SLIDES & TRANSCRIPTS
Tuesday, March 6

Premalignant Lesions: H pylori and Gastric Cancer
Steven F. Moss, MD

Just to go very briefly over the organism, helicobacter pylori is a gram negative bacterium which lives within the lumen of the stomach, closely adherent to normal gastric epithelial cells. It is acquired in childhood, probably under the age of three years, and it persists lifelong, unless the patient is given specific antibiotic therapy.

It is said to be the commonest bacterial infection in the world. In most of the developing world, where gastric cancer is really a major problem, infection is almost universal. H. pylori is mostly, however, asymptomatic. Although it does cause an intense inflammatory reaction, this is not thought to cause symptoms in most people.

So, it has a very wide disease spectrum. On the one hand, it causes a very brisk histological gastroenteritis, as we have seen. On the other hand, only a minority of individuals infected are thought to suffer adverse effects. Maybe one in 10 patients will get a peptic ulcer over the course of their lifetime. Maybe up to two percent of individuals infected will get a malignancy, either a lymphoma or a carcinoma, over many decades. Based on the evidence which was outlined this morning, the WHO classified h. pylori as a definite carcinogen in 1994. I want to just very briefly go through the evidence behind that association.

TOP

There are really three strands of evidence pointing to a link between h. pylori and gastric cancer. The first is the pathological association, that h. pylori is commonly found in stomachs which have gastric cancer or the precursor lesions of gastric cancer. The second is, in general, geographical linkage between countries or populations with high h. pylori seroprevalence and high gastric cancer incidence. This is by no means perfect, but is a nevertheless reasonably good, significant correlation. Really, the strongest evidence were the data reviewed by Dr. Correa this morning, in the prospective seroepidemiological studies,

TOP

which have clearly shown that h. pylori is a risk factor for the development of distal gastric cancer. These studies, which came out about 10 years ago, looked at three different cohorts in the United Kingdom and the United States, followed for a decade or so. You can see that the matched odds ratio was somewhere between two and six. In other words, individuals who had serum banked and were followed for a decade had a two to six-fold increased risk of developing gastric cancer.

TOP

In 1998, there was a meta analysis done of 19 case controlled cohort studies, which reviewed 2,500 patients and nearly 4,000 controls.
In the overview, the odds ratio for developing gastric cancer if you are h. pylori positive was around two, with confidence intervals as shown here. When they broke out the data according to the site within the stomach where the carcinomas were, you can see that the odds ratio was close to three for non-cardia or distal gastric cancers, but was not significantly increased for cardia cancers. I am going to come back to that point later.

Rather interestingly, they found that the odds ratio was really similar, whether they looked at intestinal or diffuse histological subtypes, which is maybe a little surprising, based on what we know about h. pylori being found in parts of the world where intestinal cancer is really the common one.

As has been mentioned, the odds ratio is particularly increased in patients who had either early gastric cancer, or who presented with cancer at a young age. This may be an effect of the fact that the matched control for age had a relatively low incidence of gastric cancer. It has been calculated that h. pylori is responsible for something close to two thirds of all gastric cancer worldwide. As has also been mentioned this morning, the hard evidence that h. pylori could directly produce cancer came from this animal model.

TOP

Two groups from Japan published about the same time two years ago that they could directly orally inoculate toxigenic strains of h. pylori into the Mongolian gerbil, in the absence of any co-carcinogen, and produce cancers.
They went through a stage where, over six months, they had an antral predominant gastritis with lymphoid follicles, very similar to what we see in human infection. Then, over the following year, the gastritis moved away from the antrum up toward the proximal stomach. They developed a pan-gastritis, epithelial changes, some gastric ulcers and ultimately preneoplastic lesions and cancers in about 40 percent of these animals by 18 months. Interestingly, these were the intestinal type of cancers with p53 mutations, although no metastases have been reported in these animals.

TOP

This is an image scanned up from the cancer research paper, which shows that these cancers look pretty much like human intestinal-type gastric cancers. So, this may be a good model for gastric cancer. There are some problems with the model which we can maybe discuss later. One group has gone on to ask the question, can you eradicate h. pylori from these animals that are infected, and reduce the incidence of the development of gastric cancer.

In this paper, they actually used h. pylori, together with a nitrosating carcinogen, MNU. They gave both h. pylori and MNU in the first couple of months to the gerbils, and then eradicated h. pylori at 21 weeks. They found that the tumor incidence went down drastically to about 25 percent of the animals in whom h. pylori was eradicated. Interestingly, all the animals did not develop intestinal metaplasia, also, once they eradicated h. pylori. So, here was some evidence from an animal model that perhaps some of these preneoplastic changes are reversible.

TOP

The big question, I think one that has been alluded to this morning, is why only some individuals, and really a minority of individuals affected by h. pylori, develop carcinoma. I am going to run through three possible explanations, the first one being strain variation. This is h. pylori strain variation.

TOP

We have publicly accessible now two complete h. pylori genomes. Just by comparison, of the two genomes, it is clear that h. pylori, as has been known for a while, is an extremely heterogeneous bacterium. The two genomes differ by about 10 to 15 percent. A group at Stanford, Stan Falco and his group, have developed a DNA micro array to look at whole genome analysis of h. pylori.
In a limited number of strains that they have looked at, it is clear that only about 60 or 70 percent of h. pylori's genome is common to all strains.

TOP

So, what are the virulence-associated factors of h. pylori. There are a number of putative virulence-associated genes, most of which actually have been identified in the pre-h. pylori genome era. The one at the bottom is a kind of post-genome finding. I am going to just spend some time talking about the CAG pathogenicity island. This is really the h. pylori virulence factor which has been most extensively investigated.

TOP

Around 10 years ago, it was found that certain individuals who had a higher incidence of gastric cancer and duodenal ulcer disease had antibodies in their serum against what turned out to be called the CAG-A gene products, the cytotoxic associated gene product.
CAG-A is a highly immunodominant antigen expressed on about half of all h. pylori strains worldwide. It is a good marker for what is called the CAG pathogenicity island, which is about a 40 kilobase region, containing about 30 open reading frames. The world of h. pylori can be divided into those strains which carry the CAG pathogenicity island, and those which don't. CAG-A is a marker for the presence of the island. Interestingly, the GC content of this CAG pathogenicity island is quite different from the rest of h. pylori's genome, implying that h. pylori probably acquired this from some other bacterium way back in evolution.

TOP

The function of the genes within the CAG pathogenicity island is now beginning at last to be understood. It is clear that many of the genes within CAG pathogenicity island encode for what is called a type IV secretory apparatus, which enables h. pylori to insert its own products into host epithelial cells. This is rather analogous to the type III secretory system which has been worked out in the last couple of years for enteropathogenic e. coli.

So, for EPEC, it translocates its intemin receptor into the intestinal epithelial cell and then it allows the intemin on EPEC to be attached. This is a virulence factor for EPEC. H. pylori has evolved a similar kind of strategy. In fact, one of the translocated products is the CAG-A gene product. This is translocated into gastric epithelial cells, where it is then phosphorylated and has actions in the gastric epithelial cell. So, h. pylori can signal two gastric epithelial cells. One of the effects of this signalling is actin polymerization, which alters the motility and the migration of the gastric epithelial cells, and enables an attachment pedestal to be pushed out by the epithelial cell, allowing h. pylori to be in much closer contact with it.

TOP

Probably other genes within the pathogenicity island, or elsewhere in h. pylori's genome, are also capable of being translocated within the host epithelial cell, and a number of signalling pathways have been recently identified which are activated by h. pylori attachment, including MAT kinase cascades and f. capa B activation. This seems to be critically important in, for example, IL-8 transcription in the epithelial cell. So, you have cross talk here between the epithelial cell and the bacterium. It is no simply a bystander.

TOP

So, we talked somewhat about strain variation. Now I want to briefly go over differences in susceptibility related to the topographical distribution of h. pylori within the stomach.

TOP

This really follows on from what Dr. Correa was saying early this morning. There are at least two patterns of gastritis and h. pylori-associated gastritis, in patients infected with this organism. In patients with duodenal ulcer, as we have heard, the gastritis is predominantly antral. The rest of the stomach is unaffected. So, the parietal cell secretions are normal or even increased. That is why, in duodenal ulcer patients, they have high acid secretion. As we have heard, this patients actually have a lower-than-normal risk of gastric cancer. In contrast, people who go on to develop a chronic multifocal gastritis, their proximal stomach is also infected by h. pylori and the associated inflammatory response. They tend to have low acid secretion. These are the people at higher risk of gastric cancer and the premalignant changes that accompany or precede gastric cancer.

TOP

Now, the third line of investigation is why do only some people affected by h. pylori develop cancer. This may be related to genetically determined host responses to infection. This has been shown in a number of mouse models of infection in the last few years. Certain strains of mice, if you infect them with h. pylori, they get a very intense inflammatory response, a reactive epithelial response.
Other strains of mice, although they have equal numbers of bacteria in their stomach, have a much lower pro-inflammatory response. There have been really a very limited amount of studies looking at the same kind of phenomena in human infection.

TOP

I think there was an important paper published last year in Nature that came out of a group in Glasgow, Scotland and collaborators here at the NCI, that looked at IL-1 polymorphisms and the risk for gastric cancer after infection by h. pylori. The reason why they were interested in IL-1 polymorphisms is because it is known that IL-1 is a potent inhibitor of acid secretion and, on a molar basis, is much more potent even than the proton pump inhibitors.

Secondly, IL-1 is known to be increased in the mucosa of patients infected by h. pylori. Thirdly, it is known that there are genetic polymorphisms for IL-1 in the human population. So, to study this, they looked at two cohorts, first of all, a Scottish cohort with 100 first degree relatives of gastric cancer patients. These gastric cancer patients could be subdivided physiologically into those with gastric atrophy and a low acid secretion, and half of them had normal acid secretion and no atrophy. So, they represented two different sort of geographical distributions and different physiological outcomes to h. pylori infection.

They had 100 population controls, so they had a background knowledge of the IL-1 genotype. They also looked at 400 gastric cancer patients from Poland and a similar number of controls. In all these patients, they did IL-1 genotyping by PCR and they established an in vitro inflammatory assay, so they could determine whether particular IL-1 genotypes had a relationship with inflammation.

TOP

What they found was that the IL-1 genotype of the individual did not influence the risk of h. pylori infection, in other words, whether they were h. pylori positive or not. However, the IL-1 genotype, the pro-inflammatory IL-1 genotype, was associated with low acid secretion in first degree relatives, and gastric cancer in the cases. So, the model here, from this study, is that h. pylori infection, in certain individuals who have a pro-inflammatory IL-1 polymorphism, leads to a situation of low acid secretion and subsequently increased cancer risk.

TOP

I think this is an interesting study which tended to correlate physiology with genetics, and outcome following infection with h. pylori. What are the mechanisms by which h. pylori can promote cancer? There is no shortage of putative mechanisms. It may, for example, have a direct effect on the epithelial cell. We have seen it certainly has the capacity to pass genes into epithelial cells, although as far as we are aware, there is no evidence that h. pylori DNA integrates into the host genome. It may cause a non-specific damage or mutagenesis. There are some studies showing increased DNA adducts, for example, in the gastric mucosa of patients infected by h. pylori. It may alter the normal cell cycle of the gastric epithelium, and I will show you some data for that in a moment. Clearly, as we have heard this morning, the inflammatory response to h. pylori may also be very important in promoting transformation in the gastric mucosa. We heard something about reactive species and also certain cytokines may have pro-carcinogenic effects.

\There are many indirect mechanisms, by which the presence of h. pylori in the stomach alters normal gastric physiology in such a way as to promote carcinogenesis. It tends to deplete vitamin C, both within the lumen and the gastric mucosa itself. As we have heard, certain individuals develop hypochlorhydria after long-term h. pylori infection, which causes the build-up of nitrosyl compounds in the stomach related to overgrowth of other bacteria. Finally, h. pylori alters growth factors, both mucosal growth factors -- EGA, TGF alpha -- and also circulating growth factors. For example, serum gastrin concentrations are elevated in h. pylori infection. In some animal models, gastrin is clearly a cofactor in causing mucosal growth in the colon and stomach and elsewhere.

TOP

H. pylori is also associated not only with increased proliferation but also with increased cell death by apoptosis. Here, for example, a group of patients with duodenal ulcer disease, if you eradicate h. pylori from these individuals, the percentage of apoptotic epithelial cells goes down dramatically. Many studies have now shown that the number of apoptotic cells were also increased in the presence of h. pylori.

TOP

So, we have increased proliferation and increased apoptosis. It is questionable which is the chicken and which is the egg. Is it the cell death caused by h. pylori which stimulates a hyperproliferative response, or does h. pylori primarily stimulate proliferation and, out of uncontrolled proliferation you get apoptosis as a defensive reaction?
Certainly, in cell cultures, apoptosis seems to be the predominant effect that we see, much more than proliferation, but this may be due to some of the artifacts, some of the limitations of cell culture studies.

Obviously, it is impossible to do long-term natural history studies in humans, but you can do them in animals. We have been involved in some studies of the Mongolian gerbils. The gerbils, they get a peak in apoptosis about four weeks or so after experimental inoculation, and then it decreases. Then proliferation comes in a couple of months later. So, for the animal models, it would seem like apoptosis is the principal event, and the hyperproliferation may be a response. It is interesting to speculate whether, in fact, apoptosis goes right down later in life in the gerbil, as does proliferation. Perhaps there is an adaptive response in the stomach, where you get reduced apoptosis in time, and this may be relevant to carcinogenesis. Certainly in cell culture we can show that you can challenge gastric epithelial cells with h. pylori. Out of the initially dying cells come the selection of derivatives which are apoptosis resistant. What are the molecular changes involved? We don't know. P27 may be relevant.

TOP

Now, to return to a couple of clinical issues. It is the hope and belief of most gastroenterologists that, if you eradicate h. pylori, you will reduce the risk of gastric cancer. Certainly, if you eradicate h. pylori, the inflammation resolves, although this takes months and up to two years before the inflammation goes completely away. Is this associated with a decreased cancer risk? There was one study published two or three years ago from Japan -- Yumora was the first author -- in which they selected patients who had early gastric cancer.

The cancer was removed by endoscopic mucosal resection. These patients are known to have a fairly high incidence of a second early gastric cancer, so they were followed closely. The patients were randomized after their first early gastric cancer into eradication versus no eradication. From this study they found that eradication of h. pylori after the first gastric cancer reduced the incidence of the second early gastric cancer. This study has been criticized on the basis of an extremely short follow-up and questions about the blinding. As far as I know, no one else has reproduced this. Dr. Correa may know further information about this.

Now, there are some other small limited studies which show that if you follow endoscopic cohorts of patients and eradicate h. pylori, their mucosa improves in terms of these preneoplastic lesions, such as intestinal metaplasia. That is really a problem with small studies which don't take enough biopsies. You may have tremendous sampling variation, which accounts for some of these supposedly positive results.

Clearly, what we need are large, randomized prospective studies, to look at populations of patients randomized into h. pylori eradication versus no eradication, and to look really at the primary outcome being gastric cancer incidence. Now, those kinds of studies have been started, mainly in the far east. We don't have any data about gastric cancer incidence yet.

TOP

We are starting to see the first generation of studies reporting back on changes in preneoplastic lesions.
This is the study which Dr. Correa published a couple of months ago in JNCI. Probably it would be better for him to explain further details, but I just go through this briefly. This was, I think, an important study. He had a group of patients in southeast Columbia who were at high risk of gastric cancer, over 800 patients.

At entry into the study, they had atrophic gastritis, and many of them also had intestinal metaplasia. These patients were then randomized into h. pylori eradication. Some of them also had beta carotene. Some of them also had ascorbic acid, in a two-by-three placebo-controlled factorial design. So, some individuals got all three interventions and some got none, and some got in between. These patients were then followed up with annual carbon 13 urea breath tests, to determine whether h. pylori had been eradicated, and biopsied up to six years, using four gastric biopsies at each time the patients were endoscoped. This is kind of the minimum number of biopsies you really need to do for this kind of study, because of the tremendous sampling variation. This was a nicely blinded study. The way that the data was analyzed was for individual patients, whether they had a progression of their lesion or a regression. For example, if a patient had atrophic gastritis at entry into the study, six years later had dysplasia, this patient would be classified as having progression.

If, on the other hand, at six years they had a normal-looking mucosa, this would be regression. The placebo group had an overall 26 percent rate of progression, worsening histology over six years, and a seven percent regression. So, the overall progression was something like 20 percent. The calculated odds ratio for regression, in comparison to the placebo arm, for each of these interventions, was greater than one. So, each of these interventions had a positive effect in reducing the progression down the gastric precancerous pathway. So, h. pylori or beta carotene or ascorbic acid odds ratio was around five in patients who initially had atrophy and around three in patients who initially had atrophy plus intestinal metaplasia. If they did a subgroup analysis, looking at only those patients in whom h. pylori had been successfully eradicated, which was about 75 percent of those who were given h. pylori eradication therapy, then these numbers were increased.

Some interesting things came out of the study. There appeared to be no synergy between h. pylori eradication and either of these antioxidants. Fifteen percent of the patients, over six years, had a spontaneous apparent eradication of h. pylori.

TOP

So, is there a down side to eradicating h. pylori from populations at risk for gastric cancer? As we know, over the course of this century, in the developing world, the incidence of distal gastric cancer has come down quite steadily, coincident with a natural decline in h. pylori incidence in this country, even prior to people being given antibiotic therapy for h. pylori.
As you know, this has coincided with an increase in proximal gastric or GE junction cancers.
So, one can speculate that they have something to do with each other. Of course, someone pointed out that this increase is also associated with the increased number of people undergoing endoscopy, which is probably fair enough. Whether the decline in h. pylori is really causative remains an open question, but I want to show you a couple of pieces of speculative data.

TOP

It is known that there were a few reports that came out two or three years ago, looking at patients with duodenal ulcer disease, after they irradiated h. pylori from patients with duodenal ulcer disease. A group of the patients developed gastro-esophageal reflux disease. Then if you look at cross sectional studies at h. pylori incidence in different populations, you tend to find that in patients with esophagitis and the complications of esophagitis, they have a lower incidence of h. pylori compared with endoscopic controls. Furthermore, this trend to a decline is even stronger if you look at only h. pylori CAG-A positive strains. So, CAG-A positive strains, we think, are bad players for distal gastric cancer, but appear to protect you, at least on this kind of soft epidemiology, in terms of proximal gastric cancer and GE junction cancer.
Now, there is some supportive data to suggest this might not be as ridiculous as you might first think. This was a substudy from an NIH multi-center retrospective case controlled study of patients who had GE junction cancer.

TOP

The cases were obviously the cancer patients. The controls were the spouses. If you looked in this population, the incidence of h. pylori appeared to be a risk factor for non-cardia cancer, which of course, is what you would expect, but would appear to protect or was negatively associated with a GE junction cancer, with a p value that was close to statistical significance.
If you then subdivided the h. pylori positive individuals according to CAG-A positive or CAG-A negative strains, you can see that there appeared, in this study at least, to be a fairly strong negative correlation between h. pylori CAG-A positive seroprevalence and GE junction cancer.

TOP

So, in conclusion, I think that h. pylori clearly is an important etiological factor in distal gastric cancer, probably the main statistically important factor in distal gastric cancer. It may be protective against proximal gastric cancer. It is unclear how h. pylori promotes carcinogenesis, but I have talked about some of the ways in which it may do so. Animal models may be helpful in unraveling this association and also testing whether h. pylori eradication will reduce the incidence of gastric cancer, while we wait for the human studies to come back. It remains a very open question why only a minority of individuals go on to develop gastric cancer after infection with h. pylori. Although we intuitively think that eradication should lower the risk of distal gastric cancer, we don't have too much evidence at the moment to support that. We have to think about a possible cost, a detrimental cost, in terms of eradicating h. pylori in certain populations. Thanks very much.

TOP