Potential conflict of interest: Nothing to report.
The 1-year spontaneous mortality rate in patients with Budd-Chiari syndrome (BCS) approaches 70%. No prospective assessment of indications and impact on survival of current therapeutic procedures has been performed. We evaluated a therapeutic strategy uniformly applied during the last 8 years in a single referral center. Fifty-one consecutive patients first received anticoagulation and were treated for associated diseases. Symptomatic patients were considered for hepatic vein recanalization; then for transjugular intrahepatic portosystemic shunt (TIPS), and finally for liver transplantation. The absence of a complete response led to the next procedure. Assessment was according to the strategy, whether procedures were technically applicable and successful. At entry, median (range) Child-Pugh score and Clichy prognostic index were 8 (5–12), and 5.4 (3.1–7.7), respectively. A complete response was achieved on medical therapy alone in 9 patients; after recanalization in 6, TIPS in 20, liver transplantation in 9, and retransplantation in 1. Of the 41 patients considered for recanalization, the procedure was not feasible in 27 and technically unsuccessful in 3. Of the 34 patients considered for TIPS, the procedure was considered not feasible in 9 and technically unsuccessful in 4. At 1 year of follow-up, a complete response to TIPS was achieved in 84%. One- and 5-year survival from starting anticoagulation were 96% (95% CI, 90–100) and 89% (95% CI, 79–100), respectively. In conclusion, excellent survival can be achieved in BCS patients when therapeutic procedures are introduced by order of increasing invasiveness, based on the response to previous therapy rather than on the severity of the patient's condition. (HEPATOLOGY 2006;44:1308–1316.)
Many treatments have been proposed for primary Budd-Chiari syndrome (BCS),1–4 a disease whose spontaneous mortality was reported to approach 70% at 1 year and 90% at 3 years.5 Routine anticoagulation has been recommended.5, 6 Increasing expertise in percutaneous recanalization of short-length stenosis on inferior vena cava (IVC) or hepatic veins has lead to a systematic search for these correctable lesions.7, 8 Surgical portosystemic shunting has long been used,9 but recent data failed to show its favorable impact on survival.5, 6, 10 Recently, transjugular intrahepatic portosystemic shunt (TIPS) has shown encouraging results.11–13 In patients with BCS, liver transplantation has achieved survival rates similar to that in patients with other chronic liver diseases.14–17 Transplantation was suggested to be preferable to surgical portosystemic shunting.18 However, the indications, benefits, and clinical impacts of these treatments for BCS remain unclear because no randomized controlled trials are available, and only limited attempts have been made to adjust the comparisons between procedures for the baseline severity of the disease.
In 1997, based on the analysis of our experience and of reported data,6–8, 14, 19–23 we adopted a strategy using these therapeutic procedures successively, by order of increasing invasiveness. This strategy has been based on the response to previous treatment, not on the actual severity of the patient's condition. The aim of this study is to assess the outcome of patients managed through this strategy.
All patients with BCS admitted in our unit from October 1997 to October 2004 were enrolled into the study if a diagnosis of primary BCS was established, age was greater than 15 years, portal vein was patent, and no malignant tumor was identified. Data at the time of diagnosis and of new treatment were considered. Date of diagnosis was the date of the first investigation where the criteria for diagnosis were fulfilled, based on imaging showing an obstructed venous outflow tract.3 Inherited and acquired risk factors for thrombosis were systematically investigated according to methods reported elsewhere.2, 24 The Clichy and Rotterdam prognostic scores were calculated.5, 6 Follow-up was from the date treatment was initiated, regardless of whether diagnosis had been established at that time.
BCS was considered asymptomatic when there was no evidence of unusual abdominal pain, jaundice, ascites, hepatomegaly, portal hypertension-related bleeding, or encephalopathy; and serum transaminase levels were persistently below twice the upper limit of normal values.10, 21 BCS was considered symptomatic otherwise.
Clinical failure to therapy (treatment failure) was considered when criteria for complete or ongoing response were lacking. Complete response was considered when all of the following six criteria were met and stable: (1) absence of clinically detectable ascites, with normal serum sodium and creatinine levels, in the absence of diuretic therapy, or on low-dose diuretics (spironolactone 75 mg/d or furosemide 40 mg/d) and moderate NaCl intake (<6 g/d); (2) increase in coagulation factor V to a level above 40% of normal value; (3) decrease in conjugated serum bilirubin to a level below 15 μmol/L; (4) absence of first or recurrent portal hypertension–related bleeding on primary or secondary prophylaxis with nonselective beta-blockers or with endoscopic therapy; (5) absence of spontaneous bacterial infection; and (6) BMI > 20 kg/m2 after substraction of ascites and edema. Ongoing response was considered when all of the following three criteria were met on a 2-weekly evaluation basis: (1) in the presence of ascites, a negative sodium and water balance was achieved using low-dose diuretics and moderate sodium intake, together with normal serum sodium and creatinine levels, or with increasing serum sodium if initially low and decreasing serum creatinine levels if initially high; (2) factor V level was increasing if initially low; and (3) serum conjugated bilirubin level was decreasing if initially high.
Treatment response (treatment failure or complete response) was assessed, taking into account not the feasibility or the achievement of the treatment procedure, but the therapeutic strategy described below. Technical success was defined as follows: for recanalization with or without stenting, when the pressure gradient across IVC or hepatic vein stenosis fell to below 5 mmHg; and for TIPS, when porto-hepatic pressure gradient decreased to a level below 12 mmHg.
The therapeutic strategy consisted of the following four successive steps:
Step 1: On diagnosis of BCS, all patients were given immediate and permanent anticoagulation therapy and were referred to a hematologist for management of a possible blood disease. The anticoagulation consisted of low-molecular-weight heparin, unfractionated heparin being used only when heparin therapy would have to be immediately discontinued. Vitamin K antagonists were substituted to heparin within a couple of weeks, targeting an international normalized ratio 2 to 3.
Step 2: All patients in whom step 1 failed (with BCS without ongoing response within 2 weeks of anticoagulation) were considered for percutaneous recanalization of hepatic veins or IVC. Recanalization was attempted when a short-length stenosis was found. A stent was inserted when stenosis was long and rigid.
Step 3: Patients with failure of these therapies were considered for transplantation and listed when younger than 65 years. TIPS insertion was then considered. Up to 2001, patients were considered non-eligible for TIPS when the caudate lobe was larger than 9 cm or when the portal vein section diameter was smaller than the hepatic arterial section diameter. From 2001 on, these criteria were abandoned. TIPS insertion was performed under ultrasound guidance through a hepatic vein stump if any, and through the inferior vena cava otherwise, according to reported procedures.12
Step 4: Patients with treatment failure of TIPS, or in whom TIPS insertion was judged unfeasible or was unsuccessful, underwent liver transplantation as soon as a graft became available. All patients received lifelong anticoagulation therapy, starting from the first day after liver transplantation.
Follow-up evaluation, from diagnosis and after each procedure, was at least every 2 weeks for 1 month then monthly for 2 months, every 3 months for a year, and every 6 months thereafter.
Quantitative variables are expressed as mean (±standard error), or median and range, or IQR (interquartile range), and qualitative variables as absolute and relative frequencies. Comparisons between groups of quantitative and qualitative variables were made by the Wilcoxon and chi-squared tests, respectively. Overall survival and incidence rates of new procedures were assessed by the Kaplan-Meier method. Comparisons of survival rates between groups were made by the log-rank test. All tests were two-sided and used a significance level of .05. Data handling and analysis were performed with SPSS 12.0 (SPSS Inc., Chicago, IL).
Characteristics of Patients at the Time of Diagnosis
Among 57 patients with primary BCS referred to our unit, six were excluded because the therapeutic strategy could not be applied (extensive portal vein thrombosis in three, previous surgical portosystemic shunt in one, and no possible follow-up in two foreign patients). Fifty-one patients were thus included. As shown in Table 1, their main characteristics are typical for primary BCS. In three asymptomatic patients (6%), diagnosis had been made fortuitously. Median Clichy prognostic index and Rotterdam score were 5.4 and 1.3, respectively. The acute, chronic, and acute-on-chronic variants10 were present in 3, 35, and 13 patients, respectively. Twelve patients had severe hepatic insufficiency25; two of these had acute liver failure.26 Underlying prothrombotic disorders are presented in Table 1.
Table 1. Characteristics of 51 Patients With BCS at Study Entry
Myeloproliferative disease/antiphospholipid syndrome/nocturnal paroxysmal hemoglobinuria N (%)
Median follow-up was 35 months (IQR, 17–56). Figure 1 shows outcome and time to failure according to strategy steps. Overall 1-, 3-, and 5-year survival are 96% (95% CI, 90–100), 89% (95% CI, 79–100), and 89% (95% CI, 79–100), respectively (Fig. 2). According to Clichy prognostic index, patients scoring less than 5.4 had 1-, 3-, and 5-year survival rates of 100%, 92%, and 92%, respectively; and patients scoring greater than 5.4 had a 1-, 3-, and 5-year survival rates of 92%, 86%, and 86%, respectively (P = .82, log-rank test). According to Rotterdam prognostic classification, 5-year survival was 100% in class I patients, and 1-, 3-, and 5-year survival rates were 94%, 84%, and 84%, respectively, in the combined classes II and III. Between diagnosis and end of follow-up, overall Clichy prognostic index fell from 5.4 (3.1–7.7) to 2.9 (2.2–5.8) in the 46 surviving patients, meaning an excellent recovery in hepatic function and complete control of symptoms in most (Table 2).
Table 2. Evolution of Clichy Prognostic Index* According to Therapy
Anticoagulation and Treatment for Associated Conditions.
In addition to treatment for any associated condition, all 51 patients were administered anticoagulation therapy. Anticoagulation was initiated a median of 1 day, (IQR 0–19 days; range, −334 to 7,494 days) from diagnosis. Median follow-up on anticoagulation alone was 25 months, (IQR 16–48 months). Eight patients required substitution with human antithrombin concentrate. Nine of the 51 patients (18%) had a complete response and thus underwent no additional procedure. Among these nine patients, at diagnosis, three were asymptomatic, and two had a Clichy score greater than 5.4. One of these nine patients died of other complications of paroxysmal nocturnal hemoglobinuria. Thus, for anticoagulation alone, treatment failure occurred in 42 of 51 patients, corresponding to a 1-year incidence rate of 55%. Twelve (23%) patients experienced bleeding complications: hematoma in seven patients (with bilateral adrenal involvement in one); vaginal bleeding in three; gastrointestinal bleeding in three; and heparin-induced thrombocytopenia in seven patients (three of whom developed thrombosis as a complication thereof).
Recanalization was considered in all 41 surviving patients without ongoing response BCS. Recanalization was actually attempted in 14 patients (34%) in whom a short-length stenosis was demonstrated in terminal IVC (in two patients) or large hepatic veins (in 12). Recanalization was technically successful in 11 of these 14 patients (81%). Three patients had angioplasty, and eight patients had stenting associated to angioplasty of the vein. Reasons for failure are shown in Table 3. Recanalization was carried out a median of 3 months (IQR: 1–10) after initiation of anticoagulation therapy, at the level of IVC in two patients, and hepatic veins in 12. A complete response was observed in seven patients (two with IVC obstruction and five with hepatic vein obstruction). As to the first two steps of the strategy, 3-month and 1-year complete response rates were 57% and 50%, respectively. Among the seven patients without complete response after recanalization, TIPS was considered in all, attempted in five, and technically successful in four. Thus, orthotopic liver transplantation was performed in two of the seven patients without complete response to recanalization, because they could not have TIPS for technical contraindication (no hepatic vein stump and obstruction by intracaval stent in one patient, and excessive length of intrahepatic tract in one). No procedure-related complications were found in any of the patients treated with recanalization.
Table 3. Reason for Failure of Interventional Therapy*
Number of patients
Patients have concomitant reasons for failure.
No indication (no patent portion in IVC or hepatic vein)
No hepatic vein stump
Excessive length of intrahepatic tract
Small or non-visible intrahepatic portal veins
Several of the above reasons
Uncorrectable TIPS dysfunction
Death of septic shock
Recurrent thrombosis leading to graft failure
TIPS was considered in the 34 patients who failed to achieve a complete response to anticoagulation therapy and recanalization. TIPS was attempted in 25 (49%), and was technically successful in 21 (83%). Reasons for TIPS failure in the other nine patients are presented in Table 3. Median delay to TIPS after initiation of anticoagulation was 4 months (IQR: 3–13 months). The patients who were submitted to TIPS more than 1 year after the beginning of anticoagulation alone either worsened after a period of improvement under anticoagulant therapy alone or had anticoagulation for deep vein thrombosis before BCS was diagnosed (two patients). Sixteen transcaval (75%) and five transhepatic TIPS were performed. Eleven patients (50%) had e-PTFE–covered stent graft (Viatorr, Gore, Evry, France).
Among the 21 patients who had TIPS, only one (4%) required liver transplantation for clinical failure 1 month post-TIPS. According to intention-to-treat analysis, 3-month and 1-year complete response rate are 84% and 84%, respectively. Failure led to liver transplantation in 2 (8%) of the 25 TIPS patients. Twenty patients having a complete response to TIPS became asymptomatic. The 1-year mortality rate was 8%, and the complication rate was 40%. Immediate complications occurred in 10 of the 25 patients. They are shown in Table 4. Procedure-related complications after technical failure were responsible for death in two patients and for emergency liver transplantation in two. Over the study period, the complication rate decreased whereas the proportion of patients considered non-eligible decreased (Fig. 1), and there was no change in the proportion of technical failure or in median time from referral to TIPS (4 months [IQR, 1–8 months] before 2001, and 4 months [IQR, 2–7 months] after 2001. Long-term revision was necessary in eight patients, including in 7 of 10 patients with bare stents (median follow-up, 32 months; IQR, 1–49 months) and 1 of 11 patients with covered stents (P = .08, chi-squared test) (median follow-up, 10 months; IQR, 1–21 months).
Eleven (22%) patients underwent liver transplantation a median of 12 months (IQR, 1–29 months) from anticoagulation initiation. The patients who were submitted to orthotopic liver transplantation more than 1 year after the beginning of anticoagulation alone either had anticoagulation for deep vein thrombosis before BCS was diagnosed (in two patients), or worsened after a period of stabilization or improvement under anticoagulant therapy alone.
Among the 11 transplant patients, TIPS had been considered unfeasible in nine and had failed in two. One transplant patient died of sepsis. Overall 1-, 2-, and 3-year posttransplantation survival rates were 91%, 91%, and 91% (95% CI, 74–100), respectively. One-, 2-, and 3-year retransplantation-free survival was 91% (95% CI, 74–100), 81%, and 81% (95% CI, 57–100), respectively. Complications occurred in nine patients: bleeding in six; hepatic artery thrombosis in three; portal vein thrombosis, recurrent hepatic vein thrombosis, and severe heart failure in one each. One patient had retransplantation for ischemic cholangiopathy in relation with hepatic artery thrombosis.
The rarity of BCS has precluded therapeutic trials. In this study, the relatively small sample size and the absence of randomization are compounded by the following strengths: First, the enrollment spans a period of only 7 years, during which uniform diagnostic and therapeutic algorithms have been used routinely, whereas previous surveys spanned a period of 15–25 years during which many aspects of diagnosis and treatment have evolved. Study duration was short enough to cope with the need for homogeneity and long enough to cope for the need for clinically significant sample size and follow-up. Indeed, almost all of BCS-related mortality takes place during the first 2 years after diagnosis (Table 5). Second, biases resulting from procedure availability remained minimal, because this study is based on a single referral center where all previously proposed diagnostic and therapeutic procedures for BCS could be performed. Third, a homogeneous strategy for treatment was defined at the start of the study, based on a review of the data available by 1996, as reported elsewhere.6–8, 14, 19–23 Furthermore, also at the start of the study period, precise although empirical criteria defining treatment failure and response were elaborated based on our experience. Future studies may help in simplifying or refining these definitions. Actually, this strategy is in agreement with recent recommendations by expert panels.3, 27
Table 5. Reported Mortality in Patients' Budd-Chiari Syndrome According to Time
The main finding in this study is an excellent outcome, because overall survival rates were 93%, 89%, and 89%, respectively, at 1, 3, and 5 years. Table 5 shows the survival rates previously reported in patients that were not selected on the basis of treatment received. Our findings compare favorably with all previous reports, including those from this center.5, 6, 10, 28–32 Although statistical analysis cannot be performed to test superiority, these findings are as or more significant because the present median Clichy prognostic index was identical to that of the three largest and most recent retrospective surveys.5, 6, 10 The current survival rate is similar to recently reported data from another referral center where interventional radiology was also used in a stepped manner before any consideration for liver transplantation.33 Thus, excellent medium-term survival can be achieved with a strategy that is based not on disease severity but on the response to treatment procedures of increasing invasiveness. Moreover, data presented in Table 3 show that end-of-follow-up condition was excellent, because Clichy prognostic index was at the lowest possible value in the surviving patients. This strategy minimized mobilization of health care resources, although this has to be checked as well. Considering the rarity of the disease, it will be difficult to gain further support for this strategy by means of randomized controlled trials. In addition, such studies will be difficult to design considering the need to test a whole strategy.
Although the evidence supporting routine anticoagulant therapy in patients with BCS is circumstantial, recent recommendations have emphasized immediate and prolonged anticoagulation as first-line treatment.3, 27 This recommendation is supported by the fact that 9 of 51 patients in this cohort, including two with a severe disease, did well on anticoagulation alone. However, anticoagulation alone was insufficient in the rest of the patients. Clearly, early indicators of anticoagulation therapy failure are needed. Such indicators would allow for shortening overall management. The hazards of anticoagulation were particularly noticeable in the patients from this cohort undergoing interventional procedures, because the 50% rate of anticoagulation-related complications was observed, although none of these was lethal. It is noteworthy, but still unexplained, that heparin-induced thrombocytopenia occurred in 15% of patients, a rate 10 times higher than reported in trials of prophylactic heparin after surgery.34–37
In agreement with the experience of other referral centers, we found recanalization to be safe and feasible in one third of the patients with symptomatic BCS.7, 33, 38, 39 Fifty percent of patients with technically successful recanalization had a complete response. However, recanalization as a whole was of clinical benefit in only 15% of patients, mainly because of non-eligibility (64% of all patients), clinical failure in 45% of patients with technically successful angioplasty, and, marginally, to technical failure (21% of eligible patients). Here again, early predictors for angioplasty failure, when available, would allow proceeding to TIPS without waiting undue time.
Our findings show that TIPS can be placed successfully in a high proportion of selected patients. TIPS resulted in a complete response in most of them. Secondary dysfunction was amenable to revision with prolonged efficacy. Indeed, TIPS allowed hepatic transplantation to be avoided in 85% of our patients. Still, a quarter of our patients were considered non-eligible for TIPS placement based on pre-intervention analysis of imaging data. This proportion appears to be higher than in other reported series where the number of patients was lower, or where the selection criteria were less clear.11–13, 33 Conversely, data presented in Fig. 3 clearly indicate that a learning curve effect occurred at our center, because there was a higher proportion of eligible patients and a lesser proportion of complications in recent years as compared with the to initial period. Overall, these data support the implementation of expert centers for TIPS in patients with BCS, preferably in close relationship with a transplant center. In the last study year, 90% of the patients had a successful TIPS, and the Clichy prognosis score improved from 6.4 before TIPS to 2.8 after TIPS insertion. Thus, once inserted, TIPS is highly effective in BCS patients. Recently reported data from a smaller cohort do suggest that medium-term liver transplantation can be avoided in almost all patients with TIPS insertion.33
In agreement with recent data,33, 40 covered stents had an apparently higher rate of primary patency as compared with bare stents. However, differences in follow-up limit interpretation of our data. Transcaval approach proved effective. Yet, the risk of fatal IVC injury is illustrated by one patient in this series. Furthermore, as emphasized by others,11 we experienced an unusually high rate of bleeding originating from the liver (intrahepatic hematoma, and hemobilia), as compared with our experience in inserting TIPS in patients with cirrhosis.41 These specific difficulties could be explained by a high degree of portal phlebosclerosis and hepatic arterialization occurring in BCS patients.42 Our protocol involved resumption of anticoagulation within minutes of the completion of TIPS deployment to prevent immediate thrombosis. The benefit/risk ratio of such an immediate anticoagulation can be questioned. Thus, after TIPS placement, optimal timing for resumption of anticoagulation and routine search for false aneurysm have to be reconsidered.
The observed survival after transplantation in this series compares favorably with those reported in previous studies.14, 43–45 In other words, submitting patients with severe BCS to transplantation only when other procedures failed did not compromised their outcome. The severe thrombotic complications occurred despite routine early posttransplantation anticoagulation, and the high incidence of bleeding. Again, the benefit/hazard ratio of routine anticoagulation in patients undergoing transplantation for BCS should be improved.
Thus, this study in BCS patients shows that excellent medium-term survival rates can be achieved using a strategy in which procedures are introduced by order of increasing invasiveness, based on treatment response rather than on the severity of the patient condition. Although the algorithm here assessed yielded a high survival rate, several strategy modifications should be considered to further improve overall results and, specifically, to decrease resort to transplantation.33 First of all, criteria for response must be refined to allow accurate and early prediction of procedure outcome. Second, protocols for anticoagulation must be improved to decrease the risk of bleeding without compromising prevention for thrombosis. Third, angioplasty and stenting could prove more efficient when inserted early in patients with short-length venous stenosis, even at a stage where symptoms are absent or minimal. Fourth, TIPS could similarly prove more efficient when inserted early in patients with little symptoms. Procedure-related complications and feasibility, the main limiting factors, are greatly influenced by technical skill and experience so that improved results can be expected in the near future. Controlled trials are needed to address these issues. Hopefully, collaborative studies will permit gathering a sufficient number of patients to perform these clinical studies.