Would eradication of Helicobacter pylori infection reduce the risk of gastric cancer?
Prof. R.A.Feldman, The Royal London Hospital, Whitechapel Road London, E1 1BB, UK E-mail: firstname.lastname@example.org
This article reviews the data on the epidemiology of gastric cancer, to determine if treatment of an asymptomatic individual can be justified. It reviews retrospective and prospective case-control studies of gastric cancer in Italy and other countries. Mucosa-associated lymphoid tissue lymphoma is associated with Helicobacter pylori infection. The risk of noncardia gastric cancer is higher (4-fold or greater) in those with H. pylori infection. Although no studies have shown prevention following treatment, erradication of asymptomatic H. pylori infection in an individual in the age group 40 or lower may be expected to reduce the risk of gastric cancer.
There are epidemiologic data concerning the association of Helicobacter pylori and gastric cancer and mucosa-associated lymphoid tissue (MALT) lymphoma that allow a decision whether to consider treating asymptomatic H. pylori infection in order to reduce the risk of gastric cancer. There are other reasons for considering whether or not to treat an H. pylori infection, with or without symptoms, which will not be considered here.
Epidemiologic studies of the gastric cancer–H. pylori association are most often either retrospective case-control studies or prospective cohort (nested case-control) studies. In retrospective case-control studies of the disease–infection association of H. pylori and gastric cancer, it is important to accept the fact that both for an elderly case and for the control, when they are studied at the time of diagnosis, seropositivity may have reverted to negative in previously seropositive persons as a result of loss of infection due to gastric atrophy. With prospective cohort studies of the disease–infection association of H. pylori and gastric cancer, an important element of the study is the length of follow-up after the initial sera were obtained. Since H. pylori infection leads to progressive changes, and ultimately gastric atrophy with the loss of seropositivity, even if it were once positive, the longer the follow-up, the more reliable the resulting measures of risk.
With these two study designs in mind, a look at the available data will assist in deciding whether it is possible to reduce gastric cancer risk by treatment of the infection in an individual. We begin with a review of gastric cancer data from Italy.
Gastric cancer in Italy
In Italy, age-standardized death certification rates from cancer of the stomach decreased during the 15-year period from 1955 to 1979, from 47 to 31 per 100,000 in males and from 35 to 19 per 100,000 in females. The province-specific gastric cancer mortality in 1975–77 was about 10% lower for both sexes in the 14 provinces that included the largest urban concentrations (over 250,000 inhabitants), than in 81 other provinces. There were lower gastric cancer mortality rates in both sexes in the southern compared with the central and northern areas of Italy. And there were several areas of exceedingly high gastric cancer mortality scattered in northern and central Italy.1 One example is San Marino, which has a high incidence of gastric cancer, with 9% of all deaths from 1969 to 1983 and 33% of all cancer deaths attributed to gastric cancer.2
In a case-control study of dietary histories in gastric cancer patients in northern Italy, after adjusting for socio-economic status, a 4-fold lower risk of gastric cancer rates was found for those whose green vegetables intake was greater than the controls, while there was an increased risk for those with a greater use of polenta or ham.3 A more recent, larger and geographically more diverse case-control study found that gastric cancer risk rose with increasing consumption of protein, and decreased in proportion to the intake of ascorbic acid, alpha-tocopherol and vegetable fat. Ascorbic acid showed the strongest correlation with geographical area, with highest consumption in low-risk areas. The findings suggest that the protective effects reported for consumption of fresh fruit, fresh vegetables and olive oil may be linked to the vitamins C and E contained in these foods. The findings are consistent with the hypothesis that N-nitroso compounds are involved in gastric cancer risk, since elevated risks were apparent for agents (nitrites, protein) that promote nitrosation, while decreased risks were found for nutrients (ascorbic acid and alpha-tocopherol) that inhibit the process. Studies of risk factors in the diet, occupation, geographical origin and lifestyle have shown that although the consequences of H. pylori infection may themselves be a prominent underlying cause, differences in the prevalence of the infection itself is not sufficient to explain all the variations seen in gastric cancer incidence in Italy.4–11
Association of H. pylori and MALT lymphoma
Gastric lymphomas are the commonest of the extranodal lymphomas, but they comprise less than 5% of all gastric cancers, with the overall rates of disease in the age group 60–79 being about two per 100,000 per year.12 The gastric lymphomas have clinicopathologic features more closely related to the structure and function of so-called MALT than to peripheral lymph nodes. In contrast to peripheral lymph nodes, which are adapted to deal with antigens carried to the node in afferent lymphatics, MALT appears to have evolved to protect mucosal tissue, which is directly in contact with antigens in the external environment. Normally there is no lymphoid tissue in the stomach. The presence of lymphoid follicles in the gastric mucosa is virtually pathognomonic of H. pylori infection, and the growth and development of a gastric MALT lymphoma growth is subject to immunological stimuli related to H. pylori.
Four independent sources of evidence link prior H. pylori infection and MALT lymphoma.13,14 These include: (1) the presence of H. pylori in more than 90% of cases of gastric MALT lymphoma; (2) a high incidence of gastric MALT lymphoma in north-east Italy, in association with a high seroprevalence of H. pylori; (3) case-control data showing an association between previous H. pylori infection and the development of primary gastric lymphoma; (4) studies showing that treatment of the H. pylori infection leads to resolution of low-grade MALT lymphoma, with approximately 70% of cases in stage IE of gastric MALT lymphoma regressing following successful treatment.
It is still unclear which risk factors identify those at greatest risk of developing MALT lymphoma among those who are infected, although one of the factors appears to be characteristics of the infecting strain.15
Association of H. pylori infection and risk of gastric cancer
Internationally gastric cancer rates increase with age, and rise sharply after age 65 to over 250 per 100,000 at age 80, with males more frequently affected than females. It has been hypothesized that an early age of acquisition of H. pylori may be a significant factor associated with gastric cancer occurrence, but data concerning the age of infection can only be inferred from characteristics in the family. Since infection occurs predominantly in children, then those of higher birth orders, and in large families, should have earlier acquisition of infection. A nested case-control study of a cohort of Japanese-American men in Hawaii16 matched controls with cases for sibship size and birth order. Helicobacter pylori-infected but not H. pylori-uninfected men from larger sibships (odds ratio, 2.06) and of higher birth order (odds ratio, 1.67) were at increased risk for developing gastric cancer. There was no relation between birth order or sibship size and the occurrence of duodenal ulcer.
Because gastric cancer is most frequent in persons over 65, the case-control studies were conducted with individuals, in both case and control groups, in which gastric atrophy was not infrequent, and gastric atrophy by itself is associated with a loss of seropositivity. This, in itself, can mean that persons who were once seropositive will be seronegative when studied, and this can bias the result of the study. However, this bias is minimized among gastric cancer cases occurring at an early age. In such a study in Japan17 with all cases under 40 years of age, the odds ratio for increased risk associated with H. pylori was 13 (95% CI: 5.3–36.0).
Another way to reduce the problems of bias introduced by loss of seropositivity is to observe the effect of having a long period of follow-up. If the loss of seropositivity is important, the observed risk should be greater when there is a long period of observation after collection of the sera. A Norwegian prospective study18 of over 100,000 male and female subjects followed them from between 1972 and 1977 until 1992. The median age of study subjects when sera were collected was 45 years, while the median age at which the cancer diagnosis was made was 57 years. The study included over 200 cases and showed an odds ratio of 5.1 (95% CI: 2.8–9.4) of an association of H. pylori seropositivity and noncardia gastric cancer. The pooled odds ratio for association in three earlier cohort studies was 3.8 (95% CI: 2.3–6.2). However, when the data were stratified by duration of follow-up, the odds ratio increased to 8.7 (95% CI: 2.7–44.7) for the group with 15 or more years between the assessment of infection and a diagnosis of gastric cancer.19 The data from one of the three studies (the UK study) were updated with a longer follow-up and more cases, and the odds ratio increased from 2.7 to 4 (95% CI: 1.9–8.2).20 In a Finnish21 study, the odds ratio for association increased from 0.3 with 2 years of follow-up to 3.3 for at least 7 years of follow-up. With the availability of serologic results at an earlier age than those available for case-control studies, the prospective studies give a more valid measurement of risk.
An interesting and unexplained observation in the Norwegian study was that when comparing risks of gastric cancer in those observed for the longest interval (14+ years), there was a lower odds ratio than those observed for about 10 years.
There are conflicting data concerning the significance of the association of gastric cancer with the Cag-A status of H. pylori. The risk of gastric cancer in young adult Japanese22 was equal for bacteria that were Cag-A–(odds ratio 15, CI: 6.4–35.2) and Cag-A+ (odds ratio 14.6, CI: 6.7–31.9). However, an increased association of gastric cancer with the H. pylori Cag-A genotype was found in young Italian patients.23
Do the data justify treating H. pylori in an adult to reduce their risk of gastric cancer?
The question of whether treatment in an adult will reduce the risk of gastric cancer in that adult is not one that can easily be answered by controlled trials. You can reduce the numbers of people in an adequately designed study by starting recruitment at an older age (since there will be a larger number of cases expected in an older age group), but the older the group recruited, the less likely it is that therapy will affect the existing pathology in the stomach, and so the older the age group recruited, the less likely is a positive result. Perhaps, on an ‘island’ population, such as a South Pacific island with a high gastric cancer rate, or an ‘island’ in Italy, such as San Marino,24 with a high gastric cancer rate, it might be possible to effectively treat the adults, but there would still be a difficulty in identifying a comparison group, in order to show that if there were a reduction in gastric cancer rates, the change was due to the intervention.
The Norwegian study18 suggested that because there was an increased risk of noncardia gastric cancer associated with Helicobacter, for all three Lauren histological subtypes, while there was no increased risk for cardia gastric cancer, the two entities are separate disease entities with different causation. The authors suggested that the challenge would be to find individuals at high risk of developing noncardia gastric cancer, and to target them for treatment before embarking on a large-scale prevention program. The intention then would be to study further the causation of cancers of the cardia, so that solving one problem produces another.
Hansen et al. observed that those with the longest interval of time between serum collection and onset of cancer had a lower odds ratio for association with H. pylori infection than those whose follow-up was around 10 years. Although this observation does not decrease the significance of an association of gastric cancer and H. pylori, it does deserve further study. Factors such as changes in the dietary patterns of those infected may be important in understanding this finding.
To answer the question about treatment of an asymptomatic H. pylori infection in an individual, it might be that for the oldest age group, treatment of an asymptomatic infection would hardly affect the risk of gastric cancer. However, for a young individual, which would include patients in their forties, the data provide justification in attempting to reduce their risk of gastric cancer by treating an H. pylori infection, even without symptoms.