The term ascites is derived from the Greek word askitos (bladder, belly, bag) and denotes the presence of excessive fluid in the peritoneal cavity. It may present as an isolated clinical finding or be seen in the context of generalized oedema. Ascites is typically obvious: small amounts of fluid can be detected by diagnostic imaging.

Ascites may be caused by hepatic, renal, or cardiac decompensation; by hepatic venous outflow obstruction; and by malignant, inflammatory, or infectious processes in the peritoneum. Several different classification schemes have been proposed: one discriminating between transudative and exudative ascites has been adopted widely (Table 1) 378.

Ascites is not the only cause of marked abdominal swelling. Gross obesity, gaseous distension, visceromegaly or abnormal masses can all be confused with fluid excess. A massive ovarian cyst, hydatid cyst, and, exceptionally, pregnancy with polyhydramnios or pregnancy alone may be confused with ascites.

Ascites occurs when there is an imbalance of factors that favour the flow of fluid from the vascular space or when there is exudation of fluid through infection or malignant implantation on the peritoneum, and represents a state of excess total body sodium.

Normally, the higher pressure at the arteriolar end of the capillary allows passage of fluid without protein into the pericapillary space, while reabsorption takes place at the venous end of the capillary where hydrostatic pressure is lower than oncotic pressure. In cirrhosis, portal venous hypertension increases the filtration pressure at the capillary level with transudation of fluid, while low albumin levels decrease vascular oncotic pressure. Two potential mechanisms have been proposed to explain the pathogenesis of ascites. The ‘underfilling hypothesis’ suggests that the primary abnormality is a sequestration of fluid within the splanchnic vascular bed secondary to portal hypertension and reduced oncotic pressure. These events cause a decrease in effective circulating blood volume, which leads to compensatory renal retention of sodium by increasing aldosterone production.

The ‘overflow hypothesis’ suggests that renal dysfunction (perhaps mediated by a hepatorenal reflex) is the primary event leading to sodium and water accumulation in the absence of intravascular volume depletion. This hypothesis is supported by the fact that acute constriction of the portal vein in dogs is associated with renal vasoconstriction, renal sodium and water retention, and renin stimulation. These effects can be abolished in the homolateral, denervated kidney, but not in the contralateral innervated kidney. Not all data, however, substantiate the ‘overflow’ hypothesis. For example, patients with decompensated cirrhosis show an activation of the sympathetic nervous and renin–angiotensin–aldosterone systems and non-osmotic release of vasopressin, which should be suppressed, not stimulated, by renal sodium and water retention with volume expansion.

A new hypothesis that integrates the underfill and overfill hypothesis into one is that of peripheral arterial vasodilation. The underlying assumption of this hypothesis is that patients have peripheral vasodilation as a consequence of arteriovenous shunts in the splanchnic, dermal, and pulmonary circulations. The peripheral vasodilation that accompanies arteriovenous shunts causes renal sodium retention. Because sodium retention may not be sufficient to fill the enlarged vascular compartment, renin, angiotensin, aldosterone, and vasopressin are not suppressed.

Patients may initially not be aware of the presence of fluid within the abdomen. As the amount increases, the patient may become aware of distension, a sensation of fullness, or of tight-fitting clothes. Tense ascites may be associated with respiratory distress, anorexia, nausea, early satiety, pyrosis, ventral hernia, or abdominal pain. Body weight generally increases, but if the ascites is associated with alcoholism, poor nutrition, neoplasia or the wasting of end-stage chronic liver disease, weight may be stable or decreased. Some patients with tense ascites complain of numbness or paraesthesias with sensory loss in the distribution of the lateral cutaneous nerve of the thigh, presumably as a result of pressure where it emerges behind the inguinal ligament (meralgia paraesthetica).

Physical examination of patients with massive ascites reveals a tensely distended abdomen with tightly stretched skin, bulging flanks, and an everted umbilicus. Smaller amounts of fluid (in excess of 120 ml) may be detected by eliciting shifting abdominal dullness, a fluid wave, or periumbilical dullness with the patient on hands and knees; this last manoeuvre is known as a positive ‘puddle sign’. When massive ascites is present, it is not uncommon to find presacral, scrotal, and pedal oedema, as well as pleural effusions, especially on the right side. Ascites is usually accompanied by other signs of chronic liver disease, such as jaundice, spider angiomata, hepatic or splenic enlargement, or prominent abdominal wall veins (‘caput medusa’). Radiographic imaging examinations are helpful for confirming or extending impressions gained on physical examination. Ultrasonography and computed tomography can detect as little as 30 ml of ascitic fluid. Even routine plain radiographs of the abdomen may be helpful, demonstrating generalized haziness and loss of the psoas shadow, as well as centralization and separation of bowel loops by ascites.

Diagnostic paracentesis is a simple, relatively safe procedure that should be done in every patient with new onset ascites and can help determine the cause of ascites. Points of entry into the peritoneal cavity with the patient in the supine position include the left or right lower quadrant and the avascular linea alba below the umbilicus. Previous scar sites should be avoided. Regardless of the point of entry patients should empty their bladders before paracentesis.

After selection of a needle insertion site, the area is disinfected with iodine solution, and skin and subcutaneous tissues can be infiltrated with a local anaesthetic. The operator's gloved hand retracts the skin caudad in relation to the abdominal wall while a needle is inserted (needle size is dependent on thickness of the panniculus). The direction of the needle is changed three times, in ‘Z’ fashion, to minimize the likelihood of post-tap ascites leak. The character of the fluid (clear, straw-coloured, thick, odorous, cloudy, milky, bloody) should be noted, and laboratory analysis should be performed. Traditionally, the concentration of protein in the ascitic fluid and the ascites: serum lactate dehydrogenase ratio are used to determine whether the ascites is exudative (protein > 2.5 g/100 ml, ascites serum lactate dehydrogenase ratio > 0.6) or transudative (protein < 2.5 g/100 ml, ascites serum lactate dehydrogenase ratio < 0.6). Unfortunately, the distinction between exudative and transudative ascites is not absolute. Some patients with infectious and malignant ascites have a transudate, while patients with cirrhotic ascites may have an exudate. The serum:ascites albumin concentration gradient (a difference rather than a ratio) is felt by some authorities to be more accurate than the exudate/transudate distinction in classifying ascites. This method is based on the concept of oncotic–hydrostatic balance; patients with portal hypertension have a large difference (≥ 1.1 g/dl) between their serum and ascitic fluid albumin concentrations, while patients with ascites secondary to other processes have serum: ascites albumin gradients less than 1.1. For this method of classification to be accurate, however, the specimens must be obtained simultaneously (at least on the same day), the patient should be haemodynamically stable, and the specimens should be tested using the same assay system. The assay technique should also be accurate at low albumin ranges.

The serum: ascites albumin gradient provides information only about the presence or absence of portal hypertension, not about other causes of ascites. Patients with portal hypertension will have a high albumin gradient regardless of the presence of other superimposed conditions.

Ascitic fluid white blood and differential counts should be determined. An absolute count lower than 250 polymorphonuclear cells/mm³ usually denotes uninfected ascites, while counts of above this suggest infection. Ascitic fluid lactic dehydrogenase (compared to serum level), albumin, total protein, glucose and triglyceride (if chylous ascites is suspected) concentrations should be measured; lactate levels may also be obtained when peritonitis is suspected. Cytology and bacteriological analysis should also be performed. Enough fluid should be obtained to perform Gram stains, routine cultures and staining for acid-fast bacilli. Mycotic and mycobacterial cultures should be performed if indicated, and amylase levels should also be determined.

An ascitic fluid polymorphonuclear count above 250/mm³ has the accuracy for the diagnosis of spontaneous bacterial peritonitis. Although ascites pH and lactate levels have been proposed as helpful tests, they are not sufficiently discriminating to be of diagnostic value: ascitic pH values below 7.32 and lactate levels above 32 mg/dl are seen not only in spontaneous bacterial peritonitis, but also in malignant ascites, pancreatic ascites, and tuberculous peritonitis.

Chylous ascites refers to a milky or creamy appearance of peritoneal fluid resulting from the presence of thoracic or intestinal lymph. Chylous ascites contains Sudan-staining fat globules that can be detected microscopically and an increase in triglyceride level that can be documented by chemical examination.

Some patients may respond to simple measures, while others may require a series of therapeutic approaches that will be only palliative. Generally speaking, the more advanced the liver disease, the worse the prognosis, and, therefore, the less effective therapy will be.

As in any disease process, attention should be directed to treating the underlying cause. Attempts to improve liver function are worthwhile: patients should abstain from alcohol and receive optimal nutrition. Bedrest is usually recommended but is of little practical value, especially in patients whose urine sodium concentration is < 10 mEq/l. The theoretical benefit of bedrest derives from the fact that an upright posture activates the renin–angiotensin–aldosterone and &agr;-adrenergic systems; however, the effect of bedrest is negligible compared with that of conventional therapy. Sodium intake should be restricted to 1.5 to 2 g day; lower limits are rarely practical in patients unless they are admitted to hospital. If renal function is normal and hyponatraemia is absent, fluid restriction is not indicated.

In cirrhotic patients with ascites, prostaglandins are felt to be helpful in preserving renal function by maintaining glomerular filtration rate and free water clearance. Administration of drugs that inhibit prostaglandin synthetase, such as non-steroidal anti-inflammatory drugs, should be avoided in such patients.

Most patients require diuretic therapy. Dye dilution studies have demonstrated that, in the absence of peripheral oedema, only 0.5 kg per day of ascitic fluid can be mobilized without compromising the intravascular compartment; however, many authorities advocate an approximate limit of 1 kg (1 litre) a day for medical diuresis. The presence of peripheral oedema with ascites permits a more liberal mobilization of fluid with diuretics. In the current climate which favours large volume paracentesis removing several litres of fluid, the applicability of these medical limits needs to be re-evaluated; however, excessive diuresis may result in azotaemia, hyponatraemia, encephalopathy, and hepatorenal syndrome.

Spironolactone is an aldosterone antagonist that promotes natriuresis and potassium retention. The initial dose ranges from 50 to 200 mg/day given in two or four doses. A useful parameter to gauge the efficacy of natriuresis is the concentration of urinary sodium relative to potassium. It is practical to measure this in the first morning urine specimen, after the patient has abstained from sodium containing food for 8 to 12 h. If the sodium concentration of the urine remains lower than the potassium concentration, the diuresis is suboptimal and the dose of spironolactone should be increased by 50 to 100 mg every 2 to 3 days until urinary sodium exceeds urinary potassium. Theoretically, there is no absolute dose limit; however, hyperkalaemia and azotaemia usually prevent unlimited dose escalation. Most patients respond to a dose of less than 300 mg/day. Serum electrolytes should be monitored closely, depending on the responses to and duration of diuretic therapy. Gynaecomastia and galactorrhoea are recognized complications of spironolactone therapy and can be dose-limiting. If spironolactone causes hyperkalaemia, a potassium-wasting diuretic can be added.

Amiloride is another potassium sparing diuretic which, although not an aldosterone antagonist, has been used successfully to treat ascites. It increases renal excretion of sodium and chloride without significantly affecting the glomerular filtration rate.

These agents act by increasing the delivery of sodium to the distal tubule, where ability to reabsorb sodium is exceeded. The most common loop diuretic employed is frusemide and an initial dose of 20 mg/day may be sequentially increased, especially in patients with peripheral oedema. Loop diuretics should be used as supplements to spironolactone, not as the primary diuretic. Hypokalaemia may develop, despite administration of potassium-sparing diuretics, and hyponatraemia may also develop or worsen. Metabolic alkalosis and hepatic encephalopathy are other potential complications.

These are usually ineffective when given alone, and they may increase renal production of ammonia; they can, however, be used in conjunction with spironolactone either to reduce potassium retention or to provide a synergistic effect.

During the 1950s, before diuretics were available, large volume paracentesis was employed widely; this was, however, later implicated as a cause of hypotension, renal insufficiency, symptomatic hyponatraemia and encephalopathy, and the procedure was virtually abandoned for the next two decades. The procedure was resurrected in the 1980s by Rodés and his colleagues in Barcelona and by Kao and coworkers in the United States and was demonstrated to be safe and well tolerated, without haemodynamic, renal, or hepatic consequences. Numerous groups have since performed randomized studies comparing large volume paracentesis (with or without albumin infusion) with diuretic therapy and peritoneovenous shunting. Most studies have concluded that large volume paracentesis associated with intravenous albumin infusion is a fast, effective, and safe treatment for cirrhotic ascites. In general, patients undergoing paracentesis have a shorter hospital stay than those treated with diuretics. Recently, the Barcelona group has demonstrated the safety and clinical value of total volume paracentesis, performed over 1 to 2 h, with no effect on renal, hepatic, or hormonal features, provided that intravenous albumin is administered. Some groups have questioned the necessity for expensive salt-poor albumin infusion after paracentesis and have been treating patients quite successfully and safely without albumin infusions.

Paracentesis per se is quite simple. It should be performed in sterile conditions and under local anesthesia. The Barcelona group uses a needle (preferably a sharp metal needle within a 7-cm long, 17-gauge metal blunt-edged cannula with sideholes) that is inserted in the left lower abdominal quadrant. Once the needle has entered the peritoneal cavity, the inner part is removed and the fluid can be mobilized with the aid of a large volume capacity suction pump. After the procedure, patients remain recumbent for at least 2 h on the side opposite to the paracentesis to avoid leakage of fluid. Typically, after several days, diuretic therapy is resumed; paracentesis may be repeated when the clinical condition dictates—usually after several weeks but not uncommonly after a few days. In Barcelona, therapeutic paracentesis is ‘first-line’ therapy for patients with tense cirrhotic ascites; however, although quite widely used, large volume paracentesis has not been embraced by all centres. The need for rapid paracentesis is often not compelling, medical therapy is quite effective in most patients, the need to return for frequent procedures is difficult for many patients, and complications do occur. In addition, levels of opsonins have been shown to be reduced after paracentesis: this could theoretically predispose these patients to spontaneous bacterial peritonitis. Complications of paracentesis include infection, peritoneal dissection of fluid along fascial planes to the scrotum or pleural space, haemorrhage, perforation of the bowel with generalized peritonitis or abdominal abscess, and retention of catheter fragments in the abdominal cavity; however, these complications are rare in experienced hands.

A few clinicians resort to this modality when conservative measures have failed. Head out of water immersion decreases plasma renin activity, aldosterone, vasopressin, and noradrenaline as well as increasing the percentage of water load and urinary sodium excretion. Obviously, encephalopathic patients cannot be subjected to this approach.

In 1974, LeVeen and colleagues developed a pressure activated one-way valve for uses as a peritoneovenous shunt. One limb of the shunt lies free in the peritoneal cavity, and the venous opening of the efferent limb inserts into the superior vena cava near its entrance into the right atrium. Flow in the shunt is maintained if there is a 3 to 5 cm H&sub2;O pressure gradient between the valve and its venous end. If the gradient falls below this level, the valve closes, preventing blood from flowing back into the shunt tubing. Two additional shunts are available: the Denver Shunt and the Cordis-Hakim shunt, which incorporate a pumping mechanism at the abdominal end. Most patients, however, find ‘manual pumping’ of these shunts difficult, and they represent little improvement over the LeVeen shunt. Even with a peritoneovenous shunt, most patients require diuretics at a reduced dose. These shunts are also associated with concomitant improvement in renal blood flow.

A Veterans Administration Cooperative Study of LeVeen shunts diminished initial enthusiasm for this procedure. Over 3000 alcoholic patients with cirrhotic ascites were enrolled in this study; those refractory to diuretics were randomized to continued diuretic therapy plus therapeutic paracentesis versus LeVeen shunt placement. Patients randomized to LeVeen shunt therapy did not show improved survival, the number of infections increased, and the need for sodium restriction and diuretics continued, even in those with functioning shunts. Excessive shunt failure was also observed. A study published in 1989 compared medical treatment with peritoneovenous shunt therapy: the latter relieved the ascites promptly, reduced the length of hospital stay, and delayed recurrence of ascites. The duration of survival and number of complications in the medical and shunt groups were similar. This study suggests that peritoneovenous shunting is indicated in patients whose ascites is not treated satisfactorily by medical therapy, including those with recurrent ascites after multiple large volume paracentesis procedures and who have acceptable operative risk. If shunting is undertaken, current practice is to remove all or nearly all the ascitic fluid during the operation and to replace it with 5 l of isotonic sodium chloride solution.

Early complications of peritoneovenous shunts are valve malfunction, intravascular volume overload, pulmonary oedema, disseminated intravascular coagulation, hypokalaemia, air embolism, peritonitis, sepsis, and ascites leakage. Late complications are shunt occlusion (in about 30 per cent of patients), hepatic vein thrombosis, variceal haemorrhage, and peritoneal fibrosis with bowel obstruction. At least half of the patients will experience one or more complications. Mortality rates for these patients with end-stage cirrhosis and refractory ascites are quite high; about half succumb within the first postoperative year.

This procedure is contraindicated in patients with sepsis, heart failure, and active peritonitis. The best results are obtained in a limited subset of patients with diuretic-resistant ascites but relatively well preserved hepatic function.

Use of portasystemic shunts for treatment of cirrhotic ascites is warranted only in those patients in whom all other forms of therapy have failed. Many randomized trials have been done in the last 15 years comparing medical therapy with different types of portasystemic shunts, but most such studies were designed to evaluate shunts as therapy for bleeding oesophageal varices. The shunt operation that is most effective at relieving ascites has not been established; in fact, ascites is reduced after any type of portasystemic shunt as a consequence of decreased portal flow and decreased intrahepatic congestion. Among the most commonly performed shunts, splenorenal and portacaval shunts and their variants have been proven effective in relieving ascites. Ascites may actually increase transiently after distal splenorenal (Warren) shunt surgery, but, ultimately both distal and proximal (Linton) shunts lower the tendency to accumulate ascites. Between the two types of portacaval anastomosis, the end-to-side portacaval shunt decompresses the entire splanchnic bed but not hepatic sinusoids; after this shunt, ascites may persist. The side-to-side portacaval shunt, which decompresses both the splanchnic and hepatic sinusoidal beds, is very effective in relieving ascites; it is the decompressive shunt of choice for the treatment of ascites in the Budd–Chiari syndrome (when hepatic function is well-preserved).

The choice of portasystemic shunt should be individualized, taking into accounts factors such as the experience and preference of local surgeons.

Angiographically guided insertion of a metal stent between an intrahepatic hepatic vein and portal vein has been used to reduce portal hypertension in patients with bleeding oesophageal varices. Such transjugular intrahepatic portasystemic shunts (TIPS) have been shown to relieve intractable ascites as well. The role of TIPS in the management of refractory ascites—the object of current investigation—remains to be more full defined.

Although beyond the scope of this chapter, liver transplantation is the treatment of choice for patients with liver failure and otherwise intractable ascites, unless contraindications exist.

An important complication of ascites is spontaneous bacterial peritonitis, defined as an infection of ascites fluid in the absence of any obvious intraabdominal source.

Spontaneous bacterial peritonitis can occur in patients with ascites of any cause; the mechanisms underlying it remain unknown. Plausible contributing factors include transmural migration of bacteria secondary to increased permeability of the gut to enteric organisms, derangement in abdominal lymphatic circulation, haematogenous seeding from a distant source, decreased function of the reticuloendothelial system in patients with chronic liver disease, and impaired chemotaxis by blood monocytes and neutrophils in patients with cirrhosis.

A single organism is the cause of over 90 per cent of cases. The most common organisms are Escherichia coli and Streptococcus species; other organisms may also be implicated (Table 2) 379.

Prompt recognition of spontaneous bacterial peritonitis requires not only an awareness of the presenting symptoms and signs but also a high index of suspicion in any patient with ascites and clinical deterioration. Spontaneous bacterial peritonitis can vary in its presentation from being clinically dramatic to totally asymptomatic. Fever about 100°F is the most common presenting feature and occurs in 50 to 80 per cent of patients. Abdominal pain, usually diffuse, occurs in 25 to 72 per cent of patients; rebound tenderness is elicited in over 50 per cent of patients. None of these findings is specific for spontaneous bacterial peritonitis, however; abdominal complaints are common in cirrhotic patients with ascites in the absence of this complication, and spontaneous bacterial peritonitis can occur in the absence of fever or abdominal pain.

Other subtle indicators suggestive of spontaneous bacterial peritonitis include diarrhoea, worsening renal insufficiency, refractoriness to diuretics, hypothermia, and unexplained encephalopathy. Peripheral blood leucocytosis is common but may be masked by hypersplenism. Despite negative ascitic fluid cultures (which may occur in up to 50 per cent of patients), the presence of more than 250 polymorphonuclear cells per mm² in ascites suffices to establish a diagnosis and to mandate therapy.

Once the diagnosis is entertained, paracentesis is indicated. Ascitic fluid should be submitted promptly to the laboratory for aerobic and anaerobic cultures, cell count with differential, and Gram stain of centrifuged fluid, as well as other routine laboratory assessments. To increase the likelihood of a positive ascites fluid culture, the operator should inoculate 10 ml of ascitic fluid directly into blood culture bottles at the bedside. ‘Bedside’ inoculation increases the sensitivity of ascites culture and decreases the time between inoculation of the culture and appearance of bacterial growth.

The diagnosis of spontaneous bacterial peritonitis is based on a polymorphonuclear cell concentration of 250 cell/mm² or higher; ascitic pH below 7.3, a lactate level above 32 mg/100 ml. Glucose, specific gravity, and protein concentrations are not helpful. Most cases of peritonitis in cirrhotics are due to spontaneous bacterial peritonitis; therefore, the threshold for surgery, which can be fatal, should be very high. Conversely, peritonitis secondary to an intra-abdominal process such as a perforated ulcer or diverticulitis can occur in cirrhotics with ascites; failure to recognize such a source of infection and to intervene surgically and promptly is almost universally fatal. Hallmarks of ‘surgical’ peritonitis include polymorphonuclear cell counts above 10 000, polymicrobial peritonitis, low ascites glucose and high levels of lactate dehydrogenase and protein. A recent report suggested a diagnostic algorithm for patients with peritonitis: an ascitic fluid total protein above 1 g/dl, glucose below 50 mg/dl, and lactate dehydrogenase greater than the upper limit of normal for serum in the presence of polymorphonuclear cells provides evidence for non-spontaneous peritonitis (secondary to gut perforation).

Once a presumptive diagnosis of spontaneous bacterial peritonitis has been made, prompt and appropriate intravenous antibiotic therapy should be instituted, long before culture results are known, and even if cultures are negative. Empirically, a third-generation cephalosporin should be used, unless a Gram stain suggests that a narrower-spectrum antibiotic will suffice. Aminoglycosides should be avoided; they are associated with a high risk of nephrotoxicity and have unpredictable distribution in cirrhotics. When antibiotic sensitivities of the organism are known, antibiotic coverage can be narrowed. Direct intraperitoneal instillation of antibiotics is not necessary. Treatment should continue for 10 to 14 days, although, shorter duration therapy with third-generation cephalosporins is effective. Response to antibiotics can be judged clinically; repeat paracentesis in 48 h may be helpful but is not usually necessary. Once an episode is treated, the frequency of recurrence is high, and the mortality is also high, a reflection of the underlying severity of the liver disease in these patients. Prophylaxis with daily quinolone therapy has been shown to reduce the frequency of recurrent spontaneous bacterial peritonitis.

Progressive ‘functional’ renal failure occurs in patients with advanced liver disease, all of whom have decompensated cirrhosis, and most of whom have tense ascites. The kidneys are anatomically and histologically normal; if hepatic deterioration is reversed, for example by liver transplantation, the renal failure is also reversible.

Altered renal blood flow appears to be the primary abnormality. There is vasoconstriction of arterioles of the outer renal cortex with shunting of blood to the renal medulla, which results in decreased glomerular filtration rate and urine flow. These abnormalities are probably a result of decreased ‘effective’ blood volume and sympathetic overdrive. Accumulation of false neurotransmitters at nerve endings has also been proposed. Derangements in the renin–angiotensin and in the kallikrein–kinin system as well as altered synthesis of renal prostaglandins, vasoactive intestinal peptide, and endotoxins are also thought to be involved in the pathogenesis of the hepatorenal syndrome.

Progressive azotaemia, oliguria (<500 ml/day), a concentrated urine with a urine: plasma osmolarity above 1.0, and a urinary sodium concentration below 10 mEq/l in the presence of a normal urinalysis are the typical features (Table 3) 380. The urine may contain small amounts of protein, hyaline, and a few granular casts. Oliguria may occur spontaneously but usually follows diuretic therapy, diarrhoea, paracentesis, gastrointestinal haemorrhage, or sepsis. To establish this diagnosis, decreased intravascular volume must be excluded by fluid loading.

There is no effective therapy for hepatorenal syndrome; however, precipitating factors should be eliminated. Diuretics should be withheld, blood volume replaced, serum electrolyte abnormalities corrected, infections treated promptly, and drugs known to inhibit prostaglandin synthesis as well as other nephrotoxic drugs discontinued.

A fluid challenge to increase effective plasma volume should be attempted; a combination of saline and salt-poor albumin should be administered while close monitoring, preferably including central venous pressure measurement, is undertaken. Numerous vasodilatory drugs (e.g. phentolamine, papaverine, metaraminol, phenoxybenzamine) have also been administered but have not been effective.

If the liver disease is reversible (for example, fulminant hepatitis), dialysis may be helpful, but dialysis is of no value in patients with end-stage chronic liver disease. Recovery from hepatorenal syndrome has been reported anecdotally following portasystemic shunts and liver transplantation. Nevertheless, most patients with this disorder are often too decompensated to be candidates for these surgical interventions. Occasionally peritoneovenous shunting reverses the hepatorenal syndrome.

Akriviadis EA, Runyon BA. Utility of an alogrithm in differentiating spontaneous from secondary peritonitis. Gastroenterology 1990; 98: 127–33.

Bories P, et al. The treatment of refractory ascites by the LeVeen shunt; a multicenter controlled trial of 57 patients. J. Hepatol 1986; 3: 212–18.

Conn HO. Transjugular intrahepatic portal-systemic shunts: the state of the art. Hepatology 1993; 17: 148–58.

Epstein M. Derangements of renal water handling in liver disease. Gastroenterology 1985; 89: 1415–25.

Epstein M. Hepatorenal syndrome. In: Epstein M, ed. The Kidney in Liver Disease. 3rd edn. Baltimore: Williams and Wilkins, 1988: 92.

Gines P, et al. Comparison of paracentesis and diuretics in the treatment of cirrhotics with tense ascites. Gastroenterology 1987; 93: 234–41.

Gines P, et al. Randomized comparative study of therapeutic paracentesis with and without intravenous albumin in cirrhosis. Gastroenterology 1988; 94: 1493–1502.

Kao HW, Rakov NE, Savage E, Reynolds TB. The effect of large volume paracentesis in plasma volume, a cause of hypovolemia? Hepatology 1985; 5: 403–7.

Malt RA. Portasystemic shunts. N Engl J Med 1976; 295: 24–8 and 80–5.

Nicholls KM, Shapiro MD, Kluge R, Hsaio-Min C, Bichet DG, Schrier RW. Sodium excretion in advanced cirrhosis: effect of expansion of central blood volume and suppression of plasma aldosterone. Hepatology 1986; 6: 235–8.

Pinto PL, Amerian J, Reynolds TB. Large volume paracentesis in nonedematous patients with tense ascites: its effect in intravascular volume. Hepatology 1988; 8: 207–10.

Quintero E, et al. Paracentesis versus diuretics in the treatment of cirrhotics with tense ascites. Lancet 1985; i: 611–2.

Ring-Larsen, Henricksen JM. Pathogenesis of ascites formation and hepatorenal syndrome: humoral and hemodynamic factors. Semin Liver Dis 1986; 6: 341–52.

Runyon BA, Canawati HN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology 1988; 95: 1351–5.

Runyon BA, Antillos RM, Montana AA. Effects of paracentesis on ascites fluid opsonic activity and serum complement. Gastroenterology 1989; 97: 158–62.

Schrier RW. Pathogenesis of sodium and water retention in high-output and low-output cardiac failure, nephrotic syndrome, cirrhosis and pregnancy. N Engl J Med 1988; 319: 1065–72 and 1127–33.

Shear L, Ching S, Gabuzda GI. Compartmentalization of ascites and edema in patients with hepatic cirrhosis. N Engl J Med 1970; 282: 391–96.

Smajda C, Franco D. The LeVeen shunt in the elective treatment of intractable ascites in cirrhosis. Ann Surg 1985; 201: 488–93.

Stanley MM, et al. Peritoneovenous shunting as compared with medical treatment in patients with alcoholic cirrhosis and massive ascites. N Engl J Med 1989; 321: 1632–38.

Tito L, et al. Total paracentesis associated with intravenous albumin in management of patients with cirrhosis and ascites. Gastroenterology 1990; 98: 146–51.