Vascular Liver Disorders, Portal Vein Thrombosis, and Procedural Bleeding in Patients With Liver Disease: 2020 Practice Guidance by the American Association for the Study of Liver Diseases

New Features and Changes From the Previous American Association for the Study of Liver Diseases Guidelines

Preamble

Previous guidelines reviewing vascular disorders of the liver^(1-3) focused on thrombosis of large vessels, such as the portal vein (PV) and hepatic vein (HV). However, increased understanding and research in hemostasis and bleeding in patients with cirrhosis has led to more diagnostic and therapeutic opportunities targeting prevention of bleeding, especially procedural bleeding. Therefore, this Guidance covers bleeding related to disturbances in the hemostatic system predisposed by chronic liver disease as well as thrombosis, especially in the splanchnic vasculature.

This American Association for the Study of Liver Diseases (AASLD) Guidance provides a data-supported approach to the management of vascular liver disorders, portal (PVT) and hepatic vein thrombosis (HVT), and procedural bleeding in patients with liver disease. It differs from AASLD Guidelines, which are supported by systematic reviews of the literature, formal rating of the quality of evidence and strength of the recommendations, and, if appropriate, meta-analysis of results using the Grading of Recommendations Assessment Development and Evaluation system. In contrast, this document was developed by consensus of an expert panel and provides guidance statements based on formal review and analysis of the literature on the topics with oversight provided by the AASLD Practice Guidelines Committee at all stages of Guidance development. The AASLD Practice Guidelines Committee chose to commission a guidance on this topic because a sufficient number of randomized controlled trials were not available to support the development of a meaningful guideline.

Current Understanding of Coagulation and Hemostasis in Cirrhosis

Patients with cirrhosis have multiple alterations in their hemostatic system.⁽⁴⁾ First, defective hepatic synthetic capacity results in decreased plasma levels of coagulation factors, inhibitors of coagulation, and fibrinolytic factors. Second, circulating platelet numbers are decreased because of a combination of decreased thrombopoietin (TPO) synthesis, splenomegaly with sequestration, accelerated platelet turnover, and decreased megakaryocyte production.⁽⁵⁾ Third, plasma levels of hemostatic proteins that are primarily synthesized by endothelial cells are substantially elevated.⁽⁶⁾ Fourth, low-grade intrahepatic and/or systemic activation of the hemostatic system results in consumption of hemostatic proteins.⁽⁷⁾ Fifth, acquired disorders in platelet function and posttranslational modification of hemostatic proteins (e.g., fibrinogen) result in altered function.⁽⁸⁾ There are many more hemostatic changes in patients with cirrhosis that are as yet unknown or not adequately characterized. Such changes include newly discovered roles of neutrophils⁽⁹⁾ and red blood cells⁽¹⁰⁾ in hemostasis and thrombosis; endothelial changes, including alterations in the endothelial glycocalyx^{(11, 12)}; and alterations in circulating levels of procoagulant microparticles.⁽¹³⁾ Figure 1 provides an overview of the current understanding of the cellular-based hemostasis system in cirrhosis.

Hemostatic changes promoting bleeding and clotting occur simultaneously in any individual patient, and some of these changes counteract each other. For example, thrombocytopenia in cirrhosis is alleviated by elevated levels of the platelet adhesive protein, von Willebrand factor (VWF),⁽⁶⁾ decreased levels of procoagulant proteins are counteracted by decreased levels of natural anticoagulant proteins,⁽¹⁴⁾ and decreased levels of antifibrinolytic proteins are counteracted by decreased levels of profibrinolytics,⁽¹⁵⁾ except in critically ill patients.⁽¹⁶⁾ In view of these changes and clinical observations, cirrhosis is no longer considered a condition associated with an overall bleeding tendency, but rather both prohemostatic and antihemostatic pathways are disturbed in complex ways, such that it is difficult to determine in an individual patient exactly where the balance is using currently available clinical tests.

Traditional laboratory measures of coagulation, such as prothrombin time (PT), international normalized ratio (INR), and activated partial thromboplastin time (aPTT), have proven inadequate and often misleading in the evaluation of hemostasis in cirrhosis because these measures only partially evaluate the hemostasis system and neglect the counterbalanced factors mentioned above (Table 1). Global tests of hemostasis, including plasma-based thrombin generation tests, plasma-based fibrinolysis assays, and whole-blood viscoelastic tests of clot formation, have been instrumental in developing the current concept of rebalanced hemostasis. For example, although traditional laboratory measures of coagulation may suggest a bleeding tendency, more advanced hemostasis tests, including plasma-based thrombin generation tests and whole-blood viscoelastic tests such as thromboelastography (TEG) or rotational thromboelastometry (ROTEM), have shown that patients with cirrhosis have normal to elevated thrombin-generating capacity and near-normal whole-blood clot formation.⁽⁴⁾ Details of strengths and weaknesses of TEG and ROTEM in liver disease have been published.⁽¹⁷⁾ Nevertheless, all currently available global hemostasis tests have limitations and should be interpreted cautiously because the predictive value of all available tests for spontaneous or procedural bleeding is unproven (Table 2). Whether global hemostasis tests are able to identify patients at risk for procedural bleeding has been poorly studied. Published data on viscoelastic tests, in particular, are contradictory because of a heterogeneous collection of subjects and procedures tested, a predominance of study procedures with a low risk of bleeding, poorly defined bleeding endpoints, and the use of fresh frozen plasma (FFP) transfusion in the standard-of-care arms.^(18-22) However, because these tests are frequently within normal ranges in patients with cirrhosis, they may be useful in selecting which patients do not need preprocedural intervention. Because of the systematic underestimation of coagulation capacity by currently available viscoelastic testing methods,⁽²³⁾ abnormal preprocedural viscoelastic testing may still not require correction. Further study of viscoelastic testing with standardized clinical bleeding endpoints is needed.⁽²⁴⁾

Although there are simultaneous changes in prohemostatic and antihemostatic pathways that have been interpreted to result in a reset in the hemostatic balance,⁽²⁵⁾ there are poorly understood distinct hypercoagulable and hypocoagulable features that may contribute to thrombosis or bleeding in any individual patient (Table 1).

Procedural Bleeding in Patients With Cirrhosis

Procedure-related bleeding in patients with cirrhosis may be a consequence of hemostatic failure, but many bleeding events arise as a consequence of unrelated factors, such as portal hypertension or vessel rupture/puncture. These types of bleeding events are therefore not preventable with preprocedural manipulation of the hemostasis system, although major nonhemostatic bleeding events may eventually turn into hemostatic bleeds because of the consumption of hemostatic proteins. Consequently, clinicians routinely face complex dilemmas when assessing procedural bleeding risk because bleeding can occur from different underlying causes (Fig. 2). Interventions directed at preventing and treating bleeding should specifically target the expected or actual source of bleeding (e.g., vessel rupture, portal hypertension, or hemostatic failure).

Bleeding Risk Stratification

Assessment of bleeding risk for any particular procedure is dependent on numerous factors that can be broadly grouped into procedural- or technical-related factors, disease-state characteristics inherent to cirrhosis, and systemic factors.^(26-32) Although these factors are interdependent and vary over time, understanding each component’s relative contribution to bleeding risk is important when estimating risk. Past studies have grouped both low- and high-risk procedure types together to derive broader conclusions with respect to an overall bleeding risk,^{(18, 33-35)} but procedure risk stratification should be based on specific procedure type when data are available.

Procedural and Technical Factors

In previous guidance statements and expert opinion documents, procedure bleeding risk has been categorized as low or high (Table 3).^{(26, 28, 29)} Low-risk procedures constitute all procedures whereby major bleeding is expected to occur in <1.5% of cases and/or significant bleeding, if it occurs, can be easily controlled. Conversely, high-risk procedures have increased risk of bleeding (estimated >1.5% risk of major bleeding) and/or hemorrhage, if it occurs, can be difficult to control or may lead to catastrophic consequences, even in small amounts (e.g., central nervous system bleeding).^{(26, 28)} This risk dichotomy is based on expert opinion and generally grounded in the context of periprocedural therapeutic anticoagulation, but provides a useful framework when engaging in multidisciplinary planning before procedures.

An individual procedure will involve different technical risks. Both provider experience and technique directly affect outcomes. For example, the use of imaging with ultrasound guidance for thoracentesis in patients with abnormal coagulation parameters reduces the risk of bleeding.⁽³⁶⁾ Similarly, the use of ultrasound guidance for PV access during transjugular intrahepatic portosystemic shunt (TIPS) reduces the risk of capsule perforation, a known precipitant of intraperitoneal hemorrhage.⁽³⁷⁾ Although factors such as technique and operator experience are often discussed, they are difficult to quantify and study systematically, but should be considered in procedure risk stratification. Another important factor inherent to each individual procedural technique is the ability to control active bleeding if it occurs. For example, significant bleeding after dental extractions is often prevented with local hemostatic measures, even in patients with advanced cirrhosis, significant thrombocytopenia, and elevated INR.^{(38, 39)}

Liver Disease Factors

Patients with cirrhosis can present with a wide spectrum of diseases, from well-compensated cirrhosis to chronic decompensation to acute decompensation with multiorgan failure.⁽⁴⁰⁾ Advanced liver disease is associated with thrombocytopenia attributable to portal hypertension and a prolonged INR as a result of an impaired synthesis of coagulation factors. Although these coagulation indices are historically used to assess bleeding risk, they are misleading in patients with cirrhosis.

Advanced portal hypertension is an independent risk factor for procedure-related bleeding, even in low-risk procedures,^{(41, 42)} but there are factors that may confound this finding. The hemostatic system is affected by hepatic decompensation, and patients with cirrhosis are at risk of becoming critically ill, requiring exposure to multiple procedures and potentially increasing the risk of bleeding complications.^{(43, 44)} Acute decompensation of cirrhosis and acute-on-chronic liver failure (ACLF) directly alter the hemostatic system.^{(16, 45, 46)} A retrospective case-control study demonstrated that patients with ACLF had a higher risk of hemorrhage after paracentesis, and low fibrinogen levels were associated with this risk.⁽⁴⁷⁾ Of note, there were no significant differences between platelet and INR levels between matched controls and patients with bleeding, and neither factor was predictive of bleeding on multivariable analysis. The systemic inflammatory response and association with concurrent infection alter hemostatic function, as measured by global hemostasis testing, and may be associated with an increased risk of bleeding, particularly in the setting of invasive procedures. Overall, data remain sparse on the independent influence of advanced portal hypertension, acute decompensation, and ACLF in hemostasis; however, it is likely that patients with these disorders are at higher risk of procedural bleeding, and careful consideration in preprocedural planning is recommended.

Systemic Factors

Patients with cirrhosis often have multiple medical comorbidities that may affect overall bleeding risk. Among these, acute and chronic kidney disease (CKD) have been most extensively studied as risk factors for bleeding. Underlying alteration in coagulation factors may play a role in the association of bleeding risk and acute kidney injury (AKI).^{(48, 49)} One study examined patients with postparacentesis hemorrhage and found an association between AKI and bleeding, independent of infection.⁽⁵⁰⁾ In a large retrospective analysis of large-volume paracentesis, nine bleeding events out of 4,729 procedures were reported, and renal dysfunction was reported in 89% of patients (8 of 9 patients) with bleeding events.⁽⁵¹⁾ A study examining postbanding ulcer bleeding after endoscopic variceal ligation (EVL) demonstrated that patients who developed bleeding (n = 6) compared to those who did not (n = 259) had significantly higher serum creatinine (2.2 mg/dL bleeders vs. 1.0 mg/dL nonbleeders; P = 0.001).⁽⁵²⁾ In this study, there was no significant difference in platelet level and INR between patients with bleeding and those without bleeding. Based on these data, addressing acute and/or chronic kidney dysfunction as much as possible before elective procedures is recommended.

Medications, in particular antiplatelet therapy and anticoagulation, alter the risk of bleeding during procedures and should be carefully considered in the periprocedural period. In a retrospective cohort analysis comparing hospitalized patients with cirrhosis receiving venous thromboembolism (VTE) prophylaxis to those who did not, there was a trend of increased nongastrointestinal bleeding events in patients receiving VTE prophylaxis (64% of nongastrointestinal bleeding events were procedure-related bleeding).⁽⁵³⁾ On multivariable analysis, the use of VTE prophylaxis in this cohort (OR, 2.36; CI, 1.12-4.97; P = 0.025) was associated with overall in-hospital bleeding, yet platelet count, INR, and Model for End-Stage Liver Disease (MELD) score were not predictive of bleeding. However, in a study examining post-EVL ulcer bleeding, the use of therapeutic low-molecular-weight heparin (LMWH) was not associated with bleeding.⁽⁵²⁾ More data are needed to better understand the effect of anticoagulation on bleeding risk with specific procedures in patients with cirrhosis.

Examples of Common Procedure Risk Assessment in Patients With Cirrhosis

Procedures are routinely performed in patients with cirrhosis (e.g., paracentesis and endoscopy) with very low risk of bleeding in the presence of elevated INR and thrombocytopenia.^{(41, 54, 55)} Therefore, the use of prophylactic transfusions of FFP and/or platelets based on preprocedural laboratory tests is not routinely indicated.⁽⁵⁶⁾ For example, a study of 1,100 therapeutic paracenteses in 628 patients (513 with cirrhosis) and no prophylactic transfusion before the procedure showed no bleeding or complications requiring hospitalization in any patient (292 patients with INR > 2; 598 patients with platelets <50,000/µL).⁽⁵⁵⁾ EVL can lead to delayed postbanding ulcer bleeding in patients with cirrhosis. In one study of 150 patients with cirrhosis undergoing EVL, traditional hemostasis testing (elevated INR and thrombocytopenia) was not predictive of EVL bleeding, but advanced cirrhosis (Child-Turcotte-Pugh [CTP] C) was the most important factor associated with risk of post-EVL ulcer bleeding.⁽⁴²⁾ This finding has been reproduced in other studies.⁽⁵⁷⁾ Because EVL bleeding is typically delayed by several days, prophylactic measures before or during the procedure will likely not alter post-EVL bleeding risk. As such, the use of prophylaxis, based on traditional measures of hemostasis before EVL, is not recommended. The risk of bleeding with liver biopsy, either percutaneous or transjugular, is low in the general population, but may be higher in patients with cirrhosis.^(58-61) Historically, traditional hemostasis studies do not correlate with bleeding risk in patients undergoing liver biopsy.^{(58, 62)} One study of 2,740 percutaneous liver biopsies performed in patients with advanced fibrosis or well-compensated cirrhosis documented a 0.6% bleeding rate with no bleeding-related deaths.⁽⁶¹⁾ Data from this study suggest an association with thrombocytopenia (<60,000/µL) and elevated INR (>1.3) and postbiopsy hemorrhage.⁽⁶¹⁾ Because of the conflicting data in the literature, current AASLD guideline recommendations on liver biopsy do not identify a specific INR or platelet cutoff in which the risk of bleeding is substantially increased, but suggest individualized approaches and assessment of other risk factors that may increase bleeding risk,⁽⁶⁰⁾ as discussed above.

Preprocedural Interventions to Prevent Bleeding

The preparation and intraoperative management of patients with cirrhosis undergoing major surgery, including liver transplantation (LT), has been described.⁽⁶³⁾ However, it is important to note that data suggest that LT surgery can be safely performed without prophylactic or protocol-driven preprocedural administration of blood, FFP, or platelets.⁽⁶⁴⁾ In the general medical population, it has been well established that preprocedural hemostasis laboratory testing serves little purpose unless patients have a history of bleeding or an inherited bleeding disorder or are using antithrombotic drugs.⁽⁶⁵⁾ Because patients with cirrhosis often do have a bleeding history and because of the known hemostatic abnormalities of these patients, preprocedural hemostasis testing is very common. In the following sections, we argue that there is little evidence that correction of hemostatic laboratory abnormalities before common procedures decreases bleeding risk.

Platelet Interventions and Therapies Targeting Primary Hemostasis

In vitro data have been interpreted to suggest platelet levels >55,000/μL⁽⁶⁶⁾ improve hemostasis in patients with cirrhosis. However, these in vitro data have only assessed platelet procoagulant activity and have not accounted for potential compensation by VWF and other endothelial-based components. In addition, this threshold has not been validated clinically. Data suggest that platelet transfusions do not substantially improve thrombin generation capacity or viscoelastic markers of bleeding risk.⁽⁶⁷⁾ Furthermore, despite modest rises in absolute platelet counts, they carry a potential for the transfusion-related lung injury syndromes.⁽⁶⁸⁾

The data on a threshold platelet level for bleeding risk minimization before procedures are mixed. Some studies have shown no predictive value of preprocedural peripheral platelet count on procedural bleeding complications.^{(69, 70)} In a prospective cohort study of critically ill patients, of whom 211 had cirrhosis, the most common bleeding events were spontaneous gastrointestinal or variceal bleeding; however, 10 events (occurring in 4.7% of the entire cohort of patients with cirrhosis) were postprocedural or postoperative.⁽⁷¹⁾ In this cohort, although a peripheral platelet count of <30,000/μL was associated with bleeding, most of the bleeding was attributable to portal hypertension. Another study of 50 LT candidates with platelet counts <125,000/μL (52% with counts <75,000/μL) were followed for procedural complications.⁽³³⁾ There were 10 (20%) bleeding complications after procedures, all of which occurred in patients with a platelet count <75,000/μL. Despite this finding, patients in this study who received prophylactic platelet transfusion before the procedures were paradoxically more likely to experience bleeding. Therefore, this suggests that the low platelet count may have been merely a reflection of advanced portal hypertension and not a causative risk factor for bleeding.

There are three medications approved by the U.S. Food and Drug Administration (FDA) for increasing platelet counts in patients with cirrhosis. All of these agents are TPO receptor agonists and stimulate bone marrow production of platelets. Eltrombopag has the obsolete indication for treatment of thrombocytopenia related to interferon-based hepatitis C therapy and is now rarely used in the prophylactic role. However, an early study of this agent designed to produce platelet counts in the “healthy normal” range before invasive procedures in patients with cirrhosis was discontinued early because of excess thrombotic events, particularly PVT, in the treatment arm, possibly related to excessively high platelet counts.⁽⁷²⁾ Avatrombopag⁽³⁵⁾ and lusutrombopag⁽³⁴⁾ are indicated for the treatment of thrombocytopenia in adult patients with chronic liver disease who are scheduled to undergo a procedure. Each of these oral agents requires completion of a 2- to 8-day course preceding the scheduled procedure. These agents are superior to placebo in achieving a target platelet count ≥50,000/μL before the procedure, with no statistical differences in thrombotic complications compared to placebo. Of note, there were no statistical differences in postprocedural bleeding events in these studies between treatment arm and placebo, and therefore routine use of these agents to prevent procedure-related bleeding cannot be recommended.

There are no high-quality data on appropriate platelet thresholds before procedures, and general interventions to increase platelet counts to prevent bleeding are not evidence based and cannot be recommended. Given the low bleeding risk of many common procedures, potential risks of platelet transfusion, lack of evidence that elevating the platelet count reduces bleeding risk, and ability to use interventions, including transfusion and hemostasis procedures on an as-needed basis if bleeding occurs, it is reasonable to perform both low- and high-risk procedures without prophylactically treating the platelet count. This recommendation deviates from recommendations by other societies (Table 4), and, ideally, studies with clinical bleeding endpoints and placebo controls are required to examine the best strategy for periprocedural platelet management. Understanding the inconsistency of various societal recommendations and multidisciplinary nature of procedural health care, an individualized approach to patients with severe thrombocytopenia before procedures is recommended because of the lack of definitive evidence for safety and efficacy of interventions intended to increase platelet counts in patients with cirrhosis.

1-Deamino-8-d-arginine vasopressin (DDAVP) increases release of endothelial VWF. A randomized controlled trial using intranasal administration of DDAVP showed equivalent postprocedural bleeding rates compared with platelet and FFP transfusions preceding dental extractions in patients with cirrhosis with INR between 2 and 3 and/or platelet count between 30,000 and 50,000/μL.⁽⁷³⁾ The mechanism of action of DDAVP in this study was not clear, and DDAVP was shown to not affect primary hemostasis or platelet adhesion in another study of patients with cirrhosis.⁽⁷⁴⁾

Coagulation Interventions and Therapies Targeting Procoagulant Factor Deficiencies

As discussed above, the PT and its derivative, the INR, are widely available traditional measures of coagulation. However, because the INR only assesses quantitative defects in procoagulant clotting factors (Tables 1 and 2), it is not a reliable indicator of hemostatic balance in patients with cirrhosis and is unable to predict procedural bleeding risk. The literature on the INR’s poor predictive value in procedural bleeding is extensive.^{(42, 55, 60, 75-77)} However, INR is a surrogate indicator of protein synthetic functional capacity of the liver and may correlate with bleeding risk related to severity of liver disease, rather than bleeding risk related directly to hemostatic failure.

FFP transfusions carry important risks: (1) a potential for developing transfusion-related lung injury syndromes⁽⁶⁸⁾ and (2) increases in portal pressures directly proportional to the volume transfused.⁽⁷⁸⁾ Importantly, FFP transfusions show minimal efficacy in improving thrombin generation capacity in vitro.⁽⁷⁹⁾ Outside of advanced malnutrition states or chronic cholestasis, vitamin K replacement has no measurable effect on the INR in patients with cirrhosis. Determination of factor V levels can aid the practitioner in differentiating vitamin K deficiency from impaired liver function if the diagnosis is in question. Given the lack of effect of FFP on hemostatic capacity in the population of patients with cirrhosis, we advise against prophylactic FFP transfusion before common procedures and await studies with clinically relevant endpoints to yield further guidance. This recommendation is consistent with the AASLD practice guidelines for management of ascites⁽⁵⁶⁾ and portal hypertensive bleeding⁽⁸⁰⁾ as well as other professional society guidelines.^{(3, 28, 81, 82)}

Recombinant factor VIIa is effective in normalizing the INR in patients with cirrhosis, yet a randomized clinical trial showed no benefit in controlling esophageal variceal hemorrhage.⁽⁸³⁾ Its use in a prophylactic setting before invasive procedures is not recommended, and the use of this agent in nonhemophilia patients has been associated with increased risks of thrombosis, including arterial thromboses.⁽⁸⁴⁾ Prothrombin complex concentrates targeting both the deficiency of the procoagulant system as well as the acquired deficiency of the innate anticoagulant system are under study in patients with liver disease,^{(85, 86)} but efficacy and safety data are lacking to date.

Fibrinogen and Therapies Targeting Excessive or Inappropriate Fibrinolysis

Plasma fibrinogen levels <100 mg/dL are associated with spontaneous and procedure-related bleeding in patients with cirrhosis,^{(71, 87)} but causal relationships are not established. Low fibrinogen levels may reflect critical illness and do not directly cause bleeding in most cases. In a case series of critically ill surgical patients without cirrhosis, a fibrinogen level of ≥100 mg/dL resulted in greater mechanical clot stability compared to lower levels of fibrinogen, but the relationship was relatively linear without an obvious threshold effect.⁽⁸⁸⁾ The literature on fibrinogen replacement for bleeding is derived from trauma literature or advanced fibrinolytic states during cardiac surgery and LT. Diagnostic laboratory tests for hyperfibrinolysis are not readily available in clinical practice, and current viscoelastic testing is not sensitive for moderate or mild hyperfibrinolysis. The diagnosis of hyperfibrinolytic postprocedural bleeding is made clinically because laboratory testing is not reliable for this condition. Typical manifestations include continuous venous oozing from skin puncture sites and persistent mucosal or submucosal bleeding. Epsilon-aminocaproic acid (EACA), a lysine analogue that prevents plasminogen and tissue plasminogen activator (tPA) from binding to fibrin, has shown benefit in stopping fibrinolytic bleeding without major toxicity in an uncontrolled case series.⁽⁸⁹⁾ In patients without cirrhosis, some benefits of antifibrinolytic therapy are observed after dental extractions in patients on continuous therapeutic anticoagulants.⁽⁹⁰⁾ There are no definitive data supporting the use of EACA for prevention of bleeding in cirrhosis, although elective procedures should be delayed during states of clinically apparent pathological fibrinolysis. Tranexamic acid has antifibrinolytic activity through a similar mechanism as EACA. A recent international multicenter, randomized, double-blind, placebo-controlled trial of high-dose tranexamic acid as an adjunctive therapy in patients presenting with acute gastrointestinal bleeding showed no reduction in death or bleeding in the tranexamic acid arm.⁽⁹¹⁾ This study population included a minority of patients with cirrhosis, and a subgroup analysis showed the risk of VTE events to be higher in patients with cirrhosis receiving the tranexamic acid. Until further supportive data are published, the prophylactic use of EACA or tranexamic acid to prevent procedural bleeding cannot be recommended.

Because fibrin and its precursor, fibrinogen, are key components of functional clot formation, until definitive clinical trial data are available, it is logical to maintain blood levels of this factor to promote the ability to form strong and functional clots. Therefore, correction of plasma fibrinogen to levels >100 mg/dL using cryoprecipitate or fibrinogen concentrate before high-risk procedures could be considered for patients with cirrhosis, but data to support this strategy reducing bleeding complications are lacking. Regardless, levels of fibrinogen of <100 mg/dL are rare in patients with cirrhosis outside of acute critical illness.

Portal Vein Thrombosis

Terminology and Classification Systems

PVT is a heterogeneous condition with respect to etiology, manifestations, natural history, and therapeutic options. For this reason, terminology and classification systems also vary extensively in the literature. It is desirable to establish standardized terminology in describing PVT to allow comparison and external validation of future studies. In regard to the time course of thrombosis, the term “recent” is preferred to the term “acute” because the latter implies both clinical symptoms as well as recent-onset thrombosis, yet not all patients with a recent PVT develop symptoms. Precise dating can be impossible in patients without recent suggestive symptoms or previous abdominal imaging and is often done retrospectively. The term “chronic” should be applied to a PV obstruction persisting longer than 6 months after presumed onset. Although somewhat arbitrary, 6 months has some justification based on a pivotal prospective, longitudinal study in 100 patients with recent PVT followed up until 1 year. In this study, recanalization that did not take place within 6 months did not occur between 6 and 12 months, and cavernous transformation developed in most such patients despite anticoagulation therapy.⁽⁹²⁾ Following the onset of a PV obstruction that is not followed by regression, cavernous transformation may develop within a few weeks regardless of the cause of obstruction. Therefore, when portal cavernous transformation is present, this term should be preferred to “chronic portal vein thrombosis” for descriptive and systematic categorization. Table 5 outlines a simplified nomenclature system that can be used in future publications and research to allow comparison between studies and classification systems. In addition to time course, extent and response to treatment are major areas in which consistency in nomenclature is desirable, but quantitative assessments for these endpoints are largely lacking in current clinical practice.

TABLE 5. Recommended Standardized Nomenclature for Description of PVT in Both the Clinical and Research Setting

Descriptor	Definition
Time course
Recent	PVT presumed to be present for <6 months
Chronic	PVT present or persistent for >6 months
Percent occlusion of main PV
Completely occlusive	No persistent lumen
Partially occlusive	Clot obstructing >50% of original vessel lumen
Minimally occlusive	Clot obstructing <50% of original vessel lumen
Cavernous transformation	Gross portoportal collaterals without original PV seen
Response to treatment or interval change
Progressive	Thrombus increases in size or progresses to more complete occlusion
Stable	No appreciable change in size or occlusion
Regressive	Thrombus decreases in size or degree of occlusion

Various classification systems have been proposed to group PVT variants into broad categories. An overview of the major published PVT classification systems is presented in Table 6. The majority of these systems were developed exclusively in the LT population, and clinical outcome correlations are mostly limited to short-term LT recovery and survival.^(93-98) Some researchers have correlated PVT location and extent with clinical outcomes in patients with cirrhosis outside of pure transplant outcomes,^{(99, 100)} but long-term outcome data are also lacking in nontransplant recipients. A recently proposed system combines multiple components, including location, degree of occlusion, time course, and presence of underlying liver disease, but is complex and lacks any clinical outcome correlation.⁽¹⁰¹⁾ Further longitudinal, prospective studies using standardized categorization systems are needed. Figure 3 depicts some of the important points to note when defining the location, extent, and percent occlusion of a PVT. When available, a morphometric assessment of residual portal venous lumen is desirable. Technologies for assessing total volume of clot burden should be developed and made clinically available, especially in the setting of treatment response assessment.^{(102, 103)}

TABLE 6. Summary of Major Published PVT Classification Systems and Their Strengths and Weaknesses

System	Classification Type	Categories	Outcome Correlation	Strengths	Weaknesses
Stieber et al., 1991(93)	Anatomical	Partial Complete PV occlusion Extensive with mesenteric involvement	Posttransplant survival	First systematic report; correlates with perioperative death in early LT series	Only anatomical classes and exclusive to LT patients
Nonami et al., 1992(94)	Anatomical	Intrahepatic Primary PV branch Main PV trunk Mesenteric plus main trunk	Correlated with severity of portal hypertension pretransplant	Large series of early LT patients with anatomical PVT location described	Only patients evaluated at the time of LT
Gayowski et al., 1996(95)	Anatomical	Intrahepatic Primary PV branch Main PV trunk Mesenteric plus main trunk	Posttransplant patient and graft survival	Longer follow-up posttransplant and more complete survival information	Only patients presenting for LT
Yerdel et al., 2000(96)	Anatomical and degree of occlusion	<50% occlusion >50% occlusion Complete PV/SV occlusion Complete PV occlusion with SMV extension	Posttransplant survival	Well-defined long-term posttransplant survival; investigation into pretransplant risk factors; more clinically pertinent anatomical definitions	Only patients presenting for LT
Charco et al., 2005(97)	Anatomical and collateral presence	Confined to PV Extending to proximal SMV Diffuse involvement of mesenteric system with large collaterals Diffuse with fine collaterals	None based on this system	System based on collateral formation and surgical approach	Only patients presenting for LT; no outcome assessment of classification system
Bauer et al., 2006(99)	Anatomical and degree of occlusion	<25% occlusion of PV 26%-50% occlusion 51%-75% occlusion 76%-100% occlusion All stratified by location of clot and cavernous transformation presence	Change in grade after TIPS placement	First series dedicated to patients not necessarily undergoing LT; extreme detail on clot location and percentage occlusion	No long-term outcome data and only patients who underwent TIPS and endovascular intervention
Ma et al., 2014(100)	Anatomical; degree of occlusion; and presence of collaterals	Duration of clot Presence of symptoms Degree of portal hypertension	Correlation with symptoms and portal hypertension complications	Nontransplant cohort with careful imaging classification and symptom measurement	No long-term outcome assessment
Sarin et al., 2016(101)	Anatomical; degree of occlusion; duration; and presence of underlying liver disease	Location of clot Degree of occlusion Time course Extent of occlusion Presence of underlying liver disease	None	PVT classified in multiple characteristics and scenarios	System is complex and nomenclature can be cumbersome; no outcome correlation.
Bhangui et al., 2019(98)	Anatomical and presence of collaterals	Complex Noncomplex	Requirement of portal inflow reconstruction at the time of LT	Simplifies anatomical considerations for PV management during LT	Only applies to LT techniques; no natural history outcomes

• Abbreviations: SMV, superior mesenteric vein; SV, splenic vein.

PVT Prevalence and Risk Factors: Patients With Cirrhosis

Determining the incidence and prevalence of PVT is difficult because of the heterogeneity of risk factors in the population and lack of a universal classification system. As a result, the reported prevalence and incidence of PVT vary widely. The most studied cirrhosis population with regard to PVT is LT candidates. One report, based on 849 patients with cirrhosis who underwent LT with well-defined explant thrombosis classification, described a prevalence of 9% with either partial or complete obstruction of the main PV.⁽⁹⁴⁾ Another study from Japan showed autopsy PVT prevalence of 0.05% in patients without cirrhosis and 6.59% in patients with cirrhosis.⁽¹⁰⁴⁾ Recent reports, including national databases of hospital discharge and transplant registries, have listed prevalence rates in patients with cirrhosis between 1.3% and 9.8%.^{(96, 105-110)} Fewer prospective studies documenting incidence rates in the cirrhosis population have been published, but a prospective incidence ranging between 3.2% and 4.1% at 1 year after diagnosis with cirrhosis is often cited.^(111-114)

Severity of portal hypertension and liver disease are major predictive factors for PVT in patients with cirrhosis.^(113-118) Decreased velocity of PV flow at baseline has been recognized as a risk factor for PVT in patients with cirrhosis in several studies, albeit with varied predictive value.^(117-120) Some studies have identified specific thresholds of portal flow associated with increased risk of PVT, ranging from 10 to 15 cm/s, as determined by Doppler ultrasonography.^{(117, 119, 121)} A meta-analysis demonstrated an increased incidence in PVT in patients with cirrhosis taking nonselective beta-blockers (OR, 4.62; 95% CI, 2.50-8.53; P < 0.001; I² = 80%; P < 0.001).⁽¹²²⁾ However, analysis of confounding variables in this study attributable to the presence of esophageal varices, extent of portal hypertension, and nonselective beta-blocker use as secondary variceal bleeding prophylaxis has raised questions about the independent effect of the nonselective beta-blockers in development of PVT.^{(114, 115, 123)} The potential role of an altered endothelium has been underinvestigated⁽¹²⁴⁾ and data on the role of inflammation/bacterial translocation are still limited.^{(107, 125)} Obesity, metabolic syndrome, and NASH cirrhosis are also recognized as independent risk factors for PVT.^(126-128) There are contrasting data on the prevalence of factor V Leiden and G20210A prothrombin gene mutation in this population, and testing for these disorders in the cirrhosis population is rarely useful and does not change management.^{(113, 129)} A detailed thrombophilia workup in the patient with cirrhosis is not generally recommended unless specific concerns are raised during the history, routine laboratory and imaging workup (Table 7). HCC invading the PV and/or HV is common in patients with cirrhosis; therefore, it is mandatory to rule out malignant venous obstruction with appropriate dynamic contrast-enhanced studies because the treatments and prognosis are vastly different compared to those with spontaneous bland thrombosis. Detailed reviews of imaging the portal system and the strengths and weaknesses of the various modalities are beyond the scope of this guidance document and are published elsewhere.^(130-132)

PVT Prevalence and Risk Factors: Patients Without Cirrhosis

PVT in patients without cirrhosis is a rare disease. An autopsy study from Japan showed a PVT prevalence of 0.05%,⁽¹⁰⁴⁾ but this study likely overestimated PVT prevalence in the general population because of postmortem thrombosis. A combined evaluation by an experienced hematologist and hepatologist of every patient without cirrhosis who develops PVT is recommended. In the patient without cirrhosis, PVT and HVT share many common risk factors, and the thrombophilia workup is similar in both groups (Table 7). In contrast to the patient with cirrhosis, a detailed thrombophilia workup is warranted in most patients without cirrhosis who develop a PVT (Table 8).

Obstruction of the extrahepatic portal venous system in patients without cirrhosis is mostly related to myeloproliferative neoplasia, surgery, or inflammatory conditions affecting the digestive system organs or the spleen (Table 7).⁽¹⁾ Bariatric surgery, splenectomy, pancreatitis, inflammatory bowel disease, appendicitis, and diverticulitis are important contributors to extrahepatic portal venous obstruction as well, and ~30% of these patients also have a systemic thrombophilic disorder.⁽¹³³⁾ Among the inherited thrombophilias, the G20210A prothrombin gene mutation is most prevalent in this setting. Overall, risk factors are absent in up to 35% of patients. A combination of several low-level risk factors appears to be more common in patients without cirrhosis with PVT than expected in the general population.^{(134, 135)} Table 8 outlines the strengths and weaknesses of thrombophilia testing in this population.

Influence of PVT on Mortality in Patients With Cirrhosis

The influence of PVT on cirrhosis disease progression and mortality is not fully elucidated. There is a clear association of PVT with more severe portal hypertension; however, the temporal relationship is difficult to discern.^{(108, 136)} Whether PVT is merely a manifestation of progressive disease or an actual cause of disease progression is still unknown, and the literature is contradictory. In a prospective, longitudinal study of 1,243 patients with cirrhosis followed over a mean of 47 months,⁽¹¹³⁾ PVT was deemed to follow hepatic decompensation temporally as opposed to being an etiological factor and was not independently associated with disease progression. Other prospective studies have arrived at similar conclusions.^{(114, 137)} In contrast, a small, prospective, randomized controlled trial of enoxaparin compared to no anticoagulation therapy in prevention of PVT in high-risk patients with cirrhosis⁽¹²⁵⁾ showed not only efficacy in prevention of PVT in the enoxaparin group, but also a survival and decompensation benefit. Although this study has not been replicated, the implication of prolonging decompensation-free survival with prevention of PVT implies etiological potential. Anticoagulation therapy in this study was also associated with findings (decrease in bacterial translocation) believed to have hepatic benefits beyond prevention of PVT.

In relation to LT, there is a distinct technical advantage to maintaining a patent main PV in the recipient up to the time of surgery that is associated with improved posttransplant survival. Nonanatomical PV reconstructions during LT add technical difficulty and increase graft ischemic times. These technical complications add risk, and there are inferior survival and more complications in the LT recipient who presents for transplant with PVT.^{(138, 139)} In the United States, LT regulatory agencies have acknowledged this increased risk by incorporating the presence of pretransplant PVT as a strong independent predictive variable in models of posttransplant survival.⁽¹⁴⁰⁾ Large transplant database analyses and other case series confirm these findings.^{(94, 96, 108)} However, it should be noted that these databases lack granularity to determine the size and extent of PVT, and therefore it is unknown whether there is a threshold at which outcomes are worse. There are no randomized controlled trials of PVT therapy before transplant to determine whether regression of PVT before transplant improves posttransplant survival. Some researchers have suggested that PVT is again simply a measure of more advanced portal hypertension before transplant, and not an independent risk factor, in all but the extreme cases of extensive PV reconstruction.⁽⁹⁶⁾ Limited data suggest higher posttransplant vascular complication rates, including hepatic arterial thrombosis, in NASH patients with pretransplant PVT,⁽¹⁴¹⁾ but this area needs further investigation.

Goals of Therapy and Rationale for Treatment: Patients With Cirrhosis

In patients with underlying cirrhosis and portal hypertension, congestive intestinal ischemia is much less frequent compared to patients without cirrhosis because of the presence of previously formed portosystemic collaterals. In the patient without ischemic symptoms, the aim of treatment is not to prevent portal hypertension development (that already exists), but to prevent worsening and avoid progression of thrombosis that may hinder a future LT. Management of complications of portal hypertension should be as recommended in other patients with cirrhosis.⁽¹⁴²⁾ In a small cohort of patients with PVT without cirrhosis, the incidence (7.4%) and severity of bleeding in patients in whom EVL was performed without stopping anticoagulation was not significantly different to that in patients who did not receive anticoagulants.⁽¹⁴³⁾ Similarly, one study in patients with cirrhosis suggests that LMWH continued through prophylactic EVL does not increase the risk of bleeding or death.⁽⁵²⁾ These results suggest that EVL can be performed without stopping anticoagulation, but more safety studies are needed to make a formal recommendation on this issue. Based on the available data, it seems unnecessary to delay anticoagulation until variceal eradication or adequate beta-blockade is achieved, and we recommend initiating these treatments as soon as possible. Figure 4 depicts the suggested management of occlusive or partially occlusive PVT in a patient with cirrhosis.

Goals of Therapy and Rationale for Treatment: Patients Without Cirrhosis

In patients with an underlying healthy liver, the aims of therapy for recent PVT are to: (1) prevent thrombus extension to mesenteric veins; (2) prevent complications of intestinal ischemia; and, ideally, (3) achieve recanalization to prevent development of portal hypertension. When present, portal hypertensive complications of PVT in patients without cirrhosis are managed in a similar fashion as those with cirrhosis. As mentioned above, the diagnosis of a thrombophilic condition or hematological malignancy is much more likely in the patient without cirrhosis, but this infrequently results in short-term management differences. If septic pylephlebitis is diagnosed, prolonged treatment with antibiotics adapted to the isolated bacteria or anaerobic digestive flora is necessary. Limited retrospective data suggest that patients with septic pylephlebitis also benefit from concurrent anticoagulant therapy, with higher complete resolution rates and lower long-term portal hypertension complications.⁽¹⁴⁴⁾

PVT Treatment Options

Medical Therapies for PVT

In patients with PVT, the choice of anticoagulation historically has been limited to UFH, LMWH, and VKAs, such as warfarin. With the advent of DOACs, clinicians face greater complexity of treatment decisions with more therapeutic options (Table 9).⁽¹⁵⁶⁾ Data remain considerably limited in this population; therefore, recommendations rely on smaller cohort studies and extrapolation from anticoagulation experience in other populations.^{(1, 2, 31, 157-159)}

Medical Therapies: Patients With Cirrhosis

Direct Oral Anticoagulants

Patients with cirrhosis were excluded from clinical trials comparing DOACs versus VKAs and LMWH for prophylaxis and treatment of VTE or atrial fibrillation. Very little is known regarding the pharmacodynamics of DOACs in cirrhosis. In vitro studies conducted using plasma from patients with decompensated cirrhosis show differences in anticoagulant potency when measured by thrombin generation assay, and these findings have been confirmed in vivo using edoxaban.^(174-177) These studies highlight concerns with these agents and emphasize the need for larger in vivo studies. Clinical experience with DOACs in patients with cirrhosis remains sparse and is generally limited to highly select cohorts with well-compensated cirrhosis.^{(156, 178-181)} Overall, DOACs appear to have a similar safety profile in patients with compensated cirrhosis compared to patients without cirrhosis, and their use is expanding in patients with all indications for anticoagulation, including atrial fibrillation, VTE, and PVT.^{(182, 183)} The availability of direct reversal agents may allay some fears concerning bleeding. For example, successful use of idarucizumab for reversal of dabigatran in a patient undergoing LT has been reported.⁽¹⁸⁴⁾ Two studies have directly compared DOACs to VKAs in different cohorts of patients with cirrhosis and PVT. One study involved 80 patients with recent PVT after elective splenectomy,⁽¹⁸⁵⁾ and the other involved 50 patients converted to oral therapy in a nonrandomized fashion after 2 weeks of injectable LMWH therapy.⁽¹⁸⁶⁾ Acknowledging the low level of evidence produced by both of these studies, bleeding rates were not significantly higher in patients treated with DOACs.

Future studies are needed in this population to establish the safety and efficacy of DOACs in patients with cirrhosis and PVT. Use of standardized classification systems (Table 6) for PVT and standardized approaches to control for spontaneous recanalization will improve our understanding of anticoagulation in this population. Bleeding definitions are highly variable in these studies, and therefore use of validated and accepted bleeding definitions is also recommended.

Medical Therapies: Patients Without Cirrhosis

Direct Oral Anticoagulants

The use of DOACs as therapy for PVT is now becoming commonplace despite the fact that large clinical trials of DOACs have excluded patients with PVT in the past.⁽¹⁹³⁾ Several retrospective, small studies are now published demonstrating the successful use of DOACs for this indication in patients without cirrhosis.^{(94, 194, 195)} A large, retrospective, single-center cohort examined 330 patients with PVT without cirrhosis treated with VKAs (n = 108), LMWH (n = 70), DOACs (n = 93), and no anticoagulation (n = 57).⁽¹⁹⁶⁾ In this cohort, DOAC therapy had superior efficacy (rates of thrombus resolution) and less major bleeding when compared to warfarin. DOAC use offers several advantages over warfarin, including the lack of a need for monitoring and predictable anticoagulant effect (Table 9). Although prospective studies are needed, the use of DOACs in this population will likely continue to expand, and careful analysis of outcomes requires multicenter collaboration and larger database registries. Specific features common to these patients, such as hematological malignancy, are important to consider when using DOACs, and consultation with hematology and hepatology experts in the field is advised.⁽¹⁹⁷⁾

Hepatic Vein Thrombosis: Budd-Chiari Syndrome

HVT, or Budd-Chiari syndrome (BCS), is characterized by obstruction of the hepatic venous outflow anywhere from the small HVs to the junction of the inferior vena cava (IVC) and the right atrium in the absence of intracardiac or pericardial obstruction.⁽¹³⁴⁾ This practice guidance refers to primary HVT/BCS, in which an endoluminal venous lesion (usually a thrombus or a vascular web) causes the obstruction.

Prevalence and Etiology

Primary HVT/BCS is a rare disease mainly affecting young people (median age of 46 years) with a reported incidence below 1 case per million person-years (range, 0.17-0.88) and a prevalence of 1.40-7.69 per million.⁽¹⁹⁸⁾ An underlying prothrombotic disorder can be found in ~75% of patients with BCS, and more than one prothrombotic condition can be identified in at least 35% of patients (Table 7). No risk factors are found in 15%-30% of patients.^{(134, 135)} In Chinese HVT/BCS patients, the underlying etiology differs whether there is an isolated hepatic venous type, an isolated IVC type, or a combined type. In these two latter cases, there is a low prevalence of prothrombotic disorders, and most cases are associated with a low socioeconomic status and assumed to be triggered by abdominal infections.^{(199, 200)}

Clinical Presentation and Diagnostic Considerations

Presentation varies from asymptomatic cases incidentally discovered on imaging obtained for other indications to acute liver failure, depending on the extent and rapidity of vein obstruction and the development of decompressive venous collaterals. Abdominal pain (61%), hepatomegaly (67%), and ascites (83%) are the most frequent symptoms at diagnosis,⁽¹³⁴⁾ and esophageal varices can be detected in >50% of patients.⁽¹³⁴⁾ Acute liver failure may arise if thrombosis is extensive and rapidly formed. Conversely, ~15%-20% of patients^{(134, 201)} are asymptomatic and incidentally diagnosed. Usually, these cases represent patients with partial thrombosis accompanied by the formation of decompressive venous collaterals with frequent atrophy of the affected liver and hypertrophy of those segments, which are well drained.^{(134, 201, 202)} Severe acute manifestations can appear in patients with imaging signs of chronic liver disease, which likely reflects recurrence of thrombosis in a previously unrecognized patient with HV occlusion. Typical laboratory findings are bilirubin and aminotransferase elevations with increase in PT in severe cases. Ascites analysis typically shows high protein content.⁽¹³⁴⁾

In cases with obstruction of the IVC, either by thrombosis or compression by a hypertrophied caudate lobe, abdominal varices, lower limb edema, or ulcers are a frequent manifestation.⁽²⁰⁰⁾ Around 40% of patients with HVT/BCS develop nodular liver lesions during follow-up, likely as a result of the vascular changes induced by the HVT/BCS.⁽²⁰³⁾ Although these are usually benign regenerative nodules representing nodular regenerative hyperplasia, hepatocellular adenomas⁽²⁰⁴⁾ and HCC⁽²⁰⁵⁾ may also develop. Benign lesions are usually multiple (>10 lesions), small in size (<4 cm), hypervascular, and disseminated throughout the liver.⁽²⁰³⁾ In one study from France, the cumulative incidence of HCC was 6% at 7 years of follow-up.⁽²⁰⁵⁾ Patients with long-term IVC obstruction seem to have a higher risk for developing HCC.^{(205, 206)} HCC diagnosis in HVT/BCS is a challenge and must always rely on histology for diagnosis. A level of alpha-fetoprotein >15 ng/mL has been suggested as a predictive biomarker for HCC in HVT/BCS patients^{(205, 207)}; however, this threshold cannot be recommended in clinical practice until validated in larger studies. Although specific data are lacking, surveillance for HCC in patients with chronic HVT/BCS with ultrasound with or without alpha-fetoprotein levels every 6 months is recommended. Suspicious lesions should escalate to contrast-enhanced cross-sectional imaging and/or biopsy if the nature of the lesion is not certain based on imaging characteristics.⁽²⁰³⁾ It should be noted that the widely adopted Liver Reporting and Data System (LI-RADS) criteria for CT/MR⁽²⁰⁸⁾ interpretation of liver lesions specifically exclude patients with HVT/BCS and should not be applied to this population because the altered vascular perfusion of the liver leads to arterially perfused lesions, which often are not HCC.

Clinical manifestations of HVT/BCS are heterogeneous, justifying suspicion in any patient with acute or chronic liver disease of unknown origin and/or with an underlying prothrombotic condition. A thrombophilia workup is generally required in these patients, and consultation with a hematologist is warranted (Table 8). The diagnosis requires demonstration of hepatic venous flow obstruction, and noninvasive imaging techniques (Doppler ultrasonography, CT, or MR) are the mainstay of diagnosis. Doppler ultrasound, performed by an experienced operator, has a sensitivity >75%.⁽²⁰⁹⁾ Typical ultrasound features of venous obstruction are identification of thrombus, nonvisualization of the HV, collateral veins and transformation of the HV into a cord lacking flow signals, caudate lobe hypertrophy, and a caudate vein >3 mm in diameter. Usually, the role of contrast-enhanced MR or CT is for diagnosis confirmation and is the diagnostic study of choice in the absence of an experienced ultrasound operator. CT and MR can additionally depict rapid clearance of contrast from the caudate lobe and patchy hepatic enhancement attributable to uneven portal perfusion. Hepatic venography may be helpful in cases of uncertain diagnosis, and the most typical sign is the presence of a “spider web” pattern of collateral circulation. Liver biopsy is not necessary for diagnosis unless there is concern for small intrahepatic vein obstruction when histology is the only way to achieve definitive diagnosis.

Treatment Options

The goals of HVT/BCS treatment include management of portal hypertension, treatment of the underlying thrombotic or malignant disease, and restoration of hepatic venous outflow. The most recommended and supported treatment approach is a progressive therapeutic strategy,^{(201, 210)} stepping from less invasive (medical) to more invasive (LT) treatment according to the clinical response. A major challenge is recognizing the appropriate time to step up to the next treatment choice in a given patient. Because of this difficulty, these patients should routinely be managed in tertiary referral centers.

Medical Therapies

Medical treatment must be initiated early in the course and include treatment of the complications of portal hypertension. In addition to therapeutic anticoagulation, early recognition and treatment of an underlying prothrombotic disorder is also critical. All patients with HVT/BCS, even in the absence of a recognized prothrombotic disorder, should receive anticoagulation with the aim of achieving recanalization and preventing thrombosis progression. Immediate LMWH, followed by a VKA, is the most frequent anticoagulation approach. Long-term UFH should be avoided because of the risk of heparin-induced thrombocytopenia.⁽²¹⁰⁾ Although promising and clinically attractive, data with DOACs are very limited.⁽¹⁷⁸⁾ One quarter of HVT/BCS patients will be alive at 5 years with medical therapy alone.⁽²⁰¹⁾

Interventional Vascular Procedures

Sinusoidal Obstruction Syndrome

Sinusoidal obstruction syndrome (SOS), formerly known as hepatic veno-occlusive disease, is a distinct and potentially lethal form of liver injury characterized by toxic/inflammatory damage to sinusoidal endothelial cells. In this syndrome, these cells undergo necrosis and extrusion into the sinusoids, resulting in partial or complete occlusion of small hepatic venules, and thus SOS is a postsinusoidal form of portal hypertension.⁽²²⁸⁾ It occurs most commonly as a complication of myeloablative chemotherapy regimens (high-dose with or without total-body irradiation) used before hematopoietic stem cell transplantation (HSCT), particularly following allogeneic (as opposed to autologous) HSCT.⁽¹⁾ Chemotherapeutic agents known to cause SOS are alkylating agents (cyclophosphamide, busulfan, and melphalan) and the platinum complexes (oxaliplatin, carboplatin, and cisplatin). Thiopurines (azathioprine, mercaptopurine, and thioguanine) have also been implicated. Even though targeted therapy is less toxic, gemtuzumab and inotuzumab have been associated with SOS. Unrelated to HSCT and cancer chemotherapy, pyrrolizidine alkaloids found in plants or bushes from which tea is made or that contaminate seeds are well-documented causes of SOS.^{(229, 230)}

Development can be acute, presenting 1-3 weeks after exposure, with sudden onset of right upper quadrant pain, weight gain attributable to edema, hepatomegaly, and ascites.^{(231, 232)} Aminotransferase elevations can be marked with no or mild elevations of alkaline phosphatase, although some cases may present with jaundice. Standardized diagnostic criteria for SOS (Seattle criteria⁽²³³⁾ and Baltimore criteria⁽²³¹⁾) have been established, but presentations can be atypical. Subacute SOS is a poorly defined syndrome and overlaps with portosinusoidal vascular disorder (PSVD) because it occurs months or even years after exposure and presents with fatigue, ascites, hepatic encephalopathy, or varices, and symptoms may be attributable to nodular regenerative hyperplasia.⁽²²⁸⁾ This most commonly occurs with chronic low-dose ingestion of pyrrolizidine alkaloids or oral chemotherapeutic agents.

Definitive diagnosis is histological, but performing a liver biopsy is challenging in these patients because of cytopenias and other comorbidities at the time the diagnosis is in question, especially in the HSCT population. Early-stage SOS is characterized by sinusoidal dilatation with red blood cells escaping through the space of Disse as well as perivenular necrosis. Later-stage SOS shows fibrous obliteration of central venules. In cases in which the diagnosis is uncertain, a transjugular liver biopsy with hepatic venous pressure gradient (HVPG) measurement should be performed. In the proper clinical scenario, clinically significant portal hypertension (HVPG ≥ 10 mm Hg) without an alternative cause is highly specific to rule in SOS.⁽²³⁴⁾ The overall mortality rate from SOS averages ~5%, but depends on the severity of disease; in mild presentations (8% of cases), mortality is 1%; in cases that resolve with therapy (64%), mortality is 18%; and in severe cases (28%), mortality can be as high as 67%.⁽²³⁵⁾

Given its potential high mortality rate, early identification of high-risk patients (e.g., allogeneic HSCT, preexisting liver disease, or second HSCT) is critical. Data are sparse regarding preexisting liver disease as a risk factor for SOS because HSCT is uncommonly performed in patients with significant preexisting liver disease. A small study showed that any previous liver disease independently increased the risk for SOS (OR, 3.35; 95% CI, 1.71-6.58; P < 0.001).⁽²³⁶⁾ Preventive strategies are unproven, not specifically data driven, and usually based on local practices or expert opinion.⁽²³⁷⁾ However, adjusting the conditioning regimen may reduce SOS risk, and the frequency of SOS has decreased over recent years partially because of the use of less aggressive myeloablative regimens (e.g., no cyclophosphamide, lower radiation doses).⁽²³⁶⁾ Myeloablative regimens are strongly discouraged in patients with cirrhosis,⁽²³⁷⁾ and before HSCT, workup should focus on excluding cirrhosis. Liver biopsy may be needed for this purpose before HSCT, if feasible. Data on liver stiffness measurement and general hepatic outcomes following HSCT are sparse,^{(238, 239)} and correlation of liver stiffness with specific risk of SOS is limited to a small case series.⁽²⁴⁰⁾ Care should be taken in the interpretation of liver stiffness measurements because of potential confounding from liver involvement of hematological malignancy or secondary iron overload, both of which can elevate liver stiffness.

Data from randomized controlled trials regarding ursodeoxycholic acid for prophylaxis of SOS have been mixed, with some studies showing benefits and others none^(241-244); however, a meta-analysis of pooled study results suggested an overall benefit (pooled relative risk, 0.34; 95% CI, 0.17-0.66).⁽²⁴⁵⁾ Thus, ursodeoxycholic acid is recommended as prophylactic therapy for SOS in patients undergoing allogeneic HSCT.^{(244, 246, 247)} The suggested dose is 12 mg/kg divided in two doses starting the day before conditioning and continuing for 3 months after HSCT.

Defibrotide is the only FDA-approved treatment for SOS and is generally reserved for moderate or severe cases.^(246-248) In the past, some authorities have recommended it for prophylaxis in high-risk cases, but the benefit of this approach is unproven, and the acquisition costs for the drug are very high. Its mechanism of action is uncertain, but it has both anti-inflammatory and anticoagulant activities, although bleeding complications are uncommon. Defibrotide is recommended at a dose of 25 mg/kg/d for at least 21 days and until resolution of the SOS. Side effects are not different than best supportive care in large clinical trials.⁽²⁴⁸⁾ Management of ascites and fluid retention in SOS is the same as for patients with cirrhosis. TIPS has been performed for treatment failures, published in case reports or very small case series, with some success in portal decompression, but no impact on overall survival.⁽²⁴⁹⁾

Liver Vascular Malformations and Hereditary Hemorrhagic Telangiectasia

Hereditary hemorrhagic telangiectasia (HHT), or Osler-Weber-Rendu disease, is a rare (1 in 5,000-8,000 persons) autosomal-dominant genetic disease characterized by widespread arteriovenous malformations (macroscopic and microscopic) that can involve the skin, mucous membranes (oral, nasal, and conjunctival), lungs, brain, gastrointestinal tract, and/or liver. Because HHT is inherited in an autosomal-dominant pattern, a family history of recurrent bleeding in any of the target organs may aid in the diagnosis. Liver vascular malformations (LVMs) are present in 40%-70% of patients with definite HHT, more commonly in those with a mutation in the activin receptor-like kinase type 1 gene.⁽²⁵⁰⁾ Mortality is higher in patients with LVMs versus those without them. Symptoms from LVMs occur at a median of 90 months after imaging diagnosis and typically appear around age 30, occurring predominantly in females.⁽²⁵¹⁾ Symptoms result from three types of shunting within the liver: hepatic artery (HA) to HV, HA to PV, and PV to HV.⁽²⁵²⁾ Shunting leads to different clinical presentations, the most common being high-output heart failure, followed by biliary ischemia, portal hypertension (ascites more common, variceal hemorrhage rarer), portosystemic encephalopathy, and intestinal ischemia⁽²⁵³⁾ (Fig. 5). Symptoms may occur concurrently or transition from one presentation to another. LVMs can present with a nodular liver (nodular regenerative hyperplasia), varices, ascites, and splenomegaly, leading to an erroneous diagnosis of cirrhosis. Liver synthetic function and platelet count are usually normal unless end-stage secondary biliary cirrhosis or severe hepatic congestion is present.⁽²⁵³⁾ Prevalence of focal nodular hyperplasia is greater with LVMs (2.9%) than in the general population, and the liver may be nodular in the presence of LVMs. This may lead to the erroneous diagnosis of cirrhosis with HCC, neither of which is directly associated with HHT.⁽²⁵⁴⁾ A liver biopsy is not recommended in patients with LVMs because findings are typically nonspecific, and bleeding risk may be substantial depending on the type of shunting.⁽²⁵⁵⁾ Presence of LVMs should be confirmed by cross-sectional imaging showing heterogeneous enhancement or hypervascularization of the liver and common HA enlargement.

No treatment is recommended for asymptomatic LVMs.⁽²⁵⁵⁾ Patients with symptomatic vascular malformations should be managed at specialized centers using a multidisciplinary approach. Standard therapy includes management of congestive heart failure, including correction of arrhythmias and anemia, which is frequent in these patients because of recurrent epistaxis or bleeding from gastrointestinal vascular malformations. Abdominal pain from biliary ischemia is managed symptomatically and/or with percutaneous drainage and antibiotics in the case of biloma infection; endoscopic retrograde cholangiography should be avoided unless a clear dominant stricture amenable to therapy is present. Management of complications of portal hypertension should be the same as for cirrhosis. Symptoms are controlled with these therapies in >60% of patients.⁽²⁵¹⁾

In nonresponders to standard therapy, several alternative treatments can be contemplated. The least invasive is the intravenous infusion of bevacizumab, a monoclonal antibody that inhibits the activity of vascular endothelial growth factor in endothelial cells, at a dose of 5 mg/kg intravenously every 2 weeks for six doses, with maintenance cycles in those with persistent or reappearing symptoms.^{(256, 257)} Bevacizumab improves epistaxis, cardiac index in heart failure patients, and ischemic cholangiopathy.^{(258, 259)} However, long-term efficacy and tolerance require further evaluation.⁽²⁶⁰⁾ Hepatic arterial embolization or surgical ligation are only transiently effective and can be potentially harmful because they can lead to biliary and/or hepatic necrosis and are generally not recommended.⁽²⁵⁵⁾ LT has been associated with 82.5% 10-year patient and graft survival in the largest European cohort⁽²⁶¹⁾ and 86% 4-year survival in a USA cohort.⁽²⁶²⁾ However, a high rate of perioperative complications has been reported.⁽²⁶¹⁾ Additionally, LVMs may recur as soon as 6 years post-LT.^{(260, 262, 263)}

Idiopathic Noncirrhotic Portal Hypertension

Idiopathic noncirrhotic portal hypertension (INCPH), or idiopathic portal hypertension, is characterized by portal hypertension in the absence of cirrhosis, PV obstruction, or HVT/BCS.^{(264, 265)} INCPH has been further characterized by the absence of sarcoidosis, congenital hepatic fibrosis, and schistosomiasis, all three conditions being well-documented causes of noncirrhotic portal hypertension. The precirrhotic stage of conditions that progress to cause cirrhosis (e.g., hepatitis B or C, fatty liver disease, or metabolic liver disease) have usually also been excluded from INCPH diagnostic criteria. This disorder is a heterogeneous collection of diseases of the liver, and the term INCPH is best considered as a broad category of diseases with similar manifestations.

Histopathology

In addition to the absence of cirrhosis, various vascular histopathological lesions have been described in patients with INCPH, including intrahepatic PV stenosis (also called obliterative portal venopathy), nodular regenerative hyperplasia, or incomplete septal cirrhosis, as well as less specific but suggestive histological alterations, such as herniated PVs, hypervascularized portal tracts, abnormal periportal vessels, sinusoidal dilatation, or regenerative hepatocytes.⁽²⁶⁶⁾ An overarching entity called portosinusoidal vascular disease (PSVD) has been proposed to account not only for INCPH, but also for the increasingly reported similar histopathological anomalies in patients without portal hypertension.^(267-269)

Diagnosis

INCPH should be considered in any patient with unexplained liver disease and portal hypertension with mild or no primary liver dysfunction. Furthermore, mild or no elevation of liver stiffness^{(270, 271)} or HVPG often contrasts with the clinical features of marked portal hypertension.

INCPH should be suspected when the conditions listed in Table 10 are present⁽²⁶⁹⁾ along with clinical evidence of portal hypertension. Liver biopsy is required to exclude cirrhosis and is beneficial in the recognition of specific or suggestive lesions. Liver biopsy can be obtained through the transvenous route in patients with severe thrombocytopenia prohibiting the percutaneous route. The transvenous route offers the added benefit of allowing HVPG measurement, but percutaneous biopsy can effectively rule out cirrhosis and give diagnostic information as well. Reticulin stains on liver biopsy specimens are critical for assessing disorders of the liver architecture responsible for portal hypertension without cirrhosis.

Associated Diseases and Etiology

Several uncommon systemic conditions (Table 10) appear to be unexpectedly more frequent in patients with INCPH.^{(269, 272-276)} These underlying conditions appear to bear prognostic significance and may require a specific management other than treatment related specifically to the portal hypertension. Therefore, a careful consideration of the conditions is recommended in light of medical history, clinical characteristics, imaging, and laboratory features.

Natural History and Outcome

Gastrointestinal bleeding related to portal hypertension is the most frequent and severe complication, much less frequently followed by ascites and portosystemic encephalopathy.^{(269, 272-276)} The latter two complications are most often triggered by bleeding and are transient in nature. Based on cross-sectional imaging, focal nodular hyperplasia-like lesions have been reported in 14% of patients.⁽²⁷⁷⁾ However, hepatocellular adenoma and HCC are extremely rare. LT is needed in some patients, mostly because of encephalopathy, refractory ascites, or shunting disorders such as hepatopulmonary syndrome.

The cumulative incidence of PVT in INCPH appears to be higher (32%) than in patients with cirrhosis (18%).⁽²⁷⁶⁾ Risk factors have not been reproducibly recognized. Evidence implicating long-standing, well-controlled HIV infection or prothrombotic conditions needs replication. The independent effect of PVT on natural history is unclear in this disorder. The impact of INCPH on overall survival has been variably reported across surveys; however, a consistent finding has been low risk of liver-related death. Indeed, the associated conditions and general medical disorders appear to be the major determinants of overall survival.

Therapy

There is no established independent therapy for INCPH or PSVD. There are no specific data on the treatment of established PVT in this setting, and it is unknown whether data for patients with cirrhosis or without underlying liver diseases are applicable. Similarly, there are no data on prophylaxis for gastrointestinal bleeding related to portal hypertension in INCPH. It is reasonable to expect a favorable risk-benefit balance in applying the recommendations for cirrhosis or chronic PVT to INCPH. A single retrospective series of these patients undergoing TIPS indicates that the main determinants of outcome are related to concurrent severe comorbidities, particularly renal dysfunction.⁽²⁷⁸⁾

Hepatic and Splenic Artery Aneurysms

Hepatic (HAAs) and splenic artery (SA) aneurysms (SAAs) are uncommon.

True arterial aneurysms involve all three layers of the arterial wall, whereas pseudoaneurysms are characterized by a localized arterial wall disruption in which blood is confined by perivascular tissue, usually attributable to trauma or procedural complication. Unless otherwise specified, all commentary and guidance in this document is related to true aneurysms, not pseudoaneurysms.

Hepatic Artery Aneurysms

Most HAAs are true arterial aneurysms.⁽²⁷⁹⁾ The primary causes are atherosclerosis, mediointimal degeneration, trauma, and, less commonly, infection. Vasculitides (e.g., polyarteritis nodosa, systemic lupus erythematosus, Takayasu arteritis, and Kawasaki disease) and connective tissue disorders (e.g., Marfan syndrome, Ehlers-Danlos syndrome, and HHT) are rarely implicated.^{(279, 280)} Pseudoaneurysms are commonly intrahepatic and usually result from trauma related to a liver biopsy, transhepatic biliary drainage, cholecystectomy, hepatectomy, or LT.⁽²⁸⁰⁾ Most true HAAs are asymptomatic and found incidentally on cross-sectional imaging obtained for other reasons. Patients may present with rupture into the biliary tract, with hemobilia, epigastric pain, and jaundice; rupture into the PV, with portal hypertension and variceal bleeding; or rupture into the peritoneal cavity, with abdominal pain and shock. The reported incidence of rupture in patients with true HAAs is up to 25%.^(279-285) Such estimates are based on retrospective case series from clinical units and may be biased by a predominant inclusion of symptomatic patients. Large surveys in a general population are lacking. The mortality rate from rupture of an HAA is >30%.^{(279, 280, 284, 285)} Nonatherosclerotic aneurysms and multiple HAAs appear to carry an increased risk of rupture.⁽²⁸¹⁾ However, a published case series could not associate the risk of rupture of any individual HAA with its maximum diameter.⁽²⁸³⁾

Doppler ultrasound studies and CT readily demonstrate HAAs.^{(280, 286, 287)} Multidetector CT arteriography is the optimal tool for defining these lesions, assessing the collateral circulation, and planning treatment. In a retrospective review of 22 patients with a mean HAA diameter of 2.3 cm (range, 1.5-5.0) who were followed up and untreated for a mean of 68.4 months (range, 1-372), no complications were identified. Overall, open surgery and endovascular approaches have been proposed for elective management, but their comparison is not possible because of selection bias by indications. In addition, patients treated using endovascular approaches appear to have less morbidity and mortality.^{(280, 282-285)} The actual benefit-to-risk ratio of treating aneurysms deemed to be at higher risk of rupture has not been evaluated.^{(280, 281, 288)} There are no data on which to base recommendations for surveillance of HAAs in the setting of expectant management.

Splenic Artery Aneurysms

A majority of SAAs are true aneurysms of a mean size of ~3 cm upon diagnosis.^{(279, 289)} Female subjects account for ~70% of patients, and a positive relationship with multiparity has been suggested.^(290-292) A majority of patients are in their sixth or seventh decade of life at the time of diagnosis.^{(290, 293)} Associations with pregnancy, portal hypertension, or pancreatitis have been frequently reported.^{(279, 290-292, 294, 295)} However, there is undoubtedly a detection bias in the published literature attributable to repeated abdominal imaging performed in such conditions.^{(290, 294)} Commonly reported causes include arteriosclerosis and fibromuscular dysplasia.^{(280, 290)} Recently, attention has been drawn to HHT⁽²⁹⁶⁾ and the combination of alpha-1 antitrypsin deficiency and chronic liver disease requiring LT as risk factors.⁽²⁹⁷⁾ Pseudoaneurysms of the SA are mostly related to pancreatitis, infection, and trauma.⁽²⁸⁹⁾

Most SAAs are asymptomatic and found incidentally.^{(290, 292)} Patients may present with penetration into the stomach with gastrointestinal bleeding or rupture into the peritoneal cavity with abdominal pain and shock. Incidence of rupture in patients with SAAs is ~3%, and mortality related to rupture is 20%-36%.^{(288, 290, 298, 299)} The risk of rupture of true aneurysms appears to be negligible, below a size of 2 cm, which is the size of the majority of SAAs upon detection.^{(288, 290, 298)} Growth of SAAs with a diameter of ≤2 cm is rare and, if present, typically develops very slowly,^{(290, 300)} suggesting that unusually large aneurysms may be attributable to a different yet unexplained pathophysiology.^{(301, 302)} The actual risk of rupture for aneurysms larger in diameter than 2 or 3 cm, although unknown, is likely <5% after a mean observation duration of 70 months.^{(288, 290, 293)} Data on the natural history of SAAs larger than 5 or 10 cm are lacking because interventional therapy has been proposed for nearly all such patients.^{(301, 302)} Risk factors for rupture are not clearly known.

Pregnancy has accounted for a high proportion of previously reported patients with rupture, suggesting that it could be a triggering factor, although a reporting bias cannot be ruled out.^{(291, 292)} With the advent of endovascular therapies, modern surveys of the natural history of SAAs in pregnancy are difficult to interpret because it has become standard practice to propose preemptive interventional management in affected women who plan pregnancy.^{(290-292, 298, 300)} However, when a large obstetric population was evaluated, no statistical support was found for an increased risk of rupture during pregnancy.⁽³⁰³⁾ The prevalence of SAAs in patients with portal hypertension with or without cirrhosis is ~15%^(304-308) and could be even higher when large portosystemic collaterals are present.^{(305, 308)} Whether LT is a specific risk factor for rupture in this population remains unclear^{(266, 304, 307)} because the overall incidence (4%) is close to that in the general population.⁽³⁰⁷⁾ It has been generally accepted that the risk of rupture and mortality is higher for pseudoaneurysms than for true aneurysms, likely attributable to the independent effect of causal conditions (pancreatitis, infection, or trauma).^{(289, 307)}

Open surgical repair, endovascular repair, and conservative management have been proposed, but prospective head-to-head comparisons are not available. A systematic review with meta-analysis attempted to account for the differences in the patient populations undergoing therapy.⁽²⁹³⁾ In this pooled analysis, the proportion of patients treated for ruptured aneurysm was higher in the surgical than in the endovascular repair group (18.4% vs. 8.8%; P < 0.001), respectively. In addition, 30-day mortality was higher (5.1% vs. 0.6%; P < 0.001), but repeat intervention (0.5% vs. 3.2%) and minor complications were less common in the surgical repair group compared with the endovascular repair group, respectively. Comparison with the patients undergoing conservative management is difficult because in the latter group, aneurysms were smaller in size and comorbidity was higher than in the other two groups. A subsequent report using a large data set demonstrated that endovascular repair was associated with lower postprocedural complication and infection rate, less resource utilization, and no difference in mortality compared to surgical treatments (3%; P = 0.99).⁽³⁰⁹⁾ A laparoscopic approach has also been proposed, but available data are still inconclusive as to the superiority of one surgical approach to another.⁽³⁰²⁾ There has been no evaluation of the optimal modality and schedule for surveillance in patients with SAAs. There is no consensus on a size criterion requiring intervention on an asymptomatic true SAA,⁽²⁸⁸⁾ but traditional surgical teaching based on case series suffering from significant selection bias has historically recommended intervention at 2 cm in size.⁽²⁹²⁾ In adult living donor LT recipients, routine interventional therapy, regardless of size, has been proposed to prevent early posttransplant bleeding.⁽³⁰⁵⁾ However, the low incidence of SAA rupture, the challenging surgical treatment, and risk of complications from endovascular therapy have led some to propose cautious recommendations to treat SAAs only when they are symptomatic, >30 mm in size, or show enlargement during close follow-up with angiograms.⁽³⁰⁶⁾

Issues Specific to Children and Congenital Disorders

Hepatic Hemangiomas

In recent years, there have been a number of changes in classification and treatment recommendations of hepatic hemangiomas in children.⁽³¹⁷⁾ These may be classified into two types: (1) congenital (rapidly involuting congenital hemangioma and noninvoluting congenital hemangioma) and (2) infantile (focal, diffuse, and multifocal). In general, the distinction between the congenital and infantile types of hemangiomas is blurred. Symptoms can range from none to cardiac decompensation attributable to high-output heart failure. The focal type rarely causes symptoms, and most can safely be observed. Many involute or remain stable. Hemangioma histopathological glutathione expression may occur in as many as 30%-50% of cases of infantile hemangiomata, ⁽³¹⁸⁾ and these have been shown to respond to nonselective beta-blockers such as propranolol.^{(319, 320)} The multifocal or diffuse types may require interventions, ranging from medical therapies, such as beta-blockers or embolization, to surgical resection or LT in the most extreme examples. Other treatment depends primarily on symptoms and includes observation for many, medical therapies for heart-related symptoms, embolization, or hepatic arterial ligation as a means to reduce cardiac output, surgical resection for local or lobar disease, and LT primarily for those not responding to therapy or unresectable symptomatic cases.

Hepatic Arteriovenous Malformations

Hepatic arteriovenous malformations are rare lesions that can be acquired or congenital. Acquired lesions are generally secondary to external trauma, such as a penetrating injury to the liver. Iatrogenic injuries may also be possible secondary to percutaneous liver biopsies. Congenital fistulas lead to portal hypertension and reversal of flow in the PV, which may cause inflammatory changes in the gut drained by the portal venous system. Treatment options include embolization or resection of the anatomical sector of the liver that is involved with the vascular malformation.⁽³²¹⁾ Adjunctive medical therapies may include anti-inflammatory agents in the presence of rarely associated autoimmune disorders that coexist with hepatic arteriovenous malformations.⁽³²²⁾

Congenital Portosystemic Shunts

Congenital extrahepatic portosystemic shunts were first described by Abernethy in the 18th century and were much later classified into type I and II lesions depending on the presence or absence of an accompanying hypoplastic PV. The largest cohort to date reports on 66 patients from multiple centers across Europe.⁽³²³⁾ The hallmark of these shunts is hyperammonemia from shunting of mesenteric blood into the systemic circulation. Treatment with LT has been proposed, but is unnecessary in the vast majority of children because of modern endovascular shunt closure techniques.^{(324, 325)} A significant proportion of these patients may have severe congenital cardiac, genitourinary, and musculoskeletal defects.⁽³²³⁾ Acquired portopulmonary hypertension (15% of cases) and hepatopulmonary syndrome (3%) require careful management⁽³²³⁾ of pulmonary hypertension preoperatively and postoperatively.