Traditional views of the pathogenesis of hepatic encephalopathy (HE) have emphasised the role of neurotoxins arising from the gut. These compounds gain access to the systemic circulation as a result of the combined effects of hepatocellular failure and portal-systemic shunting. In the past years, a new mechanistic paradigm has been developed, emphasising changes in brain astrocyte function with subsequent alterations in glial-neuronal communications.1 ,2 Astrocyte dysfunction results from several factors. These include a decrease in plasma osmolarity, effects of circulating cytokines, and the changes brought on by accumulation of glutamine, the product of ammonia detoxification in glial cells. After many years of research, ammonia continues to play a central role in the pathogenesis of HE.

Under normal conditions, ammonia is generated in both the small bowel (from the effects of glutaminase on glutamine) and large intestine (from the urease activity of the colonic flora). The relative contribution of each site is difficult to quantitate but hyperammonaemic coma can still occur in hepatectomised germ free rats3 where the main source of ammonia is presumably the small bowel. Neomycin, an antibiotic active against urease containing bacteria in the colon, also inhibits small bowel glutaminase in the rat.4 Regardless of the origination site, portal vein ammonia levels are markedly elevated, fuelling the subsequent synthesis of urea and glutamine in the liver. An efficient hepatic extraction of ammonia, more than 0.8 in humans,5 contributes to the narrow range of ammonia levels seen in the periphery.

Several factors shape the anatomical and functional derangement of this pathway in cirrhosis with portal hypertension. Variables include alterations in the intestinal flora as a result of altered small bowel transit, extra- and intrahepatic portal-systemic shunts, as well as alterations in periportal (site of urea synthesis) and perivenous (glutamine synthesis) hepatocyte function. In the past decade, several studies have implicated gastric infection withHelicobacter pylori as an additional pathogenic factor in hyperammonaemia. The basis for this tenet is seemingly logical: the urease activity of this organism would generate enough gastric ammonia to increase portal levels and subsequent peripheral ammonia values with worsening of the course of HE. Critical to this line of reasoning is whether the density of the urease containing bacterial population in the colon, being several log-fold higher than the gastric counterpart in cases of H pylori infection, makes the contribution of the latter to ammoniagenesis quantitatively less important. However, qualitative differences in the urease activity of H pylori may be present, such as a marked increase in activity at a lower pH.6

Investigators have employed two frameworks to prove or refute a pathogenic role for H pylori in HE: an epidemiological analysis and an assessment of the effects of eradication. In the first approach, subjects with alcoholic hepatitis7 and cirrhosis8 were examined according to serology or antral biopsy results: HE was more evident in patients with H pylori infection. Other cross sectional studies have failed to confirm this finding.9 In the second approach, several groups noted a beneficial effect on ammonia levels and in clinical symptomatology after eradication of H pylori.8 ,10 Concomitant changes in the colonic bacterial flora brought on by antibiotics cloud the interpretation of such positive results. In a careful study examining cirrhotic patients with minimal encephalopathy, no improvement in clinical or biochemical parameters of HE could be seen two months after eradication.11 However, the extent of portal-systemic shunting and hepatocellular failure may be less prominent in these compensated patients. A preliminary analysis of patients with cirrhosis after placement of TIPS, a procedure that results in maximal portal-systemic shunting, indicated no increase in encephalopathy in those subjects infected with H pylori.12

In the current issue of Gut, a third approach is reported in an experimental animal: assessment of the effects of de novo infection with H pyloriin a model of cirrhosis (see page 605).13 H pylori grows in the gastric mucosa of the Mongolian gerbil, resulting in inflammation and even gastric cancer.14 The authors noted a significant increase in portal and peripheral ammonia levels coupled with decreased survival in infected gerbils made cirrhotic with a choline deficient diet. In their view, data support their previous results on H pylori eradication in humans,10 while providing additional evidence for a pathogenic role of this infection in HE.

The study is not without flaws. While hepatocellular function was similar in cirrhotic gerbils with and without infection, the extent of portal-systemic shunting was not described, a variable that could independently influence the results. The two end points of the study were also confusing: the increased death rate reported in infected cirrhotic gerbils was not well analysed and suggests additional pathology beyond HE, as the degree of hyperammonaemia was not as severe to explain a neurological death. Furthermore, ammonia levels, reported in this study from a peripheral vein, display less accuracy than arterial values in view of muscle extraction. Even if arterial levels were used, nocturnal eating habits and retained gastric contents can affect ammonia results in rodents.

Despite these criticisms, the authors deserve recognition for tackling the question in a paradigm that provides a fresh perspective of the problem. Behavioural tools have been used in rodent models to quantitate HE15 and should be the end point in future studies rather than the use of single measurements of ammonia blood levels. In the meantime, the practising clinician faces the therapeutic dilemma of whether to search for and treat H pylori in patients with cirrhosis. The sum of clinical evidence does not support a recommendation for therapy at this time. Additional studies along the lines of the current report and careful eradication studies in patients with advanced cirrhosis and portal-systemic shunting (the group most likely to be influenced by an increased ammonia load) should offer definitive answers on this controversy.