TIPS in the Management of Refractory Ascites
Kevin B. Mercure, M.D.
Wake Forest University Baptist Medical Center
Department of Internal Medicine
Resident Grand Rounds
April 6, 2004
PATHOPHYSIOLOGY OF ASCITES
Ascites is a collection of excess fluid in the peritoneal cavity. The pathogenesis of ascites in patients with cirrhosis involves several mechanisms. Portal hypertension results in decreased perfusion of hepatocytes with portal blood. This leads to increased reabsorption of sodium and water by the kidney resulting in increased plasma volume and increased portal flow. Portal inflow increases but resistance within the liver is relatively fixed secondary to underlying hepatic fibrosis. This too results in portal hypertension. Hepatic fibrosis also results in sinusoidal hypertension altering starling forces and driving fluid into the Space of Disse. Recall that the Space of Disse is a perisinusoidal space into which microvilli of the hepatocytes protrude.
Patients with cirrhosis are often hypoalbuminemic and hypoalbuminemia also works to promote this movement of fluid into the Space of Disse. This fluid is removed by hepatic lymphatics. Hepatic lymph flow increases dramatically in response to tissue edema. Normal thoracic duct lymph flow is 800-1000 cc per day. In patients with cirrhosis, hepatic lymph flow may approach 20 liters per day exceeding the capacity of the thoracic duct and percolate from the liver capsule into the peritoneal cavity.
Also as a result of decreased perfusion of hepatocytes, splanchnic vasodilation occurs. This triggers the release of sympathetic neurotransmitters further activating the renin-angiotensin-aldosterone axis. This further stimulates sodium and water retention by the kidney. It is important to note that renal function in patients with cirrhosis is disturbed long before ascites forms.
The International Ascites Club, a panel of experts representing the spectrum of clinical practice from North America to Europe, defined ascites as “uncomplicated” and “refractory.” Uncomplicated ascites was defined as ascites that is not infected and that is not associated with hepatorenal syndrome (HRS).
Refractory ascites was defined as ascites that cannot be mobilized or the early recurrence of which cannot be satisfactorily prevented by medical therapy.
refer to the included table from the Report on the Consensus Conference of the International Ascites Club.
TIPS in the Management of Refractory Ascites
THERAPY FOR ASCITES
Bed Rest: Many practitioners believe that bedrest may improve the response to diuretics. In patients with cirrhosis and ascites, upright posture activates sodium-retaining systems and impairs renal perfusion and sodium excretion. No clinical trials have shown that bed rest actually improves the efficacy of medical treatment.
Sodium Restriction: Sodium retention is one of the main mechanisms behind ascites formation. The goal is to obtain a negative sodium balance. It is recommended to restrict sodium to less than 2000 mg/day. Unfortunately severe sodium restricted diets are unpalatable and lead to poor patient compliance.
Diuretic Therapy: Both controlled and uncontrolled trials have shown that spironolactone is the initial drug of choice for ascites. Secondary hyperaldosteronism is a major factor promoting renal sodium retention in the distal tubules and collecting ducts of the nephron.
The loop diuretic furosemide is often used as well. Furosemide used alone is often less effective because sodium that is not absorbed in the Loop of Henle is taken up avidly in the distal and collecting tubules of the kidney because of the hyperaldosteronism of cirrhosis.
The effectiveness of diuretic therapy is best assessed by weighing the patient daily. The rate of weight loss should not exceed 0.5 kg/day in the absence of edema, or 1.0 kg/day when edema is present. The maximum recommended dose of spironolactone is 400 mg/day and 160 mg/day of furosemide. Side effects of these medicines include dehydration, muscle cramping, hyponatremia, gynecomastia and hepatic encephalopathy. Controlled clinical trials have demonstrated that the combination of sodium restriction, aldactone and furosemide achieve a response rate of nearly 90% in patients without renal failure.
Paracentesis: Large volume paracentesis is often a temporizing treatment. Paracentesis if often employed in patients that have not responded to medical therapy. Albumin is often given concomitantly to support effective arterial volume. Decreased effective arterial volume will lead to further sodium retention by the kidney further worsening ascites. There has been controversy over the effectiveness and necessity of volume expansion with albumin during a large volume paracentesis. There is data, however, indicating that patients that received paracentesis alone had a higher rate of complications.
Repeated paracentesis presents an increased risk of infection and bleeding and over time depletes total body protein stores.
Liver Transplantation: Any patient with cirrhosis that develops ascites should be considered as a potential candidate for liver transplantation. There are several limitations to liver transplantation, however. At this time there is a tremendous shortage of organs to transplant in relation to the overwhelming demand. The average wait for a liver transplant is about 500 days. In addition, there are several criteria to be eligible for liver transplantation and many patients with cirrhosis do not meet the criteria. In particular, patients must abstain from alcohol for at least six months to be eligible for a liver transplant.
TRANSJUGULAR INTRAHEPATIC PORTOSYSTEMIC SHUNT
Transjugular Intrahepatic Portosystemic Shunt (TIPS) was devised as a less invasive method to achieve portal decompression. The potential advantages of this technique include avoidance of general anesthesia, decreased procedural morbidity and mortality rates, and avoidance of surgery in the region of the hepatic hilum, which may be important in potential liver transplantation candidates.
The technique of placing the stent is as follows: The right internal jugular vein is accessed percutaneously. A vascular sheath is then advanced into the inferior vena cava and then into the hepatic vein. Next, a transjugular needle is advanced through a sheath into the liver parenchyma to form a tract.
The portal vein is then accessed and confirmation is made by injection of a contrast medium. Portal pressure is then measured and a guidewire is manipulated into the main portal vein. Angioplasty is then done to dilate the tract between the hepatic and portal veins. An 8-10mm expandable metallic stent is then deployed across the tract. Portal venography is then repeated and post shunt portal and vena caval pressures are determined.
Procedural complications are generally seen in 10% or fewer patients after the placement of TIPS.
Severe life threatening bleeding has been reported in 1-2% of patients secondary to puncture of the liver capsule resulting in hemoperitoneum or inadvertent puncture of the biliary tree resulting in hemobilia.
Other possible complications include infection, contrast induced renal failure, heart failure, stent migration, arrhythmias, and damage to adjacent organs. Although procedural complications are relatively infrequent, hepatic encephalopathy is a common complication after TIPS, with a frequency of 20-30%.
Another complication that is important to note is that of accelerated liver failure. The placement of TIPS may result in complete diversion of portal flow as well as a proportion of hepatic arterial flow into the shunt. Accelerated liver failure occurs as a result of a critical loss of hepatic perfusion. According to texts there is a wide variation in the 30 day mortality rate (a range of 3-45%) but is typically 10-15%. Appropriate patient selection is the key.
Unfortunately because of the nature of TIPS, the patients in whom TIPS is most clearly indicated often have advanced liver disease and are at the greatest risk of not surviving the procedure. Another challenge clinicians must face with TIPS is occlusion or stenosis of the shunt as a result of thrombosis or as a result of pseudointimal hyperplasia.
It is recommended to monitor shunt patency every 3-4 months by Doppler ultrasound and to proceed to venography if an ultrasonographic abnormality is found. There is a great deal of ongoing research to attempt to decrease the rate of shunt stenosis or thrombosis. One strategy that has been discussed but has yet to be investigated is the use of pharmocologic coated stents to prevent such complications.
THE EVIDENCE BEHIND TIPS FOR REFRACTORY ASCITES
A comparison of paracentesis and transjugular intrahepatic portosystemic shunting in patients with ascitesxii
Rossle et al. published this prospective, randomized study of TIPS versus large volume paracentesis for the treatment of refractory ascites in the New England Journal of Medicine in June, 2000.
Methods: The primary endpoint of the study was survival without transplantation. Patients with cirrhosis who had refractory ascites or recurrent ascites were included. Refractory ascites was defined as ascites that did not respond to four weeks of sodium restriction, 300-400 mg of spironolactone/day and 120 mg of furosemide/day or could not tolerate medical therapy. Recurrent ascites was defined as tense ascites that recurred on at least three occasions within a 12 month period on standard treatment. Patients were excluded if they had grade 2 or higher hepatic encephalopathy (indicated by somnolence), bilirubin>5mg/dL, advanced cancer, serum creatinine > 3mg/dL, portal vein thombosis, or hepatic hydrothorax.
A “complete response” was defined as elimination of ascites. A “partial response” was the presence of ascites not requiring paracentesis. “Absence of a response” was defined as persistence of ascites requiring paracentesis.
In those patients that received a shunt, the doses of diuretics were adjusted according to clinical need. Patients were seen after 1, 3, 6, 9, and 12 months and then every 6 months or whenever clinically indicated. The shunts were examined with duplex sonography. The patients that were assigned to paracentesis received dietary treatment and treatment with diuretics given at tolerable doses. Large volume paracentesis was followed by the administration of albumin only when clinically indicated. These patients had the same follow up schedule as the shunt group. All patients were instructed to abstain from alcohol.
Results: 29 patients were in the shunt group and 31 patients were in the paracentesis group. The two groups had very similar baseline characteristics. No patients were lost to follow up. Of note, during follow up 13 patients had shunt insufficiency and 11 underwent re-establishment of the shunt after about 10 months. Fifteen patients in the shunt group and 23 patients in the paracentesis group died. The most common cause of death was multiorgan failure. The probability of survival without transplantation was 69% at 1 year and 58% at 2 years in the shunt group. The probability of survival without transplantation was 52% at 1 year and 32% at 2 years in the paracentesis group (p=0.11). In addition, 10 patients in the paracentesis group required rescue shunt treatment.
At both 3 and 6 months the rate of complete response was significantly higher in the shunt group than in the paracentesis group (see table below). In addition, fewer patients in the shunt group had no response to treatment than in the paracentesis group.
Among patients that survived to 6 months after the initiation of therapy, measurements of hepatic function had not changed significantly in either group. With each group the Child-Pugh score improved significantly.
In the shunt group, the incidence of hepatic encephalopathy increased from 46% (12/26 patients) at baseline to 58% (15/26 patients) during follow up. In the paracentesis group, the incidence of hepatic encephalopathy increased from 39% (12/31 patients) at baseline to 48% (11/23 patients) during follow up. During follow up, the patients in the shunt group spent a total of 52 +/- 29 days in the hospital as compared with 72 +/- 48 days among those in the paracentesis group. (p=0.33)
Conclusions: This study found that as compared with large volume paracentesis, TIPS can improve the rate of survival without transplantation among patients with refractory or recurrent ascites. Although there was not a difference between the two groups with a univariate analysis, the difference was significant in a multivariate analysis. The multivariate analysis adjusted for slight differences in clinical variables that have a prognostic effect and that favor the paracentesis group. The response to treatment was evaluated at 3 months or more after randomization.
It was felt that 3 months was enough time to allow both stabilization of the patient’s condition as well as assessment of the out-of-hospital course of the disease. Of note, 10 patients between the two groups died prior to 3 months and were not included in the analysis. This paper was not clear about how many patients in each group died prior to 3 months. In contrast to the treatment of variceal bleeding, shunt treatment was not associated with a significantly increased incidence of hepatic encephalopathy.
When interpreting the results of this paper one should be cautious because although there was a significant difference in the multivariate analysis between the TIPS group and the paracentesis group this data is based on a small number of patients. In addition, the investigators failed to routinely use albumin with large-volume paracentesis in the medical therapy arm, this could potentially cause postparacentesis circulatory dysfunction.
Transjugular Intrahepatic Portosystemic Shunting Versus Paracentesis Plus Albumin for Refractory Ascites in Cirrhosisxiii
A randomized, mulitcenter study was undertaken by Gines et al. to compare TIPS and repeated large-volume paracentesis plus albumin in the management of patients with cirrhosis and refractory ascites. This study was published in Gastroenterology in December, 2002.
Methods: A total of 119 patients with cirrhosis and refractory ascites, as defined by lack of a response to low sodium diet and spironolactone 400 mg/day plus furosemide 160 mg/day or the appearance of diuretic-induced complications was evaluated for inclusion into the study. Exclusion criteria included age below 18 or greater than 75 years, serum bilirubin >10mg/dL, PT time lower than 40%, platelet count <40,000/mm3, serum creatinine >3mg/dL, hepatocellular carcinoma, complete portal vein thrombosis, cardiac or respiratory failure, bacterial infection, organic renal failure and chronic hepatic encephalopathy. Seventy patients were included in the study and randomly assigned to treatment with repeated large-volume paracentesis plus intravenous albumin or TIPS. There were 35 patients in each group. Of note, patients with refractory ascites and renal failure were randomized independently from those without renal failure. Renal and liver biochemical tests were measured at baseline and at least one, 3, 6, and 12 months after inclusion. Recurrent ascites was defined as the development of ascites either tense or large enough to produce discomfort or impair the mobility of the patient.
The analysis of costs was made by calculating the costs of all procedures performed in each treatment group. The primary endpoint of the study was survival without liver transplantation. Secondary endpoints were recurrence of ascites, hepatorenal syndrome, hyponatremia, gastrointestinal bleeding, spontaneous bacterial peritonitis, hepatic encephalopathy, and cost.
Results: No significant difference existed between the two groups with respect to the number of patients who died or were transplanted during the study period. (See table 2). Major causes of death were similar in the two groups. In the TIPS group the probability of survival at 1 year was 41% and at 2 years 26%. In the paracentesis plus albumin group the survival at 1 year was 35% and at 2 years 30%. The P value was 0.51. The univariate analysis demonstrated that several factors were associated with prognosis. In the multivariate analysis only baseline Child-Pugh score and BUN levels were independent predictors of survival.
In the TIPS group, 3 patients developed an early rise in transaminases associated with signs of liver failure (9%), 4 developed cardiac failure (12%), and 3 developed hemolytic anemia (9%). No paracentesis-related complications were observed. A TIPS was inserted in 3 of the 35 patients randomized to treatment with paracentesis (2, 3, and 12 months after inclusion) because of the need for very frequent paracentesis or the development of hepatorenal syndrome.
Recurrence of ascites was less common in patients treated with TIPS compared with patients treated with paracentesis plus albumin. (Table 2) In addition, the median time to the first recurrence of large ascites was markedly longer in the TIPS group than in the paracentesis group (171 vs. 20 days, respectively, P<0.0001) The number of patients developing de novo HRS or progressing to HRS during follow-up was significantly lower in the TIPS group compared with that in the paracentesis group
Upper GI hemorrhage occurred in 8 patients in each group, the source being esophageal varices in 2 patients from each group. Spontaneous bacterial peritonitis occurred in 2 patients from the TIPS group and in 4 from the paracentesis group.
In this study the number of patients who developed hepatic encephalopathy during follow-up did not differ between the two groups, however, the number or patients developing severe encephalopathy (grades III or IV) and the number of episodes of severe encephalopathy per patient were significantly greater in patients treated with TIPS.
In a multivariate analysis including baseline variables and treatment assignment, only treatment with TIPS and baseline BUN levels were independent factors predictive of the development of severe hepatic encephalopathy.
Theoretically, TIPS has a tremendous advantage over large-volume paracentesis in efficacy because TIPS acts on both mechanisms of ascites formation, sinusoidal portal hypertension and sodium retention. Large-volume paracentesis is a local therapy that does not modify any of the mechanisms that lead to ascites formation. TIPS, by theoretically preventing other complications of portal hypertension and ascites (variceal hemorrhage, HRS, and spontaneous bacterial peritonitis) could have a beneficial effect on survival. However, as a result of increased rates of hepatic encephalopathy and high occlusion rates TIPS could result in decreased survival and higher health care costs.
The authors of this paper felt that rather than focusing on efficacy it was necessary to determine which therapy resulted in higher survival rates and lower health care costs. This study showed that survival, the main endpoint, was not significantly different between patients treated with TIPS and those treated with repeated paracentesis plus albumin. The theoretical beneficial effect on survival that would have been obtained by preventing ascites recurrence rate and decreasing the risk of HRS was probably overridden by the detrimental effects of increasing portosystemic shunting.
study demonstrated a lower rate of ascites recurrence in patients treated with TIPS. Additionally, TIPS significantly reduced the risk of HRS occurrence, an observation not reported in previous studies. These beneficial effects are likely related to the suppression of the renin-angiotensin-aldosterone system and increase in atrial natriuretic peptide levels observed in patients that received TIPS. It is important to note that approximately 40% of patients treated with TIPS required angioplasty and/or restenting during follow-up.
In conclusion, this study demonstrated that TIPS lowers the rate of ascites recurrence and the development of HRS, however TIPS did not significantly improve survival and the cost of treating patients with refractory ascites with TIPS is substantially greater than the cost of treating patients with large-volume paracenteses plus albumin. Although the overall number of patients that developed hepatic encephalopathy during follow-up was high and not significantly different between the two groups, the frequency of severe hepatic encephalopathy was significantly higher in the TIPS group. That fact has a great deal of impact on quality of life.
The North American Study for the Treatment of Refractory Ascites
An international, multicenter, prospective, randomized controlled trial undertaken by Sanyal et al. comparing the clinical utility of repeated paracentesis, restriction of sodium, and treatment with diuretics versus medical therapy plus TIPS in patients with cirrhosis and refractory ascites. This study was published in Gastroenterology in March, 2003.
The inclusion criteria for the study included cirrhosis and symptomatic, refractory ascites with stable renal function. Cirrhosis was documented in each case by either a biopsy or abnormal liver function and a serum-to-ascites albumin gradient >1.1. Refractory ascites was defined according to the International Ascites Club criteria.
The exclusion criteria included failure to obtain consent, pregnancy, causes of ascites other than cirrhosis, incurable cancers or nonhepatic systemic diseases that were likely to limit life expectancy to <1 year, advanced liver failure, acute renal failure, INR>2, congestive heart failure, portal vein thrombosis, active sepsis, active encephalopathy (grade 2 or higher), hepatocellular carcinoma, and GI hemorrhage within 6 weeks of randomization.
109 subjects were randomized to TIPS (n=52) or medical therapy (n=57). The two arms were comparable in terms of age, sex, ethnic distribution, cause of liver disease, degree of baseline encephalopathy, quality of life, liver function, and renal function. Once eligibility was confirmed, treatment with diuretics was stopped for 5 days and studies were performed to evaluate renal function and measure hormones involved in sodium homeostasis. Paracentesis was then performed along with intravenous infusion of albumin in the paracentesis group and patients randomized to the TIPS group underwent their TIPS procedure.
The two main endpoints were recurrence of ascites requiring therapeutic paracentesis and mortality. Secondary end points included frequency of worsening encephalopathy, liver and renal function, frequency of other complications of cirrhosis and quality of life.
TIPS was successfully performed in 49 subjects. A TIPS could not be placed in 3 subjects for technical reasons. These 3 subjects were managed medically. Two subjects randomized to medical therapy received TIPS during follow-up for a ruptured umbilical hernia and a variceal hemorrhage, respectively. During the first year after randomization, 48 of the 57 (84%) patients randomized to medical therapy developed recurrent tense ascites requiring therapeutic paracentesis compared with only 22 of 52 (42%) patients randomized to TIPS. At 12 months, 70% of evaluable subjects in the TIPS group had developed stent stenosis. During follow-up, of 109 patients, 42 died and 33 underwent liver transplantation. The number of deaths was equally distributed across the two study groups. Also, the number of patients in each arm who underwent transplantation was not significantly different.
The prevalence of mild encephalopathy was similar across the two study arms. However, the incidence of episodes of acute worsening of mental status with at least grade 2 and grade 3-4 (severe) encephalopathy was higher in those receiving TIPS. The frequency of unscheduled hospitalizations and emergency-department visits was similar across the study arms.
A general health-related quality of life questionnaire with questions regarding general health, vitality, and overall functioning was employed. In both arms of the study the scores improved significantly compared with baseline and there were no significant difference across study arms. Thus, there were no demonstrable improvements in quality of life in those randomized to the TIPS arm compared with those randomized to the medical therapy arm.
The results of this study demonstrate that TIPS does not improve survival or improve quality of life. Both transplant-free and overall survival was essentially identical across the 2 arms of this trial. It is also noteworthy that no baseline parameters clearly showed a subset of patients who did better or worse after placement of a TIPS.
Also of note, three patients in the TIPS arm who had completely resolved their ascites died of hepatocellular carcinoma compared with none in the paracentesis arm. In all 3 instances, unresectable carcinoma was detected while patients were undergoing screening sonography and alpha-fetoprotein measurements.
It could have been predicted that the treatment group with better control of ascites would have a greater improvement in quality of life. The TIPS group had an advantage in control of ascites based on data demonstrating that the TIPS group had a lower rate of recurrent ascites and a significantly lower rate of paracenteses performed per subject during the first year. Interestingly, the two groups had similar quality-of-life scores by 6 months.
One must consider that the intense medical follow-up schedule in this trial may have contributed to an improved perception of quality of life by all subjects. The increased rates of encephalopathy in the TIPS arm may also have offset any improvement in quality of life due to better control of ascites in this group. It is also important to note that a general quality-of-life questionnaire was used in this study. If a disease-specific questionnaire had been employed it may have yielded different results.
According to the results of this trial, TIPS did not demonstrate an improvement in survival or in quality of life. The authors recommend that factors such as cost, candidacy for liver transplantation, underlying degree of liver failure, or the presence of other complications of cirrhosis should determine whether TIPS is employed instead of large-volume paracentesis plus albumin in the management of patients with refractory ascites.
It is important to note that liver transplantation is the only definitive treatment of cirrhosis with refractory ascites. The authors of this study concluded that TIPS should be reserved in most instances as second-line therapy or a bridge to liver transplantation.
TIPS was developed as a less invasive, nonsurgical method of creating a portalsystemic shunt. Theoretically TIPS should be superior to repeated large-volume paracentesis because TIPS acts on the mechanisms of ascites formation whereas large-volume paracentesis is a local therapy that does not modify the mechanisms that lead to ascites formation.
Rossle’s study in the New England Journal of Medicine revealed that patients receiving TIPS had better control of ascites. However, there was not a significant difference in hepatic function or in mortality. Moreover, the TIPS group had a higher rate of hepatic encephalopathy. Rossle’s study is frequently referred to in literature regarding TIPS, however, this study had several shortcomings. There was not enough power in this study and the patients in the paracentesis group were not routinely given albumin.
The other studies reviewed also suggested that TIPS provides better control of ascites but this did not translate into lower mortality, lower costs, or improved quality of life. In addition, the data indicate that TIPS results in a higher rate of hepatic encephalopathy. Perhaps, in the future, with improved techniques and equipment TIPS may have a more clearly advantageous role in the management of patients with refractory ascites. One such development on the horizon is the development of pharmacologically coated stents that may decrease the rate of shunt thrombosis and occlusion.
At this time one cannot argue that TIPS is superior to repeated large-volume paracentesis in the management of patients with refractory ascites.
i Crawford, James: Robbins Pathologic Basis of Disease, 5th ed. Philadelphia, WB Saunders, 1994. 836.
ii Crawford, James: Robbins Pathologic Basis of Disease, 5th ed. Philadelphia, WB Saunders, 1994. 836.
iii Moore et al. The Management of Ascites in Cirrhosis: Report on the Consensus Conference of the International Ascites Club. Hepatology. July 2003; 38: 259.
iv Moore et al. The Management of Ascites in Cirrhosis: Report on the Consensus Conference of the International Ascites Club. Hepatology. July 2003; 38: 259.
v Moore et al. The Management of Ascites in Cirrhosis: Report on the Consensus Conference of the International Ascites Club. Hepatology. July 2003; 38: 260.
vi Friedman, Scott, McQuaid, Kenneth, Grendell, James: Current Diagnosis and Treatment in Gastroenterology, 2nd ed. New York, McGraw Hill, 2003.
vii Moore et al. The Management of Ascites in Cirrhosis: Report on the Consensus Conference of the International Ascites Club. Hepatology. July 2003; 38: 259.
viii Feldman, Mark. Friedman, Lawrence. Sleisenger, Marvin: Sleisenger and Fordtran’s Gastrointestinal and Liver Diseases, 7th ed. Philadelphia, WB Saunders. 2002.
ix Feldman, Mark. Friedman, Lawrence. Sleisenger, Marvin: Sleisenger and Fordtran’s Gastrointestinal and Liver Diseases, 7th ed. Philadelphia, WB Saunders. 2002.
x Feldman, Mark. Friedman, Lawrence. Sleisenger, Marvin: Sleisenger and Fordtran’s Gastrointestinal and Liver Diseases, 7th ed. Philadelphia, WB Saunders. 2002.
xi Feldman, Mark. Friedman, Lawrence. Sleisenger, Marvin: Sleisenger and Fordtran’s Gastrointestinal and Liver Diseases, 7th ed. Philadelphia, WB Saunders. 2002.
xii Rossle et al. A Comparison of Paracentesis and Transjugular Intrahepatic Portosystemic Shunting in Patients with Ascites. New England Journal of Medicine, June 2000; 342.
xiii Gines et al. Transjugular Intrahepatic Portosystemic Shunting Versus Paracentesis Plus Albumin for Refractory Ascites in Cirrhosis. Gastroenterology, 2002; 123: 1839-1847.
xiv Sanyal et al. The North American Study for the Treatment of Refractory Ascites. Gastroenterology, 2003; 124: 634-641.