Most people with stomachs colonised byHelicobacter pylori do not develop peptic ulceration, gastric adenocarcinoma or gastric lymphoma in their lifetime. Development of disease is likely to be determined by a combination of factors: the virulence of the infectingH pylori strain, the susceptibility of the host and environmental co-factors. The best studied of these disease determinants is strain virulence, and the two most studied virulence markers are expression of vacuolating cytotoxin activity and production of a non-toxigenic protein of unknown function called CagA.1 Colonisation with CagA+ strains ofH pylori is easy to detect as CagA provokes a strong systemic antibody response in the human host. In Europe and the USA, more than 60% of those colonised by H pylori have antibodies to CagA: these people have more intense gastric inflammation than those colonised by CagA−H pylori alone and are more likely to develop peptic ulcers and distal gastric adenocarcinoma.1However, most people colonised by CagA+ strains do not develop these conditions and it may be more useful to regard CagA− strains as having reduced pathogenicity than to regard CagA+ strains as pathogenic. In Japan, where Maeda et al’s study (see page 336) was conducted, not only is H pylori colonisation more prevalent than in the West, but also colonisation by CagA− strains is rare.2 ,3 Both of these factors may contribute to the high prevalence of H pylori associated disease in Japan.

Advances in H pylori genetics have led to a better understanding of why CagA+ strains cause more inflammation and disease than CagA− strains. The CagA protein is named after the gene encoding it,cytotoxin-associatedgene A(cagA). cagA is misleadingly named, in that it does not encode the cytotoxin (the cytotoxin gene is vacA) and its artificial disruption does not affect cytotoxin production or activity.4 ,5 However, mapping of theH pylori chromosome aroundcagA has shown that it is at one end of a group of about 30 genes which have a different nucleotide content to the rest of the chromosome. This has features of a pathogenicity island (PAI), a group of genes acquired by a bacterium relatively recently in its evolution which render it more pathogenic,6 and has been named the cag PAI.7 ,8 Thecag PAI, but notcagA itself, is thought to be important for the enhanced inflammation associated with CagA+ strains. The evidence for this is that individual artificial disruption of manycag PAI genes (although notcagA) reduces the ability ofH pylori to induce cultured epithelial cells to release the pro-inflammatory cytokine interleukin 8.7-9 Exactly what the cag PAI encodes is still unclear, but comparisons with similar genes in other bacteria suggest that it may be a secretion system for other proteins or a mechanism for interacting closely with epithelial cells.7-9 The cytotoxin gene,vacA, is not on thecag PAI and the reason for the close genetic association between certain types of vacAand the presence of the cag PAI is still unknown.1 ,10

The paper by Maeda et al assesses the extent to which production of the CagA protein or possession of thecagA gene is a marker for the presence of an intact cag PAI in Japanese strains. It confirms the high prevalence of cagA+ strains in Japan (100% in this study) and thus the low prevalence ofcagA− strains compared with Europe and the USA. Among the cagA+ strains, six of 63 did not produce the CagA protein, a similar proportion to that found in a European study.5 Two of these strains had an intactcag PAI, induced enhanced interleukin 8 release from epithelial cells and were isolated from patients with peptic ulcers, suggesting that they were fully pathogenic strains. The other four had sections of the cag PAI missing, induced low levels of interleukin 8 production and were isolated from patients without ulcers, suggesting that they had reduced pathogenicity. For one of these four strains with partialcag PAI deletions, the reason for lack ofcagA expression was sought: it was found that the missing section of the cag PAI included the promoter and 5′ end of cagA, rendering the gene non-functional. Although numbers are small, data on these strains are consistent with the hypothesis that enhanced pathogenicity is dependent on the presence of an intactcag PAI.

The authors further explored the mechanism underlyingcag PAI partial deletions by testing for the presence of insertion sequence (IS) 605, a stretch of DNA which may be present anywhere in theH pylori chromosome and is thought to be involved in rearranging the order of H pylori genes. It has been hypothesised that IS605 was acquired by H pylori later than the cag PAI in evolutionary time and is a prerequisite forcag deletions.7 Maedaet al found that two of the four strains with partial cag PAI deletions lacked IS605 and repeated the observation that some Western strains lacking the cag PAI possessed IS605. Although not disproving the original explanation of cag deletion, these findings suggest that deletion without IS605may be possible and further research is needed to clarify this. Particular interest surrounds whether cagdeletion is an ongoing phenomenon, perhaps allowing better co-adaption between H pylori and its host, or whether the deletion events occurred distantly in evolutionary time. Further characterisation of the strains in this study may help address this point, especially if serum samples from the hosts are also available.

The authors’ main conclusion is that neither thecagA gene nor the CagA protein are invariable markers of the intact cag PAI. From their data on Japanese strains, othercag genes may be better markers, although whether this will be the case in different populations remains to be determined. In practical terms, serological testing is needed for large studies and whether these other cag genes encode sufficiently immunogenic proteins for this to be possible is unknown. The current study does not assess the accuracy of serological testing for CagA, but it is reassuring that 57/59cag PAI+ strains were CagA+ (and so would be expected to provoke an antibody response) and that all four strains with cag PAI partial deletions were CagA− (and so would be expected not to provoke such a response). In the West, colonisation by CagA− strains, most of which lackcagA and the whole cagPAI, also produces CagA negative serological tests.5 ,7 Thus, although the case is not completely proved, a positive serological test for CagA is likely to be a reasonable marker for colonisation with an H pylori strain possessing an intactcag PAI.