Potential conflict of interest: Nothing to report.
The initial short-term follow-up EN-Vie study was financially supported by the Fifth Framework Program of the European Commission (contract no.: “QLG1-CT-2002-01686”). CIBERehd is funded by the Instituto de Salud Carlos III. The center in Bonn has been additionally funded by the German Research Foundation (DFG; SFB-TRR57-P18, to J.T.).
Budd-Chiari syndrome (BCS) is a rare, life-threatening disease caused by obstruction of hepatic venous outflow. The aim of the study was to assess long-term outcome and identify prognostic factors in BCS patients managed by a step-wise approach using anticoagulation, angioplasty/thrombolysis, transjugular intrahepatic portosystemic shunting (TIPS), and orthotopic liver transplantation (OLT). We reviewed long-term data on 157 patients previously included by the European Network for Vascular Disorders of the Liver, a multicenter prospective study of newly diagnosed BCS patients in nine European countries. Patients were followed for a median of 50 months (range, 0.1-74.0). During the study, 88 patients (56%) received at least one invasive intervention (22 patients angioplasty/thrombolysis, 62 TIPS, and 20 OLT) and 36 (22.9%) died. Most interventions and/or deaths occurred in the first 2 years after diagnosis. The Rotterdam score was excellent in predicting intervention-free survival, and no other variable could significantly improve its prognostic ability. Moreover, BCS-TIPS prognostic index (PI) score (based on international normalized ratio, bilirubin, and age) was strongly associated with survival and had a discriminative capacity, which was superior to the Rotterdam score. Conclusions: The current study confirms, in a large cohort of patients with BCS recruited over a short period, that a step-wise treatment approach provides good long-term survival. In addition, the study validates the Rotterdam score for predicting intervention-free survival and the BCS-TIPS PI score for predicting survival. (HEPATOLOGY 2013;)
Budd-Chiari syndrome (BCS) is an uncommon, life-threatening disorder arising as a consequence of obstruction to hepatic venous outflow regardless of its causal mechanism or level of obstruction. This obstruction, usually caused by thrombosis, can occur from the small hepatic venules up to the entrance of the inferior vena cava into the right atrium.1, 2 In the vast majority of cases, it is possible to identify at least one inherited or acquired prothrombotic risk factor as the underlying cause of thrombosis. Therapeutic options include pharmacological management with anticoagulants and diuretics as well as invasive procedures, such as thrombolysis, percutaneous transluminal angioplasty (PTA), transjugular intrahepatic portosystemic shunting (TIPS), surgical portosystemic shunting, and orthotopic liver transplantation (OLT).1, 3 As a consequence of these therapies, especially anticoagulation, TIPS, and OLT, the prognosis of these patients has markedly improved over recent decades.4-7
However, because of the low incidence of the disease,4, 8 studies showing improvement in prognosis were mostly retrospective.6, 7, 9-11 In fact, only one prospective study exists, albeit with a short follow-up (median, 17 months).4 Hence, there are scarce data on the current long-term prognosis of BCS. Given that most patients included in the prospective cohort4 are being actively followed in their original centers, we have been able to evaluate the long-term prognosis of patients with BCS.
AUC, area under the curve; BCS, Budd-Chiari syndrome; BCIS score, BCS-intervention-free survival prognostic score; BCS-TIPS PI, BCS-TIPS prognostic index; CI, confidence interval; CRF, clinical record form; EN-Vie, European Network for Vascular Disorders of the Liver; GI, gastrointestinal; HE, hepatic encephalopathy; INR, international normalized ratio; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; PH, portal hypertension; PTA, percutaneous transluminal angioplasty; TIPS, transjugular intrahepatic portosystemic shunting.
Patients and Methods
Study Design and Data Acquisition.
The current study involves extended follow-up of the prospective European Network for Vascular Disorders of the Liver (EN-Vie) study that included 163 consecutive incidental patients with BCS diagnosed between October 2003 and October 2005 in academic and large regional hospitals in nine European countries.4 To standardize patient management, all participating centers had received guidelines with instructions on diagnostics tests and general indications for invasive procedures, such as TIPS, portosystemic shunting, and OLT, that were previously agreed upon by the EN-Vie steering committee. Further details on the study design of this original study can be found elsewhere.4
For the purpose of the present study, all previous participating centers were contacted again and agreed to participate in the extended follow-up study. Data were collected on a new, specifically designed CRF (clinical record form) where significant clinical events—defined as clinical deterioration (any new hospital admission or any clinical event), new liver-related imaging study, or any BCS-related intervention—were recorded from the end of the previous study (May 2006) until death or the end of the current study (June 2009). One investigator per country reviewed all CRFs before its inclusion in the database.
Patients were considered lost to follow-up if the last visit preceded the 9 months before study closure. All national and, if necessary, local ethics committees approved the study, and all patients provided written informed consent to participate.
Child-Pugh and Model for End-Stage Liver Disease (MELD) scores were calculated using the data at diagnosis of BCS, as previously reported.12, 13 The Rotterdam score was previously published to predict survival and is defined as follows: 1.27 × encephalopathy + 1.04 × ascites + 0.72 × prothrombin time + 0.004 × bilirubin (where ascites was scored as present “1” or absent “0”). The 5-year survival rate was 89% (95% confidence interval [CI]: 79-99) for class I, 74% (95% CI: 65-83) for class II, and 42% (95% CI: 28-56) for class III.9 The BCS-TIPS prognostic index score (TIPS-BCS PI score) was developed to predict OLT free survival in patients that received TIPS and is defined as follows: age (years) × 0.08 + bilirubin (mg/dL) × 0.16 + international normalized ratio (INR) × 0.63. The cutoff of 7 points had a sensitivity of 58%, a specificity of 99%, a positive predictive value of 88%, and a negative predictive value of 96% for death or OLT 1 year after TIPS.6
Results are expressed as N (proportions) for categorical variables and as medians (range) for continuous variables. Actuarial transplantation-free and intervention-free survival rates were calculated by using Kaplan-Meier's method. Uni- and multivariable Cox's regression analysis was used to explore the association between different variables and prognosis. New prognostic scores were constructed by combining (in a linear equation) those variables independently associated with the event multiplied by their regression coefficients. To add potential advantages to these models, we did not include subjective parameters (e.g., presence or absence of hepatic encephalopathy; HE) or INR in patients that may have initiated anticoagulation that were integrated in the previously described scores. Statistical significance was defined as a P value less than 0.05. All statistical analyses were conducted with the PASW Statistics 18 program (SPSS, Inc., Chicago, IL).
All 163 patients included in the previous EN-Vie study were eligible, and all centers, except one, that took part in the first study agreed to participate. Finally, 157 patients were included in the current study (Belgium, N = 5; France, N = 35; Germany, N = 14; Great Britain, N = 29; Italy, N = 18; Spain, N = 33; Switzerland, N = 4; The Netherlands, N = 19).
Overall median follow-up of these 157 patients was 50 months (range, 0.1-74.0). Twenty-six patients (17%) were lost to follow-up after a median time of 25 months (range, 0.3-61.0). The remaining 131 patients were followed until death (n = 36; 23%; median time to death: 10 months [range, 0.1-41.0]) or study closure (n = 95; 61%; median follow-up: 57 months [range, 43-74]).
Table 1 describes the baseline characteristics. Median age at diagnosis of BSC was 37 years (range, 16-83), and 90 patients (57.3%) were female.
Table 1. Baseline Characteristics of the 157 Patients Included in the Study
n (%) or Median and Range
Abbreviation: EGD, esophageal gastroduodenoscopy.
In 3 patients, the Rotterdam score was not possible to be calculated.
Supporting Table 1 describes the etiology for the total study population. With reference to the original EN-Vie study, we found additional causal factors in 12 patients: myeloproliferative neoplasms in 7; celiac disease in 2; and antiphospholipid syndrome, factor V Leiden mutation, and hyperhomocysteinemia in 1 each.
One hundred and thirty-nine patients (88.5%) received long-term anticoagulation. Twenty-eight bleeding complications occurred in 24 patients (17%) during the study. Main causes of bleeding were portal hypertension (PH) related (n = 14; 2 died), intracranial hemorrhage (n = 3; 1 died), and abdominal wall bleeding (n = 2), genital bleeding (n = 2), bronchial bleeding (n = 1), and peptic ulcer (n = 1; all alive). Figure 1 shows the flowchart of treatments received by patients.
Twenty-two patients underwent angioplasty (n = 13), thrombolysis (n = 7), or both (n = 2) as first invasive treatment. In 6 of these 22 patients, a vascular stent was placed at the time of angioplasty. After this initial intervention, 14 patients (64%) required further treatment with either TIPS (N = 12) or OLT (N = 2) after a median time of 1.5 months (range, 0.2-19.0) (Fig. 1). The remaining 8 patients were only treated with angioplasty/thrombolysis (in 2 patients more than once). Seven of them are alive and free of ascites with a median follow-up of 47 months (range, 32-61), but 1 died 6 months later as a result of liver failure.
Sixty-two patients underwent TIPS (39.5%). Main indications were refractory ascites (69%), liver failure (13%), and variceal bleeding (7%). Four of these (6.45%) had rescue OLT a median of 1.8 months after TIPS (range, 0.03-13.0) for the following reasons: HE (n = 1); fulminant liver failure (N = 1); and TIPS thrombosis with refractory ascites (N = 2). Three of these four patients died a median of 35 months after OLT (range, 7-45) as a result of liver failure (N = 2) and extrahepatic malignancy (N = 1). Of the remaining 58 patients, 10 (17%) died within 5.8 months (range, 0.2-39) and 48 (83%) were alive after a median follow-up of 51 months (range, 0.3-69.0).
Thus, overall, 13 patients died, 9 of them resulting from a liver-related cause. One, 3-, and 5-year actuarial survival and OLT-free survival of patients treated with TIPS was 88%, 83%, and 72% and 85%, 78%, and 72%, respectively (Fig. 2). Similar results were found if deaths clearly unrelated to liver disease were removed from the analysis or considering the date of TIPS as time zero (data not shown).
Median time from diagnosis to TIPS was 1 month (range, 0-38). Indeed, 50% of TIPS were placed in the first month, 60% in the first 3 months after diagnosis, 73% in the first 6 months, and 84% in the first 12 months. Patients who underwent TIPS in the first month had more-severe liver disease at diagnosis, as shown by a worse Rotterdam score (1.54 ± 0.59 versus 1.18 ± 0.77; P = 0.017) and Child-Pugh score (9.3 ± 1.7 versus 7.8 ± 1.9; P < 0.000). However, no differences in overall survival or OLT-free survival were observed in patients with TIPS performed before or after the first month after diagnosis. Similar results were observed when comparing patients receiving TIPS before or later than 3 or 6 months from diagnosis (data not shown).
On univariable analysis, only age and BCS-TIPS PI score (either as continuous or categorical variable [≥7 points])6 were significantly associated with survival or OLT-free survival (Supporting Tables 2 and 3). At multivariable analysis, only BCS-TIPS PI score was shown to be independently associated with survival and OLT-free survival. Because BCS-TIPS PI score was obtained at diagnosis, we performed a sensitivity analysis including only the 45 patients receiving TIPS in the first 6 months after diagnosis, obtaining similar results. No additional variables could improve the predictive ability of BCS-TIPS PI score in multivariable or classification and regression tree models (data not shown).
Three patients underwent a side-to-side portocaval shunt (2%), in 2 after an attempt at TIPS was unsuccessful. One patient developed shunt thrombosis and died soon thereafter, and another patient underwent OLT 9.8 months after shunt placement as a result of refractory ascites, despite shunt patency, and is alive at the end of follow-up. The third patient was alive and free of ascites at the end of follow-up.
Twenty patients received OLT (12.7%) a median of 2.3 months (range, 0-24) after BCS diagnosis. Sixty percent and 85% of OLT were performed in the first 6 and 12 months after diagnosis, respectively. Main indications for OLT were liver failure (40%), refractory ascites (35%), and variceal bleeding (10%). One, 3-, and 5-year actuarial survival after OLT was 95%, 89%, and 78%, respectively.
In 15 patients, OLT was the first-line proposed treatment (n = 14) or after angioplasty failure (n = 1). These 15 patients had more-frequent HE (P = 0.006) as well as higher Rotterdam score (P = 0.004) and class (P = 0.002) at diagnosis than the 62 patients receiving TIPS (n = 50 as first-line treatment and n = 12 after initial angioplasty failure) (Supporting Table 4). Despite this, no significant differences in survival were observed among groups (Supporting Fig. 1). Similar results were found when comparing TIPS or OLT as first-line intervention after excluding those patients with previous angioplasty/thrombolysis (50 TIPS versus 14 OLT; P = 0.29).
Long-Term Post-Therapeutic Prognosis of BCS.
Figure 3 shows the cumulative overall, OLT-free, TIPS-OLT–free and (any) intervention-free survival.
Sixty-nine patients did not undergo any invasive intervention during the study. Twenty died after a median time of 11 months (range, 0.10-40.0), only 2 as a result of non-liver-related death. The remaining 49 were alive after a median of 55 months (range, 0.7-69.0).
Uni- and multivariable analysis for intervention-free survival is detailed in Supporting Table 5. The Rotterdam score had an excellent prognostic value, and no further variable could significantly improve its prognostic ability. This validates the Rotterdam score as a useful prognostic tool in this post-therapeutic series of BCS. Supporting Fig. 2 shows survival curves for Rotterdam class I, II, and III.
Because the Rotterdam score includes the INR, which could not be calculated in a substantial number of patients (already on oral anticoagulants), we performed a multivariable analysis without including scores or INR. Baseline ascites, bilirubin, and creatinine were independently associated with intervention or death (BCS-intervention-free survival prognostic score [BCIS score]: ascites [yes = 1, no = 0]*1.675 + ln creatinine [umol/L]*0.613 + ln bilirubin [umol/L]*0.440). This data-driven new score showed an adequate discrimination (area under the curve [AUC] = 0.819), but it did not outperform the Rotterdam score (AUC, 0.821)9 (Supporting Fig. 3). The probability of intervention-free survival among different intervals of the BCIS score is shown in Supporting Table 7.
Thirty-six patients (23%) died during the study. Median time to death was 10 months (range, 0.1-41.0). Main causes of death are reported in Table 2. Factors associated with mortality are shown in Supporting Table 6. The BCS-TIPS PI score was strongly associated with the risk of death, so that no other variable could improve its predictive capacity. Supporting Table 8 shows survival among different ranges of BCS-TIPS PI scores. Because this score includes the INR, we performed a multivariable analysis excluding all scores and INR. Age, bilirubin, and creatinine were independently associated with survival [BCSurvival score: age/10*0.370 + ln creatinine [umol/L]*0.809 + ln bilirubin [umol/L]*0.496). The discriminative capacity was comparable to that of the BCS-TIPS PI score and better than the Rotterdam score (Supporting Fig. 4).
Table 2. Causes of Death
Related or Probably Related Liver Deaths (n = 30)
Non-Liver-Related Deaths (n = 6)
Liver failure (n = 12)
Extrahepatic malignancy (n = 1);
Multiorgan failure (n = 4)
Complication/progression of hematological disease (n = 4);
GI bleeding (n = 2)
Intracraneal hemorrhage (n = 1)
Sepsis (n = 4)
Hepatobiliary malignancy (n = 2)
Unknown (n = 6)
BCS is a rare, life-threatening disorder caused by obstruction of hepatic venous outflow. Until recently, most evidence regarding BCS was generated in small retrospective studies of patients diagnosed over long periods and managed using heterogeneous strategies.7, 9, 14 However, an international initiative, funded by the Fifth Framework Program of the European Commission, entitled the EN-Vie, was able to prospectively gather a large multicenter cohort of consecutive patients with BCS diagnosed and treated following homogeneous criteria.4 Previous retrospective studies evaluating prognosis in BCS showed that fatal events occur throughout the first 5 years after diagnosis.7, 9, 10, 14 Therefore, it is feasible that prognostic predictors were underestimated in the initial EN-Vie study where the median follow-up was short (only 17 months) and the number of events relatively low.4 Indeed, in the present extended EN-Vie study, surviving patients were followed up more than 3 additional years, and during this additional period, 8 patients received TIPS, 2 OLT, and 7 died. Thus, the present study was able to evaluate long-term outcome of BCS patients (median follow-up of almost 5 years, with a minimum of 43 months).
Our updated data confirm that, in Western countries, a step-wise therapeutic strategy confers good long-term survival in patients with BCSurvival score. Most of our patients (88.5%) received long-term anticoagulation. Interestingly enough, the rate of bleeding complications in patients receiving anticoagulation was lower than that previously reported.15 This is most likely the result of more adequate prevention of PH complications as well as careful management of anticoagulation during invasive procedures.15
Only 22 patients (14%) underwent angioplasty/thrombolysis as primary invasive therapy, and only 8 of them did not require further intervention, such as TIPS, surgical shunt, and/or OLT. It seems that angioplasty/stenting, although an attractive, minimally invasive technique with the potential of restoring physiological sinusoidal flow, has low applicability in the treatment of our BCS patients. These results contrast with a recent retrospective study from China showing a great applicability and efficacy of angioplasty/stenting in a large cohort of patients with BCS.16 In our opinion, these differences could be most likely explained by different pathogenic mechanisms of hepatic venous outflow obstruction,8 because hepatic vein stenoses are less frequent in the Western world than in Eastern countries. Therefore, angioplasty/stenting remains a potentially valuable treatment of the BCS subtype with short-length stenosis and investigation of the patients' suitability for this approach is mandatory, because the benefits are potentially significant.
Strikingly, no additional patient received a surgical shunt during the extended follow-up period, and thus only 3 patients (2%) received this therapeutic modality. TIPS has emerged as the preferred derivative treatment in Europe. The fact that two recent small retrospective studies from North America have shown excellent outcomes of BCS patients after surgical shunts does, in our opinion, not change the trend in current practice to prefer less-invasive over more-invasive procedures.17, 18 Moreover, we would like to emphasize that previous multicenter retrospective studies were unable to demonstrate a solid survival advantage in BCS patients treated with surgical shunts.7, 19-22 The low number of patients treated with surgical shunting in our data set precludes shedding more light on this issue.
Sixty-two patients required TIPS as rescue therapy after failures of medical or minimally invasive treatments (angioplasty/stenting/thrombolysis). Overall survival and OLT-free survival was comparable to that observed in a previous retrospective multicenter European study including 124 BCS patients treated with TIPS.6 These results confirm that TIPS is an effective, safe rescue therapy in patients with BCS. Interestingly, although most TIPS were placed during the first year after diagnosis, the timing was not uniform, ranging from 0 to 38 months. One of the major concerns in the management of patients with BCS is whether delaying the use of a rescue TIPS could influence outcome. Our data showed a good outcome after TIPS, regardless of whether the procedure was performed soon after diagnosis or later during follow-up. This outcome, which requires further confirmation, suggests that the approach of close clinical surveillance while reserving TIPS for those patients who progress or fail to respond to medical treatment does not have a deleterious effect on outcome. Furthermore, the current study validates our previously reported BCS-TIPS PI score >76 as the only independent factor associated with poor survival and OLT-free survival after TIPS. Whether the initial use of OLT in these patients with a high BCS-TIPS PI score may improve outcome needs to be proved. Comparing the subgroup of patients that received TIPS to those with OLT as first invasive therapy, we found that both groups had similar long-term outcome, despite the OLT subgroup of patients having had worse hepatic disease at presentation. Unfortunately, our current data do not allow us to asses the potential role of OLT as an initial procedure in these sickest patients.
Fifty-six percent of our patients underwent an invasive therapeutic procedure, most of them within the first year after diagnosis. In contrast with the population from which the Rotterdam score was defined,9 TIPS and OLT have been more widely used. Nevertheless, our study validates the use of the Rotterdam score for predicting the need of invasive intervention and death in this more-recent, prospectively studied cohort of BCS patients.
The new score (BCIS score) has an almost identical discrimination capacity to that obtained with the Rotterdam score, but with some potential advantages, including the exclusion of subjective parameters, such as the presence or absence of HE and INR in patients that may have initiated anticoagulation.9 We cannot dismiss the influence of more-rapid intervention in the sickest patients, which may have influenced our findings in relation to predicting intervention-free survival.
Another important finding of our study was that the BCS-TIPS PI score showed adequate accuracy in predicting mortality in the overall cohort of patients and better predictive capacity than the Rotterdam score. In addition, in the present study, we have identified a new survival score (BCIS score) that has an almost identical discrimination capacity to that obtained with the BCS-TIPS PI score, but with the potential advantage of not including the INR within its determinants. This may be important, because many patients may already be on anticoagulation when they arrive at referral centers.
In contrast to previous studies, validation of previous scores and identification of new ones has been done in a large cohort of patients, prospectively recruited in a short period of time and managed in a homogeneous step-wise invasive strategy.
In summary, our study validates a therapeutic algorithm aimed at providing a general framework for evidence-based decision making in patients with BCS. In addition, the present study validates the Rotterdam score for predicting intervention-free survival and BCS-TIPS PI score for survival. Furthermore, we report on two new prognostic scores that may help to better inform the choice of treatment strategy in any given BCS patient, but which need to be validated in future prospective multicenter studies.