Anticoagulation in Cirrhosis

Evidence for the Treatment of Portal Vein Thrombosis and Applications for Prophylactic Therapy

Abstract

The clinical utility of anticoagulation for patients with cirrhosis and asymptomatic portal vein thrombosis (PVT) is widely debated. Complex hemostatic derangements in cirrhosis that increase risk of both bleeding and thrombosis, as well as a lack of randomized controlled data, limit conclusive assessments regarding optimal management of anticoagulation in this setting. In this review, we summarize the relevant literature pertaining to PVT in cirrhosis, including the effect of untreated PVT on the natural progression of liver disease and the overall impact of anticoagulation on clot burden and other relevant clinical outcomes. Apart from patients who are symptomatic or listed for liver transplantation, data supporting anticoagulation for the treatment of PVT is limited and without clear consensus guidelines. In patients with cirrhosis without PVT, emerging evidence for the role of prophylactic anticoagulation to mitigate the progression of fibrosis suggests an optimal risk-benefit tradeoff with decreased rates of liver decompensation and mortality, without a heightened risk of bleeding. In summation, as our understanding of the role of both prophylactic and therapeutic anticoagulation in cirrhosis continues to evolve, ongoing risk stratification of patients with asymptomatic PVT demands further attention.

Cirrhosis is associated with an increased risk of portal vein thrombosis (PVT). Most estimates identify the prevalence of PVT in cirrhosis between 0.6% and 15%,^1,2 although some studies show a prevalence as high as 26% in end-stage liver disease.^3–5 Management of anticoagulation in this population remains challenging due to the uncertain benefit of treatment, concerns regarding heightened risk of bleeding, and a paucity of randomized controlled trials (RCTs). Indeed, although PVT is associated with features of advanced cirrhosis, whether PVT contributes to liver disease and increased mortality, or is simply a reflection of advancing cirrhosis, is widely debated.⁶ As such, although therapeutic anticoagulation is associated with improved rates of portal vein recanalization, the clinical utility of treatment remains unknown. Similarly, while a potential benefit of prophylactic anticoagulation in patients with cirrhosis has also been suggested, definitive data is lacking.^7,8 Ultimately, the guidelines for management of PVT in patients with cirrhosis vary without a clear consensus. While the American College of Gastroenterology (ACG) and recently published 2020 Practice Guidance by the American Association for the Study of Liver Diseases recommend anticoagulation for PVT primarily in candidates for orthotopic liver transplantation (OLT) and in those with symptomatic PVT, the European Association for the Study of the Liver (EASL) and Japanese Society of Gastroenterology (JSGE) guidelines extend the recommendation of anticoagulation to all patients with cirrhosis and PVT.^9–12 Notably, the recent Baverno VII workshop updated their previous consensus statement to suggest “anticoagulation should not be discouraged in patients with cirrhosis and an approved indication for anticoagulation,” marking a change from previous more restrictive guidelines.¹³

Underlying the complexity of anticoagulation in patients with cirrhosis are the hematologic derangements of liver failure.¹⁴ Abnormalities including reduced clotting factors, thrombocytopenia, and hyperfibrinolysis contribute to a bleeding diathesis, which can exacerbate threats to hemostasis posed by vascular defects and portal hypertension.^15–21 Conversely, competing derangements including elevated von Willebrand factor, resistance to thrombomodulin, reduced protein C, elevated factor VIII, and enhanced platelet-mediated thrombin production, interact with systemic processes including increased intrahepatic vascular resistance, reduced venous flow velocity, venous stasis, and endothelial activation to also precipitate thrombosis.^3,22 Together, these multiple hematologic dyscrasias result in a “rebalanced,” although tenuous, hemostasis in which patients are at elevated risk for hemodynamically significant bleeding and thrombosis, albeit the balance is thought to weigh more heavily towards hypercoagulability.^14,20,23,24 For patients with cirrhosis, optimal management of thrombosis (most commonly PVT) is consequently complex and incompletely understood. This review highlights the current data regarding the clinical significance of PVT in cirrhosis, including the natural history of untreated PVT, the consequences of untreated PVT on liver disease progression and mortality, the efficacy of and options for anticoagulation in PVT treatment, and the utility of prophylactic anticoagulation in cirrhosis. Below each section entitled “bottom line,” we have provided a summary statement and level of evidence based on the number and types of trials available for review.

UNTREATED PVT IN CIRRHOSIS: RISK OF CLOT AND FIBROSIS PROGRESSION

The Natural History of Untreated PVT in Cirrhosis

The detection of asymptomatic PVT in patients with cirrhosis has steadily risen in the advent of routine liver screening for hepatocellular carcinoma (HCC), and the natural history of untreated PVT varies amongst available data (Table 1). In a retrospective analysis of 42 patients with extrahepatic partial, untreated PVT, routine multidetector computed tomography scans every 3 to 6 months for a mean follow-up duration of 27 months were used to evaluate clot evolution.²⁶ Among untreated patients, 45% had spontaneous decrease in clot burden, 7% remained unchanged, and 48% demonstrated PVT progression, though only 14% demonstrated complete occlusion at 2 years. An additional prospective analysis of 22 patients with cirrhosis and untreated PVT found improvement in 23% of patients, stability in 50%, and progression in 27% using a combination of ultrasound and computed tomography imaging with mean follow-up of 20 months.²⁷ Two retrospective analyses demonstrate similar rates of spontaneous PVT improvement with significantly lower rates of progression: Maruyama et al²⁵ reports a 47.6% rate of spontaneous improvement with only 7.2% progression primarily assessed by ultrasonography, and Acuna-Villaorduna et al²⁹ found 48% of untreated patients had spontaneous recanalization with 3% progression, though imaging modality was not reported. In a recent meta-analysis reviewing 14 available cohort and RCTs, Qi et al³⁵ estimated a pooled incidence of spontaneous PVT recanalization of 39.8%, though there was marked heterogeneity among the study designs, patient characteristics, and severity of liver disease and PVT included in the overall analysis.

TABLE 1 - An Overview of Cohort Studies Comparing Rates of PVT Recanalization in Patients Who Did Not (White) and Did Receive (Gray) Anticoagulation, Listed by Increasing Severity of Cirrhosis (Defined by Mean MELD)

References (Region)	Study Design	Enrollment Period	Target Population	Mean Follow-up Duration (mo)	Sample Size by Treatment	PVT Characteristics	Severity of Cirrhosis*	PVT Recanalization Rate [n (%)]	PVT Stability Rate [n (%)]	PVT Extension Rate [n (%)]
Nery et al6 (France)	Multicenter, prospective	June 2000-March 2006	CP A and B cirrhosis with de novo PVT, did not differentiate treated vs. untreated	47	101 nonocclusive, ∼6 treated	Partial	CP A/B	70/101 (70%)	NR	14/101 (14%)
Maruyama et al25 (Japan)	Single-center, retrospective	January 1998-May 2009	Cirrhosis with de novo untreated PVT	66.3	Untreated n=42	Partial and occlusive	Untreated: CP A/B/C: 14/22/6; MELD 10.6	Untreated: 20/42 (48%)	Untreated: 19/42 (45%)	Untreated: 3/42 (7%)
Luca et al26 (Italy)	Single-center, retrospective	January 2004-December 2009	Cirrhosis with nonmalignant, partial untreated PVT	27	Untreated n=42	Partial, extrahepatic	Untreated: CPS 8.1±1.9; MELD 12.1±2.9	Untreated: 19/42 (45%)	Untreated: 3/42 (7%)	Untreated: 20/42 (48%)
Girleanu et al27 (Romania)	Single-center, prospective	January 2011-October 2013	Cirrhosis with nonmalignant, partial untreated PVT	20.22	Untreated n=22	Partial	Untreated: CP A/B/C: 7/9/6; MELD 12.73±4.34	Untreated: 5/22 (23%)	Untreated: 11/22 (50%)	Untreated: 6/22 (27%)
Hall et al28 (UK)	Single-center, observational	December 1997-December 2010	PVT, excluding those with cirrhosis, malignancy, or liver transplant	43.2	Untreated n=8 Treated n=11	NR	NA	Untreated: 3/8 (38%) Treated: 9/11 (82%)	Untreated: 1/8 (13%) Treated: 1/11 (9%)	Untreated: 4/8 (50%) Treated: 1/11 (9%)
Acuna-Villaorduna et al29 (United States)	Single-center, retrospective	2005-2015	PVT, mixed population	At least 12 mo	Untreated: n=60 Treated: n=25	NR	64% patients with cirrhosis, not further defined	Untreated: 29/60 (48%) Treated: 7/25 (28%)	Untreated: 29/60 (48%) Treated: 8/25 (32%)	Untreated: 2/60 (3%) Treated: 1/25 (4%)
Chen et al30 (China)	Single-center, retrospective	January 2002-June 2014	Cirrhosis with PVT	Untreated: 26 Treated: 33	Untreated: n=36 Treated: n=30	Partial and occlusive	Untreated: CP A/B/C: 8/21/2; MELD 8.9±3.01 Treated: CP A/B/C: 6/17/5; MELD 9.9±4.04	Untreated: 4/16 (25%) Treated: 15/22 (68%)	Untreated: 6/16 (38%) Treated: 4/22 (18%)	Untreated: 6/16 (38%) Treated: 3/22 (14%)
Senzolo et al31 (Italy)	Multicenter, prospective	May 2008-January 2012	Cirrhosis with splanchnic vein thrombosis (88% PVT)	20.4	Untreated: n=56 Treated: n=92	NR	Untreated: CPS† 7 (5-8); MELD† 12 (10-17) Treated: CPS† 7 (5-8); MELD† 12 (10-15)	Untreated: 36.1% Treated: 54.8%	NR	NR
Pettinari et al32 (Italy/Romania)	Multicenter, retrospective	January 2008-March 2016	Cirrhosis with PVT	19	Untreated n=101 Treated n=81	Partial and occlusive	Untreated: CP A/B/C 37/45/19; MELD ≤9: n=14 (14%) Treated: CP A/B/C 43/33/5; MELD ≤9: n=29 (36%)	Untreated: 26/101 (26%) Treated: 46/81 (57%)	NR	NR
Senzolo et al33 (Italy)	Single-center, prospective	January 2007-January 2008	Cirrhosis with nonmalignant PVT	22.53	Untreated: n=21 Treated: n=35	Partial and occlusive	Untreated: CP A/B/C 5/9/7; MELD 13.7±3.6 Treated: CP A/B/C 11/16/8; MELD 12.6±3.7	Untreated: 1/21 (5%) Treated: 12/33 (36%) complete; 9/33 (27%) partial	Untreated: NR Treated: 7/33 (21%)	Untreated: 15/21 (71%) Treated: 5/33 (15%)
Noronha Ferreira et al34 (Portugal)	Single-center, prospective	January 2002-December 2017	Cirrhosis with PVT	29	Untreated: n=32 Treated: n=35	Partial and occlusive	Untreated: CP A/B/C 9/18/16; MELD 16 (SD=7%) Treated: CP A/B/C 12/16/9; MELD 14 (SD=6%)	Untreated: 6/32 (19%) Treated: 18/35 (51%)	NR	NR

In general, rates of recanalization in untreated PVT diminish with increasing severity of cirrhosis.

^*CP data are number of patients; CPS and MELD data are mean±SD (if reported), except where otherwise indicated.

^†CPS and MELD data are median (interquartile range).

CP indicates Child Pugh; CPS, Child Pugh Score; MELD, Model for End-stage Liver Disease; NA, not available; NR, not recorded; PVT, portal vein thrombosis.

The discrepancy in reported outcomes between these analyses may in part be due to differences in imaging modalities and variations in definitions used to quantify clot burden. Though ultrasonography holds a high sensitivity and specificity for PVT detection ranging from 80% to 100% in the majority of studies, the reliability of detection and follow-up measurements is limited by operator experience and patient-specific factors such as overlying bowel gas and obesity that may hinder appropriate interpretation.³⁶ Furthermore, though several grading sytems have been proposed to estimate the extent of PVT occlusion, these classifications, such as the commonly used Yerdel grading system, are largely based on gross measurements of occlusion (<50% or >50% vs. complete occlusion).^37,38

More attention has been drawn recently to identifying predictors for spontaneous recanalization of PVT, termed “transient PVT,”³⁵ for whom anticoagulation may not be necessitated. Unfortunately, data remains sparce and no significant factors associated with spontaneous recanalization have been identified despite multivariate analyses evaluating baseline patient characteristics, severity of liver disease, and thrombus acuity and severity.^26,30 Maruyama et al²⁵ did provide raw data to suggest a possible predictive role of ultrasonic parameters in 42 patients with PVT. This analysis found no significant difference in the natural course of thrombosis based on the degree of obstruction or location of thrombus. Rather, the diameter and flow volume in the largest collateral vessel at the time of thrombus detection was significantly smaller in the patients with spontaneous PVT improvement compared with those with unchanged or worsened thrombosis (3.6 vs. 7.7 mm; 141.1 vs. 451.6 mL/min). Flow volume reflects both mean flow velocity and the cross-section diameter of the collateral vessel. Perhaps these measurements of collateral vessels serve as a surrogate marker of thrombus obstruction, in that greater flow volume through a collateral vessel indicates greater thrombus obstruction and lower likelihood for spontaneous clot recanalization.

Ultimately, further external validation is required to fully understand the predictive utility of ultrasound parameters in patients with PVT. Future large, prospective studies are essential to help stratify patients with asymptomatic PVT who may benefit from anticoagulation versus those for which observation alone could be considered. Last, underlying clinical uncertainty regarding the effects of asymptomatic PVT on morbidity and mortality in cirrhosis further complicates accurate consideration of the risks and benefits of anticoagulation, as discussed below.

Bottom Line

• Incidentally found PVT may spontaneously resolve, though measures to predict spontaneous recanalization are lacking (level of evidence=low).

Proposed Theories Suggesting a Role for Thrombosis in the Progression of Liver Disease

The notion that hypercoagulability contributes to the pathogenesis of liver fibrosis is widely supported in the literature. Most broadly, prothrombotic conditions including factor V Leiden have been shown to accelerate fibrosis, while hemophilia is associated with less severe fibrosis.^39–42 For example, in mice with cirrhosis induced via carbon tetrachloride, warfarin reduced the burden of fibrosis in wild-type models, but failed to confer a benefit in mice with the factor V Leiden mutation, suggesting resistance to treatment in the setting of hypercoagulability.⁴⁰ In a separate study using a similar mouse model, treatment with enoxaparin reduced the portal pressure, fibrosis, and other fibrosis-associated endpoints.⁴³ Interestingly, in mice with inferior vena cava ligation (a model for Budd-Chiari syndrome, another known cause of fibrosis), fibrosis development and stellate cell activation were significantly mitigated with warfarin treatment.^44,45 Last, in a study in which cirrhotic livers were assessed histologically posttransplantation, evidence of intimal fibrosis, which is highly suggestive of reorganized thrombus, was detected in 70%.⁴⁶

Mechanistically, although it was initially postulated that the pathology of cirrhosis involved hepatic microvascular thrombosis and ischemia,^39,42 the “direct stellate cell activation” hypothesis, has a stronger evidence base and is now more widely accepted.⁴⁷ Namely, hypercoagulability is thought to mediate fibrosis through direct thrombin-mediated and Xa-mediated activation of quiescent stellate cells via protease-activated receptors (PAR-1 and PAR-2).^47,48 Upon thrombin binding to a stellate cell PAR, quiescent cells are converted to wound-healing myofibroblasts with robust profibrotic activity.⁴⁷ PAR gene polymorphisms have been shown to influence rates of thrombosis.⁴⁹ Ultimately, there is strong evidence to support an interplay between hypercoagulability and cirrhosis pathogenesis, suggesting a possible role for prophylactic anticoagulation in cirrhosis.

Bottom Line

• Preclinical and histologic data suggest that thrombosis may play a role in the pathophysiology of cirrhosis (level of evidence=low).

THE TREATMENT OF PVT: CLINICAL EFFICACY AND CONSEQUENCES

Efficacy of Anticoagulation in Reducing Thrombus Burden in PVT

Although the clinical implications of PVT in cirrhosis remain somewhat unclear, retrospective studies have consistently demonstrated that anticoagulation results in higher rates of thrombus resolution and lower rates of progression without increasing bleeding risk^50–53 (Fig. 1). Several summative meta-analyses have also evaluated the efficacy of anticoagulation for PVT in patients with cirrhosis and have demonstrated statistically significant higher rates of recanalization and lower rates of progression, without a significantly increased risk of overall bleeding, including variceal bleeding.^53–55

FIGURE 1: The benefits of anticoagulation for both prophylaxis and treatment of PVT in patients with cirrhosis. AC indicates anticoagulation; PAR, protease-activated receptor; PVT, portal vein thrombosis.

In addition to these meta-analyses, a nonrandomized, prospective study involving 98 patients with cirrhosis and splanchnic vein thrombosis treated with anticoagulation showed a pooled partial and complete recanalization rate of 55%. This rate is lower than indicated by the above analyses but demonstrated a 3.33-fold higher rate of recanalization in treated patients without a significant increase in bleeding.³¹ Another systematic review used a population-specific approach, evaluating 210 patients who received anticoagulation for PVT in cirrhosis while awaiting OLT and found partial and complete recanalization rates of 56% and 44%, respectively.⁵⁶ Prompt timing of anticoagulation initiation and treatment adherence were important predictors of recanalization. Additional studies have also identified chronicity of clot as an important predictor of recanalization with anticoagulation. Two prospective analyses showed that anticoagulation is more efficient when given within 6 months from estimated diagnosis of PVT (defined as recent PVT), and treatment-based recanalization rarely occurs in patients with thrombus present for >12 months.^33,57 Though heterogeneity exists between study populations and design, data consistently points towards improved reduction in clot burden and PVT recanalization with anticoagulation. Importantly, as outlined later in this review, bleeding rates do not seem to differ significantly between treatment groups and controls.

Bottom Line

• Anticoagulation leads to higher rates of partial and complete PVT recanalization and lower rates of progression without increasing risk of bleeding, though clinical implications beyond recanalization remain unclear (level of evidence=moderate).
• Timing of anticoagulant initiation is an important predictor of efficacy—higher rates of PVT recanalization are seen when anticoagulants are started within 6 months of diagnosis (recent PVT), whereas recanalization rates diminish with PVT present or persistent for >6 months (chronic PVT) (level of evidence=moderate).

Consequences of PVT and Impact of Anticoagulation on Liver Disease Progression and Mortality

While anticoagulation appears to be safe and effective in reducing splanchnic clot burden, the significance of this outcome in terms of mortality and hepatic decompensation remains unknown. Most available studies exploring this topic are contradictory and limited by design. Among those that have found no significant difference in clinical outcomes in patients with and without PVT is a prospective, multicenter study by Nery et al⁶ which followed the development of PVT in 1243 patients with cirrhosis undergoing routine ultrasound screening for HCC. The 5-year cumulative incidence of PVT was 10.7% and only 6 patients with PVT were treated with anticoagulation. In multivariate analysis, the development of PVT was not associated with progression of liver disease, though notably was associated with greater disease severity as defined by a higher prevalence of esophageal varices. A 2-year retrospective transplant database analysis of 66,506 patients with cirrhosis, not distinctly defined, additionally showed no difference in clinical outcomes or mortality in patients with PVT [mean Model for End-stage Liver Disease (MELD): 18.4±8.2] compared with those without PVT (mean MELD: 17.2±7.9).⁵⁸ A meta-analyses on this topic also produced equivocal results, finding that patients with cirrhosis and PVT had slightly elevated risk of 1-year mortality compared with those without PVT [odds ratio (OR)=0.12; 95% confidence interval:0.04-0.34; P < 0.001], but slightly lower 3-year mortality (OR=1.04; 95% confidence interval: 1.00-1.08; P= 0.06), with similar cumulative survival rates between the 2 groups at 5 and 10 years.⁵⁹

Among studies that have attempted to identify potential clinical benefit, beyond recanalization, of anticoagulation in patients with PVT and cirrhosis is a recent nonrandomized retrospective registry of 80 patients with cirrhosis, two third of whom belonged to Child-Pugh B or C classes at the time of PVT diagnosis, which found that anticoagulation was associated with improved OLT-free survival in a subset of patients with MELD >15.³⁴ However, disparate baseline characteristics between patients who received anticoagulation and those who did not [notably higher mean MELD (16 vs. 14), higher baseline rates of prior variceal bleeding and portal cavernoma in the nonanticoagulation group] complicates interpretation of these results. A prospective study of 22 untreated patients with PVT (mean MELD: 12.7±4.3, mean Child-Pugh: 7.7±1.8) found that progression of PVT on follow-up imaging was associated with significantly higher rates of mortality and hepatic decompensation (although further multivariate analysis found that MELD score at PVT diagnosis was the only independent predictor of survival and decompensation).²⁷ Another prospective study analyzed 2-year follow-up data from a group of 149 patients with cirrhosis and PVT, of which 92 patients received anticoagulation at the discretion of their medical team.³¹ Patients in both study arms had a median MELD score of 12 and there were no statistically significant differences in baseline characteristics, though patients who did not receive anticoagulation had nonsignificantly higher baseline rates of esophageal varices and gastrointestinal bleeding at the time of clinical presentation. Mortality rates were higher in patients with no evidence of vessel recanalization (15.4 deaths per 100 patient years) compared with those who achieved partial or complete resolution of thrombosis (6.8 deaths per 100 patient years), though again, this was only apparent in patients with more advanced (Child-Pugh B and C) cirrhosis. Finally, a retrospective study of 182 patients with cirrhosis and PVT identified an overall survival benefit in patients treated with anticoagulation versus not (hazard ratio=0.30).³² At baseline, patients who received anticoagulation were generally healthier with significantly lower Child-Turcotte-Pugh class, lower MELD (≤9), lower bilirubin levels and lower baseline international normalized ratio. Child-Turcotte-Pugh classes B and C were identified as independent risk factors for mortality (HR 3.09 and 9.27, respectively).

There are several important takeaways from the results of the studies summarized above. First, though most studies demonstrate potential morbidity and mortality benefits of treating PVT (Fig. 1), nonuniform inclusion criteria and study methodology, as well as small study sizes, limit overall interpretation of results. Second, it should be highlighted that the majority of studies did not routinely include patients with more advanced, decompensated cirrhosis (MELD >16). The small number of studies that included this patient population (either nonrandomized prospective trials or retrospective analyses) demonstrate that the nonanticoagulated group typically had higher MELD scores compared with the anticoagulated group contributing to potential confounding. Furthermore, the paucity of available data limits general recommendations regarding safety and efficacy of anticoagulation for management of PVT in patients with advanced cirrhosis. Ultimately, prospective, randomized trials are desperately needed to adequately compare PVT management across the spectrum of disease severity to provide adequate guidance to clinicians.

More definitive data is available to suggest poor prognoses among certain subsets of patients with PVT, namely those listed for OLT and those with HCC. In a large, retrospective analysis of 134,109 patients with cirrhosis awaiting OLT, those with PVT were more likely to be removed from the transplant waitlist and have higher MELD scores.⁶⁰ In addition, PVT at the time of listing or actual transplant has been associated with both worsened patient and graft survival in several transplant database analyses, with increased mortality rate cited as high as 30%.^60–62 This is in part due to surgical challenges, such as portal vein reconstructions in patients with thrombus that add technical difficulties and increase graft ischemic times perioperatively.¹² Thus, although PVT is no longer an absolute contraindication for OLT, its presence requires significant surgical considerations and can often lead to delisting. Similarly, thrombus associated with HCC has also been associated with a significantly worse prognosis, with an estimated median overall survival of 2 to 4 months.⁶³ The poor prognosis results from a combination of multiple factors including aggressive tumor behavior, reduced hepatic reserve in part due to portal hypertension, and limitations imposed by an occluded portal vein on available treatment options, including surgery, transarterial chemoembolization, and radiofrequency ablation.^63–65 Owing to these considerations, anticoagulation to prevent or treat PVT in patients on or awaiting transplant or those with HCC may be more clearly indicated.

Beyond mortality outcomes, there is some evidence to suggest that patients with cirrhosis and PVT who are treated with anticoagulation may have decreased rates of variceal bleeding.^33,51 A meta-analysis which comprised of 4 studies with 158 patients found a significantly lower rate of variceal bleeding (2% vs. 12%, OR=0.232) in patients who were treated with anticoagulation compared with those who were not. Similar findings were shown in 2 other small studies: a prospective cohort of 56 total patients which found 1 episode of variceal bleeding in the treatment arm compared with 5 episodes in the control arm (P=0.09) and a retrospective cohort of patients with noncirrhotic PVT which found a significantly higher rate of portal hypertension–related complications (ascites, varices) in patients who failed to recanalize compared with those who did (83.3% vs. 27.3%, respectively).^28,33 The theorized mechanism driving this association suggests that treatment of PVT leads to decreased clot burden, thus reducing portal hypertension and preventing variceal rupture. More robust conclusions are limited as many studies did not directly measure this endpoint, therefore future prospective analyses are needed to provide further validation.

In general, reasonable selection criteria are available to identify optimal candidates for anticoagulation in PVT, including thrombus location, acuity, etiology (malignant vs. nonmalignant), and variceal burden/prophylaxis.⁶⁶ Efficacy and safety of anticoagulation in patients with advanced cirrhosis is overall unknown given the small number of studies that have included patients with MELD >16. Based on current available evidence, some professional societies recommend anticoagulation primarily for PVT in candidates for OLT and in those with symptomatic PVT, while others extend the recommendation of anticoagulation to all patients with cirrhosis and PVT.^9–13 Overall, the trend appears to be moving toward recommending treatment in all patients with PVT (as evidence by the recently updated Baveno VII consensus), despite conflicting evidence on the clinical benefits of treatment. Figure 2 provides a suggested algorithm for treatment of PVT based on available data. Future studies with additional focus on understanding the complexities of monitoring response to anticoagulation and determining optimal treatment duration are warranted.

FIGURE 2: Recommended algorithm for treatment of PVT based on available data. MELD indicates Model for End-stage Liver Disease; PVT, portal vein thrombosis; SMV, superior mesenteric vein.

Bottom Lines

• There is clinical equipoise on the benefits of treating asymptomatic PVT due to the heterogeneity of available studies and baseline cofounders such as disease severity and other comorbidities. Future randomized, prospective trials are needed to define if there is clinically meaningful benefit beyond reducing clot burden (level of evidence=low).
• Data on the use of anticoagulants in patients with MELD > 16 and Child-Turcotte-Pugh class C is lacking, therefore results cannot be generalized to patients with more advanced cirrhosis (level of evidence=low).
• Some analysis shows decreased rates of variceal bleeding when anticoagulation is used to treat PVT (level of evidence=low).

Options for Anticoagulation in Patients With Cirrhosis

Most studies reviewing anticoagulation in patients with PVT have used vitamin K antagonists (VKAs) and low–molecular-weight heparin (LMWH). LMWH is favored over vitamin K agonists given challenges with a counterproductive reduction in protein C and reliance on poorly representative international normalized ratio monitoring with VKAs, which can result in significant time spent above or below the therapeutic range. Notably, heparin products are also challenging in liver disease given variable levels of anti-Xa and subsequent challenges with monitoring via the anti-Xa assay.^67,68 Although patients with cirrhosis were not included in original trials demonstrating direct oral anticoagulant (DOAC) safety and efficacy, evaluation of traditional anticoagulation strategies (VKA and LMWH) versus the use of DOACs in patients with liver disease are beginning to emerge. These studies as well as a meta-analysis have suggested DOACs to be safe and effective in patients with cirrhosis, with similar rates of clot recurrence and progression as well as an equal or even reduced bleeding risk, particularly in the central nervous system, without the need for stringent monitoring.^69–74 Indeed, RCTs in patients with PVT have demonstrated DOAC superiority over warfarin in terms of reducing clot burden and overall mortality.^52,75–78 Notably however, all DOACs undergo some form of hepatic metabolism, and while evidence suggests that they do not carry an elevated risk of serious hepatic injury, further adequately powered studies are warranted to fully assess safety.^79,80 Currently approved anticoagulants, including the pros and cons of each agent, are summarized in Table 2. In sum, although more robust data is needed to better examine these effects, DOACs can be considered for the treatment of PVT in patients with cirrhosis. Apixaban may be considered as the optimal agent as its metabolism is not significantly altered in chronic liver disease.

TABLE 2: Anticoagulant Agents in Cirrhosis

Bottom Line

• There is increasing data to suggest that DOACs can be used in patients with chronic liver disease, though LMWH remains a viable option (level of evidence=moderate).

ANTICOAGULATION FOR PROPHYLAXIS IN CIRRHOSIS

Several analyses have also suggested potential benefits of prophylactic anticoagulation in patients with liver disease (Fig. 1). In a nonblinded, randomized, prospective single-center study, 70 patients with cirrhosis and patent portal veins were randomized to daily prophylaxis with enoxaparin versus placebo for 48 weeks.⁸ All included patients were Child-Pugh B7-C10 without evidence of decompensation, which was defined by the presence of ascites, spontaneous bacterial peritonitis, portal hypertensive-related bleeding, or portosystemic encephalopathy. The presence of varices was not an excluding factor. Patients were followed for over 144 weeks by imaging and routine hepatology visits and only 1 patient discontinued the intervention due to heparin-induced thrombocytopenia. There were no significant differences in baseline characteristics between treatment and intervention arms. At 48 and 96 weeks, no patients in the intervention group developed PVT compared with 16% and 27.7% of the controls, respectively. At the conclusion of follow-up at 144 weeks, 8.8% of patients in the intervention arm developed PVT compared with 27.7% of the controls. Furthermore, rates of survival (23.5% vs. 36.1%) and liver decompensation (38.2% vs. 83.0%) were significantly improved in those treated with prophylactic enoxaparin. Patients in the intervention arm also experienced significantly lower rates of documented bacterial infections and reduction in markers of enterocyte damage and microbial translocation. Based on these findings, it is theorized that prophylactic enoxaparin may exert a direct anti-inflammatory effect and potentially reduce bacterial translocation via improved intestinal microcirculation.⁸² This study was criticized for having a higher than usual incidence of PVT (more commonly cited ~10%^6,83,84), although inclusion of more advanced cases of cirrhosis compared with other studies may explain this difference.⁸⁵ Regardless, significant improvements in survival and lower rates of decompensation in the intervention arm signify the relevance of hypercoagulability and thrombosis as precipitants of fibrosis.

An active clinical trial, CIRROXABAN (NCT02643212), is currently evaluating prophylactic rivaroxaban versus placebo in 160 patients with Child-Pugh 7 to 10 cirrhosis without previous or current splanchnic thrombosis.⁸⁶ This is a phase 3, double-blind, and multicenter study with primary outcomes of transplant free survival and liver decompensation. Secondary outcome measures include cirrhosis progression, splanchnic thrombosis development, incidence of HCC, and changes in hepatic venous pressure gradient. This, along with future studies assessing additional clinical questions that include the effect of anticoagulation on intestinal microcirculation parameters and optimal type and dose of anticoagulation in cirrhotic patients, will help further define the role and management of thrombosis in the progression of cirrhosis.⁸²

Bottom Line

• Small studies suggest that the use of prophylactic anticoagulation in patients with chronic liver disease prevents PVT and possibly may improve other clinically meaningful outcomes including the incidence of hepatic decompensation (level of evidence=low).

CONCLUSIONS

The role of anticoagulation in patients with cirrhosis and asymptomatic PVT remains widely debated. Part of the difficulty in identifying those who may benefit from anticoagulation is identifying thromboses that may spontaneously recanalize from those that may progress. Though anticoagulation has consistently demonstrated reduction in clot burden and progression without significantly increased risk of bleeding, in most patients there is unclear clinical benefit in regard to mortality or disease progression. Overall, anticoagulation for the treatment of acute PVT should be considered on an individualized basis after weighing concurrent risks and benefits of treatment alongside patient-specific factors, such as severity of disease, transplant candidacy, presence of symptoms, degree of clot burden and extension, and other relevant comorbidities.

An area of active and exciting research lies in the role of prophylactic anticoagulation in patients with cirrhosis, the foundation of which was built on the known association between hypercoagulability contributing to fibrosis progression. Human and animal models that address this theory using anticoagulation as a form of secondary prevention in cirrhosis demonstrate promising results from fibrosis and outcome prevention data. In a disease that lacks standard, universally accepted treatment to halt progression, prevent decompensation and improve mortality, the role of prophylactic anticoagulation marks a potentially promising step forward.