Article Text

Download PDFPDF

Budd–Chiari syndrome and veno-occlusive disease/sinusoidal obstruction syndrome
  1. D-C Valla
  1. Professor D-C Valla, Service d’Hépatologie, Hôpital Beaujon, 100 boulevard Général Leclerc, 92118 Clichy, France; dominique.valla{at}

Statistics from

The hepatic venous system extends from the central veins of the hepatic lobule, up to the hepatic vein ostia into the inferior vena cava (IVC). There are numerous hepatic veins. The three major hepatic veins open into the IVC close to the right atrium. The hepatic venous system can be involved in arterio-venous or porto-venous fistulas, mostly of congenital origin, a topic which has been recently reviewed1 and will not be discussed here. Most of the other disorders of the hepatic venous system eventually cause its obstruction. Significant hepatic venous outflow block occurs either from a diffuse involvement in small vein diseases, or from a focal or extensive obstruction in diseases of the largest veins. So-called Budd–Chiari syndrome (BCS) is the liver disease resulting from the hepatic venous outflow block. The hepatic venous system can be invaded or compressed by malignant tumours of the liver, primary leimyosarcoma of the veins, or hepatic echinococcal disease. These diseases cause secondary BCS, which will not be discussed here. Primary BCS relates to the primary venous lesions including phlebitis, thrombosis or fibrosis.

Veno-occlusive disease (VOD) is characterised by particular, non-thrombotic, microvascular lesions without involvement of the large veins, occurring in a particular context of exposure to endothelial toxins. By convention, VOD has been separated from primary BCS, although VOD also causes hepatic outflow blockade, and mimics most clinical and laboratory features of primary BCS. Due to the particularities of VOD, and in order to make the distinction from BCS more clear, it was recently proposed to replace the denomination of VOD by sinusoidal obstruction syndrome (SOS).2




Primary BCS is a rare disease. Accurate estimates of incidence are lacking for western countries. In Japan, in 1989, an incidence estimate derived from a questionnaire survey and an autopsy registry was 0.2 per million inhabitants per year, while the prevalence estimate was 2.4 per million inhabitants.3 In Asian patients, the terminal portion of the IVC is generally obstructed by a fibrous or membranous material, usually incorporating the ostia of the major hepatic veins.4 In contrast, cases reported form western countries have mainly been due to pure involvement of the major hepatic veins. The nature of short-length, fibrous or membranous narrowing on the IVC or major hepatic veins has long been debated, a congenital malformation being admitted by many clinicians. However, ample evidence was obtained in the late 1990s to indicate that thrombosis is the initial pathogenetic event whatever the aspect of IVC obstruction: an obvious thrombus or a short-length stenosis so thin as to simulate a membrane or a “web”.4


The association of primary BCS with various underlying prothrombotic disorders has long been known. Associations with polycythaemia, Behcet’s disease, paroxysmal nocturnal haemoglobinuria, pregnancy and oral contraceptive use have been among the first recognised.5 In the 1980s, several groups independently showed that an atypical myeloproliferative disease (MPD) was frequently present, affecting mainly young females, and presenting without criteria for myeloproliferation in peripheral blood. The main clue to the presence of the MPD has been the “endogenous” formation of erythroid colonies in cultures of peripheral or bone marrow precursors of the red blood cells.6 This feature suggested independence from, or hypersensitivity to, the growth effect of erythropoietin. As a matter of fact, serum erythropoietin levels have been characteristically low in such patients. However, the specificity of these endogenous erythroid colonies has been questioned as no gold standard for MPD diagnosis was available, and the molecular mechanisms for these clonal disorders remained unclear. These particular culture studies have remained confined to a few specialised centres for several reasons: uncertainties regarding the significance of spontaneous colonies; non-standardised, demanding laboratory techniques; the absence of marked changes in blood cell counts; and scarce data on the course of the underlying MPD.6 The relative risk of BCS in patients with MPD could not be quantified as available case–control studies had insufficient power with regard to the low prevalence of MPD in a general population. In large cohorts of MPD patients, the incidence of BCS has been in the order of only 1%.7

Budd–Chiari syndrome (BCS)

  • Many patients with BCS harbour several prothrombotic disorders.

  • Liver disease blurs the diagnostic features of most prothrombotic disorders.

  • Myeloproliferative disease account for 50% of BCS patients. Their recognition largely relies on the assessment of the V617F JAK2 mutation in peripheral blood.

  • Treatment for BCS consists of the successive implementation of anticoagulation, percutaneous angioplasty, transjugular intrahepatic shunt and, finally, liver transplantation.

  • Using this progressive therapeutic strategy, overall 5-year survival rates currently approach 90%.

Some major inherited prothrombotic disorders (deficiency in protein C, protein S and antithrombin) have long been implicated. However, their precise role in primary BCS has been difficult to clarify. Indeed, as these coagulation inhibitors are produced by the liver, their plasma level decreases as a consequence of BCS-related liver dysfunction. Unfortunately, molecular testing has been of little help in this area as most of the numerous responsible mutations are private ones. In Asia or South Africa, an increased risk of IVC obstruction has been shown in patients with a very poor standard of living, although the mechanisms underlying this increased risk remain unknown.8 Last, but not least, it has been shown that BCS was a rare complication in patients affected with a known MPD,7 or with other known prothrombotic factors. Therefore, the reason why only rare patients with these underlying conditions develop BCS has remained unexplained.

Pathophysiology and manifestations

The clinical manifestations of BCS have been well characterised.911 Abdominal pain, ascites, liver and spleen enlargement, and portal hypertension are cardinal features, as well as a prominent dilation of subcutaneous veins of the trunk in those patients with long-standing IVC obstruction. Liver function tests are altered to various extents according to patients. Various forms of presentation and course have been described, ranging from fulminant to chronic, the latter being the most frequent presentation. Asymptomatic forms have been recognised.10 A major characteristic of these asymptomatic forms is the maintenance of a significant hepatic venous outflow, either through preservation of some hepatic veins and the IVC, or through the development of large intrahepatic or extrahepatic collaterals. A universal observation has been a discordance between the apparent duration of signs and symptoms and the actual age of venous or hepatic lesions found at pathological examination: most patients presenting with acute manifestations actually have evidence of a long-standing hepatic outflow obstruction, as indicated by extensive fibrosis or cirrhosis.9 Many patients develop numerous, regenerative macronodules that are highly enhanced at the arterial phase of contrast injection, and occasionally resemble focal nodular regenerative hyperplasia.12 The relationship of these frequent benign nodules to the otherwise known risk of developing hepatocellular carcinoma (HCC)13 has remained unclear.

Thus, there is a marked clinical and pathological heterogeneity among BCS patients. This heterogeneity has remained poorly understood, in part due to the limitations in the assessment of hepatic haemodynamics, liver histopathology and prothrombotic factors.


Advances in non-invasive vascular imaging with Doppler ultrasound, MRI, or CT with intravenous injection of contrast medium have allowed improved diagnosis, including the recognition of asymptomatic disease. A diagnostic algorithm is proposed in fig 1. Hepatic vein collaterals have proven a specific feature for making a diagnosis of hepatic vein obstruction. Finding hepatic vein collaterals is particularly useful in differentiating BCS from other chronic liver diseases.14 Therefore, angiography of the IVC and hepatic veins is needed for diagnostic purposes only in a minority of patients. Likewise, these progresses in imaging have limited the place of liver biopsy. This invasive procedure is needed only in the small subset of patients in whom BCS is related to pure thrombotic involvement of the small hepatic veins,15 where major hepatic veins and the IVC appear patent at imaging. Even using liver biopsy, it remains difficult to differentiate BCS from liver involvement due to heart failure or constrictive pericarditis, and SOS. All these conditions share congestion, liver cell loss, and fibrosis predominating in the centrilobular area.

Figure 1 Proposed diagnostic strategy for Budd–Chiari syndrome (BCS). IVC, inferior vena cava.

Outcome and prognosis

Spontaneous outcome of patients with BCS is extremely poor. Three-year survival was reported to be as low as 10%, at a time when no specific therapy was yet available.16 The main causes of death are intractable ascites leading to emaciation, gastrointestinal bleeding related to portal hypertension, and/or hepatic failure. A less severe disease, of more protracted course, was suggested in patients with obstruction of the terminal IVC compared with pure hepatic venous obstruction.4 Portal vein thrombosis, which occurs in about 20% of patients, is associated with a particularly poor outcome.17 Prognostic models elaborated in large cohorts of patients became available in the late 1990s and early 2000s.1820 Components of the Child–Pugh score were consistently found to be major prognostic indicators. Moreover, it was shown that liver biopsy findings provided no additional prognostic information to that of Child–Pugh score components. The prognostic impact of underlying prothrombotic diseases could not be ascertained in these early prognostic studies. Likewise, the independent prognostic impact of the level of hepatic venous outflow obstruction, and of the associated portal vein thrombosis, has remained unclear.


Anticoagulation, hepatic decompression—by means of angioplasty or end-to-side portosystemic shunting—and liver transplantation had long been proposed. The rarity of the disease has precluded adequate clinical trials. However, the advent of prognostic models in the late 1990s has paved the way to a more rigorous assessment by permitting adjustment of the observed survival for initial prognostic features.18 19 Using such an approach, it was possible to show (1) a non-detrimental, probably beneficial, effect of routine anticoagulation on the overall outcome; and (2) a lack of beneficial impact of surgical portosystemic shunting. However, it could not be clarified whether portosystemic shunting was not effective because of a high operative mortality—approaching 25% in many series—or because of a deleterious influence on liver function. Indeed, the absence of a beneficial effect contrasted with pathological and clinical evidence for the relief of hepatic congestion and the control of ascites. Furthermore, liver transplantation could not be evaluated in these studies. Thus, the place of liver transplantation as an alternative to first-line portosystemic shunting, or as a second-line procedure for failed surgical shunting has remained debated.21 The advent of transjugular intrahepatic portosystemic shunt (TIPS) only partially quenched the debate as procedure-related mortality was probably decreased, but the feasibility of TIPS in patients with obstructed hepatic veins or IVC was unclear. BCS recurrence in the graft was rapid and frequent in the absence of anticoagulation, but could be prevented by immediate anticoagulation.22 Overall, reported 5-year survival in patients seen in the 1990s was about 75% in independent series.11 1820 Thus, improved general care for patients with liver disease, earlier diagnosis with non-invasive imaging, anticoagulation and therapy for underlying disease, portosystemic shunting—surgical or radiological—and liver transplantation, all participated in an improved outcome, although each to an unknown extent.

Recent advances

The first years of this century have witnessed considerable advances in the overall understanding and practical management of primary BCS. Many of these advances have been made possible by the input of new knowledge from haematology, by technical improvement in interventional radiology and by international collaborative efforts. Causes and pathogenesis, pathophysiology and therapy have been the areas where the most important progress has been achieved.

Causes and pathogenesis

Recent data from several centres consistently show that primary BCS is a multifactorial disease where several prothrombotic disorders must concur for the development of thrombosis at this uncommon location.2325 This paradigm accounts for some of the previously raised issues regarding epidemiology and causes. First, the rarity of the disease is explained by the requirement for several concurrent factors for the disease to develop. Secondly, the differing features among various areas or populations can be explained by different combinations of various causes, depending on their background prevalence in the specific area or population. The impact of this multifactorial paradigm for clinical practice is that identifying one causal factor should not stop the search for the other factors involved.

Table 1 summarises the prevalence and odds ratio (OR) of the major causal factors reported in patients with primary BCS. The clarification of the major role of MPDs, accounting for about 50% of the cases, represents a significant advance. In this respect, a crucial step has been the identification of a particular somatic mutation (V617F) in the Janus tyrosine kinase-2 (JAK2) gene in myeloid cells of patients with MPD.26 JAK2 is coupled to the growth factor receptor on the cells of the myeloid lineage. Activation by the ligand (erythropoietin, thrombopoietin or other growth factors) elicits the signal for proliferation and differentiation of the myeloid precursors into mature cells through JAK2 phosphorylation. Part of the JAK2 molecule normally acts as an autoinhibitor of this phosphorylation, which limits the activation of signal transduction. V617F JAK2 mutation occurs at this inhibitory site, and results in an abrogation of the autoinhibitory function. Hence, there is constitutive activation of signal transduction resulting in independence from, or hypersensitivity to, growth factors. This discovery has provided not only an explanation for myeloproliferation but also a convenient tool for diagnosis, as this single somatic mutation can be detected in peripheral granulocytes or other blood cells of the myeloid lineage. V617F JAK2 has been found in about 80% of patients with polycythaemia vera, and in half of the patients with essential thrombocythaemia or idiopathic myelofibrosis.27 In those with primary BCS, it has been detected in 37–45% of patients.25 2832 Although such patients have increased blood cell counts as compared with those with undetectable V617F JAK2, their peripheral blood remains within normal values when BCS is present.

Table 1 Approximate prevalence and OR of major causal factors in patients with primary BCS

Clusters of dystrophic megakaryocytes have recently been shown to be a specific diagnostic feature for MPD at bone marrow biopsy.33 About two-thirds of BCS patients with an MPD harbour the V617F JAK2 mutation.25 29 However, still one-third of them lack the mutation, and evidence for the underlying MPD is derived from bone marrow biopsy findings. In practical terms, this means that the first diagnostic procedure should be the assessment of V617F JAK2 in peripheral granulocytes, followed by bone marrow biopsy in those patients that test negative for the mutation, as depicted in fig 2. This diagnostic algorithm should be implemented in any patient with BCS, whatever the results of peripheral blood cell counts.29 Taken together, V617F JAK2 mutation and bone marrow features of MPD almost completely overlap with the so-called spontaneous erythroid colony formation.25 29 32 Hence this demanding and non-standardised technique can be abandoned, at least for a diagnosis of MPD in BCS patients. Other somatic mutations of the JAK2 gene or other genes have been identified. However, preliminary data suggest that they account for only a minor proportion of the cases.29

Figure 2 Proposed algorithm for a diagnosis of myeloproliferative disease (MPD) in patients with primary Budd–Chiari syndrome (BCS). Numbers in parentheses are approximate prevalence among patients with BCS. JAK2, Janus tyrosine kinase-2.

Most recent data, from a large study in pregnant women from South France, indicate that the prevalence of V617F JAK2 mutation would be 0.2% in apparently healthy young women.34 Extrapolating these data to patients with primary BCS—most of whom are young, previously healthy women, too—would suggests the risk of BCS in child-bearing aged women to be 70- to 100-fold higher when the mutation is present that when it is absent.

Identification of factor V Leiden as a risk factor for venous thrombosis has been another major advance for the understanding of BCS pathogenesis. Factor V Leiden accounts for about 25% of patients with primary BCS, a proportion similar to that found in patients with deep venous thrombosis of the lower limbs.23 24 35 In most BCS patients, factor V Leiden is associated with other risk factors for thrombosis, as expected from its relatively weak thrombotic potential.35 In contrast, the G20210A prothrombin gene mutation, another recently discovered inherited prothrombotic disorder, appears to be less over-represented among BCS patients than the former prothrombotic conditions.

The increasing amount of available data lends support to the concept of site specificity for thrombosis according to the underlying prothrombotic disorder.23 31 36 MPD is clearly more common among BCS patients than among patients with portal vein thrombosis, and even more so than among patients with venous thrombosis at other sites. Factor V Leiden is more strongly associated with BCS than with portal vein thrombosis, whereas the converse appears to apply to G20210A prothrombin. Further site specificity might also be present within the hepatic venous outflow tract itself. Indeed, factor V Leiden appears to be particularly common in patients with IVC obstruction.35 Furthermore, oral contraceptives and pregnancy have been specifically associated with hepatic vein involvement.37 Finally, pure IVC obstruction has been almost exclusively reported from areas where a very poor standard of living is prevalent. In such an area, Nepal, the incidence of pure IVC obstruction is several times higher than in Japan, representing the leading cause of hospital admission for liver disease.8 Although these considerations are of little practical significance at present, in the future they may help to understand an important observation: the usual lack of a local—mechanical or inflammatory—factor explaining why thrombosis develops at such an unusual site in the context of a general prothrombotic condition.23

Important issues relevant to common clinical problems remain unanswered at present. When a patient presents with BCS, there is still no simple means to qualify what was the primary condition that caused the disease: a decreased plasma level of protein C, protein S, antithrombin or plasminogen; a dysfibrinogenaemia; an increased serum homocysteine level; or moderate fluctuating titres of anticardiolipin antibodies.5 Indeed, all these features are common in patients with liver disease of other origin. Likewise, emerging risk factors for venous thrombosis cannot be simply assessed once liver disease has developed: increased factor VIII levels, deficiency in thrombin-activatable fibrinolysis inhibitor (TAFI) and other alterations in recently identified members of the coagulation and fibrinolytic pathways.38 In practical terms, this frustration should translate into an opportunity for not performing useless testing. In the individual patient with decreased coagulation factor levels, and without a family history of idiopathic thrombosis, it would appear futile to test for diagnostic purpose either plasminogen, protein C, protein S or antithrombin deficiency, serum homocysteine or plasma factor VIII levels. Tables 2 and 3 detail the diagnostic features for prothrombotic disorders in patients with BCS.

Table 2 Proposed work-up for acquired conditions in patients with primary Budd–Chiari syndrome
Table 3 Proposed work-up for inherited conditions in patients with primary Budd–Chiari syndrome


A deleterious role for a loss of intrahepatic portal perfusion has been suggested by previous histopathological studies.12 This was fully confirmed in patients going to transplantation, by combining data from histopathological examination of the native liver with those of pretransplant vascular imaging.39 In these patients with end-stage liver disease, there is a close relationship between deprivation of portal blood supply, and liver cell loss in the corresponding areas. Portal venous deprivation can result either from portal vein thrombosis or from focally retrograde portal blood flow. Thus, maintenance of the portal blood supply might be crucial for preventing progression of liver disease. Increased hepatic arterial blood flow is evidenced by an enlarged hepatic artery. Moreover, regenerative macronodules, occasionally mimicking focal nodular hyperplasia, are found in arterialised liver, in those areas that are deprived of portal perfusion but apparently well drained by hepatic venous collaterals.39 Furthermore, nodular regenerative hyperplasia is commonplace in these arterialised livers.39

Extrahepatic portal vein thrombosis superimposed on BCS has been revisited recently.40 As compared with patients having a patent portal vein, manifestations at diagnosis were not more severe and, although survival tended to be decreased, the difference was not statistically significant. Therefore, it might be that superimposed extrahepatic portal vein thrombosis is mainly a consequence of severe outflow block, and resulting portal blood stasis, in a context of multiple prothrombotic factors. According to this view, superimposed extrahepatic portal vein thrombosis might be a reflection of severe liver disease rather than a factor involved in its aggravation.

BCS-related hepatic fibrosis differs from that related to alcoholic or viral liver disease, not only by its zonal distribution but also by its pathways. As compared with these parenchymal liver diseases, for a comparable stage of fibrosis, there is a several fold increase in the RNA expression of platelet-derived growth factor, connective tissue growth factor, collagen I and collagen IV. Furthermore, there is a marked increase in angiopoietin 1 expression.41

In BCS patients, HCC appears to be as significant as a long-term complication as it is in other chronic liver diseases. HCC developed in 11 of 97 patients of a recent cohort followed-up for a mean of 5 years.42 Serum α-fetoprotein appeared to be more specific for a diagnosis of HCC in patients with BCS than in those with other liver diseases. Patients with long-standing IVC obstruction carried a risk of developing HCC that was 70-fold higher than those with pure hepatic vein involvement. Multivariate analyses in larger cohorts, as well as molecular studies are needed to clarify HCC pathophysiology further in this non-inflammatory disease of the liver.


Although randomised clinical trials of treatment options are still lacking, advances have been derived from two sources. First, two international expert panel conferences have permitted a consensual elaboration of recommendations based on comprehensive reviews of evidence, and on the confrontation of the experience of referral centres.15 43 Secondly, these recommendations happened to be fully supported by subsequent reports of independent or collaborative studies. Figure 3 illustrates the therapeutic strategy that has been recommended and is discussed below.

Figure 3 Recommended treatment strategy in patients with primary Budd–Chiari syndrome. Numbers on the right represent the approximate proportion of patients that can be permanently controlled at each step of the strategy. With experience in transjugular intrahepatic portosystemic shunt (TIPS) insertion, the proportion of patients requiring liver transplantation may decrease. Data are based on recent cohort studies.44 45 48 51

Prompt recognition of underlying prothrombotic disease and early initiation of their specific therapy might translate into rapid improvement of liver disease. Although such advances are expected in the near future, particularly for MPD and paroxysmal nocturnal haemoglobinuria, data on these aspects are still lacking.

Early initiation of anticoagulation therapy is recommended for all patients regardless of whether an underlying prothrombotic disorder has been identified or not.15 43 It is recognised that the evidence on which this recommendation is based is only circumstantial. In the last two decades, administration of anticoagulation has been considered sufficiently sound by clinicians, so that most BCS patients have received this therapy. This has in some way hampered a retrospective analysis of its efficacy. A surprisingly high incidence of heparin-induced thrombocytopenia has been reported with unfractionated heparin, albeit not with low-molecular weight heparins.44 Although this finding has remained unexplained, it is the basis for a recommendation that unfractionated heparin be avoided. There is no evidence that the efficacy of vitamin K antagonists differs from that of heparin. At present, therefore, the choice of the type of anticoagulation should be dictated by pharmacokinetic considerations and by context—that is, a need for rapid and short-acting anti-coagulation therapy, or long-term treatment.

The efficacy and the inocuity of percutaneous angioplasty (with or without stenting) of short-length stenoses of hepatic veins or the IVC have been confirmed, at least when a retrograde, transjugular or transfemoral route is used.44 45 Therefore, an active search for such short-length stenoses, and, when found, an appropriate percutaneous intervention, has been recommended in patients with symptomatic BCS. In asymptomatic patients, the benefit–risk ratio of this therapeutic option is still debated. Available data suggest significant procedure-related morbidity with the transhepatic approach to hepatic vein angioplasty when a transvenous approach has failed.

Anticoagulation and angioplasty appear to succeed in controlling BCS in only 20–30% of patients, at least in western series where pure hepatic vein involvement predominates.44 Consistent data indicate that in patients where the disease is not fully controlled, the next step should be TIPS insertion. Indeed, in experienced hands, the transcaval approach is technically and clinically successful in most patients.4547 Furthermore, the advent of covered stents has allowed patency to be maintained for prolonged periods.46 Increased difficulties and increased morbidity due to the procedure in BCS patients as compared with the usual cirrhotic patients should not be underestimated. Improved survival in TIPS-treated patients is most obvious among those with the most severe disease at baseline. In that subset of patients, 5-year survival rose from 45% in a reference cohort, where TIPS was only marginally used,20 to 71% in a series of TIPS-treated patients.48 Overall, 60% of current western patients may require, and be satisfactorily treated with, a covered stent TIPS.

In the remaining 10–20% of patients, anticoagulation, percutaneous angioplasty and TIPS fail due either to technical problems or to poor clinical results of a technically successful procedure. In such patients, liver transplantation is the remaining option. Two recent retrospective analyses of the outcome in large series of transplanted patients have shown 5-year survival rates reaching 80%.49 50 The true impact of liver transplantation is difficult to assess based on these two studies, as their data do not allow for a comparison with patients of similar severity given less demanding treatments. Most recent reports suggest that a previous attempt at TIPS or percutaneous stenting of the outflow tract did not compromise the results of liver transplantation.50

Patients with a blocked portal vein are poor candidates for TIPS or liver transplantation, although each of these procedures has been reported to be successful in rare patients.40 Therefore, maintenance of a patent portal vein clearly stands as a major therapeutic target. However, it should be remembered that after adjustment for severity at diagnosis of BCS, prognosis was not influenced by the presence of portal vein obstruction.20 40 This means that compensated patients with preserved liver function despite extensive blockade of the splanchnic veins usually have prolonged survival when treated only with anticoagulation, and percutaneous angioplasty of the outflow tract when possible.

Recent clinical studies in specialised centres indicate an overall probability of 5-year survival approaching 90%, and at least 75% in patients presenting with the most severe disease.44 45 The progressive improvement in survival with time in French cohorts is depicted in fig 4. In a recent prospective international study of incident cases, the 24-month survival rate was 80%, while 47% of patients had been managed non-invasively, 40% had radiological interventions and only 12% of the patients underwent liver transplantation.51 These favourable results have been achieved with a strategy of increasing invasiveness, based on the response to therapy rather than on the initial status of the patient. In other words, each therapy should be introduced as a potentially definitive option rather than as a bridge to a further, more demanding, form of therapy. The absence of progressive improvement in manifestations and liver function should lead to the consideration of more invasive therapy. Increased survival has been paralleled by clearing of all clinical signs and symptoms of liver disease and normalisation of liver function in all patients.44 Thus, the constraints or untoward effects of therapy have become the main limitation to a patient’s quality of life. The long-term development of a malignant liver or blood disease currently represents the most dreaded complication.33 42

Figure 4 Schematic representation of survival curves in different multicentre cohorts of French patients18 44 72 The period of diagnosis is indicated in the middle of each curve. Percentages on the right are 5-year survival rates.The most recent cohort corresponds to the period when the strategy depicted in fig 3 was used.44



VOD has long been known as a human and animal disease related to the ingestion or administration of certain toxins.2 Conspicuous centrilobular damage with particular, non-thrombotic, changes in central veins have led to the naming of this entity. Plant pyrrolizidine alkaloids were among the first identified toxins causing sporadic or epidemic intoxication. Drugs have also been incriminated, as has irradiation.52 The drugs involved mostly consist of chemotherapy agents, as shown in table 4.2 All these agents induce hepatic damage in a dose-dependent manner. Hence, myeloablative “conditioning regimens” for haematopoietic stem cell transplantation by combining high dose chemotherapy drugs or chemotherapy drugs plus total body irradiation have become a major cause of VOD.2 Current myeloablative regimens are associated with an incidence of VOD between 0% and 50% depending on patient risk factors and choice of regimen. Long identified risk factors include cyclophosphamide-containing regimens, high doses of total body irradiation, hepatitis C, and norethisterone treatment.

Table 4 Drugs implicated in veno-occlusive disease/sinusoidal obstruction syndrome

Manifestations of VOD mimic those of BCS.2 Severity varies from patient to patient, ranging from silent forms to fulminant hepatic failure. In the context of haematopietic stem cell transplantation, diagnosis of clinically apparent VOD has been based on weight gain with or without ascites, right upper quadrant pain of liver origin, hepatomegaly and jaundice, occurring within the first 100 days; and exclusion of differential diagnoses such as hepatic graft versus host disease, drug-induced cholestasis and sepsis. Two diagnostic classification are presented in table 5. Liver biopsy cannot be easily obtained in this setting. As compared with histological findings, clinical criteria carry a risk of overdiagnosis (10–20%), and cannot account for the high prevalence of histological lesions without clinical expression.

Table 5 Clinical criteria for diagnosis of veno-occlusive disease/sinusoidal obstruction syndrome after haematopoietic stem cell transplantation

In the context of myeloablative therapy, reported mortality rates have varied between 0% and 67%.2 However, the highest figures may be a reflection of death with VOD in fragile patients rather than death from VOD. Some patients completely recover while others develop lethal forms, usually in association with multiorgan failure. A poor prognosis is correlated with high serum transaminase levels, a high hepatic venous pressure gradient, portal vein thrombosis, renal insufficiency and decreased oxygen saturation. Nodular regenerative hyperplasia or perisinusoidal fibrosis predominating in the centrilobular area can be found long after haematopoietic stem cell transplantation in patients with or without a clinically evident VOD. They are regarded as sequellae of previous VOD.53

Several drugs have been proposed based on findings in uncontrolled studies.2 In most instances, however, either randomised controlled trials (RCTs) were negative, or comparisons were performed with historical controls. Drawing conclusions from the latter is questionable as there are marked variations in patient outcome among centres, and within a centre according to time, often without apparent reason. Furthermore, well validated prognostic models are generally lacking so that adjustment for severity is of uncertain accuracy.

Veno-occlusive disease/sinusoidal obstruction syndrome (VOD/SOS)

  • VOD/SOS is produced by a toxic injury to sinusoidal endothelial cells.

  • Many chemotherapeutic agents can induce VOD/SOS.

  • Knowledge of the metabolic determinants of drug toxicity to sinusoidal endothelium has deepened. The translation of new knowledge into prevention and treatment of VOD/SOS still needs clinical studies.

  • Reduced intensity myeloablative regimens have proved efficient in reducing the incidence of VOD/SOS.

Recent advances

The description of a reproducible animal model of VOD using monocrotaline gavage has permitted clarification of the pathogenesis of the disease.54 A major step forward has been to show that sinusoidal, not venous, endothelial cells are the target of toxic injury related to monocrotaline—a pyrrolizidine alkaloid. This finding is in line with previous data indicating that venous alterations are inconspicuous in many patients with VOD, particularly those with less severe clinical disease. In order to account for these findings, a new denomination, sinusoidal obstruction syndrome (SOS), has been proposed to replace the term VOD.2 Some intracellular steps of the toxic effect have been dissected (for further information see reviews by DeLeve2 53). In sinusoidal endothelial cells, monocrotaline metabolite binding to F-actin leads to its depolymerisation, which in turn increases synthesis and activity of matrix metalloproteinase-9 (MMP-9). Decreased nitric oxide (NO) production by sinusoidal cells contributes to the increased MMP-9. As a consequence, endothelial cells round up—which causes sinusoidal lumen obstruction—and detach from the degraded extracellular matrix in the space of Disse. Red blood cells are then seen dissecting the space of Disse. Embolisation of debris further aggravates sinusoidal obstruction by blocking their junction with the central veins. Inhibition of MMP-9, or administration of a liver-specific NO donor, completely prevented SOS in the monocrotaline rat model.

The role of drug metabolism in producing sinusoidal toxicity has been recently clarified, as recently reviewed elsewhere.53 Drugs that produce SOS appear to share the following characteristics: (1) they are transformed in the liver; (2) they are detoxified by glutathione; (3) they produce sinusoidal endothelial cell depletion of glutathione; (4) sinusoidal endothelial cells can be rescued from the toxic agents by glutathione repletion; and (5) sinusoidal cells are more sensitive than hepatocytes to the toxic compound, in part due to a more profound depletion of glutathione in sinusoidal endothelial cells than in hepatocytes. Monocrotaline, dacarbazine and cyclophosphamide are metabolised in the liver by cytochrome P450s into toxic intermediates that are detoxified through conjugation with glutathione. Busulfan is first metabolised in the liver by glutathione S-transferases into an intermediate that is further transformed into a compound toxic to hepatocytes and sinusoidal endothelial cells. Azathioprine is transformed into its active derivative, 6-mercaptopurine, by conjugation with glutathione by glutathione S-transferase. Activation to toxic intermediates may occur in sinusoidal endothelial cells for monocrotaline and dacarbazine, whereas the toxic metabolites of cyclophosphamide are produced in the hepatocytes and transported to the space of Disse.53

In line with these data, genetic polymorphism in drug metabolism enzymes55 and drug interactions with this metabolism56 have been identified as risk factors for VOD/SOS. However, these data have not yet translated into specific clinical management. In particular, metabolism-based dosing of cyclophosphamide or busulfan has not yet been clearly shown to decrease the occurrence of fatal VOD/SOS.57

New chemotherapeutic agents, particularly gemtuzumab ozogamicin, have been associated with a frequent occurrence of sinusoidal toxicity.58 Gemtuzumab ozogamicin is a monoclonal anti-CD33 antibody targeting a potent cytotoxic agent, calicheamine, as an immunoconjugate. Frequent sinusoidal toxicity has also been revealed in the setting of neoadjuvant therapy for hepatic metastases of the long-used agent, oxaliplatin.59

The incidence of VOD/SOS after haematopoietic stem cell transplantation may have decreased in relation to patient selection, or improved overall care including prophylaxis. However, the incidence remains at around 20–40% for the most liver-toxic regimens. Novel therapeutic options have mainly consisted of prophylaxis using non-myeloablative regimens, particularly in patients carrying a high risk of severe VOD/SOS with conventional regimens. These regimens are associated with a reduced risk for VOD/SOS and immediate transplant-related mortality.60 61 However, an assessment should be carried out to ensure that this immediate benefit is not countered by a longer term increase in the risk of graft versus host disease or primary disease recurrence.

Recent studies generally failed to confirm the possible benefit of previously proposed protective agents.62 A systematic review with a meta-analysis could not show a benefit from prophylaxis using anticoagulation.63 A prophylactic effect of ursodeoxycholic acid indicated by two previous placebo-controlled randomised trials was not confirmed in two subsequently reported trials.64 65 Pre-emptive antithrombin administration failed to prevent VOD.62 66

Treatment of established VOD/SOS with recombinant tissue plasminogen activator has not been evaluated in RCTs. However, it appears that patients with severe VOD/SOS fail to respond to this therapy, which is associated with significant haemorrhagic complications in these thrombopenic patients.67 Similarly, patients with severe VOD/SOS did not respond to TIPS.68 Defibrotide elicited a great interest.62 69 A main advantage of this agent is an apparent lack of severe toxicity. However, convincing evidence for its efficacy is still lacking. Thus, in practice, management of VOD/SOS in haematopoietic stem cell recipients still relies mainly on non-specific measures, including prevention and treatment of fluid overload, prevention and treatment of sepsis, and careful consideration of the risk–benefit ratio of drug administration.


View Abstract


  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.