REVIEWS Management of portal vein thrombosis in liver cirrhosis Xingshun Qi, Guohong Han and Daiming Fan Abstract | Portal vein thrombosis (PVT) is a fairly common complication of liver cirrhosis. Importantly, occlusive PVT might influence the prognosis of patients with cirrhosis. Evidence from a randomized controlled trial has shown that anticoagulation can prevent the occurrence of PVT in patients with cirrhosis without prior PVT. Evidence from several case series has also demonstrated that anticoagulation can achieve portal vein recanalization in patients with cirrhosis and PVT. Early initiation of anticoagulation therapy and absence of previous portal hypertensive bleeding might be positively associated with a high rate of portal vein recanalization after anticoagulation. However, the possibility of spontaneous resolution of partial PVT questions the necessity of anticoagulation for the treatment of partial PVT. In addition, a relatively low recanalization rate of complete PVT after anticoagulation therapy suggests its limited usefulness in patients with complete PVT. Successful insertion of a transjugular intrahepatic portosystemic shunt (TIPS) not only recanalizes the thrombosed portal vein, but also relieves the symptomatic portal hypertension. However, the technical difficulty of TIPS potentially limits its widespread application, and the risk and benefits should be fully balanced. Notably, current recommendations regarding the management of PVT in liver cirrhosis are insufficient owing to low-quality evidence. Qi, X. et al. Nat. Rev. Gastroenterol. Hepatol. 11, 435–446 (2014); published online 1 April 2014; doi:10.1038/nrgastro.2014.36
Introduction
Xijing Hospital of Digestive Diseases, Fourth Military Medical University, No. 17 West Changle Road, Xi’an, 710032 China (X.Q., G.H., D.F.).
Portal vein thrombosis (PVT) is characterized by the formation of a thrombus within the portal vein trunk and intrahepatic portal branches.1,2 Evolution of PVT mainly includes the degree (that is, partial occlusion, complete occlusion and fibrotic cord), stage (fresh thrombus, recent or old thrombus, and portal cavernoma), and extension (portal vein alone or extension into the splenic and/or superior mesenteric vein).3–5 A classification system proposed by Yerdel is frequently used, as follows: grade 1, 50% PVT with or without minimal obstruction of the superior mesenteric vein; grade 3, complete portal vein and proximal superior mesenteric vein thrombosis; and grade 4, complete portal vein and entire superior mesenteric vein thrombosis.6 In contrast to other classifications,7–9 this system considers both the degree and extension of PVT. PVT is frequently encountered in the setting of liver cirrhosis, especially at the decompensated or advanced stage. 10,11 However, the prevalence and incidence of PVT often varies among different studies owing to heterogeneous diagnostic methods and target populations (reviewed in detail elsewhere10). In a large case series including 701 patients with cirrhosis but without hepatocellular carcinoma who were undergoing routine Doppler ultrasonography, the prevalence of PVT was 11.2% (79 patients).12 In another study, the prevalence
Correspondence to: D.F. [email protected]; G.H. [email protected]
Competing interests The authors declare no competing interests.
of PVT was 8.4% (21 of 251 patients) at the time of listing for liver transplantation, and the incidence of de novo PVT in transplant recipients was 7.4% (17 of 230 patients) during a mean follow-up of 12.1 months.13 Results of a prospective study demonstrated that the incidence of de novo PVT in liver cirrhosis was 16.4% (12 of 73 patients) within 1 year.14 Similarly, a retrospective study shows that the 1‑year and 5‑year cumulative incidence of de novo PVT is 12.8% and 20.0%, respectively, in virus-related cirrhosis.15 However, several inconsistencies and drawbacks in the design and data analysis of these studies—including splenic vein thrombosis alone falsely classified as PVT, uneven distribution of incidence of PVT during follow-up, and difficulty of using Doppler ultrasonography in the regular evaluation of PVT—raise questions about the reliability of these findings.16 In 2011, a pre-planned satellite study (using data from a published multicentre randomized trial in which the incidence of small hepatocellular carcinoma was compared between patients with compensated cirrhosis receiving 3‑month versus 6‑month ultrasonographic periodicities17) reported that 101 of 898 patients with compensated cirrhosis and a prior patent portal vein developed PVT during a mean follow-up time of 47 months, and the 5‑year cumulative incidence of PVT was 11.9%.18 A large prospective Italian Venous Thrombotic Events Registry project is ongoing primarily to estimate the prevalence of PVT evaluated by power Doppler ultrasonography in a cohort of patients with liver cirrhosis of any aetiology and severity.19 This project
NATURE REVIEWS | GASTROENTEROLOGY & HEPATOLOGY © 2014 Macmillan Publishers Limited. All rights reserved
VOLUME 11 | JULY 2014 | 435
REVIEWS Key points ■■ Portal vein thrombosis (PVT) is a fairly common complication of liver cirrhosis; occlusive PVT can be associated with poor prognosis, especially in patients with a prior history of bleeding ■■ A randomized controlled trial has shown that anticoagulation is effective for the primary prevention of PVT in liver cirrhosis, and might also improve liver function and survival ■■ Further randomized trials with a larger sample size are warranted to confirm these findings ■■ Evidence from several case series has demonstrated the efficacy and safety of anticoagulation therapy and a transjugular intrahepatic portosystemic shunt for the management of PVT in liver cirrhosis ■■ Future research should aim to weigh the benefits of various treatment modalities against the risks that they will bring and to establish their different timings and indications
will provide more solid data by recruiting 1,100 patients between December 2011 and December 2014. Decreased portal flow velocity and increased flow volume in the largest collateral vessel were considered to be independent factors predicting the development of PVT in three studies (Supplementary Table 1 online).14,15,20 In addition, increased severity of liver cirrhosis is positively associated with the reduction of portal vein flow velocity, 21 thereby potentially leading to the development of PVT.14 However, it should be noted that the statistical significance of liver dysfunction on the development of PVT disappears in multivariate analysis.14 Other local factors associated with the development of PVT include portal venous endothelial injury (for example, splenectomy, hepatectomy, surgical shunt, and other intra-abdominal surgery) and inflammation (pancreatitis, cholecystitis, appendicitis, and other intra-abdominal infections). Systemic thrombotic risk factors, which are more frequently observed in patients with cirrhosis and PVT than in those without PVT, include factor V Leiden mutation, prothrombin G20210A mutation, methylenetetrahydrofolate reductase C677T mutation, hypofibrinolysis, positive anticardiolipin antibodies, and positive lupus anticoagulant (Supplementary Table 1 online).20,22–30 A meta-analysis of observational studies performed by our group showed that decreased levels of antithrombin and protein C and S are not significantly associated with the development of PVT in liver cirrhosis.31 This finding has been further supported by the findings of our case–control study.32 Another meta-analysis by Dentali suggests that prothrombin G20210A mutation, but not factor V Leiden mutation, might contribute to the pathogenesis of PVT in liver cirrhosis.28 Evidence regarding the effect of PVT on the prognosis of patients with cirrhosis is complex and contro versial.6,13,15,33–43 The heterogeneous results of these studies might be attributable to discrepancies in the selection of the target population (for example compensated or decompensated cirrhosis), degree and extension of PVT (partial or complete, and mesenteric venous involvement or not), study design (retrospective or prospective), sample size (small or large), and follow-up length (shortterm or long-term). For example, by using the Scientific 436 | JULY 2014 | VOLUME 11
Registry of Transplant Recipients data, Englesbe et al. did not identify any significant effect of PVT on mortality on the waiting list for liver transplantation.36 However, in another study in which the same investigators stratified data from the University of Michigan according to the degree of PVT, they found that the presence of occlusive PVT was an independent predictor of mortality from the time of liver transplant evaluation.35 Indeed, several studies have reported that post-transplantation mortality is similar between patients with a patent portal vein and those with partial PVT, but is significantly increased in patients with complete or more-extensive PVT.6,1337,38 Systematic reviews have also confirmed that occlusive or complete PVT affects survival after liver transplantation.44,45 Furthermore, whether or not the physiological portal inflow is re-established in patients with complete PVT during transplantation surgery also influences the prognosis.46 However, in this Review, we do not repeat previous reports of work regarding the effect of PVT on the survival of liver transplantation recipients,44,45 but instead attempt to collect evidence regarding the effect of PVT on the prognosis of patients with cirrhosis who have had a prior variceal bleed (Supplementary Table 2 online). An Italian, multicentre, prospective, cohort study identified PVT as an important predictor of 5‑day treatment failure for variceal bleeding (defined as uncontrolled bleeding, rebleeding, or death) in patients with liver cirrhosis.47 This conclusion is also in line with the results of a prospective study from 2012.48 Moreover, the presence of PVT might increase the incidence of early rebleeding,49,50 but the relationship between PVT and late occurrence of rebleeding is not unanimous.51–54 Except for the results of two studies,52,55 most studies suggest that PVT substantially effects the short-term and long-term mortality of patients with cirrhosis and variceal bleeding.50,53–56 Collectively, complete PVT could be considered a clinical marker for the severity of liver cirrhosis.45 The clinical significance of PVT in liver cirrhosis prompts clinicians to explore different treatment modal ities. However, recommendations from current practice guidelines and consensus statements are insufficient owing to limited data and poor-quality evidence. 57,58 Herein, we discuss the current evidence regarding the management of PVT in liver cirrhosis and attempt to propose an exploratory and preliminary algorithm.
Primary prophylaxis of PVT Several large case–control studies have shown a high risk of venous thromboembolism (that is, deep vein thrombosis and pulmonary embolism) in patients with liver cirrhosis in spite of an elevated international normalized ratio (INR) and/or low platelet count.59–63 In addition, experimental studies demonstrate the hypercoagulability of plasma from patients with cirrhosis.64,65 These findings potentially suggest the necessity of primary prevention of venous thromboembolism in patients with cirrhosis.66,67 A historical case–control study further confirms that the prophylactic use of anticoagulation for deep vein thrombosis is safe in hospitalized patients with cirrhosis who do not have active bleeding, and does not increase the
www.nature.com/nrgastro © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS a 100
De novo portal vein thrombosis
b
New hepatic decompensation
c
Overall death
90
Enoxaparin group No enoxaparin group
Total events (%)
80 70 60 50 40 30 20 10 0 During the year of active treatment
At the end of follow-up
During the year of active treatment
At the end of follow-up
At the end of follow-up
Figure 1 | Benefits of anticoagulation in patients with cirrhosis. Improvement in a | the primary prophylaxis of portal vein thrombosis, b | prevention of hepatic decompensation, and c | survival.
incidence of gastrointestinal bleeding or overall death.68 However, a meta-analysis did not find any decreased risk of venous thromboembolism in patients with cirrhosis receiving thromboprophylaxis, although the use of heparin was not associated with an increased rate of bleeding.69 Certainly, these studies do not accurately identify the effect of thromboprophylaxis in patients with cirrhosis who are at an increased risk of developing venous thromboembolism. In 2012, an Italian, single-centre, randomized controlled trial (RCT) for the first time explored the benefit of enoxaparin in the primary prevention of PVT in patients with cirrhosis.70 During an enrolment period of 31 months, 70 patients with cirrhosis and a Child–Pugh score of 7–10 points and without ascites, portal hypertensive bleeding, high-risk varices or a platelet count of 2.0
Adverse effect of anticoagulants (n = 0)
Abbreviations: HIT, heparin induced thrombocytopenia; INR, international normalized ratio; LMWH, low-molecular-weight heparin; NA, not available; VKA, vitamin K antagonist.
Three studies have confirmed that early initiation of anticoagulation after the diagnosis of PVT is positively associated with portal vein recanalization in univariate analysis.96,97,105 However, the optimal interval between the diagnosis of PVT and initiation of anticoagulation should be 6 months according to the study by Senzolo,96 but 14 days in the other two studies by Delgado97 and Copaci. 105 Absence of previous portal hypertensive bleeding is also identified as a significant factor associated with portal vein recanalization in the study by Senzolo,96 but not in the study by Delgado.97 Moreover, Francoz and colleagues suggest the limited usefulness of anticoagulation therapy for complete PVT in patients with liver cirrhosis.100 By contrast, Senzolo et al.96 did not find any difference in portal vein recanalization rates between partial and complete PVT. Therefore, additional well-designed studies should be performed to explore the role of anticoagulation in the treatment of complete PVT. LMWH and vitamin K antagonists (VKA) were the two main kinds of anticoagulants used in these studies. The reasons for use of LMWH, but not unfractionated heparin, are the reduced incidence of bleeding and he parin-induced-thrombocytopenia associated with LMWH, and the once-daily dosing (unfractionated heparin must be taken three times daily). Anti coagulation-related bleeding complications were analysed in a multicentre study by Delgado and colleagues,97 in which five bleeding events probably related to anticoagulation were reported. Importantly, all of these events occurred in patients treated with VKA alone, and no bleeding event was observed in any patients treated with LMWH alone or those with LMWH followed by VKA. Thus, the safety of LMWH might be 440 | JULY 2014 | VOLUME 11
superior to that of VKA.108 The main disadvantage of LMWH is the need for long-term subcutaneous injections, which substantially reduces patients’ compliance in clinical practice; for example, in the study by Delgado, anticoagulation therapy was initiated with LMWH in 47 patients, and was later switched to VKA in 45% of these patients.97 By comparison, VKA can be administered orally. In addition, a target INR of 2–3 should be maintained to ensure adequate anticoagulation in patients treated with VKA. This recommendation primarily originates from the consensus that a low INR increases the risk of recurrent venous thromboembolism, and a high INR increases the risk of major bleeding.109,110 Owing to the difficulty of maintaining the INR in the therapeutic range throughout the course of treatment, the actual proportion of time spent within the INR target range (ITTR) might be more reasonable to assess the clinical outcome.111 Notably, regular laboratory monitoring of INR for dosage adjustment is necessary, but inconvenient. At present, direct thrombin inhibitors and inhibitors of activated factor X overcome these above-mentioned disadvantages of LMWH and VKA, because they don’t require laboratory monitoring.112–114 Certainly, the inadequacy of INR in monitoring the effect of VKA should be clarified. First, both pro-coagulant and anticoagulant factors are decreased in liver cirrhosis and the haemostatic balance is set at a lower point. However, INR does not account for the decrease in the pro-coagulant factors. Second, INR is often elevated above 2 in patients with end-stage liver diseases who receive no anticoagulation. Furthermore, evidence from Bechmann and Lisman has revealed the potential limitation of traditional anti-Xa assays in monitoring the effect
www.nature.com/nrgastro © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
90 80 70 60 50 40 30 20 10 0
Ta k
at (2 ori 01 et 3 a Co ) 104 l. pa (2 ci 01 et 3) 1 al W 05 . er ne (2 r e 0 t Se 13) 9 al. 9 nz o (2 lo e 0 t De 12) 9 al. 6 lg ad (2 o e 0 M ar 12 t al. )9 uy am 7 (2 a e 01 t 2 al Be ) 98 . (2 nto 0 e Am 11) 1t al. 03 itr an o (2 e 01 t 0 al Fr an ) 95 . co (2 z 00 et Fr 8) 100al. an c (2 oz e 00 t 5) 1 al. 3
Proportion of portal vein recanalization after anticoagualtion (%)
100
No change or thrombus extension Complete or partial recanalization
Figure 2 | Outcome of anticoagulation for the treatment of portal vein thrombosis in patients with cirrhosis.
of LMWH in patients with cirrhosis, probably because of the low anti-thrombin levels caused by the reduced hepatic synthesis.115,116 In these settings, global coagulation assays, such as the thrombin generation test with or without thrombomodulin to evaluate the balance between pro-coagulant and anticoagulant factors, might be considered for monitoring the anticoagulation effect in patients with advanced liver cirrhosis.117 Collectively, we suggest that these preliminary studies support the use of anticoagulants in patients with cirrhosis and PVT. However, many unresolved problems deserve further research,118 such as the choice of anticoagulant, optimal dosage of anticoagulant, and candidates for anticoagulation therapy.
Thrombolysis Evidence regarding the use of thrombolytics for the treatment of PVT in patients with cirrhosis is extremely scarce. In an Italian, single-centre, prospective study, nine patients with cirrhosis and recent PVT were treated with systemic thrombolysis (continuous intravenous infusion of recombinant tissue plasminogen activator) combined with anticoagulation.119 Follow-up evaluations by colour Doppler ultrasonography showed complete recanalization (n = 4), partial recanalization (n = 4), and no recanalization (n = 1). No clinically significant adverse effects were observed in these patients. However, considering that systemic thrombolysis potentially induces a general and uncontrolled fibrinolysis state, it must be used with caution, especially in asymptomatic patients. Such treatment should be reserved for thrombus extension into the superior mesenteric vein and its secondary intestinal ischaemia only. In addition, thrombolytics can be indirectly infused into the superior mesenteric artery through the femoral or radical artery, or directly infused into the portal vein through the percutaneous transhepatic or transjugular intrahepatic approach.120–122 Given the small sample size and heterogeneous population of each study, the safety and efficacy of thrombolysis needs to be further ascertained.
Percutaneous portal vein recanalization Scattered case reports show the safety and feasibility of percutaneous portal vein recanalization in patients with cirrhosis and PVT. Two Korean studies reported that three patients with liver cirrhosis developed complete obstruction of the portal trunk after living donor liver transplantation.123,124 Portal vein recanalization was successful in two patients with balloon angioplasty and/or stent placement after percutaneous transhepatic puncture of the intrahepatic portal vein, but failed in one patient after percutaneous trans-splenic puncture of the perihilar splenic vein. Another study from China showed that PVT was successfully recanalized by the placement of covered stents in five patients with cirrhosis.125 In addi tion, gastro-oesophageal variceal embolization can be performed during the procedure of portal vein recanalization.125 Importantly, procedure-related bleeding complications can be life-threatening, and patients should be kept under rigorous surveillance. TIPS Theoretically, the advantages of TIPS for the treatment of PVT in liver cirrhosis are to effectively recanalize the thrombosed portal vein using endovascular techniques (balloon angioplasty, stent-placement, thrombectomy, and thrombolysis), and to simultaneously resolve symptomatic portal hypertension and prevent the thrombus recurrence or extension by the creation of a porto systemic shunt.126,127 However, the technical difficulty of TIPS limits its widespread application, and the risks and benefits from creating a TIPS should be fully balanced. To date, >50 case reports or series from 16 countries have reported the outcome of about 500 patients with PVT treated with TIPS insertions.90 The main technical strategies include: one, TIPS placement followed by portal vein recanalization via the portosystemic shunt; two, portal vein recanalization via percutaneous approaches followed by TIPS placement; and three, TIPS insertion between a hepatic vein and a large collateral vessel without main portal vein recanalization (Figure 3). The technical success rate for TIPS is relatively high in experienced hands (range: 67–100%).9,128–134 The TIPS procedure is feasible in the presence of portal cavernoma, but not in the setting of an obliterated main portal vein or fibrotic cord if no large collateral vessel is present.129 The combination of transhepatic, trans-splenic, and transmesenteric approaches to access the portal vein can facilitate the TIPS procedure.129,135 Notably, these percutaneous approaches are risky if the intrahepatic portal vein branch is thrombosed.136 The rate of portal vein recanalization after successful TIPS insertions is up to 80%.128 In addition, it should be noted that the degree of PVT is often more severe (>50% of lumen occupancy and complete occlusion) in patients treated with TIPS insertion than in those treated with anticoagulation. By summarizing the results of TIPS for the treatment of PVT in patients with cirrhosis,9,128–131,133,134, the procedure-related complication rate varies from 0–17% (Supplementary Table 3 online). Fatal complications rarely occur. In two studies, two patients died of TIPS-related
NATURE REVIEWS | GASTROENTEROLOGY & HEPATOLOGY © 2014 Macmillan Publishers Limited. All rights reserved
VOLUME 11 | JULY 2014 | 441
REVIEWS b
a
RHV
Stent
Stent LCV
LCV
MPV GEV CTPV
CTPV GEV
DSV
PSV
Trans-splenic catheterization
SMV
Figure 3 | Insertion of a TIPS into a large collateral vessel in a patient with complete portal vein obstruction. a | Direct portography. b | Schematic diagram. The images are from a 44-year-old patient with HBV-related cirrhosis with variceal rebleeding unresponsive to medical and endoscopic therapy. A previous surgical portosystemic shunt was occluded. Diffuse thrombosis within the portal venous system and cavernous transformation was found by imaging. Direct splenoportography via a percutaneous trans-splenic approach clearly showed the location of the targeted collateral vein to facilitate TIPS. A stent was successfully placed between a large collateral vein and right hepatic vein. The portosystemic pressure gradient was reduced from 23 mmHg to 11 mmHg. Abbreviations: CTPV, cavernous transformation of portal vein; DSV, distal splenic vein; GEV, gastroepiploic vein; LCV, large collateral vein around portal vein; MPV, main portal vein; PSV, ligated proximal splenic vein; RHV, right hepatic vein; SMV, superior mesenteric vein; TIPS, transjugular intrahepatic portosystemic shunt.
PVT in liver cirrhosis
No symptomatic portal hypertension
50% thrombus, ± SMV extension
Symptomatic portal hypertension
Controlled by ET, LVP and/or drug therapy
Uncontrolled by ET, LVP and/or drug therapy
Anticoagulation (or RCT)
Progressed
Improved or stable
Potential algorithm for treatment of PVT
Progressed
Follow-up
TIPS, or symptomatic treatment if unavailable or failed
Figure 4 | Algorithm for the treatment of portal vein thrombosis in liver cirrhosis. Symptomatic portal hypertension mainly represents variceal bleeding and/or ascites. Abbreviations: ET, endoscopic therapy; LVP, large volume paracentesis; PVT, portal vein thrombosis; RCT, randomized controlled trial; SMV, superior mesenteric vein; TIPS, transjugular intrahepatic portosystemic shunt.
complications (one died of an intravascular disseminated coagulopathy after TIPS, and another died of intraabdominal bleeding caused by hepatic capsule perforation during the TIPS procedure).129,130 In the second case, the investigators clearly acknowledged that they did not fully recognize the risk of hepatic capsule perforation before the lethal event.129 However, with the improvement of TIPS techniques and peri-operative management, severe procedural complications might be gradually reduced. Shunt dysfunction and hepatic encephalopathy still represent the biggest clinical challenge for the management of patients with cirrhosis who are undergoing TIPS. 442 | JULY 2014 | VOLUME 11
The study by Perarnau showed that these postoperative complications were not significantly different between cirrhotic patients with or without PVT.130 The overall incidence of shunt dysfunction and hepatic encephalopathy after TIPS is 7–32% and 0–50%, respectively.90 The incidence of shunt dysfunction has been reduced by the use of covered stents;137,138 in the study by Luca et al. the incidence of shunt dysfunction at 12 and 24 months was 38% and 85% in the bare stent group, and 21% and 29% in the covered stent group.128 Timing of TIPS insertion in patients with cirrhosis and PVT is another major concern. In patients with cirrhosis without PVT, TIPS has been recommended as the secondline treatment option for recurrent variceal bleeding and refractory ascites.139 Whether or not these recommendations should be extrapolated to patients with cirrhosis and PVT deserves further exploration, because the presence of PVT potentially changes the natural history of liver cirrhosis. An RCT140 is being conducted at our centre to compare the incidence of variceal rebleeding, progression of PVT, and survival between cirrhotic patients with PVT who are receiving TIPS and those receiving endoscopic and drug therapy.141,142 A total of 50 patients were estimated to be needed for this trial. All patients were enrolled between June 4, 2011 and January 25, 2014, and are being followed-up. Unless this trial confirms the superiority of TIPS, conventional therapy will remain the first-line treatment of choice for the prevention of variceal re-bleeding in patients with cirrhosis and PVT.142 It might be reasonable that TIPS should be performed in the case of thrombus extension despite adequate anticoagulation96 However, if the thrombus has progressed to complete occlusion or even fibrotic cord, the technical difficulty of TIPS insertion will be substantially increased. Therefore, further studies are warranted to explore the time when anticoagulation therapy should be switched to TIPS. Given that the progression of PVT might be positively associated with treatment failure,52,100,129,143 the selection of treatment options is primarily dependent on the presence of symptomatic portal hypertension and the stage and degree of PVT. On the basis of this consideration and results of existing studies, we propose an exploratory and preliminary algorithm for the management of PVT in liver cirrhosis (Figure 4). However, we acknowledge that this algorithm is not evidence-based owing to the absence of RCTs, and many controversies regarding the indications of each treatment option still exist. First, patients with a Child–Pugh score of 7–10, no ascites, no high-risk varices or previous history of variceal bleeding, and a platelet count >10,000/mm3 have been considered as the target population for a randomized controlled trial to explore the role of anticoagulation in the primary prophylaxis of PVT.70 However, whether or not patients with a Child–Pugh score of 5–6, ascites, high-risk varices, and/or a platelet count
Introduction
Xijing Hospital of Digestive Diseases, Fourth Military Medical University, No. 17 West Changle Road, Xi’an, 710032 China (X.Q., G.H., D.F.).
Portal vein thrombosis (PVT) is characterized by the formation of a thrombus within the portal vein trunk and intrahepatic portal branches.1,2 Evolution of PVT mainly includes the degree (that is, partial occlusion, complete occlusion and fibrotic cord), stage (fresh thrombus, recent or old thrombus, and portal cavernoma), and extension (portal vein alone or extension into the splenic and/or superior mesenteric vein).3–5 A classification system proposed by Yerdel is frequently used, as follows: grade 1, 50% PVT with or without minimal obstruction of the superior mesenteric vein; grade 3, complete portal vein and proximal superior mesenteric vein thrombosis; and grade 4, complete portal vein and entire superior mesenteric vein thrombosis.6 In contrast to other classifications,7–9 this system considers both the degree and extension of PVT. PVT is frequently encountered in the setting of liver cirrhosis, especially at the decompensated or advanced stage. 10,11 However, the prevalence and incidence of PVT often varies among different studies owing to heterogeneous diagnostic methods and target populations (reviewed in detail elsewhere10). In a large case series including 701 patients with cirrhosis but without hepatocellular carcinoma who were undergoing routine Doppler ultrasonography, the prevalence of PVT was 11.2% (79 patients).12 In another study, the prevalence
Correspondence to: D.F. [email protected]; G.H. [email protected]
Competing interests The authors declare no competing interests.
of PVT was 8.4% (21 of 251 patients) at the time of listing for liver transplantation, and the incidence of de novo PVT in transplant recipients was 7.4% (17 of 230 patients) during a mean follow-up of 12.1 months.13 Results of a prospective study demonstrated that the incidence of de novo PVT in liver cirrhosis was 16.4% (12 of 73 patients) within 1 year.14 Similarly, a retrospective study shows that the 1‑year and 5‑year cumulative incidence of de novo PVT is 12.8% and 20.0%, respectively, in virus-related cirrhosis.15 However, several inconsistencies and drawbacks in the design and data analysis of these studies—including splenic vein thrombosis alone falsely classified as PVT, uneven distribution of incidence of PVT during follow-up, and difficulty of using Doppler ultrasonography in the regular evaluation of PVT—raise questions about the reliability of these findings.16 In 2011, a pre-planned satellite study (using data from a published multicentre randomized trial in which the incidence of small hepatocellular carcinoma was compared between patients with compensated cirrhosis receiving 3‑month versus 6‑month ultrasonographic periodicities17) reported that 101 of 898 patients with compensated cirrhosis and a prior patent portal vein developed PVT during a mean follow-up time of 47 months, and the 5‑year cumulative incidence of PVT was 11.9%.18 A large prospective Italian Venous Thrombotic Events Registry project is ongoing primarily to estimate the prevalence of PVT evaluated by power Doppler ultrasonography in a cohort of patients with liver cirrhosis of any aetiology and severity.19 This project
NATURE REVIEWS | GASTROENTEROLOGY & HEPATOLOGY © 2014 Macmillan Publishers Limited. All rights reserved
VOLUME 11 | JULY 2014 | 435
REVIEWS Key points ■■ Portal vein thrombosis (PVT) is a fairly common complication of liver cirrhosis; occlusive PVT can be associated with poor prognosis, especially in patients with a prior history of bleeding ■■ A randomized controlled trial has shown that anticoagulation is effective for the primary prevention of PVT in liver cirrhosis, and might also improve liver function and survival ■■ Further randomized trials with a larger sample size are warranted to confirm these findings ■■ Evidence from several case series has demonstrated the efficacy and safety of anticoagulation therapy and a transjugular intrahepatic portosystemic shunt for the management of PVT in liver cirrhosis ■■ Future research should aim to weigh the benefits of various treatment modalities against the risks that they will bring and to establish their different timings and indications
will provide more solid data by recruiting 1,100 patients between December 2011 and December 2014. Decreased portal flow velocity and increased flow volume in the largest collateral vessel were considered to be independent factors predicting the development of PVT in three studies (Supplementary Table 1 online).14,15,20 In addition, increased severity of liver cirrhosis is positively associated with the reduction of portal vein flow velocity, 21 thereby potentially leading to the development of PVT.14 However, it should be noted that the statistical significance of liver dysfunction on the development of PVT disappears in multivariate analysis.14 Other local factors associated with the development of PVT include portal venous endothelial injury (for example, splenectomy, hepatectomy, surgical shunt, and other intra-abdominal surgery) and inflammation (pancreatitis, cholecystitis, appendicitis, and other intra-abdominal infections). Systemic thrombotic risk factors, which are more frequently observed in patients with cirrhosis and PVT than in those without PVT, include factor V Leiden mutation, prothrombin G20210A mutation, methylenetetrahydrofolate reductase C677T mutation, hypofibrinolysis, positive anticardiolipin antibodies, and positive lupus anticoagulant (Supplementary Table 1 online).20,22–30 A meta-analysis of observational studies performed by our group showed that decreased levels of antithrombin and protein C and S are not significantly associated with the development of PVT in liver cirrhosis.31 This finding has been further supported by the findings of our case–control study.32 Another meta-analysis by Dentali suggests that prothrombin G20210A mutation, but not factor V Leiden mutation, might contribute to the pathogenesis of PVT in liver cirrhosis.28 Evidence regarding the effect of PVT on the prognosis of patients with cirrhosis is complex and contro versial.6,13,15,33–43 The heterogeneous results of these studies might be attributable to discrepancies in the selection of the target population (for example compensated or decompensated cirrhosis), degree and extension of PVT (partial or complete, and mesenteric venous involvement or not), study design (retrospective or prospective), sample size (small or large), and follow-up length (shortterm or long-term). For example, by using the Scientific 436 | JULY 2014 | VOLUME 11
Registry of Transplant Recipients data, Englesbe et al. did not identify any significant effect of PVT on mortality on the waiting list for liver transplantation.36 However, in another study in which the same investigators stratified data from the University of Michigan according to the degree of PVT, they found that the presence of occlusive PVT was an independent predictor of mortality from the time of liver transplant evaluation.35 Indeed, several studies have reported that post-transplantation mortality is similar between patients with a patent portal vein and those with partial PVT, but is significantly increased in patients with complete or more-extensive PVT.6,1337,38 Systematic reviews have also confirmed that occlusive or complete PVT affects survival after liver transplantation.44,45 Furthermore, whether or not the physiological portal inflow is re-established in patients with complete PVT during transplantation surgery also influences the prognosis.46 However, in this Review, we do not repeat previous reports of work regarding the effect of PVT on the survival of liver transplantation recipients,44,45 but instead attempt to collect evidence regarding the effect of PVT on the prognosis of patients with cirrhosis who have had a prior variceal bleed (Supplementary Table 2 online). An Italian, multicentre, prospective, cohort study identified PVT as an important predictor of 5‑day treatment failure for variceal bleeding (defined as uncontrolled bleeding, rebleeding, or death) in patients with liver cirrhosis.47 This conclusion is also in line with the results of a prospective study from 2012.48 Moreover, the presence of PVT might increase the incidence of early rebleeding,49,50 but the relationship between PVT and late occurrence of rebleeding is not unanimous.51–54 Except for the results of two studies,52,55 most studies suggest that PVT substantially effects the short-term and long-term mortality of patients with cirrhosis and variceal bleeding.50,53–56 Collectively, complete PVT could be considered a clinical marker for the severity of liver cirrhosis.45 The clinical significance of PVT in liver cirrhosis prompts clinicians to explore different treatment modal ities. However, recommendations from current practice guidelines and consensus statements are insufficient owing to limited data and poor-quality evidence. 57,58 Herein, we discuss the current evidence regarding the management of PVT in liver cirrhosis and attempt to propose an exploratory and preliminary algorithm.
Primary prophylaxis of PVT Several large case–control studies have shown a high risk of venous thromboembolism (that is, deep vein thrombosis and pulmonary embolism) in patients with liver cirrhosis in spite of an elevated international normalized ratio (INR) and/or low platelet count.59–63 In addition, experimental studies demonstrate the hypercoagulability of plasma from patients with cirrhosis.64,65 These findings potentially suggest the necessity of primary prevention of venous thromboembolism in patients with cirrhosis.66,67 A historical case–control study further confirms that the prophylactic use of anticoagulation for deep vein thrombosis is safe in hospitalized patients with cirrhosis who do not have active bleeding, and does not increase the
www.nature.com/nrgastro © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS a 100
De novo portal vein thrombosis
b
New hepatic decompensation
c
Overall death
90
Enoxaparin group No enoxaparin group
Total events (%)
80 70 60 50 40 30 20 10 0 During the year of active treatment
At the end of follow-up
During the year of active treatment
At the end of follow-up
At the end of follow-up
Figure 1 | Benefits of anticoagulation in patients with cirrhosis. Improvement in a | the primary prophylaxis of portal vein thrombosis, b | prevention of hepatic decompensation, and c | survival.
incidence of gastrointestinal bleeding or overall death.68 However, a meta-analysis did not find any decreased risk of venous thromboembolism in patients with cirrhosis receiving thromboprophylaxis, although the use of heparin was not associated with an increased rate of bleeding.69 Certainly, these studies do not accurately identify the effect of thromboprophylaxis in patients with cirrhosis who are at an increased risk of developing venous thromboembolism. In 2012, an Italian, single-centre, randomized controlled trial (RCT) for the first time explored the benefit of enoxaparin in the primary prevention of PVT in patients with cirrhosis.70 During an enrolment period of 31 months, 70 patients with cirrhosis and a Child–Pugh score of 7–10 points and without ascites, portal hypertensive bleeding, high-risk varices or a platelet count of 2.0
Adverse effect of anticoagulants (n = 0)
Abbreviations: HIT, heparin induced thrombocytopenia; INR, international normalized ratio; LMWH, low-molecular-weight heparin; NA, not available; VKA, vitamin K antagonist.
Three studies have confirmed that early initiation of anticoagulation after the diagnosis of PVT is positively associated with portal vein recanalization in univariate analysis.96,97,105 However, the optimal interval between the diagnosis of PVT and initiation of anticoagulation should be 6 months according to the study by Senzolo,96 but 14 days in the other two studies by Delgado97 and Copaci. 105 Absence of previous portal hypertensive bleeding is also identified as a significant factor associated with portal vein recanalization in the study by Senzolo,96 but not in the study by Delgado.97 Moreover, Francoz and colleagues suggest the limited usefulness of anticoagulation therapy for complete PVT in patients with liver cirrhosis.100 By contrast, Senzolo et al.96 did not find any difference in portal vein recanalization rates between partial and complete PVT. Therefore, additional well-designed studies should be performed to explore the role of anticoagulation in the treatment of complete PVT. LMWH and vitamin K antagonists (VKA) were the two main kinds of anticoagulants used in these studies. The reasons for use of LMWH, but not unfractionated heparin, are the reduced incidence of bleeding and he parin-induced-thrombocytopenia associated with LMWH, and the once-daily dosing (unfractionated heparin must be taken three times daily). Anti coagulation-related bleeding complications were analysed in a multicentre study by Delgado and colleagues,97 in which five bleeding events probably related to anticoagulation were reported. Importantly, all of these events occurred in patients treated with VKA alone, and no bleeding event was observed in any patients treated with LMWH alone or those with LMWH followed by VKA. Thus, the safety of LMWH might be 440 | JULY 2014 | VOLUME 11
superior to that of VKA.108 The main disadvantage of LMWH is the need for long-term subcutaneous injections, which substantially reduces patients’ compliance in clinical practice; for example, in the study by Delgado, anticoagulation therapy was initiated with LMWH in 47 patients, and was later switched to VKA in 45% of these patients.97 By comparison, VKA can be administered orally. In addition, a target INR of 2–3 should be maintained to ensure adequate anticoagulation in patients treated with VKA. This recommendation primarily originates from the consensus that a low INR increases the risk of recurrent venous thromboembolism, and a high INR increases the risk of major bleeding.109,110 Owing to the difficulty of maintaining the INR in the therapeutic range throughout the course of treatment, the actual proportion of time spent within the INR target range (ITTR) might be more reasonable to assess the clinical outcome.111 Notably, regular laboratory monitoring of INR for dosage adjustment is necessary, but inconvenient. At present, direct thrombin inhibitors and inhibitors of activated factor X overcome these above-mentioned disadvantages of LMWH and VKA, because they don’t require laboratory monitoring.112–114 Certainly, the inadequacy of INR in monitoring the effect of VKA should be clarified. First, both pro-coagulant and anticoagulant factors are decreased in liver cirrhosis and the haemostatic balance is set at a lower point. However, INR does not account for the decrease in the pro-coagulant factors. Second, INR is often elevated above 2 in patients with end-stage liver diseases who receive no anticoagulation. Furthermore, evidence from Bechmann and Lisman has revealed the potential limitation of traditional anti-Xa assays in monitoring the effect
www.nature.com/nrgastro © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
90 80 70 60 50 40 30 20 10 0
Ta k
at (2 ori 01 et 3 a Co ) 104 l. pa (2 ci 01 et 3) 1 al W 05 . er ne (2 r e 0 t Se 13) 9 al. 9 nz o (2 lo e 0 t De 12) 9 al. 6 lg ad (2 o e 0 M ar 12 t al. )9 uy am 7 (2 a e 01 t 2 al Be ) 98 . (2 nto 0 e Am 11) 1t al. 03 itr an o (2 e 01 t 0 al Fr an ) 95 . co (2 z 00 et Fr 8) 100al. an c (2 oz e 00 t 5) 1 al. 3
Proportion of portal vein recanalization after anticoagualtion (%)
100
No change or thrombus extension Complete or partial recanalization
Figure 2 | Outcome of anticoagulation for the treatment of portal vein thrombosis in patients with cirrhosis.
of LMWH in patients with cirrhosis, probably because of the low anti-thrombin levels caused by the reduced hepatic synthesis.115,116 In these settings, global coagulation assays, such as the thrombin generation test with or without thrombomodulin to evaluate the balance between pro-coagulant and anticoagulant factors, might be considered for monitoring the anticoagulation effect in patients with advanced liver cirrhosis.117 Collectively, we suggest that these preliminary studies support the use of anticoagulants in patients with cirrhosis and PVT. However, many unresolved problems deserve further research,118 such as the choice of anticoagulant, optimal dosage of anticoagulant, and candidates for anticoagulation therapy.
Thrombolysis Evidence regarding the use of thrombolytics for the treatment of PVT in patients with cirrhosis is extremely scarce. In an Italian, single-centre, prospective study, nine patients with cirrhosis and recent PVT were treated with systemic thrombolysis (continuous intravenous infusion of recombinant tissue plasminogen activator) combined with anticoagulation.119 Follow-up evaluations by colour Doppler ultrasonography showed complete recanalization (n = 4), partial recanalization (n = 4), and no recanalization (n = 1). No clinically significant adverse effects were observed in these patients. However, considering that systemic thrombolysis potentially induces a general and uncontrolled fibrinolysis state, it must be used with caution, especially in asymptomatic patients. Such treatment should be reserved for thrombus extension into the superior mesenteric vein and its secondary intestinal ischaemia only. In addition, thrombolytics can be indirectly infused into the superior mesenteric artery through the femoral or radical artery, or directly infused into the portal vein through the percutaneous transhepatic or transjugular intrahepatic approach.120–122 Given the small sample size and heterogeneous population of each study, the safety and efficacy of thrombolysis needs to be further ascertained.
Percutaneous portal vein recanalization Scattered case reports show the safety and feasibility of percutaneous portal vein recanalization in patients with cirrhosis and PVT. Two Korean studies reported that three patients with liver cirrhosis developed complete obstruction of the portal trunk after living donor liver transplantation.123,124 Portal vein recanalization was successful in two patients with balloon angioplasty and/or stent placement after percutaneous transhepatic puncture of the intrahepatic portal vein, but failed in one patient after percutaneous trans-splenic puncture of the perihilar splenic vein. Another study from China showed that PVT was successfully recanalized by the placement of covered stents in five patients with cirrhosis.125 In addi tion, gastro-oesophageal variceal embolization can be performed during the procedure of portal vein recanalization.125 Importantly, procedure-related bleeding complications can be life-threatening, and patients should be kept under rigorous surveillance. TIPS Theoretically, the advantages of TIPS for the treatment of PVT in liver cirrhosis are to effectively recanalize the thrombosed portal vein using endovascular techniques (balloon angioplasty, stent-placement, thrombectomy, and thrombolysis), and to simultaneously resolve symptomatic portal hypertension and prevent the thrombus recurrence or extension by the creation of a porto systemic shunt.126,127 However, the technical difficulty of TIPS limits its widespread application, and the risks and benefits from creating a TIPS should be fully balanced. To date, >50 case reports or series from 16 countries have reported the outcome of about 500 patients with PVT treated with TIPS insertions.90 The main technical strategies include: one, TIPS placement followed by portal vein recanalization via the portosystemic shunt; two, portal vein recanalization via percutaneous approaches followed by TIPS placement; and three, TIPS insertion between a hepatic vein and a large collateral vessel without main portal vein recanalization (Figure 3). The technical success rate for TIPS is relatively high in experienced hands (range: 67–100%).9,128–134 The TIPS procedure is feasible in the presence of portal cavernoma, but not in the setting of an obliterated main portal vein or fibrotic cord if no large collateral vessel is present.129 The combination of transhepatic, trans-splenic, and transmesenteric approaches to access the portal vein can facilitate the TIPS procedure.129,135 Notably, these percutaneous approaches are risky if the intrahepatic portal vein branch is thrombosed.136 The rate of portal vein recanalization after successful TIPS insertions is up to 80%.128 In addition, it should be noted that the degree of PVT is often more severe (>50% of lumen occupancy and complete occlusion) in patients treated with TIPS insertion than in those treated with anticoagulation. By summarizing the results of TIPS for the treatment of PVT in patients with cirrhosis,9,128–131,133,134, the procedure-related complication rate varies from 0–17% (Supplementary Table 3 online). Fatal complications rarely occur. In two studies, two patients died of TIPS-related
NATURE REVIEWS | GASTROENTEROLOGY & HEPATOLOGY © 2014 Macmillan Publishers Limited. All rights reserved
VOLUME 11 | JULY 2014 | 441
REVIEWS b
a
RHV
Stent
Stent LCV
LCV
MPV GEV CTPV
CTPV GEV
DSV
PSV
Trans-splenic catheterization
SMV
Figure 3 | Insertion of a TIPS into a large collateral vessel in a patient with complete portal vein obstruction. a | Direct portography. b | Schematic diagram. The images are from a 44-year-old patient with HBV-related cirrhosis with variceal rebleeding unresponsive to medical and endoscopic therapy. A previous surgical portosystemic shunt was occluded. Diffuse thrombosis within the portal venous system and cavernous transformation was found by imaging. Direct splenoportography via a percutaneous trans-splenic approach clearly showed the location of the targeted collateral vein to facilitate TIPS. A stent was successfully placed between a large collateral vein and right hepatic vein. The portosystemic pressure gradient was reduced from 23 mmHg to 11 mmHg. Abbreviations: CTPV, cavernous transformation of portal vein; DSV, distal splenic vein; GEV, gastroepiploic vein; LCV, large collateral vein around portal vein; MPV, main portal vein; PSV, ligated proximal splenic vein; RHV, right hepatic vein; SMV, superior mesenteric vein; TIPS, transjugular intrahepatic portosystemic shunt.
PVT in liver cirrhosis
No symptomatic portal hypertension
50% thrombus, ± SMV extension
Symptomatic portal hypertension
Controlled by ET, LVP and/or drug therapy
Uncontrolled by ET, LVP and/or drug therapy
Anticoagulation (or RCT)
Progressed
Improved or stable
Potential algorithm for treatment of PVT
Progressed
Follow-up
TIPS, or symptomatic treatment if unavailable or failed
Figure 4 | Algorithm for the treatment of portal vein thrombosis in liver cirrhosis. Symptomatic portal hypertension mainly represents variceal bleeding and/or ascites. Abbreviations: ET, endoscopic therapy; LVP, large volume paracentesis; PVT, portal vein thrombosis; RCT, randomized controlled trial; SMV, superior mesenteric vein; TIPS, transjugular intrahepatic portosystemic shunt.
complications (one died of an intravascular disseminated coagulopathy after TIPS, and another died of intraabdominal bleeding caused by hepatic capsule perforation during the TIPS procedure).129,130 In the second case, the investigators clearly acknowledged that they did not fully recognize the risk of hepatic capsule perforation before the lethal event.129 However, with the improvement of TIPS techniques and peri-operative management, severe procedural complications might be gradually reduced. Shunt dysfunction and hepatic encephalopathy still represent the biggest clinical challenge for the management of patients with cirrhosis who are undergoing TIPS. 442 | JULY 2014 | VOLUME 11
The study by Perarnau showed that these postoperative complications were not significantly different between cirrhotic patients with or without PVT.130 The overall incidence of shunt dysfunction and hepatic encephalopathy after TIPS is 7–32% and 0–50%, respectively.90 The incidence of shunt dysfunction has been reduced by the use of covered stents;137,138 in the study by Luca et al. the incidence of shunt dysfunction at 12 and 24 months was 38% and 85% in the bare stent group, and 21% and 29% in the covered stent group.128 Timing of TIPS insertion in patients with cirrhosis and PVT is another major concern. In patients with cirrhosis without PVT, TIPS has been recommended as the secondline treatment option for recurrent variceal bleeding and refractory ascites.139 Whether or not these recommendations should be extrapolated to patients with cirrhosis and PVT deserves further exploration, because the presence of PVT potentially changes the natural history of liver cirrhosis. An RCT140 is being conducted at our centre to compare the incidence of variceal rebleeding, progression of PVT, and survival between cirrhotic patients with PVT who are receiving TIPS and those receiving endoscopic and drug therapy.141,142 A total of 50 patients were estimated to be needed for this trial. All patients were enrolled between June 4, 2011 and January 25, 2014, and are being followed-up. Unless this trial confirms the superiority of TIPS, conventional therapy will remain the first-line treatment of choice for the prevention of variceal re-bleeding in patients with cirrhosis and PVT.142 It might be reasonable that TIPS should be performed in the case of thrombus extension despite adequate anticoagulation96 However, if the thrombus has progressed to complete occlusion or even fibrotic cord, the technical difficulty of TIPS insertion will be substantially increased. Therefore, further studies are warranted to explore the time when anticoagulation therapy should be switched to TIPS. Given that the progression of PVT might be positively associated with treatment failure,52,100,129,143 the selection of treatment options is primarily dependent on the presence of symptomatic portal hypertension and the stage and degree of PVT. On the basis of this consideration and results of existing studies, we propose an exploratory and preliminary algorithm for the management of PVT in liver cirrhosis (Figure 4). However, we acknowledge that this algorithm is not evidence-based owing to the absence of RCTs, and many controversies regarding the indications of each treatment option still exist. First, patients with a Child–Pugh score of 7–10, no ascites, no high-risk varices or previous history of variceal bleeding, and a platelet count >10,000/mm3 have been considered as the target population for a randomized controlled trial to explore the role of anticoagulation in the primary prophylaxis of PVT.70 However, whether or not patients with a Child–Pugh score of 5–6, ascites, high-risk varices, and/or a platelet count
