Review article: exploring the link between Helicobacter pylori and gastric cancer

Aliment Pharmacol Ther 1999; 13 (Suppl. 1), 3±11 Review article: exploring the link between Helicobacter pylori and gastric cancer E. J. KUIPERS Free University Hospital, Amsterdam, The Netherlands SUMMARY Cancer of the distal stomach, both of the intestinal and diffuse type, is strongly associated with Helicobacter pylori colonization. This bacterium causes chronic active inflammation of the gastric mucosa in the majority of colonized subjects. In a considerable number of them, this will eventually lead to a loss of gastric glands, and thus the establishment of atrophic gastritis, which is associated with the development of intestinal metaplasia and dysplasia. Development of atrophy and metaplasia of the gastric mucosa are thus strongly associated with H. pylori infection, instead of a direct and inevitable consequence of ageing. Approximately 40±50% of infected subjects develop these conditions, but they are rare in non-infected subjects. The presence of these consecutive disorders leads to a 5±90-fold increased risk for cancer of the distal stomach, in particular of the intestinal type. This sequence explains the increased risk for gastric cancer in H. pylori-infected subjects, as has been shown in various cross-sectional and longitudinal studies. In a combined analysis of three longitudinal studies, a significant trend was observed towards an increased odds ratio with longer intervals between (retrospective) serological diagnosis of H. pylori infection and observation of gastric cancer, this risk being more than eight-fold increased if the interval had been at least 15 years. This is thought to reflect development of atrophic gastritis and intestinal metaplasia with loss of H. pylori colonization in the years prior to development of cancer. Atrophic gastritis and gastric cancer thus appear closely associated with the presence of H. pylori, yet not all infected subjects will eventually develop atrophy and only a small minority develop gastric cancer. Factors that influence the risks for atrophy and cancer in the presence of infection may be related to the time that infection occurred and to characteristics of the bacterial strain and the host. Evidence for the role of these factors is now increasing. Recognition of the causal role of H. pylori in the induction of gastric cancer theoretically presents tools for cancer prevention. The efficacy of screening and bacterial eradication for prevention of distal gastric cancer is being studied in a number of large-scale intervention studies in different populations. It is hoped that these studies will also provide answers to the potential preventive role of H. pylori colonization in the development of gastro-oesophageal reflux disease and associated conditions, in particular development of cancer of the proximal stomach. Infection with H. pylori plays an important role in the aetiology of atrophic gastritis and gastric cancer. Studies suggest an eight-fold increased risk for both conditions in the presence of infection. Factors that influence the risk for both conditions in the presence of infection are the age at which infection occurred and the presence of caga as a marker for more pathogenetic H. pylori strains. The efficacy and side-effects of intervention for the prevention of distal gastric cancer has yet to be established. Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start INTRODUCTION Correspondence to: Dr E. J. Kuipers, Free University Hospital, PO Box 7057, 1007 MB Amsterdam, The Netherlands. The aims of this paper are to review the following three concepts: # 1999 Blackwell Science Ltd 3

4 E.J.KUIPERS. gastric cancer is a heterogeneous disease with a poorly understood pathogenesis;. H. pylori infection increases the risk for distal gastric cancer;. H. pylori may decrease the risk for proximal gastric cancer. GASTRIC CANCER IS A HETEROGENEOUS DISEASE The fact that gastric cancer is a heterogeneous disease with a poorly understood pathogenesis is re ected in the many di erent classi cation schemes that are used for this particular tumour (Table 1). Some of these classi- cations are based on endoscopy, for example when di erentiating between distal and proximal gastric cancer. Others are based on histology; in Europe the Lauren classi cation is commonly used to di erentiate between intestinal, di use and mixed tumours. A third group of classi cations is based on surgery, for example, the TNM and the Duke's classi cations. All these classi- cation systems are merely descriptive; they do not discuss underlying aetiology or genetic abnormalities. What are the genetic abnormalities in gastric cancer? We have previously evaluated chromosomal lesions in 28 gastric cancers using comparative genomic hybridization. 1 This technique allows the search for gains and losses in the tumour genome. It appeared that all gastric cancers had a lot of di erent genetic abnormalities; in each tumour at least seven di erent abnormalities were found. The pattern of abnormalities is di erent to that observed in other adenocarcinomas of the gastrointestinal tract. An example of this is in the 18q region where there is an oncogene which is deleted in colon cancer, the DCC region; in gastric cancer no deletion is observed. For many regions where genetic abnormalities are observed, either gains or losses of DNA, it is not yet known what the corresponding oncogene or tumour suppressor gene might be. From metaplasia to cancer Regardless of any speci c knowledge about underlying genetic abnormalities, Correa suggested more than 20 years ago that there was a histological cascade leading to intestinal-type gastric cancer (Figure 1). 2 He suggested that chronic active gastritis could eventually induce a loss of gastric glands, or atrophy, with replacement by either brosis or by intestinal metaplasia, of which it is now known that there are three di erent types, and this could eventually continue to the development of dysplasia and gastric cancer. Now, more that 20 years later, we still do not know whether, for example, intestinal metaplasia is a necessary step before a person can go on to develop gastric cancer, or whether it is just an `innocent bystander' and that the development of gastric cancer can bypass the metaplastic step. Many hypothesize that metaplasia, particularly the colonic type III intestinal metaplasia, is the only histological lesion that increases cancer risk, but the evidence for this hypothesis is actually very poor. There are two prospective cohort studies that have looked at the incidence of gastric cancer amongst subjects with and without intestinal metaplasia and showed similar results: Ectors and Dixon from the UK 3 and Filipe et al. from Slovenia. 4 Table 2 shows the results of the Filipe et Table 1. Classification systems for gastric epithelial tumours Classifications based on endoscopy Proximal/distal Exophytic/linitis plastica Classifications based on histology Lauren (intestinal/diffuse/mixed) WHO Carneiro Ming Astler-Coller Classifications based on surgery TNM Duke's Figure 1. The histological cascade leading to intestinal type gastric cancer.

REVIEW: THE LINK BETWEEN H. PYLORI AND GASTRIC CANCER 5 Table 2. Incidence of gastric cancer amongst subjects with and without intestinal metaplasia (IM) 4 No. with No. with gastric histological Odds cancer findings ratio 95% CI No IM 8 283 1.0 Ð Type I IM 6 512 0.44 0.15±1.27 Type II IM 5 392 0.94 0.31±2.9 Type III IM 15 260 1.84 0.76±4.43 al. study. They followed a large group of subjects with atrophic gastritis with and without di erent types of metaplasia and observed the incidence of gastric cancer. Within a control group of 283 subjects with atrophic gastritis but without signs of metaplasia (from gastric biopsy specimens), eight developed cancer within a maximum follow-up of 20 years. For each of the three groups with types I, II and III metaplasia, the incidence of gastric cancer was very similar to that seen in the control group, thus giving odds ratios which did not di er signi cantly from that of the control group without metaplasia. It appears therefore that, to date in the published literature, there is no evidence that supports the concept that only type III metaplasia leads to cancer. This is also illustrated by the fact that type III metaplasia is actually quite rare in gastric cancer patients. Several investigators have observed that type III metaplasia is present in only 10±20% of gastric resection specimens containing adenocarcinoma. 5±7 In summary, it appears that gastric cancer is a heterogeneous disease with many di erent underlying genetic abnormalities. In addition, the pathogenesis and relation to metaplasia is poorly understood. H. PYLORI INCREASES THE RISK FOR DISTAL GASTRIC CANCER One fact we do know is that H. pylori increases the risk for di erent types of cancer in the distal stomach. The types of evidence to consider when associating H. pylori with distal gastric cancer are summarized in Table 3 and these will be discussed in turn. If we look at the odds ratios for the association between H. pylori and distal gastric cancer observed in 15 retrospective case-control studies, as shown in Figure 2, most studies indicated some association between H. pylori and cancer. 8 Values above an odds ratio of 1.0 suggest that H. pylori increases the risk of gastric cancer while values below 1.0 indicate that H. pylori protects against gastric cancer, however, 10 of the 15 studies had con dence limits that spanned 1.0. When all data were accumulated, it appeared that the presence of H. pylori increased the risk of gastric cancer two-fold. 8 Is there any di erence if the age of the subject who develops cancer is considered? The same analysis by Huang et al. showed that there was a di erence: with increasing age there was a decrease in the odds ratio (Figure 3). 8 The most likely explanation for this relates to the prevalence of H. pylori amongst cancer cases and controls. As shown in Figure 4, in cancer cases, the Table 3. Evidence to consider when associating H. pylori with distal gastric cancer. Retrospective case-control studies. Prospective cohort studies. Geographic and socio-economic associations. Experimental models. Intervention studies Figure 2. Odds ratios for the association between H. pylori and distal gastric cancer observed in 15 retrospective case-control studies. 8

6 E.J.KUIPERS Figure 3. Age-specific odds ratios for the association between H. pylori and distal gastric cancer observed in 15 retrospective case-control studies. 8 Figure 4. Age-specific prevalence of H. pylori in gastric cancer cases and matched controls. 8 prevalence of H. pylori was between 80% and 90%, irrespective of age. 8 However, among controls the prevalence of H. pylori increased with age to become close to 80% in those over the age of 70 years, a pattern re ected in many populations. Clinical evidence linking H. pylori and gastric cancer Strong evidence for an association between H. pylori and cancer comes from the results of longitudinal studies. Figure 5 shows the results of six prospective studies. 9±14 Four studies showed a signi cant correlation between the presence of H. pylori and development of distal gastric cancer; the remaining two studies found no such correlation but in these studies the follow-up time was signi cantly shorter than in the other four, which might explain this di erence. A meta-analysis by Forman and colleagues of the results of three of these longitudinal studies, two from the USA and one from the UK, showed that this was indeed the case (Figure 6). 15 They found that the interval between serum collection and diagnosis of gastric cancer correlated signi cantly with the calculated odds ratio. In subjects in whom this interval had been less than 10 years, the calculated odds ratio was very similar to that found in the retrospective studies, at around 2.0. However, if the follow-up Ð the interval between serum collection and the diagnosis of cancer Ð was more than 10 years, and in particular more than 14 years, the odds ratio increased to &9.0. The proposed explanation for this phenomenon is that during long-term follow-up, the prevalence of H. pylori decreases among the eventual cancer cases in the presence of atrophic gastritis and intestinal metaplasia. Conclusions that can be drawn from retrospective and longitudinal studies can be summarized, as shown in Table 4. All these epidemiological data in humans has until now been lacking in the very important support of data from experimental models, but this is now changing. A recent

REVIEW: THE LINK BETWEEN H. PYLORI AND GASTRIC CANCER 7 Figure 5. Odds ratios for the association between H. pylori and distal gastric cancer observed in six prospective studies. 9±14 Table 4. Conclusions from retrospective and longitudinal studies. Data are generally consistent. Retrospective studies underestimate the relative risks observed prospectively. The odds rations are higher among younger populations and increase with longer follow-up Figure 6. The interval between serum collection and diagnosis of cancer correlates significantly with odds ratio. 15 study has been published by Watanabe et al. in which they investigated the development of gastric adenocarcinoma in Mongolian gerbils infected with H. pylori. 16 They observed that after 62 weeks of infection 37% of the infected animals had developed intestinal type adenocarcinoma (Figure 7), compared to none of the animals in the uninfected control group. This supports the hypothesis that H. pylori infection may induce gastric cancer. CALCULATION OF CANCER RISK What do all these di erent relative risks mean for an infected and an uninfected individual, i.e. what is their risk of developing gastric cancer? There are various di erent methods that can be used to calculate this. As an example, one can start with a cohort of 100 000 subjects, assuming that &35% of them are infected with H. pylori, this group is divided into 35 000 H. pylori-positive subjects and 65 000 H. pylori-negative subjects (Figure 8). In The Netherlands, the incidence of gastric cancer of the distal stomach is &15 per 100 000 per year; higher than in the USA but lower than in some eastern European countries and Japan. It can be estimated that among H. pylori-positive subjects, the gastric cancer risk is eight-fold higher than for H. pylorinegative subjects. This means that out of the 15 cancers per year, 12 will develop in the H. pylori-positive group and three in the H. pylori-negative group. If these subjects are followed for most of their adult life (the period for which they are at risk for gastric cancer) after a 30-year follow-up there would be a total of 360 cancers in the H. pylori-positive group compared to 90 in the H. pylori-negative group. Therefore, &1% of the infected subjects will develop cancer compared to one in every 750 of the non-infected group. This is one of 17, 18 many models but they all give similar results. Looking again at the scheme suggested by Correa (see Figure 2) for the histological cascade leading to intestinal type gastric cancer and adding the estimated prevalence of the di erent abnormalities, based on many di erent cohort follow-up studies, it can be seen that of 100 subjects who develop chronic active gastritis, 50 of these will develop a loss of gastric glands, or atrophy, 40 will develop intestinal metaplasia (particularly the ones who have atrophy) eight will develop dysplasia and, as in the previous model, one will develop gastric cancer (Figure 9).

8 E.J.KUIPERS Figure 7. Development of gastric adenocarcinoma in H. pylori-infected Mongolian gerbils. 16 Figure 8. Estimated gastric cancer risk in H. pylori-infected and uninfected subjects. What we need to de ne is the di erence between the one person who develops cancer and the 99 others who do not, i.e. what determines cancer risk in the presence of H. pylori? There are many possible in uences including strain characteristics: caga, vaca, the duration of infection, the diet, host genetics and the pattern of gastritis and ulcer type. Looking particularly at the in uence of ulcer type, Hansson and colleagues showed that gastric ulcer (GU) patients who were characterized by reduced acid production and pangastritis had a higher cancer risk than duodenal ulcer (DU) patients with high acid production and antral dominant gastritis 19 (Figure 10). Patients with DU had a lower cancer risk than the general population. This is often interpreted as if DU protects against gastric cancer, however, the general population consists of both H. pylori-negative and H. pylori-positive patients. If the H. pylori-negative subjects only are considered, it is hypothesized that their cancer risk is actually lower than the general

REVIEW: THE LINK BETWEEN H. PYLORI AND GASTRIC CANCER 9 Figure 9. The frequency of subsequent histological steps in gastric carcinogenesis. population as a whole, and therefore DU disease by itself does not protect against the development of gastric cancer in comparison to the absence of H. pylori infection (Figure 10). Several investigators have studied the in uence of host genetics on gastric cancer development and it is suggested that certain HLA-DQ types are related to cancer risk. 20, 21 An interesting observation is that at the beginning of this century, gastric cancer was the predominant lesion in the initial description of the cancer family syndrome 22 but now colorectal cancer is much more common. 23 Thus, the same families are now diagnosed as having hereditary non-polyposis colon cancer whereas they used to have predominantly gastric cancer. The hypothesis that can be proposed from these ndings is that when infection with H. pylori was ubiquitous, the interaction between the organism and the host genetics led to gastric cancer prior to the time when they could develop colon cancer. Figure 10. Odds ratio for the development of gastric cancer in gastric ulcer (GU) and duodenal ulcer (DU) patients compared to the general population. 13 show that since 1950 there has been approximately a 50% decrease in the incidence of distal gastric cancer, re ecting a decrease in the prevalence of H. pylori 25 (Figure 12). At the same time there has been an increase in the incidence of proximal gastric cancer; it is now proportionally the most rapidly increasing tumour in The Netherlands. In the USA, among white males, proximal gastric cancer is now more common than distal gastric cancer. It is questioned whether this increase could also re ect changes in the prevalence of H. pylori. Clinical evidence for a link between H. pylori and a decreased risk of proximal gastric cancer Hansen and colleagues undertook a nested case-control study based on over 100 000 patients followed-up for over 12 years, the results of which suggested that the increase in proximal cancers could indeed be related to H. PYLORI MAY DECREASE THE RISK FOR PROXIMAL GASTRIC CANCER It is well known that the prevalence of H. pylori is decreasing in many populations. Alongside this, there is a decrease in associated conditions such as atrophic gastritis. Data from Sipponen in Finland showed that over a period of 15 years, from 1977 to 1992, for all age groups there was a decrease in the prevalence of atrophic gastritis 24 (Figure 11). Similarly, in many countries there is a decrease in the incidence of gastric cancer. Data from The Netherlands Figure 11. Age-specific prevalence of atrophic gastritis in Finland in 1977 and 1992. 18

10 E. J. KUIPERS Table 6. Retrospective case-control study of proximal gastric cancer risk in relation to H. pylori and caga status 27 Proximal Odds H. pylori caga Controls (n) cancer (n) ratio 95% CI Negative Negative 138 91 1.0 Ð Positive Negative 40 26 1.1 0.6±2.1 Positive Positive 46 12 0.4 0.2±0.9 Figure 12. Time-specific incidence of distal and proximal gastric cancer in The Netherlands. 25 H. pylori prevalence 26 (Table 5). It was found that in the presence of H. pylori the chance of developing distal gastric cancer was approximately nine-fold that of H. pylori-negative subjects. It also showed that in the presence of H. pylori, development of adenocarcinoma of the cardia was in fact lower than in H. pylorinegative subjects. Similar data were recently published by Chow et al. 27 (Table 6). They studied retrospectively the presence of H. pylori and CagA serum antibodies in a group of &130 patients with adenocarcinoma of the distal oesophagus and proximal stomach and compared them with around 220 controls. Of the control group, 138 were both H. pylori-and caga-negative, compared to 91 of the group with proximal cancer. Forty of the controls and 26 of the cancer group were H. pylori-positive but caga-negative. This was similar to the control group and thus had an odds ratio close to 1.0. However, 46 of the controls but only 12 of the cancer group were both H. pylori- and caga-positive, giving an odds ratio of 0.4 for the development of proximal gastric cancer if infected with a caga-positive strain. The underlying hypothesis at the present time is that infection with H. pylori, in particular with the more severe caga-positive type, will reduce acid secretion and thus protect against gastro-oesophageal re ux disease and its associated complications. This is only a hypothesis and data are currently accumulating. GASTRIC CANCER PREVENTION How do all these data relate to gastric cancer prevention by screening and eradicating H. pylori? There are currently many studies ongoing, both mathematical modelling and clinical trials. They have di erent starting points: some look at primary prevention, either population-based screening and treatment or evaluation of speci c risk groups (e.g. family cancer risk or cagapositive), others look at secondary prevention in patients with atrophic gastritis or those with dysplasia and early gastric cancer. CONCLUSIONS In terms of the link between H. pylori and distal gastric cancer it can be concluded that:. H. pylori gastritis increases gastric cancer risk 8±9- fold;. 60±80% of gastric cancers in developed countries are associated with H. pylori;. the lifetime risk of developing gastric cancer in H. pylori infected persons is estimated at 1%. With respect to the link between H. pylori and proximal gastric cancer it can be concluded that:. the evidence suggests that H. pylori, in particular the caga-positive type, may protect against cancer of the cardia;. the strong increase in the incidence of cardia cancer may then be explained by the decreased prevalance of H. pylori, especially the caga-positive type. Table 5. Site-specific gastric cancer risks in H. pylori-infected subjects 26 Cancer site Odds ratio 95% CI Antrum and corpus 8.7 2.5±30.2 Cardia 0.3 0.1±0.9 REFERENCES 1 van Grieken NCT, Hermsen MAJA, Meyer GA, et al. Chromosomal aberrations in H. pylori associated gastric cancer analysed by comparative genomic hybridization. Gastroenterology 1998; 114: A695(Abstract).

REVIEW: THE LINK BETWEEN H. PYLORI AND GASTRIC CANCER 11 2 Correa P, Haenszel W, Cuello C, Tannenbaum S, Archer M. A model for gastric cancer epidemiology. Lancet 1975; ii: 58±9. 3 Ectors N, Dixon MF. The prognostic value of sulphomucin positive intestinal metaplasia in the development of gastric cancer. Histopathology 1986; 10: 1271±7. 4 Filipe MI, MunÄ oz N, Matko I, et al. Intestinal metaplasia types and the risk of gastric cancer: a cohort study in Slovenia. Int J Cancer 1994; 57: 324±9. 5 Houben G, de Bruine A, Hameeteman W, StockbruÈ gger R. Subtypes of intestinal metaplasia in patients with gastric cancer. Gastroenterology 1996; 110: A532(Abstract). 6 Solcia E, Fiocca R, Luinetti 0, Villani L, et al. Intestinal and di use gastric cancers arise in a di erent background of Helicobacter pylori gastritis through di erent gene involvement. Am J Surg Pathol 1996; 20(Suppl. 1): S8±22. 7 Sipponen P, SeppaÈ IaÈ K, Varis KF, et al. Intestinal metaplasia with colonic-type sulphomucins in the gastric mucosa; its association with gastric carcinoma. Acta Pathol Microbiol Scand 1980; 88: 217±24. 8 Huang JQ, Sridhar S, Chen Y, Hunt RH. Meta-analysis of the relationship between Helicobacter pylori seropositivity and gastric cancer. Gastroenterology 1998; 114: 1169±79. 9 Forman D, Newell DG, Fullerton F, Yarnell JWG, Stacey AR, Wald N, Sitas F. Association between infection with Helicobacter pylori and risk of gastric cancer: evidence from a prospective investigation. Br Med J 1991; 302: 1302±5. 10 Nomura A, Stemmermann GN, Chyou P, Ikuki K, Perez-Perez GI, Blaser MJ. Helicobacter pylori infection and gastric carcinoma among Japanese Americans in Hawaii. N Engl J Med 1991; 325: 1132±6. 11 Parsonnet J, Friedman G, et al. Helicobacter pylori infection and the risk of gastric carcinoma. N Engl J Med 1991; 325: 1127±31. 12 Hanssen S, Vollset SE, Melby K, Jellum E. Strong association of previous Helicobacter pylori infection to distal gastric adenocarcinoma. Am J Gastroenterol 1994; 89: 1358 (abstract). 13 Lin JT, Wang LY, Wang JT, Wang TH, Yang CS, Chen CJ. A nested case-control study on the association between Helicobacter pylori infection and gastric cancer risk in a cohort of 9775 men in Taiwan. Anticancer Research 1995; 15: 603±6. 14 Webb PM, Yu MC, Forman D, et al. An apparent lack of association between Helicobacter pylori infection and risk of gastric cancer in China. Int J Cancer 1996; 67: 603±7. 15 Forman D, Webb P, Parsonnet J. Helicobacter pylori and gastric cancer. Lancet 1994; 343: 243±1. 16 Watanabe T, Tada M, Nagai H, Sasaki S, Nakao M. Helicobacter pylori infection induces gastric cancer in Mongolian gerbils. Gastroenterology 1998; 115: 642±8. 17 Parsonnet J. Helicobacter pylori in the stomach Ð a paradox unmasked. N Engl J Med 1996; 335: 278±80. 18 Kuipers E.J. Relationship between Helicobacter pylori, atrophic gastritis and cancer. Aliment Pharmacol Therap 1998; 12 (Suppl. 1): 25±36. 19 Hansson LE, Nyre n 0, Hsing AW, et al. The risk of stomach cancer in patients with gastric or duodenal ulcer disease. N Engl J Med 1996; 335: 242±9. 20 Lee JE, Lowy AM, Thompson WA, et al. Association of gastric adenocarcinoma with the HLA class II gene DQB1*0301. Gastroenterology 1996; 111: 426±32. 21 Azuma T, Ohtaki Y, Ito Y, et al. HLA-13QA gene contributes to the host's response against Helicobacter pylori infection: host's genetic analysis in two areas in Japan with contrasting gastric cancer risks. Gastroenterology 1996; 110: A54(Abstract). 22 Whartin AS. Heredity with reference to cancer. Arch Intern Med 1913; 12: 546±55. 23 Lynch HT, Smyrk T. Hereditary nonpolyposis colorectal cancer (Lynch syndrome). Cancer 1996; 78: 1149±67. 24 Sipponen P, Helske T, JaÈ rvinen P, HyvaÈ rinen H, SeppaÈ laè K, Siurala M. Fall in the prevalence of chronic gastritis over 15 years: analysis of outpatient series in Finland from 1977, 1985 and 1992. Gut 1994; 35: 1167±71. 25 Eliel MR, Schaap LF. Oncologieboek IKMN. Utrecht 1997. 26 Hansen S, Vollset SE, Ardill JES, et al. Hypergastrinemia is a strong predictor of distal gastric adenocarcinoma among Helicobacter pylori infected persons. Gut 1997; 41(Suppl. 1): A41(Abstract). 27 Chow W, Blaser MJ, Blot WJ, et al. An inverse relation between caga+ strains of H. pylori infection and risk of esophageal and gastric cardia adenocarcinoma. Cancer Res 1998; 58: 588±90.