Refractory ascites: pathogenesis, definition and therapy of a severe complication in patients with cirrhosis


Francesco Salerno, MD, Internal Medicine, Policlinico IRCCS San Donato and University of Milan, Via Morandi, 30, 20097 Milan, Italy
Tel/Fax: +00 3902 5277 4462


Ascites is a frequent complication of cirrhosis and portal hypertension, because of the increase of the sinusoidal hydrostatic pressure. Cirrhosis accounts for over 75% of episodes of ascites. Cirrhotic patients with ascites have marked alterations in the splanchnic and systemic haemodynamics, causing central hypovolaemia and arterial hypotension with consequent activation of the vasoconstrictor systems, renin–angiotensin and sympathetic systems, and with increased renal sodium re-absorption. One of the most serious complications in cirrhotic patients with ascites is the occurrence of refractoriness, that is the inability to resolve ascites by the standard medical treatment with low sodium diet and diuretic doses up to 160 mg/day of furosemide and 400 mg/day of spironolactone. Many patients with refractory ascites also have a chronic renal insufficiency that is called hepatorenal syndrome type-2. In these patients ascites may be treated with periodic paracentesis or with transjugular intrahepatic portosystemic shunt. However, only liver transplantation may improve the survival of such patients.


hepatorenal syndrome;


paracentesis-induced circulatory dysfunction;


polymorphonuclear cells;


plasma renin activity;




peritoneo-venous shunt;


renin–angiotensin–aldosterone system;


randomized control trial;


spontaneous bacterial peritonitis;


sympathetic nervous system;


transjugular intrahepatic portosystemic shunt.

Ascites is one of the most frequent complications of cirrhosis. Cirrhotic ascites accounts for over 75% of patients who are admitted to hospital with ascites, with the remaining 20% owing to malignancies (12%), cardiac failure (5%), tuberculosis (1%), pancreatitis (1%) or other rarer causes (1).

Cirrhotic patients with ascites are characterized by marked alterations in their splanchnic and systemic haemodynamics (2), leading to splanchnic congestion, central hypovolaemia, and arterial hypotension with activation of vasoconstrictor systems, such as renin–angiotensin and sympathetic systems with renal increase of sodium reabsorption. One of the most adverse consequences of such haemodynamic dysfunctions is the development of refractory ascites, which is the clinical condition whereby standard medical treatment with low sodium diet and diuretics is unable to resolve the ascites. Refractory ascites generally affects patients with advanced cirrhosis who may also develop hepatorenal syndrome (HRS) type-2 (3). Both refractory ascites and HRS type-2 are independent predictors of short survival (4).

The present paper reports an update on the pathogenesis and treatment of refractory ascites generated by a joint meeting of the International Ascites Club and the European Association for the Study of the Liver (EASL) held in Barcelona on 11 April 2007. The authors updated the review according to new investigations published after this date.

Pathogenesis and definition

Pathogenesis of ascites

Ascites formation in cirrhotic patients is related to two main pathogenetic mechanisms: portal hypertension and renal sodium retention (Fig. 1). An increased resistance to portal flow at the sinusoidal level leads to the development of sinusoidal portal hypertension and the backward transmission of this increased pressure into splanchnic capillaries (5, 6). The result is that the excess of fluid preferentially localizes in the peritoneal cavity. Sinusoidal hypertension is also associated with the development of a peripheral, mainly splanchnic, arterial vasodilation that leads to an increase in the fraction of blood flowing through the splanchnic area with a further increase of portal pressure and with a reduction of central (or essential) volaemia (2, 7, 8).

Figure 1.

 Mechanisms of determination of ascites, hyponatraemia and renal failure in cirrhosis according to the peripheral vasodilation hypothesis (2). 1=extravasation is caused by intrasinusoidal hydrostatic hypertension; 2=lymphatic drainage is a compensatory mechanism for interstitial fluid accumulation; 3=peripheral vasodilation and 4=low cardiac index are the main mechanisms causing effective hypovolaemia.

The second mechanism, renal sodium retention, is also related to portal hypertension and precedes ascites formation (9). The mechanism that initiates renal sodium retention is still debated; both a hepatorenal baroreflex initiated by portal hypertension (10–12) or a subtle state of hypovolaemia with minimal activation of the renin–angiotensin–aldosterone system (RAAS) have been implicated. The initial renal sodium retention that occurs before ascites development does not appear to occur in response to a significant reduction in arterial blood volume. At a later stage of cirrhosis, renal sodium and water retention are necessary to replenish the intravascular volume in order to maintain haemodynamic stability. Continued renal sodium retention is a prerequisite for ascites formation, as increased hydrostatic pressure alone without sodium retention will not produce any ascites.

The main consequence of these haemodynamic and renal alterations is a continuous escape of fluid from the hepatic sinusoids and from the splanchnic capillaries into the interstitial space. This spill over is initially compensated by an increased return of fluids to the systemic circulation through the lymphatic system and the thoracic duct. However, as cirrhosis progresses, the escape of fluids overcomes the lymphatic return, leading to progressive accumulation of fluid directly into the peritoneal cavity (13, 14). This is followed by a further activation of the renal mechanisms of sodium and water re-absorption that is crucial to perpetuate ascites formation.

Ascites re-absorption

Shear et al. (15) calculated that the rate of fluid re-absorption from the abdominal cavity is a function of intra-abdominal pressure, and that the maximal capacity of re-absorption through the peritoneal membrane in patients with alcoholic cirrhosis is <900 ml/day. Development of clinically significant ascites, therefore, occurs when the rate of ascites formation exceeds the rate of ascites re-absorption. Therefore, if the volume of fluid that spills into the peritoneal cavity can be reduced to below this threshold, the ascites can be easily controlled. This is the main reason why at the early stage of decompensation, ascites is responsive to a reduced intake of dietary sodium and to moderate doses of diuretics.

Potential mechanisms of progression of ascites

With progression of cirrhosis and portal hypertension, other mechanisms could make ascites more difficult to resolve, even when their roles are still not definitively supported by experimental evidence.

In the more advanced stages of decompensated cirrhosis, patients develop low arterial pressure because of a further reduction of the peripheral vascular resistances (2, 7). Two different compensatory mechanisms develop in order to maintain the haemodynamic stability. The first is a further increase in the release of vasoconstrictors to raise peripheral vascular resistance and the second is an increase in the cardiac output to refill the expanded intravascular bed. This last mechanism is operative throughout most of the clinical course of the cirrhotic process. However, when cirrhosis progresses and the arterial vasodilation becomes more marked, the heart can be no longer able to keep pace with the marked systemic vasodilatation (8, 16, 17). At this stage, arterial hypotension may be prevented only by a further increase of endogenous vasoactive compounds [RAAS, sympathetic nervous system (SNS) and antidiuretic hormone or vasopressin]. Reduced renal perfusion and further sodium and water retention are the natural consequences of this physiological response to relative arterial underfilling. This contributes to increasing ascites, and in many cases, to the development of HRS type-2. This could be the case in patients whose ascites can no longer be resolved simply by standard pharmacological therapy and dietary sodium restriction.

Refractory ascites

Five to 10% of ascitic patients per year become refractory to standard medical treatment, because of either an insufficient natriuretic effect of diuretic medications, or more often, to the development of diuretic-related severe side effects, which compel the patient to discontinue the diuretic treatment (4, 18). However, these figures are derived from observational studies performed in hospitalized patients (19) or from controlled clinical studies in highly selected populations (20, 21), and the true incidence of refractory ascites in everyday clinical practice remains unsettled. It is likely that it varies considerably in relation to the stage of cirrhosis, being far greater in patients with impaired renal function (HRS type-2). Therefore, unresponsiveness to pharmacological treatment may be more often observed in patients with cirrhosis and ascites admitted to tertiary referral centres or to liver transplant centres.

According to the International Ascites Club (4), refractory ascites is defined as ascites that cannot be mobilized or whose early recurrence after paracentesis cannot be satisfactorily prevented by medical therapy (Table 1).

Table 1.   Definition of refractory ascites (4, 18)
Refractory ascites: ascites that cannot be mobilized or the early recurrence of which (i.e. after therapeutic paracentesis) cannot be satisfactorily prevented by medical therapy.
It can be divided into:
1. Diuretic-resistant ascites: ascites that cannot be mobilized or the early recurrence of which cannot be prevented because of a lack of response to dietary sodium restriction and intensive diuretic treatment.
2. Diuretic-intractable ascites: ascites that cannot be mobilized or the early recurrence of which cannot be prevented because of the development of diuretic-induced complications that preclude the use of an effective diuretic dosage.

Clinical consequences of refractory ascites

Refractory ascites is frequently associated with HRS type-2, spontaneous bacterial peritonitis (SBP), dilutional hyponatraemia, muscle wasting and pleural effusion. It also implies an ominous prognosis, as the 2-year probability of survival among patients with refractory ascites is about 30% (22), while at least 40% of patients with responsive ascites are alive at 5 years (23, 24). Thus, the onset of refractory ascites requires specific therapeutic measures and dictates that the patient be rapidly evaluated for possible liver transplantation if not carried out before.

Mechanisms of refractory ascites

From a practical standpoint, refractory ascites refers to ascites that cannot be resolved by dietary sodium restriction and diuretic treatment. Such a simple definition, however, encompasses situations that may differ widely in terms of pathogenesis, pharmacological treatment and concurrent complications.

The severity of renal sodium retention increases throughout the natural history of cirrhosis, because of the progression of systemic and portal haemodynamic abnormalities and the associated activation of neurohumoural vasoactive systems leading to avid renal re-absorption of sodium and water (25, 26). At the same time, renal perfusion and glomerular filtration rate progressively decline. As a result, sodium re-absorption at the proximal convoluted tubule markedly increases and its delivery to distal segments of the nephron is markedly reduced (26). Thus, renal sodium retention mainly occurs proximally to the site of action of both antimineralocorticoid and loop diuretics, and this can explain why diuretic treatment becomes unsuccessful in some patients. In addition, the reduced cardiovascular responsiveness to vasoconstrictor systems (27) perpetuates the relative underfilling of the effective arterial blood volume and this compounds the hypovolaemic effects of diuretics. In such a situation, diuretic-related side effects frequently occur, precluding the continuation of effective dosages of diuretics. Accordingly, refractoriness to diuretic treatment is the consequence of the haemodynamic abnormalities characterizing advanced cirrhosis.

Conditions leading to transient or apparent refractoriness to diuretic therapy

There are, however, several other conditions of no or insufficient response to diuretic treatment that cannot be properly defined as refractory ascites. Firstly, an inappropriate diuretic therapy should be ruled out. This includes patients treated only with loop diuretics when their effect is thwarted by unopposed hyperaldosteronism or patients receiving insufficient doses of antimineralocorticoids, which should be proportional to the degree of hyperaldosteronism (28). Clearly in these patients, responsiveness to treatment can be restored by adjusting the doses of diuretics. The same applies when diuretics have evoked an exaggerated response (overdiuresis) leading to a negative fluid balance in excess of 900 ml/day with too rapid weight loss. Prerenal azotaemia ensues in these cases and refractoriness is actually induced by the treatment itself (29). This condition requires diuretic withdrawal, plasma volume expansion and a cautious restoration of a gradually increasing diuretic therapy to doses lower than those that caused prerenal azotaemia.

Transient refractoriness of ascites can also occur when the renal function is impaired by other iatrogenic causes or concurrent, but reversible, complications. Iatrogenic factors include the administration of nonsteroidal anti-inflammatory drugs which impair renal function by inhibiting the synthesis of vasodilating prostaglandins (30), the administration of angiotensin-I-converting-enzyme inhibitors or angiotensin receptor antagonists, which can impair the renal blood perfusion and reduce the glomerular filtration rate in patients with cirrhosis and ascites (25) or the administration of nephrotoxic drugs such as aminoglycosides (31). Complications precipitating renal failure include fluid loss through vomiting, diarrhoea or bleeding or bacterial infections such as SBP, which can intensify arterial vasodilatation, and therefore, worsens the mismatch between the intravascular volume and the vascular capacitance (32). In these cases, withdrawal of the offending drug or resolution of the complication combined with an appropriate plasma volume expansion may restore responsiveness to the standard treatment of low sodium intake and diuretics.

At last, we should remember the cases of the so-called refractory ascites because the patient did not comply with a low sodium diet. A 24-h urinary sodium excretion and a quick calculation using the amount of weight gained will allow one to tell whether or not the patient is compliant with sodium restriction.

Definition and diagnostic criteria of refractory ascites

In view of these issues, the term refractory ascites has often generated confusion in the past and made it difficult to compare the results of clinical studies performed by different investigators. This led to the proposal of a stringent definition by the International Ascites Club (4), which has been almost universally accepted and recently confirmed (18) (Table 1).

The term refractory ascites includes two different subtypes: diuretic-resistant ascites and diuretic-intractable ascites. The International Ascites Club also set the diagnostic criteria (4), which have since been slightly modified (18) as listed in Table 2.

Table 2.   Revised diagnostic criteria of refractory ascites (18)
  • *

    These diuretic dosages have been indicated because higher doses are unlikely to attain a greater efficacy and almost certainly will induce severe side effects. As a matter of fact, doses of furosemide such as 160 mg/day may not be reached in clinical practice, as side effects often occur at lower doses.

  • †Grade 1 ascites is mild ascites only detectable by ultrasound examination; grade 2 ascites or moderate ascites is manifest by moderate symmetrical distension of the abdomen. Grade 3 ascites is large or gross ascites with marked abdominal distension.

1. Treatment duration: Patients must be on intensive diuretic therapy (spironolactone 400 mg/day and furosemide 160 mg/day)* for at least 1 week and on a salt-restricted diet of less than 90 mmoles/day or 5.2 g of salt/day.
2. Lack of response: Mean weight loss of <0.8 kg over 4 days and urinary sodium output less than the sodium intake.
3. Early ascites recurrence: Reappearance of grade 2 or 3 ascites within 4 weeks of initial mobilization.
4. Diuretic-induced complications:
 a. Diuretic-induced hepatic encephalopathy: development of encephalopathy in the absence of any other precipitating factor.
 b. Diuretic-induced renal impairment: increase of serum creatinine by >100% to a value >2 mg/dl in patients with ascites responding to treatment.
 c. Diuretic-induced hyponatraemia: decrease of serum sodium by >10 mmol/L to a serum sodium of <125 mmol/L.
 d. Diuretic induced hypo or hyperkalaemia: change in serum potassium to <3 mmol/L or>6 mmol/L despite appropriate measures.

Massive and recidivant (or recurrent ascites)

Based on the definition and diagnostic criteria reported above, it follows that massive or tense ascites can by no means be considered as refractory, as the failure to respond to treatment has to be demonstrated. Another condition that cannot be termed refractory is recidivant (or recurrent) ascites, a peritoneal effusion that recurs on at least three occasions within a 12-month period despite dietary sodium restriction and adequate diuretic doses (4). Such a definition, however, remains somewhat ambiguous as it can encompass conditions that differ in pathophysiological and clinical terms. Because of poor compliance with medical treatment or dietary excesses, the meaning of ascites recurrence is quite distinct from that resulting, for example, from haemodynamic instability, which prevents the use of an adequate diuretic dosage to achieve a significant urinary sodium excretion rate.

Therapy of refractory ascites

The role of paracentesis

Paracentesis, the oldest treatment of ascites, is still the first-line treatment for refractory ascites (4, 18, 22, 33–35). It is also effective in accelerating the resolution of abdominal tension and enabling an easier ultrasound investigation in patients with massive ascites. Moreover, total paracentesis shortens the length of hospitalization and can be performed as an outpatient procedure.

During the 1960s, large-volume paracentesis was progressively abandoned, because of the introduction of potent and safe diuretics, such as furosemide, and to the common fear that rapid abdominal fluid evacuation could induce severe side effects such as hepatic encephalopathy, cardiac dysfunction, hypotension, hyponatraemia, renal failure and favour the relapse of tense ascites and death (36, 37). However, this fear has been shown to be unfounded when paracentesis was favourably compared with the standard diuretic therapy or with the peritoneo-venous shunt by randomized controlled trials (RCTs) (35, 38–40). Indeed, these studies showed that large-volume paracentesis combined with albumin i.v. infusion (6–8 g/L of ascites removed) in patients with tense ascites is more effective than diuretics and causes fewer complications.

In patients with refractory ascites, paracentesis needs to be frequently repeated because of the recurrence of tense ascites. The intervals between two consecutive taps are widely variable, probably according to the different rates of ascites formation, the compliance of the patient with dietary sodium restriction, the body habitus of the patient and overall his/her capacity to tolerate abdominal tension.

Complications of paracentesis

Despite the fact that all studies on paracentesis excluded patients with SBP, elevated serum creatinine, severe hepatic encephalopathy, severe thrombocytopaenia, low blood pressure or severe jaundice, there is no evidence that these clinical conditions should be considered as major contraindications for paracentesis in everyday clinical practice. Despite cirrhotic patients having coagulopathy and thrombocytopaenia, however, the incidence of clinically significant peritoneal bleeding during paracentesis was low (0.5%) in a retrospective study including >4500 procedures of paracentesis (41), and slightly higher in another investigation of 515 procedures (1% of cases) (42). Therefore, fresh-frozen plasma or platelet concentrates can be used in individual cases but they are not recommended as a standard procedure when experienced personnel performs the paracentesis (33, 43). The risk of bleeding might be higher in patients with renal failure (HRS type-2) and thus an extended post-paracentesis observation may be advised in these cases (44).

Leakage of ascitic fluid through the puncture site is a frequent complication, which should be managed by placing a purse-string suture around the opening and instructing the patient to lie with the puncture site uppermost. With the Z-tract technique (the skin is pulled 2 cm caudally before the needle is inserted and advanced), the leakage could be prevented (34, 45). The application of skin adhesive after paracentesis can also prevent such leaks.

The most common complication of large-volume paracentesis is effective hypovolaemia and renal impairment, an event called paracentesis-induced circulatory dysfunction (PICD) (46). To date, there are few studies identifying the mechanisms as well as the factors that predict the development of PICD. When the ascites is tapped, the intra-abdominal pressure rapidly falls, thereby improving the venous blood return to the right heart and a transient increase of cardiac output may occur (47). This enhances the hyperkinetic circulation causing shear stress of peripheral vessels. The consequence is a further decrease in effective arterial blood volume. This is documented by significant increases of the activity of the RAAS and the SNS (46–50). The occurrence of hyponatraemia (48) suggests that the paracentesis-induced decrease in effective volume may also stimulate vasopressin secretion with subsequent free water retention. The diagnosis of PICD is conventionally based on an increase in plasma renin activity (PRA) of >50% of the pretreatment value to above 4 ng/ml/h on the sixth day after paracentesis (50). Changes in PRA may also predict post-paracentesis hyponatraemia (51). The diagnostic value of other surrogate biomarkers (e.g. norepinephrine or vasopressin) has not yet been evaluated.

In a small series of patients treated with paracentesis not followed by any plasma expansion, the frequency of PICD was 75% (48). The incidence of PICD is thought to depend in part on the volume of the removed ascites; the greater the volume the higher the risk of developing PICD (48), but no sufficient evidence supports this opinion. Moreover, the occurrence of PICD has not yet been specifically studied in patients with refractory ascites treated with repeated paracentesis. However, it is plausible that these patients are at a high risk of developing PICD because generally they have large-volume paracentesis at each session and show low resting blood pressure values as a consequence of a much more deranged haemodynamic control.

It is important to note that PICD may persist for months (50) and may be associated with adverse clinical events such as an increased rate of recurrent ascites, the development of HRS and/or hyponatraemia in 20% of cases and reduced survival (50). These events likely result from the activation of the neurohumoural systems. Because of the risk of complications, many but not all experts believe it is necessary to use a treatment that prevents PICD in patients receiving large-volume paracentesis (52). Logically, the use of plasma expanders has been suggested to prevent PICD. Some trials, including small numbers of patients, showed a similar protective effect using human albumin solutions as well as less expensive plasma expanders such as polygeline (51) or dextran (53) or simple saline solution (54). However, a definite larger multicentre RCT showed the superiority of albumin compared with synthetic plasma expanders in preventing PICD (50). To be precise, albumin was better when a volume of more than 5 L of ascites was removed whereas this advantage was not evident when the volume removed was <5 L (50).

An analysis of data from all published studies also suggests that albumin is more effective in the prevention of hyponatraemia (8% of 482 patients) compared with other plasma expanders (17% of 344 patients) (55). Moreover, a recent double-blind pilot trial shows that the number of liver-related complications was significantly lower using albumin than using polygeline (56). These findings persuaded many to use intravenous albumin infusion to prevent PICD in patients who have a large volume of ascites removed (>5 L).

As vasodilation plays a crucial role in the development of PICD, it has been suggested that the administration of a vasoconstrictor (e.g. terlipressin or midodrine) might prevent PICD. A randomized pilot study compared the effects of terlipressin and albumin on PICD (57) showing similar levels of post-paracentesis PRA in patients treated with terlipressin and in those treated with albumin. However, this study was underpowered and thus larger trials are needed to confirm the effectiveness and safety of terlipressin. Two small studies compared midodrine with albumin to prevent PICD and gave conflicting results (58, 59). Thus, the efficacy of midodrine in these cases is not supported by sufficient evidence.

Contraindications to paracentesis

There are no absolute contraindications to perform paracentesis. However, the procedure should be avoided in patients with disseminated intravascular coagulation (41), and should be performed with caution in patients with intra-abdominal adhesions or with a distended urinary bladder. The use of ultrasound guidance may reduce the risk of iatrogenic injury in these cases.

Transjugular intrahepatic portosystemic shunt as a treatment of refractory ascites

A portocaval intrahepatic shunt effectively functions as a side-to-side portacaval shunt, and therefore, lowers the intrahepatic portal pressure (60). The introduction of an expandable flexible metal shunt prosthesis, originally designed for strictures of the biliary tract, has proven to be successful in maintaining the intrahepatic shunt patency (61). Since then, the metal prosthesis, or transjugular intrahepatic portosystemic shunt (TIPS), has enjoyed much popularity in the management of complications of portal hypertension. In suitably selected patients, TIPS is effective in the treatment of refractory ascites.

Although repeat large-volume paracentesis has been shown to be safe and effective, it is time consuming for both the physician and the patient, as ascites invariably recurs after each paracentesis. Furthermore, the presence of ascites predisposes the patient to the development of SBP, abdominal hernias and HRS. Therefore, the use of TIPS to manage refractory ascites has proven to be popular. Because one of the pathogenic factors for the development of sodium and water retention in cirrhosis is sinusoidal portal hypertension, it stands to reason that reduction of portal hypertension should result in an increased sodium excretion and an eventual elimination of ascites. In longitudinal follow-up of patients who received a TIPS, the increase in urine sodium output post-TIPS in the absence of diuretics is slow but steady, culminating in a urinary sodium excretion of approximately 100 mEq/day at 12 months post-TIPS (62). Therefore, in the post-TIPS period patients will need to stay on a low sodium diet or must take diuretic medications in order to facilitate the clearance of ascites.

To date, there are five published RCTs comparing the effects of TIPS vs large-volume paracentesis as a treatment for refractory ascites (63–67). Of a total of 330 patients enrolled, 162 patients received TIPS, while 168 underwent repeat large-volume paracentesis. TIPS was effective in about two-thirds of patients in eliminating ascites, and was superior to repeated large-volume paracentesis, which controlled ascites in only 23% of patients. A meta-analysis confirmed that TIPS is better than large-volume paracentesis in the control of ascites (68). The insertion of TIPS, however, worsens hepatic encephalopathy, but with long-term follow up, successful TIPS placement with elimination of ascites results in improved renal function (69), better nutritional status and positive nitrogen balance (70). This could translate into an improved quality of life. The meta-analysis also showed that TIPS is associated with a trend towards improved survival (68). In a further meta-analysis performed on the individual data obtained from the four larger RCTs, which provided their data (64–67), patients receiving TIPS showed a significantly better survival than patients receiving paracentesis (71), especially in the subgroup of patients younger and with minimal hepatic dysfunction. Indeed, not all these patients had refractory, but rather recidivant ascites, and the RCT that did not provide patient data (63) was that with the worse outcomes. Therefore, also considering the many contraindications to the implantation of TIPS, the superiority of TIPS is still controversial and paracentesis remains the first option in treating refractory ascites. However, in patients well-selected according to the risks discussed previously, TIPS can be used with a high probability to obtain a significant improvement of the patient's health. Moreover, all the above-mentioned trials used bare stents, and therefore, new data using covered stents are still required.

Criteria of selection of patients to TIPS treatment

Given the fact that TIPS can fail in eliminating ascites in some few patients, it is prudent to select patients appropriately for TIPS insertion to maximize its benefits. In addition, TIPS is associated with a unique set of complications, which must be thought of in every case when TIPS is being considered as a treatment for refractory ascites. The major complications of TIPS include:

  • a.TIPS stenosis is estimated to occur in up to 70% of patients with a bare TIPS within the first year (72). This is because of overgrowth of the endothelium within the stent, narrowing its lumen. TIPS stenosis can be treated by balloon dilatation of the stent, and if this is not successful, a new stent can be inserted. The advent of polytetrafluoroethylene (PTFE) covered stents has significantly reduced the incidence of TIPS stenosis, as they avoid bile leakage into the shunt lumen because of their decreased porosity, and also permit a more uniform endothelial growth owing to a smoother internal lining (73–75).
  • b.A rarer complication can be the dislocation of the stent or its migration to the right heart and the lungs, but this is an exceptional event when TIPS is placed by radiologists who are expert in the procedure.
  • c.The incidence of new or worsening hepatic encephalopathy after TIPS insertion is of the order of 20–30% (76, 77). In a meta-analysis of RCTs comparing TIPS vs large-volume paracentesis as a treatment for refractory ascites, the patients who received TIPS had an odds ratio of 2.26 for the development of hepatic encephalopathy after TIPS (68). Older age is a risk factor for post-TIPS encephalopathy, and an RCT did not support the benefit of prophylactic treatment of hepatic encephalopathy with lactulose or rifaximin (78). However, only in a very small proportion of patients encephalopathy is refractory to medical management. In such cases, occlusion or reduction of the shunt size is mandatory to control the encephalopathy (79).
  • d.TIPS can induce intravascular haemolysis. This is particularly evident with bare stents, occurring in approximately 10% of patients (80). Although haemolysis can be quite severe, in most cases it is self-limited and decreases as the stent becomes endothelialized in 12–15 weeks. Patients with severe haemolysis may require blood cell transfusions for support. With the advent of PTFE-covered stents, the incidence of intravascular haemolysis should decrease, but no data are yet available on this issue.
  • e.Other serious complications include the development of cardiac failure (2.5%), renal failure (4.3%) and liver failure (1.9%). The development of cardiac failure after TIPS may be related to the presence of an unrecognized cardiomyopathy. The development of renal failure post-TIPS may be related to the worsening of systemic arterial vasodilatation, not adequately compensated by refilling of the systemic circulation by the venous return from the splanchnic circulation, thereby leading to the continued activation of the vasoconstrictor systems with consequent renal vasoconstriction. The development of acute liver failure tends to occur in patients with advanced liver dysfunction pre-TIPS (Child–Pugh score>11 or MELD score>18) (63, 81), and this may be related to a liver ischaemia in the immediate post-TIPS period when most of the portal blood volume is directly shunted through the TIPS to the systemic circulation without passing through the rest of the liver.
  • f.Bacterial infection of the stent itself or endotipsitis is an uncommon complication, occurring in about 1.5% of patients, but it is a very serious event because it often cannot be successfully treated (82). Given all these potential complications, TIPS should not be performed in elderly patients with a history of recurrent hepatic encephalopathy. Likewise, patients with cardiac dysfunction or pulmonary hypertension should be carefully identified before TIPS is indicated, and patients with renal failure should also be carefully assessed. Although TIPS has been used as a treatment for HRS type-1 with an improvement of renal function post-TIPS, worsening of renal function post-TIPS has also been reported (83). Furthermore, all patients receiving a TIPS should be free of infections and should receive an antibiotic prophylaxis during the procedure. The presence of intrahepatic malignancy can also preclude TIPS insertion, as this can disseminate the tumour. Moreover, the patient should have a patent portal vein, and there should not be any structural abnormalities such as multiple hepatic cysts; otherwise, the insertion of TIPS may be technically unfeasible (84).

Role of vasopressin antagonists

Most cirrhotic patients with refractory ascites have hyponatraemia because of an increased release of vasopressin, a hormone that regulates the body water content by modulating the renal free water excretion (85) through specific receptors (V2 receptors) located in the principal cells of the collecting ducts of the kidney (86). Therefore, agents that antagonize V2 receptors (vaptans) increase free water excretion and serum sodium concentrations (87). Multicentre RCTs tested the effects of one of these agents (Satavaptan) in cirrhotic patients with ascites (88–89). A 14-day treatment with satavaptan, associated with low doses of diuretics, accelerated the mobilization of ascites mobilization in cirrhotic patients with no-refractory ascites (89). Based on this experience, it is conceivable that combining satavaptan with diuretics could be also beneficial in patients with refractory ascites reducing the frequency of large-volume paracentesis. This hypothesis, however, must be validated with RCTs, that include only patients with refractory ascites.

A rational clinical approach to the patient with refractory ascites

The presence of refractory ascites denotes the further progression of cirrhosis, and therefore, a worse prognosis than that of patients with cirrhosis and uncomplicated ascites (22).

Large-volume paracentesis combined with albumin i.v. infusion and TIPS are therapies that are useful in making patients comfortable and in delaying or preventing the recurrence of ascites but neither of them can induce a substantial improvement in survival (68). Therefore, all patients with refractory ascites should be evaluated for liver transplant candidacy without delay, even if we expect that some of these patients may have a good survival after TIPS, particularly alcoholics who become abstinent or patients with HBV-related hepatitis treated with effective antiviral drugs.

In the meantime, the following approach is recommended:

  • a.The first-line therapy of refractory ascites is large-volume paracentesis adding albumin if >5 L of fluid are removed. In patients in whom <5 L are being removed, another plasma volume expander or no volume expansion can be considered. To reduce the frequency of repeated paracentesis, patients may continue on maximally tolerated diuretic doses provided that their urinary sodium is >30 mEq/L. Otherwise, diuretics can be discontinued to reduce side effects (21, 22, 50, 55).
  • b.Although recent studies have shown a very low prevalence (0–3.5%) of SBP in the setting of routine outpatient therapeutic paracentesis (90), routine ascitic polymorphonuclear cells (PMNC) count is recommended with each paracentesis even in asymptomatic cirrhotic patients undergoing paracentesis for refractory ascites.
  • c.If a patient with refractory ascites is admitted or develops symptoms suggestive of SBP and/or develop renal dysfunction or hepatic encephalopathy, a diagnostic paracentesis and blood and ascitic fluid cultures should be performed. If ascites PMNC count is> 250/mm3, SBP is diagnosed and empirical antibiotic therapy should be undertaken (91). Evidence supports the use of either a third-generation cephalosporin (cefotaxime 2 g every 8 h or ceftriaxone 2 g/day) or amoxicillin/clavulanate (1 g tid). Therapy should be given for at least 5 days and, when feasible, paracentesis should be repeated after 48 h to document >25% decrease in PMNC count. If PMNC count is not decreased, antibiotic should be changed possibly according to the bacterial susceptibility in positive cultures. Patients with elevated bilirubin (>4 mg/dl) and evidence of acute renal dysfunction (defined as a BUN>30 mg/dl and creatinine>1.0 mg/dl) should receive adjunctive albumin at days 1 and 3 of acute SBP therapy as this was shown to prevent further renal deterioration and improve survival (92). The dose of albumin is 1.5 g/kg on day 1 and 1 g/kg at day 3 up to a maximum of 100 g/day.
  • d.The administration of long-term antibiotic prophylaxis with oral norfloxacin (400 mg q.d.) is indicated in:
  • Patients with a prior history of SBP (93)
  • Patients without a prior history of SBP but who have a low ascites protein (<1 g/dl) plus renal failure (creatinine>1.2 mg/dl or BUN>25 mg/dl) or a serum sodium ≤130 mEq/L or severe liver disease (Child score≥9 with serum bilirubin ≥3 mg/dl) (94).

These criteria will probably be met in many patients with refractory ascites.

  • e.TIPS should be considered in patients who require more than two large-volume paracentesis per month or when ascites is loculated and cannot be easily removed with a paracentesis, or when the patient becomes intolerant of repeated taps. However, elderly patients and patients with a Child–Pugh score >11 or a MELD score>18, and patients with elevated serum bilirubin have a high mortality post-TIPS and the main predictor of death after TIPS placement for refractory ascites is a high bilirubin level (95). Therefore, TIPS is not recommended in patients with a serum bilirubin >5 mg/dl or Child–Pugh score>11 or a MELD score>18 and/or an age>70 (84, 96). TIPS is also not recommended in patients with high risk of heart failure (cardiomegaly, low cardiac index or low ejection fraction), firstly because with a high right-sided pressures there may not be an adequate pressure gradient between the portal and the systemic venous systems for TIPS to function, and secondly because TIPS placement in this setting may precipitate overt heart failure. However, no precise heart-related criteria for excluding have been defined. TIPS is also contraindicated in patients with a significant pulmonary hypertension (>50 mmHg).

However, in patients with refractory ascites who are free of all contraindications, TIPS can be an extraordinary therapeutic resource that extinguishes ascites and improves both the survival and the quality of life. In patients who require frequent paracentesis and are not TIPS or transplant candidates, a peritoneo-venous shunt (PVS) could be considered.

In RCTs, PVS has been shown to be as effective as paracentesis+albumin and to have a similar rate of complications and a comparable survival (1, 34, 40). However, because of its high obstruction rate, PVS requires frequent admissions for shunt revision or for the management of other serious complications, such as infection. Moreover, placement of a PVS may hinder the future placement of TIPS and may complicate liver transplant surgery given its ability to produce peritoneal adhesions. Therefore, PVS has been almost completely abandoned even if it could be indicated in patients with refractory ascites who cannot receive TIPS or transplant.


The authors have no competing interests.

Financial support: The 2007 IAC/EASL join symposium was supported by an unconditional educational grant from Sanofi-Aventis.