One function of the liver is the synthesis of urea from various sources of ammonia, most of which come from protein-splitting bacteria in the GI tract. In cirrhosis, there is extensive liver cell destruction and fibrous tissue replacement of areas between nodules of irregularly regenerating liver cells. This architectural distortion also distorts the hepatic venous blood supply and leads to shunting into the systemic venous system, a phenomenon often manifested by esophageal varices. Thus, two conditions should exist for normal liver breakdown of ammonia: (1) enough functioning liver cells must be present and (2) enough ammonia must reach these liver cells. With normal hepatic blood flow, blood ammonia elevation occurs only in severe liver failure. With altered blood flow in cirrhosis, less severe decompensation is needed to produce elevated blood ammonia levels. Nevertheless, the blood ammonia is not directly dependent on the severity of cirrhosis but only on the presence of hepatic failure.

Hepatic failure produces a syndrome known as “prehepatic coma” (hepatic encephalopathy), which progresses to actual hepatic coma. Clinical symptoms of prehepatic coma include mental disturbances of various types, characteristic changes on the electroencephalogram, and a peculiar flapping intention tremor of the distal extremities. However, each element of this triad may be produced by other causes, and one or more may be lacking in some patients. The ensuing hepatic coma may also be simulated by the hyponatremia or hypokalemia that cirrhotic patients often manifest or by GI bleeding, among other causes. Cerebrospinal fluid glutamate levels are currently the most reliable indicator of hepatic encephalopathy. However, this requires spinal fluid, and in addition, the test is often not available except in large medical centers or reference laboratories. Of more readily available laboratory tests, the blood ammonia level shows the best correlation with hepatic encephalopathy or coma. However, the blood ammonia level is not elevated in all of these patients, so that a normal blood ammonia level does not rule out the diagnosis. Arterial ammonia levels are more reliable than venous ones since venous ammonia may increase to variable degree compared to arterial values. RBCs contain about 3 times the ammonium content of plasma, so that hemolysis may affect results. Muscular exertion can increase venous ammonia. Plasma is preferred to serum since ammonia can be generated during clotting. Patient cigarette smoking within 1 hour of venipuncture may produce significant elevation of ammonia. One investigator reported transient ammonia elevation at 0.5-3 hours and again at 3.5-6 hours after a meal containing protein in some normal persons, with the effect being magnified in persons with liver disease.

Blood ammonia has been proposed as an aid in the differential diagnosis of massive upper GI tract bleeding, since elevated values suggest severe liver disease and thus esophageal varices as the cause of the bleeding. However, since cirrhotics may also have acute gastritis or peptic ulcer, this use of the blood ammonia level has not been widely accepted. At present, the blood ammonia is used mainly as an aid in diagnosis of hepatic encephalopathy or coma, since elevated values suggest liver failure as the cause of the symptoms. Otherwise, ammonia determination is not a useful liver function test, since elevations usually do not occur until hepatic failure.