Ascites is the consequence of a complex set of interacting and incompletely understood neurohormonal and physiologic forces. Portal hypertension has long been recognized as a key factor, but increasingly appreciated is the role of locally produced nitric oxide, a potent vasodilator that acts on the splanchnic arterial bed lowering splanchnic artery pressure. As cirrhosis and portal hypertension advance, the increasing degree of splanchnic vasodilation is sufficient to lower systemic blood pressure, triggering sodium retention. The combination of portal hypertension, fluid retention, and splanchnic arterial vasodilation raises capillary transudation pressure in the intestinal vascular bed, leading to the accumulation of ascitic fluid.

Stages of ascites can be categorized according to degree of distension and response to diuretic therapy:

Hepatic encephalopathy is believed to be a consequence of intestinally derived toxic substances that escape hepatic detoxification as a result of portosystemic shunting and hepatocellular dysfunction. Candidates include ammonia, benzodiazepine-like substances, mercaptans, phenol, neuroinhibitors, false neurotransmitters, and short-chain fatty acids.

There are no pathognomonic clinical findings, and so the diagnosis requires ruling out other causes of altered mental status (e.g., medications, intracranial bleed, renal failure). Elevated ammonia levels correlate with the appearance of hepatic encephalopathy, but elevated levels are often seen in chronic liver disease in the absence of confusion and do not require treatment in the absence of encephalopathy. Venous levels are adequate for monitoring but can be falsely elevated when a tourniquet is left on too long at the time of blood drawing. Arterial sampling for the determination of the partial pressure of ammonia is more important in acute liver failure. Patients with encephalopathy are best assessed and followed by checking for asterixis and performing routine mental status examinations, for example, tracing a five-point star and signature testing, rather than serial ammonia levels.

This complication develops in about 10% of patients with ascites per year as bacteria spread from the bowel lumen to regional nodes, blood stream, and ascitic fluid. The most common pathogen is Escherichia coli. A polymorphonuclear cell count of greater than 250 cells/mL is considered diagnostic. Gram stain is often negative, as is culture of the ascitic fluid. Prognosis is poor with the onset of SBP, which also increases the risk of hepatorenal syndrome; the recurrence rate within 1 year is 70%.

Esophageal varices are a consequence of portal hypertension leading to portosystemic shunting. Bleeding from esophageal varices occurs in 20% to 30% of cirrhotic patients and has a poor prognosis. One third of patients with variceal bleed may die during the initial hospitalization, one third rebleed within 6 weeks, and one third survive for 1 year or more. The bleeding may be exacerbated by coagulopathy, resulting from reductions in vitamin K-dependent clotting factors (II, VII, IX, and X) secondary to decreased hepatic protein synthesis and increased plasma proteolytic activity. In addition, thrombocytopenia due to hypersplenism may worsen bleeding.

The hepatorenal syndrome occurs in the context of advanced liver disease and results from severe renal vasoconstriction, a consequence of systemic hypoperfusion. The condition is characterized by elevated serum creatinine, urinary sodium less than 10 mEq/L, the absence of urinary protein loss, and a hyperosmolar urine. Its onset heralds a poor prognosis (median survival <1 month for type 1), often precipitated by an attack of spontaneous bacterial peritonitis (SBP). Type 1 hepatorenal syndrome is associated with a rapid rise in serum creatinine and progressive oliguria; type 2 disease manifests a more indolent clinical course with modest elevations of serum creatinine often associated with diuretic refractory ascites and hyponatremia.

Patients with cirrhosis are at increased risk for hepatocellular carcinoma, especially those with underlying chronic hepatitis B and C and hemochromatosis. Up to 5% of patients with cirrhosis will develop hepatocellular carcinoma annually. Risk associated with alcoholic cirrhosis is less (1% at 5 years in a Danish population study). When detected at an early stage, hepatocellular cancer can be treated with curative intent by liver transplantation, surgical resection, or local ablative therapies. However, impact on survival is variable, dependent in part on the prognosis for the underlying liver disease. Abdominal ultrasound surveillance for hepatocellular cancer is recommended by some authorities for patients with cirrhosis; based on the doubling time of the tumor, it would need to be performed semiannually. Abdominal ultrasound is operator dependent and has a lower detection rate in the setting of ascites, underlying fatty liver, and obesity. In patients with suspicious or questionable ultrasound findings or a rising serum α-fetoprotein (AFP), contrast-based imaging such as CT or MRI can be performed. The serum a-fetoprotein level is commonly ordered as a screening test, but its sensitivity and specificity are low, and there are scant data to support its routine use. AFP as a confirmatory test may add some specificity to ultrasound imaging.

Prognosis is determined by the nature, severity, and activity of the underlying illness (Table 71-1). The traditional prognostic indicator, the Pugh modification of the Child-Turcotte prognostic classification, uses encephalopathy grade, degree of ascites, bilirubin, serum albumin, and PT. It has been supplanted for purposes of determining candidacy for liver transplantation by the better-validated model for end-stage liver disease (MELD), which predicts 3-month mortality and assigns a priority score for transplantation based on serum bilirubin, creatinine, and INR (Table 71-2).