Salt and Gastric Adenocarcinoma: A Population-Based Cohort Study in Norway

Gastric adenocarcinoma is the second leading cause of cancer death globally (700,000 deaths annually; ref. 1). Despite a declining incidence in many western countries, the absolute number of cases is predicted to increase up to the year 2050 (2). There is a need to identify modifiable lifestyle factors, such as diet, which can be applied for cancer prevention strategies. The hypothesis that high dietary salt intake could increase the risk of gastric adenocarcinoma was spawned in the 1960s, and evidence mainly from high-incidence Asian countries has been accumulating over the past decades (3-6). A high salt concentration in the stomach leads to diffuse erosion, and the induced proliferation in the inflamed environment could promote the effect of carcinogens derived from food (7). However, few prospective studies have assessed the association of salt intake with the risk of gastric adenocarcinoma, and the results from these studies have not been consistent (4). In this long-term follow-up of a large population, we have assessed the effect of dietary salt intake on the risk of gastric adenocarcinoma in a low-incidence western region.

The design of this cohort has been presented in detail elsewhere (8). In brief, in 1984 to 1986, all adult inhabitants in Nord-Trondelag County in Norway were invited to participate in a health survey. Among 85,100 eligible persons, 75,058 (88%) responded to written questionnaires and attended a clinical examination. Among these, 3,163 persons whose follow-up time after inclusion was <3 years were excluded because of possible influence of yet undetected cancers on the exposure prevalence (9). Information was collected on a range of lifestyle and health-related factors, including measures of dietary intake of salt. The unique 11-digit identity number of Norwegian citizens was used to link cohort members to information on cancer incidence at the Cancer Registry of Norway. Gastric adenocarcinoma was registered according to the International Classification of Diseases, Seventh Edition codes 151.0, 151.8, and 151.9. Virtually all gastric tumors (>98%) were histologically confirmed. This study only included adenocarcinoma, the dominating malignancy of the stomach. For accurate censoring, we performed register linkages with the Norwegian Central Person Registry, which provided information on vital status and emigration throughout the follow-up period.

The participants were asked about their average frequency of dietary intake of salted foods, with five response options (never or less than once a month, once to twice a month, up to once a week, up to twice a week, or more than twice a week). There were no lists of food items to choose between. The actual question asked was “How often do you eat salted meat or salted fish/herring?” In addition, they were asked, “How often do you sprinkle extra salt on your hot food?” (never, seldom, occasionally, or often). Furthermore, for the analysis, we constructed a summary score from these two salt-related variables. The score summarized each participant's responses according to the following equation: 1/5 × frequency category (1-5) of intake of salted foods + 1/4 × frequency category (1-4) of sprinkling extra salt. The score variable was thereafter grouped into three categories: low, moderate, or high. Persons who never consumed salted foods or consumed them never, less than once a month, or once to twice a month were categorized as having a low intake, whereas a summary score above this level and below the median was classified as moderate intake, and a score at the median or above was classified as high intake. No information was collected on salt added during cooking.

Each cohort member contributed person-time from the date 3 years after the health survey examination to the first occurrence of a cancer diagnosis (any site), death, emigration, or end of follow-up (December 31, 2002). Hazard ratios (HR) and 95% confidence intervals (95% CI) were estimated by a Cox proportional hazards regression model. Proportional hazards assumptions were tested by Schoenfeld's method (10). Because the proportional hazards assumptions were violated in some regression models, mainly by the age variable, stratified analysis by age group was also done. In multivariable models, adjustments were made for attained age, gender, smoking status (never, former, or current), alcohol use (frequency during the last 2 weeks: never, not been drinking, 1-4 times, or ≥5 times), recreational physical activity (low, medium, or high according to a summary score of frequency, duration, and intensity of physical activity; ref. 8), and occupation (higher-level employees or employers and professionals, intermediate employees, lower-level employees, other self-employed or farmers, skilled laborers, or unskilled laborers; ref. 11). Because body mass index did not influence the risk of gastric adenocarcinoma in our analyses, this variable was excluded from our final model. Nonresponders to the questions regarding salt intake were analyzed as a separate group. Trend tests for the different exposures were made by entering categorical data as ordinal variables into the Cox regression model. We also performed sex-stratified analyses. The statistical tests were two-sided. Stata (version 9.0) was used for all analyses.

The final study cohort comprised 73,133 persons who were followed up for an average of 15.4 years, contributing a total of 1,122,765 person-years at risk. The follow-up identified 313 new cases of gastric adenocarcinoma, of which 264 (84%) were located distal to the cardia and 49 in the cardia. Some characteristics of the study participants are presented in Table 1. There was a male predominance (60%) among gastric adenocarcinoma patients. The mean age at baseline of cohort members was 49 years and that of gastric adenocarcinoma patients was 65 years. The mean body mass index and frequency of current smoking were equal among all cohort participants and the cases, whereas formal education was lower among the cases (Table 1).

There were no statistically significant associations between different levels of intake of salted foods and risk of gastric adenocarcinoma. The point estimates were generally close to unity and did not change materially from the crude to the fully adjusted models (Table 2). Persons reporting having a high intake of salted foods (more than twice a week) had an adjusted HR of 1.1 (95% CI, 0.6-1.8) for gastric adenocarcinoma compared with cohort members who reported never, or almost never, consuming salted foods. No indication of a dose-risk effect was revealed (P_trend = 0.39). The habit of always or almost always sprinkling extra salt on food rendered no statistically significantly increased risk of gastric adenocarcinoma (HR, 1.4; 95% CI, 0.7-2.6). There was no indication for any association between the highest salt intake category and the lowest in summary score variable and risk of gastric adenocarcinoma (HR, 1.0; 95% CI, 0.7-1.4). The HRs for noncardia gastric adenocarcinoma were similar to those for overall gastric adenocarcinoma (Table 2). All analyses were also conducted specifically by gender, but no gender-specific effects were seen.

Because the assumptions of proportional hazards were violated in some analyses, stratification for age was carried out. However, almost no differences were found between the stratified analyses and those derived from the Cox regression models, and no level of statistical significance changed. Therefore, only the results from the latter models are presented (Table 2). The proportion of missing values for both of the two primary salt variables among the study participants was ∼1 of 6. When the group with missing information was analyzed as a separate group, no evidence of an increased risk of gastric adenocarcinoma for this group was revealed. To further identify potential characteristics of nonresponders, we conducted univariate analyses with every variable chosen as a potential confounder in the multivariable analyses. No clear pattern was observed, although nonresponders were on average 1 year younger than the cohort mean age, and nonresponders tended more often to be never users of cigarettes and alcohol, to have lower education, and to have a less qualified occupation (data not shown).

This investigation, conducted in a western low-incidence population in Norway, provided no evidence of an association between dietary salt intake and risk of gastric adenocarcinoma.

A public health goal recommended by the World Cancer Research Fund/American Institute for Research on Cancer is that the population average consumption of salt from all sources should be <5 g/d (6). The WHO states that salt “probably” increases the risk of gastric cancer (12), but most investigations have assessed this association retrospectively, with an inherent problem with recall bias, and have been conducted in high-incidence populations (3). Few prospective studies have been undertaken and their results are conflicting. In a U.S. cohort study of White men, a positive association was found between salted fish and bacon and risk of gastric cancer (13). In a Japanese cohort study, identifying 116 cases of gastric cancer, high intake of salted foods was not associated with an increased risk (14). In another Japanese cohort, comprising 93 patients, it was, however, found that dietary salt intake was associated with risk of gastric cancer (15). Another Japanese cohort study, including 486 cases, showed a dose-dependently increased risk in men after adjustment for potential confounders, not including Helicobacter pylori (16). In the Netherlands Cohort Study, 282 incident cases of gastric cancer were used in a case-cohort analysis, and salt intake was not associated with an increased risk (17).

Advantages of this study include the prospective design, large sample size, complete follow-up, information on several potentially confounding factors, and information on anatomical localization and histologic type of the gastric tumors. Moreover, this study population is considered to have a diet rich in salt (has a high exposure prevalence). The a priori decision to exclude the first 3 years of follow-up was based on several considerations. First, detection bias of a yet undiagnosed gastric cancer could represent a possible problem. Second, the induction time between a harmful dietary exposure and the detection of an invasive gastric adenocarcinoma is usually substantial, and any dietary risk exposures occurring only close in time to the diagnosis might not influence the cancer development. Third, estimates of the prevalence of potentially confounding factors might be biased (e.g., a person might stop smoking before a gastric cancer is diagnosed due to the prediagnostic symptoms; ref. 9).

Disadvantages include the lack of information on some potential confounding factors, including H. pylori infection and intake of fruit and vegetables. However, the lack of association between salt and gastric adenocarcinoma in the current study cannot conceivably be explained by negative confounding by H. pylori or a low intake of fruit and vegetables, because these exposures tend to be positively linked to both gastric adenocarcinoma and salt intake (2, 3, 18-20). Based on available information, salt intake is unlikely to be confounded by H. pylori infection. If a positive association between salt intake and gastric cancer would have been revealed, the lack of information on H. pylori infection would have been a greater threat to the validity of the study. There is no information available on potential changes in the prevalence of salt intake, but it is unlikely that material changes would have occurred during the study period. The proportion of people with missing information on the salt intake variables is another concern, but there were no indications of confounding when the group with lacking information was analyzed separately.

Some of the previously reported positive findings regarding the association between salt and gastric cancer might reflect confounding by infection with H. pylori (21), a strong and well-established risk factor for stomach cancer (18), and this infection could act as an effect modifier regarding the effect of sodium on the gastric mucosa (22, 23). High salt intake might also be associated with an increased risk of H. pylori infection (19, 20). Salt intake seems to be associated with an increased dietary intake of foods rich in nitrites (17), which could be converted in the stomach to nitrosamines with an established carcinogenic effect on the gastric mucosa (24). These reactions are favored by a high pH, which occurs in a stomach with mucosal atrophy and intestinal metaplasia due to H. pylori infection. The higher prevalence of H. pylori in high-incidence Asian countries might explain the link between salt and gastric cancer in those populations.

In conclusion, in this prospective cohort study with complete cancer follow-up of a large low-incidence Western population, high intake of dietary salt was not associated with an increased risk of gastric adenocarcinoma.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the Norwegian Institute of Public Health and the HUNT Research Centre (Verdal, Norway) for performing the HUNT survey and the HUNT Research Centre and the Medical Faculty of the Norwegian University of Science and Technology (Trondheim, Norway) for allowing access to the database.