleeding [1, 16].
unclear and no definitive recommendations have been reported
in clinical guidelines or consensus conferences. The therapeutic
strategies available are anticoagulation therapy and, in some tech-
nically suitable patients, transjugular intrahepatic portosystemic
shunt (TIPS). Although anticoagulation therapy is associated with
a high rate of recanalization, the indications for treating PVT in
cirrhotic patients are still not strictly defined and neither is the
optimal treatment duration.
Accordingly, not all PVT patients have been treated with anti-
coagulants in the past decade [1]. In fact, patients with cirrho-
sis are considered at risk of bleeding events due to the common
findings of a low platelet count and prolonged prothrombin time,

1
Department of Medical and Surgical Sciences, University of Bologna, Azienda Ospedaliero Universitaria S.Orsola Malpighi, Bologna, Italy. 2Gastroenterology
Department, University of Medicine and Pharmacy “Iuliu Hatieganu”, Cluj-Napoca, Romania. 3Unit of Internal Medicine, Azienda Ospedaliero Universitaria S.Orsola
Malpighi, Bologna, Italy. These authors contributed equally and share first authorship: Pettinari I, Vukotic R. The members of the group are listed above references.
Correspondence: F.P. (email: fabio.piscaglia@unibo.it)
Received 27 February 2018; accepted 27 September 2018

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 2 Pettinari et al.
although the quantification of this risk does not appear to be well
defined. The safety of anticoagulation in cirrhosis is still a matter
of debate. At the same time, the pro-thrombotic potential of cir-
rhotic patients is often underestimated and, in the clinical practice,
Liver
it is difficult to assess where on the bleed/clot spectrum a patient
is at any time [17]. The authors took advantage of this gray area of
clinical management to collect a large series of cirrhotic patients
with non-tumoral PVT and to compare those who were treated
with anticoagulants to those who were left untreated based on the
decision of the physician.
The aim of this study was to retrospectively analyze the course
of PVT in cirrhotic patients, and the occurrence and pattern of
bleeding events in a large cohort of patients, either untreated or
treated with anticoagulation therapy in order to assess its safety
and efficacy.
Patients and Methods
Patients
All adult patients with cirrhosis of any etiology and PVT defini-
tively characterized as non-neoplastic (i.e., no tumor vein inva-
sion) according to validated criteria [18] at the General and
University Hospital Sant’Orsola-Malpighi, Bologna (Italy) and the
Regional Institute of Gastroenterology and Hepatology of Cluj-
Napoca (Romania), between January 2008 and March 2016 were
retrospectively evaluated for this study.
The exclusion criteria were: a lack of clinical and demographic
variables, insufficient follow up (<3 months), active tumors apart
from hepatocellular carcinoma (HCC) at the time of the diagnosis
of the PVT, and anticoagulation use at the time of the first observa-
tion at the study center.
Patients presenting a cavernomatous transformation of the por-
tal vein were not excluded. Patients were followed until death, liver
transplantation (LT) or the end of the study (July 2016). Clinical,
epidemiologic, laboratory and radiologic data were collected at the
time of the diagnosis of the PVT, and at intervals of 3, 6, and/or 12
months ± 2 months.
The study was approved by the Ethics Committees of the Gen-
eral and University Hospital Sant’Orsola-Malpighi, Bologna, Italy
and of the University of Medicine and Pharmacy “Iuliu Hatieganu”,
Cluj-Napoca, Romania (111/2015/O/OssN).
Definition
Portal vein thrombosis was defined as the absence of flow in part
of or in the entire lumen of any site among portal vein trunk, por-
tal vein branches, superior mesenteric vein (SMV) or splenic vein
(SV) caused by the presence of solid material within the vein, as
documented by an imaging technique (Doppler ultrasound [US],
contrast enhanced ultrasound [CEUS] computed tomography
[CT], or magnetic resonance imaging [MRI]). With the concur-
rent presence of the following findings, the PVT was considered
to be non-neoplastic: the lack of vascularization of the throm-
bus at contrast imaging, the absence of mass-forming features of
PVT and the absence of evidence of disruption of vessel walls. If
uncertainty persisted after these three criteria had been applied,
The American Journal of Gastroenterology
a bioptic sampling of the thrombus was performed whenever
possible [18]. Spontaneous cavernomatous transformation of the
PVT was also considered to be suggestive of a non-neoplastic
nature [18].
Thrombosis was considered complete when the blood flow was
absent or the thrombus involved more than 90% of the vessel
diameter. Recanalization was considered complete when the por-
tal vein trunk, portal vein branches, SMV, and SV were all com-
pletely patent. Recanalization was considered partial when some
parts of the thrombus persisted but there was at least a 50% reduc-
tion in the thickness or length of the thrombus, or when complete
patency was achieved in the portal vein trunk and in at least one
of the following segments if previously thrombosed: main intra-
hepatic branches, SV or SMV. Lack of recanalization according
to the definition above was considered to be a non-response to
treatment. Thrombosis progression was considered to occur
when thrombus thickness increased > 50% or when the throm-
bosis extended to previously unaffected segments of the spleno-
porto-mesenteric axis. All patients were consecutively enrolled
from the time of the first detection of PVT, including either com-
plete or partial PVT, and regardless of the presence of symptoms
at presentation.
Splenomegaly was considered when the maximal spleen
diameter was greater than 12 cm.
Statistical Analysis
Continuous variables are expressed as mean ± standard deviation
(SD) and categorical variables as numbers and frequencies. Vari-
able distribution was assessed by the Kolmogorov–Smirnov test,
and continuous variables were compared using analysis of vari-
ance (ANOVA). Categorical variables were compared using the
χ2 test with a Yates’ correction.
Survival was calculated as the time from the PVT diagnosis to
death/LT or the last follow-up visit (censoring events) and was
expressed as medians and 95% confidence intervals (95% CIs).
Survival curves were generated by using the Kaplan–Meier method
and were compared with the log rank test.
Cox univariate analysis was carried out to assess the degree of
association between survival and the above-mentioned variables.
Variables associated (p ≤ 0.10) with survival at the univariate
analysis were tested using the Cox multivariate regression model.
Before entering into the multivariate analysis model, the variance
inflation factor (VIF) was calculated to check multicolinearity
among the variables; a VIF value < 5 was considered indicative of
no colinearity. The hazard ratio (HR) and 95% CI were calculated
for independent predictors of survival.
A two-tailed p < 0.05 was considered statistically significant. All
®
statistical analyses were performed using the SPSS 21.0 statistical
package (SPSS Inc., Chicago, Illinois, USA).
Endpoints
The primary endpoints of this study were to explore the safety of
anticoagulant treatment of PVT in terms of the rate and pattern
of the bleeding events, and its efficacy, in terms of rate of PVT
recanalization, in a retrospective cohort of cirrhotic patients. The
www.nature.com/ajg
 clinical impact and Safety of Anticoagulants for Portal Vein Thrombosis in cirrhosis 3

Table 1  Baseline characteristics of the 182 patients included in the study

Variable All patients (n = 182) Treated (n = 81) Untreated (n = 101) P value

Liver
Age (years) (mean, SD) 57.8 ± 11.2 57.9 ± 11.1 57.7 ± 11.3 0.921
Male gender (N, %) 130 (71.4) 56 (69.1) 74 (73.3) 0.621
Etiology*
 HCV (N, %) 55 (30.2) 27 (33.3) 28 (27.7) 0.414
 HBV (N, %) 21 (11.5) 11 (13.6) 10 (9.9) 0.484
 HCV + HBV (N, %) 8 (4.3) 2 (2.5) 6 (5.9) 0.469
 Virus + alcohol (N, %) 14 (7.6) 6 (7.4) 8 (7.9) 1
 Alcohol (N, %) 47 (25.8) 14 (17.3) 33 (32.7) 0.027
 NASH (N, %) 9 (4,9) 4 (4.9) 5 (4.9) 1
 NASH + alcohol (N, %) 3 (1.6) 1 (1.2) 2 (2.0) 1
 Autoimmune (N, %) 8 (4.3) 5 (6.2) 3 (3.0) 0.301
 Cryptogenic (N, %) 13 (7.1) 8 (10.0) 5 (4.9) 0.247
Unknown (N, %) 4 (2.1) 2 (2.5) 3 (3.0) 1
Liver function
 CTP class
   A (N, %) 80 (43.9) 43 (53.1) 37 (36.6) 0.035
   B (N, %) 78 (42.8) 33 (40.7) 45 (44.5) 0.653
   C (N, %) 24 (13.2) 5 (6.2) 19 (18.8) 0.015
    ≤ 9 (N, %)
MELD  43 (23.6) 29 (35.8) 14 (13.9) 0.001
   
Ascites 81 (44.5) 39 (48.1) 59 (58.4) 0.181
(N, %)
   Varices 141 (77.5) 67 (82.7) 74 (73.3) 0.155
(N, %)
   Hepatic encephalopathy (N, %) 21 (11.5) 5 (6.2) 16 (15.8) 0.060
 Laboratory tests
   Bilirubin (mg/dL) (mean, SD) 2.4 (2.1) 1.9 (1.6) 2.8 (2.4) 0.010
    Platelets (mean, SD) 92.8 (72.0) 84.8 (57.2) 99.3 (81.8) 0.209
   INR (mean, SD) 1.3 (0.5) 1.1 (0.3) 1.5 (0.5) <0.0001
 Tumor
   HCC discovered at PVT diagnosis (N, %) 30 (16.5) 13 (16.0) 17 (16.8) 1
   HCC known before PVT diagnosis (N,%) 41 (22.5) 19 (23.4) 22 (21.8) 0.859
HCV hepatitis C virus, HBV hepatitis B virus, NASH non-alcoholic steatohepatitis, CPT Child–Turcotte–Pugh, MELD Mode for End.Stage Liver Disease, INR International
Normalized Ratio, PVT portal vein thrombosis, HCC hepatocellular carcinoma. Bold font represents statistically significant p values

secondary endpoints were to explore the rate of PVT recurrence
after recanalization and to identify the clinical factors significantly
influencing survival.
Results
Of the 208 patients retrospectively identified during the pre-
defined enrollment period, 23 were excluded due to the lack of
clinical demographic variables or sufficient follow up and 3 due to
the presence of non-HCC active cancer at the time of diagnosis.
Thus, 182 cirrhotic patients with PVT were finally included in
the analysis.
Patient baseline characteristics and extent of the PVT
The median follow-up was 19 months (range 3–94) after PVT
detection. The main characteristics of the study population
at baseline are shown in Table  1. Hepatitis C virus (HCV) was
the leading cause of liver disease occurring in 55/182 patients
(30.2%), followed by alcohol-related cirrhosis (47/182, 26.4%).
HCC was present in 30/182 patients (16.5%).

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 4 Pettinari et al.

Table 2  Extent of portal vein thrombosis at diagnosis Table 3  Characteristics and type of anticoagulation

Variable All patients Treated Untreated P value Treated (n = 81)

(n = 182) (n = 81) (n = 101)
Time interval between PVT diagnosis and start of anticoagulation
Liver

Imaging for PVT assessment

   ≤6 months (N, %) 66 (81.5)
  Ultrasound (N, %) 127 (69.8) 54 (66.7) 73 (72.3) 0.422
   7–12 months (N, %) 7 (8.6)
  CT (N, %) 46 (25.3) 21 (25.9) 25 (24.7) 0.865
   ≥13 months (N, %) 8 (9.9)
   MRI (N, %) 9 (4.9) 6 (7.4) 3 (3.0) 0.162
Type of treatment
Portal trunk thrombo- 132 (72.5) 59 (72.8) 73 (72.3) 1
sis: Overall (N, %)   LMWH (N, %) 56 (69.1)

  Partial thromb 122 (67.0)/ 51 (63.0)/ 71 (70.3) 0.065   VKA (N, %) 10 (12.3)
(N, %)/    Fondaparinux (N, %) 15 (18.5)
 Complete thromb 10 (5.5) 8 (9.9) 2 (2.0) EBL before anticoagulation 22 (27.2)
(N, %)
Therapy duration (months) (mean, SD) 13.4 (14)
Intrahepatic portal 96 (52.7) 52 (64.2) 44 (43.6) 0.007
branches: Overall Beta-blockers during therapy (N, %) 60 (74.1)

 Partial thromb 77 (42.3) 41 (50.6) 36 (35.6) 0.049 PVT portal vein thrombosis, LMWH low-molecular-weight heparin, VKA vitamin
(N, %) K antagonist,
EBL endoscopic band ligation, SD standard deviation
 Complete thromb 19 (10.4) 11 (13.6) 8(7.9) 0.156
(N, %)
Splenic vein 24 (13.2) 10 (12.3) 14 (13.9) 0.828
One hundred and thirty-two patients (72.5%) had involvement
 Partial thromb 21 (11.5) 10 (12.3) 11 (10.9) 0.285
of the portal trunk (partial in 122 cases and complete in 10 cases).
(N, %)
The presence of thrombosis in the portal vein branches was signifi-
 Complete thromb 3 (1.6) 0 3 (3.0)
(N, %)
cantly more frequent in the treated group (52/82 [64%] vs 44/101
[43.6%]; p = 0.007). Portal cavernoma was more frequent in the
Superior mesenteric 41 (22.5) 19 (23.4) 22 (21.8) 0.859
vein
untreated patients (p = 0.030).
  Partial (N, %) 36 (19.8) 17 (21.0) 19 (18.8) 0.921
Anticoagulant Therapy
 Complete (N, %) 5 (2.7) 2(2.5) 3 (3.0) Anticoagulant therapy was administered to 81 patients (44.5%)
Cavernoma (N, %) 20 (11.0) 4 (4.9) 16 (15.8) 0.030 while 101 patients (55.5%) received no anticoagulation (Table 3).
Splenomegaly 166 (94.3) 76 (93.8) 90 (89.1) 0.188 Sixty-six out of 81 (81.5%) started treatment within 6 months
(N, %)a from diagnosis, 7/81 (8.6%) between 6 and 12 months, and
Best recanalization 72 (39.6) 46 (56.8) 26 (25.7) <.0001 8/81 (9.9%) after 12 months. The mean treatment duration was
by the end of the 13.4 ± 14.0 months. Prior to anticoagulation, all patients had
follow-up: Overall
received an esophagogastroduodenoscopy for variceal screening.
  Partial (N,%) 28 (15.4) 15 (18.5) 13 (12.9) <.0001 All 66 patients with high-risk varices underwent either variceal
 Complete (N;%) 44 (24.2) 31 (38.3) 13 (12.9) <.0001 prophylactic eradication by endoscopic ligation prior to start-
PVT portal vein thrombosis, CT computed tomography, MRI magnetic resonance ing anticoagulation (12/81, 14.8%) and/or were treated with beta
imaging. Bold font represents statistic ally significant p values blockers (54/81, 66.6%).
a
Data available in 176 patients
Fifty-six patients (69.1%) were treated with low-molecular-
weight heparin (LMWH), 15 patients (18.5%) with fondaparinux
and 10 patients (12.3%) with oral anticoagulants (vitamin K antag-
onists, VKAs). Of the 56 patients treated with LMWH, 32 (57%)
The Child–Turcotte–Pugh (CTP) class was significantly better received ≤ 50% of the full dose of LMWH, administered once a
in the treated than in the untreated group (CTP-A 53.1 vs. 36.6%, day whereas the remainder received 51 to 100% of the expected
p = 0.035; CTP-C 5 vs 19% p = 0.015). The Model for End-Stage full dose, administered in two injections per day.
Liver Disease (MELD) score was also lower in the treated group
(MELD < 9 35.8 vs. 13.9%, p = 0.001). Course of Portal Vein Thrombosis
No significant difference was found in the platelet count between Recanalization was observed in 46 of the 81 treated patients
groups. (56.8%, after a median of 5 months (range 1–13) whereas no
The diagnosis and the extension of PVT at baseline were response to anticoagulation was observed in 35 patients. In 31/46
obtained by Doppler examination in 127 patients (69.8%), CT in (67.4%) of the treated patients who responded to anticoagulant
46 patients (25.2%), and MR in 9 patients (4.9%) (Table  2). The treatment, the PVT recanalization was complete while it was par-
extent of the portal vein thrombosis is shown in Table 2. tial in the remaining 15/46 (32.6%).

The American Journal of Gastroenterology www.nature.com/ajg

 clinical impact and Safety of Anticoagulants for Portal Vein Thrombosis in cirrhosis
1.0 Anticoagulation
No anticoagulation
0.8
Cumulative survival
0.6
0.4
0.2
0.0
0 20 40 60 80 100
Months
Fig. 1 Survival analysis according to treatment allocation
Twenty-eight patients (60.9%) achieved recanalization at 3
months after the start of the anticoagulant therapy, 13 patients
(28.3%) after 6 months and 5 patients (10.8%) after 12 months.
Of the untreated patients, 26/101 (25.7%) had spontaneous
recanalization, complete in 13 (50%) cases and partial in 13 (50%).
Recanalization occurred after a median of 5 months.
The recanalization rate was significantly higher in the anti-
coagulation group than in the untreated group (56.8 vs. 25.7%
p < 0.0001) (Table 2). After excluding patients with cavernoma, the
recanalization rate was 59.7 and 30.6%, respectively (p = 0.0235).
The rates of partial and complete recanalization were both higher
in treated than in untreated patients (Table 2).
Multivariate analysis revealed no predictive factors of recanali-
zation in treated patients (data not shown).
Follow-up After Discontinuation of The Anticoagulant Therapy
Seventeen of the 46 recanalized patients (36%) had recurrence of
PVT after stopping the anticoagulant therapy.
Seven of the 35 patients (20%) who did not achieve recanaliza-
tion at the time of anticoagulation discontinuation showed addi-
tional progression of the PVT after treatment discontinuation.
Safety
Bleeding events in patients not treated with anticoagulants. Dur-
ing the follow-up, 22 untreated patients (22/101, 21.8%) presented
with events of bleeding. In all cases, the bleeding events were re-
lated to portal hypertension, supposedly aggravated by PVT. In
particular, 16 patients presented variceal bleeding, 1 hemorrhoi-
dal bleeding and 5 bleeding due to gastric antral vascular ectasia
(GAVE). Three patients had > 2 events of bleeding requiring hos-
pitalization; two of the latter 3 patients had a concurrent progres-
sion of the thrombosis.
© 2018 The American College of Gastroenterology
5
Bleeding events in patients treated with anticoagulants. Bleeding
events were reported in 16 treated patients (16/81, 19.7%): 12 re-
lated to portal hypertension (4 variceal bleeding, 6 hemorrhoidal
bleeding and 2 due to GAVE), and 4 probably favored by anti-
Liver
coagulant treatment (following trauma or accidental falls). There
were no significant differences in the rate of bleeding complica-
tions between the two groups (p = 0.855).
No death was reported within 30 days of the bleeding events in
either group.
Outcomes
The median overall survival (Fig. 1) for the treated group was 70
months and for the untreated group 59 months. Kaplan–Meier
curve analysis revealed higher cumulative survival in the treated
group than in the untreated group (p = 0.010). Twenty-four
patients underwent LT (18 [17.8%] untreated and 6 [7.4%] treated
patients, p = 0.0476). Univariate analysis showed that antico-
agulant treatment, liver function (expressed in CTP classes) and
infection at the diagnosis of PVT were associated with survival
(Table 4). Anticoagulant treatment and CTP B and C classes were
the only independent factors related to, respectively, longer and
shorter survival at multivariate analysis (Table 4).
Multivariate analyses carried out to identify factors predictive
of survival in treated patients and in the subgroup of the treated
patients who experienced progression or relapse of PVT after
stopping anticoagulant treatment did not identify any significant
factor (Supplementary Tables 1 and 2). Finally, a subgroup analy-
sis of the treatment outcomes according to the CTP class revealed
significantly shorter survival and higher rates of bleeding along
with the progression of the CTP class with the worst outcomes in
CTP-C patients, as expected according to the natural history of
the disease (Supplementary Table 3). However, no relevant differ-
ence in recanalization rates was observed between CPT-A and -B
patients. No case of recanalization was observed in the in CPT-C
class, which was, however, made of only 5 patients. The rate of
recurrence after treatment discontinuation was lower in CPT-B
than CPT-A class, although without reaching the statistical
significance.
Discussion
Portal vein thrombosis is a frequent complication in liver cirrho-
sis but its natural history as well as therapeutic management have
not yet been clearly addressed by either international guidelines
or consensus conferences. The present study confirmed that reca-
nalization of PVT may occur spontaneously, but it is significantly
favored by anticoagulant treatment. Of note, limited to the study
observation timeframe, anticoagulation was not associated with
an increased risk in major bleeding complications and it positively
influenced survival.
It is worth mentioning that the present study population was by
far the largest regarding cirrhotic patients with PVT, reporting not
only long-term observation of the course of PVT, but also a sur-
vival analysis of both patients who underwent anticoagulant treat-
ment and those who did not.
The American Journal of Gastroenterology
 6 Pettinari et al.

Table 4  Factors associated with survival

Univariate analysis Multivariate analysis

Liver

Variable HR 95% CI P value HR 95% CI P value

Treatment
  Untreated Reference Reference
  Treated 0.405 0.185–0.883 0.023 0.263 0.105–0.657 0.004
  Age 1.015 0.979–1.052 0.417
   Male gender 1.077 0.487–2.379 0.855
Etiology
  Virus 1.474 0.680–3.196 0.326
  Alcohol 1.081 0.477–2.448 0.852
  Other 2.880 0.681–12.179 0.151
Liver function
  Child–Turcotte–Pugh A Reference Reference
  Child–Turcotte–Pugh B 2.728 1.036–7.188 0.042 3.086 1.139.8.365 0.027
  Child–Turcotte–Pugh C 10.701 3.810–30.053 <0.0001 9.270 2.666–32.232 <0.0001
  Platelets 0.993 0.983–1.003 0.170
  Varices 1.227 0.788–1.911 0.363
  Treatment of esophageal varices 0.891 0.392–2.204 0.783
before PVT
  Treatment of gastric varices before 0.404 0.054–3.206 0.378
PVT
  Esophageal bleeding before PVT 1.359 0.421–4.623 0.586
  Bleeding complications during PVT 1.058 0.466–2.403 0.893
Tumor 1.345 0.510–3.549 0.549
 HCC at diagnosis 1.529 0.531–4.402 0.431
 HCC before diagnosis 0.036 0.000–5.626 0.198
Treatment HCC
 Documented infection at diagnosis 2.599 1.108–6.097 0.028 1.084 0.561–2.093 0.811
of PVT
Portal vein trunk (N;%) 1.929 0.735–5.063 0.182
 Partial Reference
 Complete 1.860 0.952–3.635 0.069 1.965 0.951–4.062 0.068
Intrahepatic branches 1.330 0.637–2.773 0.448
 Partial Reference
 Complete 1.572 0.350–7.055 0.555
Splenic vein 1.458 0.549–3.870 0.449
 Partial Reference
 Complete 1.520 0.678–3.410 0.310
Superior mesenteric vein 1.582 0.716–3.494 0.257
 Partial Reference
 Complete 1.489 0.763–2.904 0.243
Cavernoma (N;%) 1.460 0.346–6.150 0.606
Splenomegaly (N;%) 1.178 0.617–2.248 0.619
HR hazard rate, CI confidence interval, PVT portal vein thrombosis, HCC hepatocellular carcinoma. Bold font represents statistically significant p values

The American Journal of Gastroenterology www.nature.com/ajg

 clinical impact and Safety of Anticoagulants for Portal Vein Thrombosis in cirrhosis
The rate of spontaneous partial or complete recanalization
of PVT rates ranged from 5 to 48% in the literature [14, 19], the
wide range due to heterogeneity in terms of thrombus extension,
study populations and imaging methods. In a study by Luca et al.
involving 42 patients with cirrhotic PVT, a spontaneous reduction
in thrombosis occurred in 47.6% of those with partial thrombosis
[14] while Senzolo et al. reported a spontaneous rate of improve-
ment in thrombosis in 5% [19]. In the present study, 25.7% of
untreated subjects showed spontaneous recanalization.
The rates of recanalization under anticoagulant therapy in the
present study (56.8%) were in keeping with those reported to date
(37–93%) [20] and in particular with a recent metanalysis by Lof-
fredo et al. [21]. It should be pointed out that the population in the
present study alone was bigger than approximately half of that of
all the studies taken together in this recent metanalysis [21].
The rates of complete recanalization were reported to be 42–45%
in patients treated with VKAs [6, 22] and similarly, it ranged from
33 to 50% in patients treated with LMWH [19, 23]. In keeping with
these data, no statistically significant differences were found in the
present study in terms of rates of recanalization according to differ-
ent anticoagulant therapies. To date, no answer can be provided to
the question as to whether certain anticoagulants work better than
others. Both the decision to treat and the choice of the type of anti-
coagulant are influenced by both drug-related and patient-related
factors. In fact, LMWH has a shorter time duration of effects, and
its dose can be easily and precisely modulated. However, its phar-
macokinetics are more influenced by a reduced glomerular filtra-
tion rate, it requires daily subcutaneous injection(s) and, notably,
heparin resistance due to low levels of antithrombin III may occur
in cirrhosis. Conversely, VKAs can require up to a few days to
spontaneously restore normal clotting function. Moreover, their
daily dose is titrated based on the International Normalized Ratio
(INR) value, which unfortunately is an imperfect measure in cir-
rhotics and indeed the anticoagulant effect may oscillate, addition-
ally increasing the risk of bleeding. On the other hand, they are
not significantly affected by the renal impairment and are taken
orally. Direct oral anticoagulants (DOACs) have not yet been stud-
ied in cirrhotic patients with severely impaired liver function and
their dosage cannot easily be modulated, but the recent data are
promising when it comes to their safety profile compared to the
traditional anticoagulants [24].
Some relevant considerations were prompted by our findings in
anticoagulated subjects. Not all patients displayed recanalization
within the first 3 or 6 months, suggesting that prolongation of the
therapy should be attempted even after apparent initial inefficacy.
This is further confirmed by the finding that, after the discontinua-
tion of therapy, 36% of our patients who had a partial or a complete
recanalization subsequently reported a recurrence/progression of
PVT. These results are in line with those reported in the literature
[22, 23] and to those of Delgado et al [22]. who showed recurrence
or progression of PVT in approximately one third of patients who
discontinued treatment. Complete recanalization was achieved by
31 (38.3%) patients after a median of 6 months after starting antico-
agulant therapy. Partial recanalization was achieved in 15 (18.5%)
patients after a median of 3 months of therapy. As described in an
© 2018 The American College of Gastroenterology
7
observational study [23], where full recanalization was reached in
33% and partial in 50% of patients after 6 months of anticoagulant
therapy, continued treatment might perhaps subsequently allow
complete recanalization. Therefore, the continuation of anticoagu-
Liver
lant therapy for at least 6 months could have a role in the secondary
prophylaxis of PVT recurrence after having achieved recanaliza-
tion. Whether similar results of secondary prophylaxis could be
achieved with lower doses of anticoagulants remains to be explored.
Nevertheless, 8 (9.9%) of our patients had progression of the
thrombosis during treatment, which is again in line with the recent
metanalysis of Loffredo et al. [21] which reported a 9% rate of pro-
gression during therapy (vs. 33% of untreated patients). Whether
these are truly refractory patients or simply sub-maximally treated
patients has still not been clarified and requires prospective studies
with focused dedicated laboratory testing.
Currently, concerns regarding the use of anticoagulant therapy
in patients with liver disease are mainly related to the fear of an
increased risk of bleeding due to the compromised coagulation
function and to the risk of acute variceal hemorrhage due to portal
hypertension. In agreement with the current literature, the present
study has shown that individualized anticoagulant therapy in cir-
rhotic patients with PVT is safe as bleeding events do not occur at
higher rates than in untreated patients, and the hemorrhagic com-
plications are seldom ascribed to the anticoagulant therapy itself.
In the present study, 16 (19.7%) treated patients had hemorrhagic
complications, but only four were likely related to the anticoagu-
lant treatment. It is important to point out that all patients with
high-risk varices at the time of the diagnosis of PVT underwent
prophylactic therapy (variceal band ligation and/or beta-blockers)
prior to initiating anticoagulant therapy. The rate of hemorrhagic
complications associated with therapy for PVT reaches 18% in the
literature [21, 22]. The bleeding rate seems to correlate with severe
thrombocytopenia and the use of VKAs [22]. No such correlation
emerged from our data whereas nearly all bleeding events were
related to portal hypertension and the incidence rates were simi-
lar in both the treated (19.7%) and the untreated (21.8%) groups,
hence without any additional hazard related to anticoagulation.
Unfortunately, data about the number of hospitalizations, length
of intensive care unit admissions and number of transfusions
associated with bleeding episodes were not sufficient for statistical
reporting, since this was a retrospective real-life study and such
information could not be systematically retrieved.
Our data also show no significant difference in the rates of reca-
nalization between CPT-A and -B patients. Conversely, recanaliza-
tion occurred less in CPT-C class (Suppl Table 3) but only 5 CPT-C
patients were treated which also showed longer survival despite
40% rates of bleeding (making it difficult to dissect the beneficial
contribution of anticoagulation from the possibility that only the
less severe CPT-C patients were treated). Surely, this more severe
patient category deserves targeted investigation. Apparently CPT-B
patients were the category benefitting most of the anticoagulation,
showing a lower rate of recurrence after stopping treatment and
longer survival than the untreated subjects (Suppl Table  3), in
keeping with a previous prospective study which showed survival
benefit with enoxaparin in such patients [25].
The American Journal of Gastroenterology
 8 Pettinari et al.
In the present study population, not all the treated patients
received full-dose anticoagulation. Since this was a retrospec-
tive study, the reasons for individual treatment schedules were
not reported but, most likely, the reduced doses were prescribed
Liver
to avoid exposing fragile patients to an increased bleeding risk
while maintaining beneficial anticoagulant effect. The individually
tailored approach was probably relevant in optimizing the risk/
benefit ratio by producing significant recanalization rates without
having an additional bleeding risk. Another limitation is that the
reasons for deciding whether or not to start anticoagulation can-
not be clearly determined. However, the untreated patients in this
study displayed higher baseline INR values as well as higher base-
line CTP and MELD scores, suggesting that decompensated cir-
rhosis and/or an impaired coagulative state were conditions which
tended to deter physicians from prescribing anticoagulation ther-
apy, even following PVT occurrence. However, the present data
indicated a good safety profile for anticoagulation. Future prospec-
tive trials should also investigate attenuated anticoagulation sched-
ules in comparison to those adopted in systemic anticoagulation
in non-cirrhotic patients. This hypothesis should also be consid-
ered for DOACs as they have not yet been approved for PVT, and
very limited data exist regarding their use in cirrhotic patients [24,
26]. However, their utilization is very tempting due to the ease of
use. Accordingly, prospective data exploring safety and efficacy of
DOACs compared to the traditional anticoagulants specifically in
PVT are warranted in the next future.
Limitations of the present study are mainly related to its retro-
spective nature. One is the lack of a standardized anticoagulation
protocol, which translates into the impossibility of making clear
cut recommendations regarding anticoagulation drug choice and
doses. Moreover, the assessment of PVT extension and evaluation
of the degree of response were provided using different imaging
techniques. Ultrasound is accurate in investigating the presence
and extension of intrahepatic thrombosis; however, it is not as
accurate as CT and MRI in the case of extrahepatic involvement.
Computed tomography and MRI are expensive and require the
injection of potentially harmful contrast agents while US is less
costly and easily complemented by color/duplex Doppler and by
safe contrast agents. Computed tomography also implies exposure
to ionizing radiation. Therefore, CT and MRI are much less feasible
for the short-/mid-term follow-up of PVT to monitor progression
or regression. For this reason, the US was the technique adopted
in the majority of the patients in the present study. In the authors’
view, the different imaging techniques have complementary rather
than competing roles in this setting.
Another point to highlight is that it is often impossible to correctly
date the thrombus. In the present study, all patients were consecu-
tively enrolled at the time of the first observation of PVT, regard-
less of the presence or absence of PVT-related symptoms. Partial
PVT is commonly asymptomatic and, furthermore, symptoms may
develop only later when partial PVT progresses to complete PVT.
Finally, only the detection of a cavernomatous transformation of
the portal veins excludes acute PVT. Hence, the presence/absence
of symptoms is not a reliable index for dating the first occurrence of
PVT when imaging cannot provide this information.
The American Journal of Gastroenterology
Our study included similar rates of patients with viral hepatitis B
or C and alcoholic cirrhosis, but a limited number of cirrhosis due
to Non Alcoholic Liver Disease. Whether the present results fully
apply also to the latter etiology remains to be elucidated.
Importantly, unlike previous studies which concentrated only on
the rates of recanalization, the present study also reported survival
information and showed that patients who received anticoagulation
therapy following PVT diagnosis had greater survival rates than the
untreated patients, independently of liver dysfunction severity. This
result appeared to confirm the data of a prospective, randomized
controlled trial regarding the primary prophylaxis of PVT [25].
In conclusion, anticoagulant therapy appeared to be effective in
cirrhotic patients with PVT, reaching recanalization rates of 56.8%,
more than doubling the spontaneous recanalization rates. Reca-
nalization was commonly obtained in the first 6 months after the
start of treatment. Prior to initiating treatment, either beta-blocker
therapy was started or prophylactic endoscopic band ligation of
high-risk varices was carried out to protect from the risk of variceal
bleeding during treatment. Finally, anticoagulant treatment signif-
icantly improved the survival of cirrhotic patients with PVT.
However, although the treatment response occurred mainly
within 6 months, precocious discontinuation for the presumed
lack of treatment response should be avoided.
In patients who responded to anticoagulation treatment, its dis-
continuation was associated with a high risk of PVT recurrence.
For this reason, it could be speculated that some patients might
benefit from a secondary prophylactic strategy which should be
maintained after recanalization.
Finally, anticoagulant treatment in cirrhotic patients with PVT
was shown to be safe in the short/mid-term. These findings high-
lighted the need for randomized prospective trials to test the safety
and efficacy of a long-term secondary prophylaxis with anticoagu-
lants for PVT in cirrhosis.
Members of the Vascular Liver Disease Interest study
Group-Bologna (VALDIG-BO). L Badia, S Berardi, A Cappelli,
M Cescon, F Conti, A Cucchetti, C Galaverni, R Golfieri,
A Granito, C Mosconi, M Renzulli, Mr Tamè, G Verucchi,
G Vitale, L Bolondi.
CONFLICT OF INTEREST
Guarantor of the article: Fabio Piscaglia.
Specific author contributions: Pettinari I: conducted the study
and drafted the manuscript . Vukotic R: conducted the study,
analyzed and intrepreted the data, and drafted the manuscript.
Stefanescu H: conducted the study, analyzed and intrepreted
the data, and drafted the manuscript. Pecorelli A: analyzed and
intepreted the data. Morelli MC, Grigoras C, and Sparchez Z:
collected data. Andreone P: interpreted the data and revised the
manuscript. Piscaglia F: planned the study, interpreted data, and
drafted and revised the manuscript. All authors approved the
final draft submitted. All members of the study group helped
collecting minor data.
Financial support: None.
Potential competing interests: None.
www.nature.com/ajg
 clinical impact and Safety of Anticoagulants for Portal Vein Thrombosis in cirrhosis
Study Highlights
What is the current knowledge?
✓ There are no cutting-edge indications for the treatment of
non-neoplastic PVT in cirrhosis and its clinical manage-
ment is still variable and inhomogeneous.
✓ Few data exist regarding the impact of anticoagulation on
PVT improvement, but hardly anything is known regard-
ing the impact of therapeutic anticoagulation on overall
survival in cirrhotic patients with PVT.
What is new here?
✓ Our data highlighted the satisfactory safety and efficacy
profile of anticoagulant treatment for PVT in cirrhosis and
showed an independent beneficial effect on survival.
✓ A non-negligible rate of recurrence of thrombosis occurred
after treatment discontinuation.
✓ These real-life, large cohort data support the use of anti-
coagulation in non-neoplastic PVT in cirrhosis and produce
evidence warranting prospective studies to standardize the
doses and duration of anticoagulation in this setting.
References
1. Violi F, Corazza GR, Caldwell SH, et al. Portal vein thrombosis relevance
on liver cirrhosis: Italian Venous Thrombotic Events Registry. Intern
Emerg Med. 2016;11:1059–68.
2. Sacerdoti D, Serianni G, Gaiani S, et al. Thrombosis of the portal venous
system. J Ultrasound. 2007;10:12–21.
3. Ponziani F, Zocco MA, Garcovich M, et al. What we should know about
portal vein thrombosis in cirrhotic patients: a changing perspective.
World J Gastroenterol. 2012;18:5014–20.
4. Nery F, Chevret S, Condat B, et al. Causes and consequences of portal
vein thrombosis in 1,243 patients with cirrhosis: results of a longitudinal
study. Hepatology. 2015;61:660–7.
5. Zocco MA, Di Stasio E, De Cristofaro R, et al. Thrombotic risk factors in
patients with liver cirrhosis: correlation with MELD scoring system and
portal vein thrombosis development. J Hepatol. 2009;51:682–9.
6. Francoz C, Belghiti J, Vilgrain V, et al. Splanchnic vein thrombosis in can-
didates for liver transplantation: usefulness of screening and anticoagula-
tion. Gut. 2005;54:691–7.
7. Maruyama H, Okugawa H, Takahashi M, et al. De novo portal vein
thrombosis in virus-related cirrhosis: predictive factors and long-term
outcomes. Am J Gastroenterol. 2013;108:568–74.
8. Garcia-Pagan JC, Valla DC. Portal vein thrombosis: a predictable mile-
stone in cirrhosis? J Hepatol. 2009;51:632–4.
© 2018 The American College of Gastroenterology
9
9. Molmenti EP, Roodhouse TW, Molmenti H, et al. Thrombendvenectomy
for organized portal vein thrombosis at the time of liver transplantation.
Ann Surg. 2002;235:292–6.
10. Yerdel MA, Gunson B, Mirza D, et al. Portal vein thrombosis in adults
undergoing liver transplantation: risk factors, screening, management,
Liver
and outcome. Transplantation. 2000;69:1873–81.
11. Primignani M. Portal vein thrombosis, revisited. Dig Liver Dis.
2010;42:163–70.
12. Ponziani FR, Zocco MA, Campanale C, et al. Portal vein thrombosis:
insight into physiopathology, diagnosis, and treatment. World J Gastroen-
terol. 2010;16:143–55.
13. Shah V. Molecular mechanisms of increased intrahepatic resistance in
portal hypertension. J Clin Gastroenterol. 2007;41:S259–S261.
14. Luca A, Caruso S, Milazzo M, et al. Natural course of extrahepatic
nonmalignant partial portal vein thrombosis in patients with cirrhosis.
Radiology. 2012;265:124–32.
15. Englesbe MJ, Kubus J, Muhammad W, et al. Portal vein thrombosis and
survival in patients with cirrhosis. Liver Transpl. 2010;16:83–90.
16. Amitrano L, Guardascione MA, Brancaccio V, et al. Risk factors and clini-
cal presentation of portal vein thrombosis in patients with liver cirrhosis.
17.
J Hepatol. 2004;40:736–41.
Tripodi A. Liver diseases and hemostatic (Dys)function. Semin Thromb
Hemost. 2015;41:462–7.
18. Piscaglia F, Gianstefani A, Ravaioli M, et al. Criteria for diagnosing
benign portal vein thrombosis in the assessment of patients with cirrhosis
and hepatocellular carcinoma for liver transplantation. Liver Transpl.
2010;16:658–67.
19. Senzolo M, Sartori TM, Rossetto V, et al. Prospective evaluation of antico-
agulation and transjugular intrahepatic portosystemic shunt for the man-
agement of portal vein thrombosis in cirrhosis. Liver Int. 2012;32:919–27.
20. Qi X, De Stefano V, Li H, et al. Anticoagulation for the treatment of portal
vein thrombosis in liver cirrhosis: A systematic review and meta-analysis
of observational studies. Eur J Intern Med. 2015;26:23–9.
21. Loffredo L, Pastori D, Farcomeni A, et al. Effects of anticoagulants in
patients with cirrhosis and portal vein thrombosis: a systematic review
and meta-analysis. Gastroenterology. 2017;153:480–7.
22. Delgado MG, Seijo S, Yepes I, et al. Efficacy and safety of anticoagulation
on patients with cirrhosis and portal vein thrombosis. Clin Gastroenterol
Hepatol. 2012;10:776–83.
23. Amitrano L, Guardascione MA, Menchise A, et al. Safety and efficacy of
anticoagulation therapy with low molecular weight heparin for portal
vein thrombosis in patients with liver cirrhosis. J Clin Gastroenterol.
2010;44:448–51.
24. Intagliata NM, Henry ZH, Maitland H, et al. Direct oral anticoagulants
in cirrhosis patients pose similar risks of bleeding when compared to
traditional anticoagulation. Dig Dis Sci. 2016;61:1721–7.
25. Villa E, Cammà C, Marietta M, et al. Enoxaparin prevents portal vein
thrombosis and liver decompensation in patients with advanced cirrhosis.
Gastroenterology. 2012;143:1253–60.
26. De Gottardi A, Trebicka J, Klinger C, et al. Antithrombotic treatment with
direct-acting oral anticoagulants in patients with splanchnic vein throm-
bosis and cirrhosis. Liver Int. 2017;37:694–9.
The American Journal of Gastroenterology