Dietary factors and gastric cancer in Korea: A case-control study
To assess gastric cancer (GC) risk in relation to dietary intake in Korea, a case-control study was performed. Trained dietitians interviewed 136 patients diagnosed with GC, and the same number of controls were selected by matching sex, age and hospital. A significant decrease in GC risk was observed with increased intake of Baiechu kimchi (prepared with salted Chinese cabbage and red pepper, etc.), Baiechu kimchi-stew, garlic, mushroom and soybean milk. On the contrary, a significant increase in the risk of GC was observed with increased intake of cooked rice with bean, charcoal grilled beef, pollack soup, Kkakduki (a kind of kimchi prepared with salted radish and red pepper, etc.), Dongchimi (a kind of kimchi prepared with radish and a large quantity of salt water) and cooked spinach. In food groups, increased intake of soybean products was associated with decreased risk of GC. Intake of citrus fruits rather than total fruits was shown to have a protective effect on the risk of GC, but was not significant. In this study, intake of total vegetables was shown to have a protective effect, whereas high nitrate-containing vegetables increased the risk of GC. In conclusion, our study suggests that the risk of GC decreased with high consumption of fresh vegetables and fruits, whereas high consumption of foods rich in nitrate and carcinogenic substances produced during the cooking process increased the risk of GC. © 2001 Wiley-Liss, Inc.
Although gastric cancer (GC) incidence and mortality have been decreasing, it is still 1 of the common cancers around the world.1 GC is the most common cause of cancer death in Korea.2
The risk factors of GC are largely environmental, genetic and infection. Being infected with the bacterium Helicobacter pylori, 1 of the risk factors of GC, causes inflammation of gastric mucosa.3Helicobacter pylori infection per se, however, may not be sufficient to cause GC because there is low GC incidence despite high infection rate of Helicobacter pylori.4
Among environmental factors such as socioeconomic status, alcohol drinking, smoking, dietary habit and refrigerator use, the dietary factor is 1 of the most important risk factors of GC. Most studies on the relationship between diet and GC have suggested that foods rich in nitrate or nitrite, a high salt diet, smoked meat or fish and a high carbohydrate diet increase the risk of GC, whereas high consumption of fresh vegetables, fruits and dairy products decreases the risk of GC, although the findings are not always consistent.5–15
Dietary habits of Korean people can be characterized as a high intake of salty foods and carbohydrates and a higher intake of cooked rather than fresh vegetables. For this reason, it is necessary to investigate the effect of dietary factors on GC in Korea. Therefore, we conducted a case-control study, which focused on the association of dietary factors and GC in Korea.
MATERIAL AND METHODS
Trained dietitians interviewed a total of 157 patients highly suspected by endoscopic examination as having GC at Hanyang University Hospital and Hallim University Hospital from March 1997 to November 1998. A blind method was applied to avoid the information bias: all patients were interviewed immediately after endoscopic examination prior to histologic diagnosis in a condition to have no information about their own disease. The same number of controls were matched by sex and age (±2 years) during the same period in the same hospital. The hospital was considered as a matching factor based on the fact that most of the cases and controls appeared to reside in the area near the hospital where this study was carried out. We selected the controls among those who visited one of the clinics of orthopaedic surgery, ophthalmology, dermatology, plastic surgery and family medicine. They also were confirmed as having no gastric problems by gastroscopy. Among 154 GC patients confirmed by histologic method, 136 pairs were included in this analysis. The composition of males and females were 68.4% and 31.6%, respectively. The age group showing high GC incidence was 55–74 years of age (Table I).
Table I. General Characteristics of Subjects
The questionnaire was composed of general characteristics (age, sex, resident area, marital status, education level, job, income), familial history (gastritis, GC and other cancer), medical history of patients, smoking, drinking, duration of refrigerator use, vitamin supplement intake and dietary information.
Dietary data were collected by interviewer-administered questionnaires, using the quantitative food frequency method. All subjects were asked about the average frequency of intake and portion size of 109 food items for a 12-month period of 3 years prior to the interview. Before the dietary intake interview, trained dietitians asked subjects whether their dietary intake pattern of 3 years prior to the interview was different from that at the time of the interview. If the dietary pattern has changed, the dietitian helped subjects recall the dietary pattern of 3 years prior to the interview by reminding them of special events or changes in their personal life. The quantitative food frequency of 109 food items was revised from the semiquantitative food frequency questionnaire of 84 food items used in a large prospective cohort study (the Korean Cancer Research Survey), which was developed to consider commonly consumed food items based on nationwide data on food consumption in Korea.16 Kim et al.17 evaluated the validity and reproducibility of this semiquantitative food frequency questionnaire by administering it twice at an interval of 2 years and comparing it with 4 or 5 24 hr recalls collected at intervals of 3 months. The Pearson's correlation coefficients averaged 0.53 on the first questionnaire vs. 24 hr recalls and 0.37 on the second questionnaire vs. 24 hr recalls. The reproducibility determined by the correlation of nutrient intakes from 2 semiquantitative food frequency questionnaires was greater than 0.5. Some of food items developed by Lee et al.14 for the study on the association of GC with dietary intake in Korea and major sources of nutrients related to GC reported in other foreign studies were additionally added. Although a validation study was not additionally conducted as to whether these 109 food items were valid to our GC study, these food items are not largely different from the initial 84 food items. So it is thought that the association with GC and dietary intake of Koreans could be well assessed by these 109 food items. The information concerning portion size was asked through an open question about the intake amount according to the defined unit (cup, bowl, etc.).
Nitrate/nitrite values used in our study were taken from data analyzed by Shin and Namkung18 in Korea, since the Korea national food composition data do not contain information on nitrate. Radishes, young radish, spinach, malvaceae and beet showed a high concentration of nitrate in that study, so we regarded these foods as high nitrate-containing vegetables.
For statistical analysis, the total amount of each food intake was divided into 3 groups (<25, 25–75, >75 percentiles) under the criterion of the control group. For adjusted odds ratios (OR), corresponding 95% confidence intervals (CI) and trends were obtained by using unconditional logistic regression analysis. The factors associated with GC risk and food intake in our study were included in the regression model. All analyses were conducted using the SAS (version 8.1) program.
The total number of subjects was 136 pairs; 68.4% were males and 31.6% were females. The age group showing a high GC incidence rate was 55–74 years of age (Table I).
Table II shows OR of GC in relation to general characteristics. The risk of GC was decreased with increasing socioeconomic status but was not significant (OR = 0.52, CI = 0.25–1.08). The risk of GC was increased in the presence of the family history of GC, but was not significant (OR = 2.10, CI = 0.92–4.75). And duration of refrigerator use was shown to have a tendency to negative association (OR = 0.64, CI = 0.39–1.04).
Table II. Odds Ratios (OR) of Gastric Cancer in Relation to Socioeconomic Status, Family History and Refrigerator Use
| Medium||44||49||0.62 (0.33–1.16)|
| High||50||57||0.52 (0.25–1.08)|
|p-value for trend = 0.1736|
|Family history of gastric cancer3|
| Yes||19||10||2.10 (0.92–4.75)|
|Duration of refrigerator use|
| ≥20||51||66||0.64 (0.39–1.04)|
Table III presents ORs of smoking and drinking habits. Cigarette smoking showed the tendency to increase the risk of GC but was not statistically significant (OR = 2.05, CI = 0.84–4.96). In alcohol drinking according to the amount of alcohol consumed (gram alcohol/week), light drinking showed a protective effect on the risk of GC compared to nondrinking and heavy drinking (OR = 0.37, CI = 0.15–0.91).
Table III. Odds Ratios (OR) of Gastric Cancer in Relation to Cigarette Smoking and Alcohol Drinking History
|Cigarette smoking (pack-year2)|
| 1–11,000||59||55||1.76 (0.84–3.69)|
| ≥11,000||32||27||2.05 (0.84–4.96)|
|p-value for trend = 0.2425|
|Amount of alcohol drinking (g alcohol/week)|
| 1–45||8||22||0.37 (0.15–0.91)|
| >46||69||53||1.49 (0.84–2.65)|
|p-value for trend = 0.0115|
The relationship of food intakes and other factors, such as socioeconomic status, family history, refrigerator use, cigarette smoking and alcohol consumption was analyzed. Of these factors, socioeconomic status was associated with food intakes in our study (data not shown). So sex, age, socioeconomic status, family history and refrigerator use were included in the regression models.
Adjusted ORs of food items associated significantly and marginally with GC are shown in Table IV. A significant decrease in the risk of GC was observed with increased intake of Baiechu kimchi (prepared with salted Chinese cabbage and subingredients such as spring onion, garlic, red pepper, ginger, carrot and salted anchovy or shrimp), Baiechu kimchi-stew, garlic, mushroom and soybean milk. On the contrary, a significant increase in the risk of GC was observed with increased intake of cooked rice with bean, charcoal grilled beef, pollack soup, Kkakduki (prepared with salted radish and the same subingredients as Baiechu kimchi), Dongchimi (prepared with radish and a large quantity of salt water) and cooked spinach. Among these foods, charcoal grilled beef, pollack soup, cooked spinach, garlic, and mushroom were significant in trend.
Table IV. Odds Ratios (OR) of Gastric Cancer in Relation to Each Food Intake
|Cooked rice with bean2||1.00||2.01 (1.08–3.74)||1.70 (0.92–3.15)||0.0631|
|Charcoal grilled beef||1.00||3.43 (1.18–9.95)||2.11 (1.17–3.82)||0.0100*|
|Pollack soup||1.00||1.52 (0.84–2.77)||2.10 (1.16–3.82)||0.0454*|
|Baiechu kimchi stew||1.00||0.55 (0.31–0.99)||0.62 (0.29–1.35)||0.1294|
|Baiechu kimchi3||1.00||0.51 (0.28–0.94)||0.50 (0.25–1.01)||0.0686|
|Kkakduki4||1.00||1.98 (1.03–3.80)||1.78 (0.85–3.73)||0.1218|
|Dongchimi5||1.00||1.44 (0.79–2.63)||1.96 (1.01–3.83)||0.1413|
|Cooked spinach||1.00||0.87 (0.47–1.61)||1.85 (1.01–3.46)||0.0361*|
|Raw cucumber||1.00||0.57 (0.32–1.03)||0.87 (0.45–1.67)||0.1543|
|Garlic||1.00||0.50 (0.27–0.91)||0.53 (0.27–1.02)||0.0470*|
|Mushroom||1.00||0.38 (0.21–0.66)||0.30 (0.15–0.62)||0.0003**|
|Pear||1.00||0.57 (0.32–1.01)||0.63 (0.33–1.23)||0.1373|
|Fruit juice||1.00||0.71 (0.39–1.31)||0.55 (0.28–1.07)||0.1838|
|Soybean milk||1.00||0.72 (0.26–1.98)||0.50 (0.27–0.93)||0.0878|
According to Tables V and VI, food groups of meat, fish and shellfish by the cooking method were not shown to have an association with GC. Increased intake of soybean products was significantly associated with decreased risk of GC. Intake of citrus fruits rather than total fruits was shown to have a protective effect on the risk of GC but was not significant. In our study, intake of total vegetables was shown to have a protective effect, whereas high nitrate-containing vegetables increased the risk of GC.
Table V. Odds Ratios of Gastric Cancer in Relation to Food Groups of Meat, Fish and Shellfish by the Cooking Method
|Charcoal grilled meats2||1.00||1.28 (0.72–2.30)||1.58 (0.80–3.10)||0.4037|
|Total beef3||1.00||0.78 (0.42–1.44)||1.67 (0.86–3.27)||0.0660|
|Total pork4||1.00||0.80 (0.43–1.50)||0.94 (0.45–1.97)||0.7426|
|Meat and fish soups5||1.00||0.86 (0.46–1.59)||1.52 (0.72–3.23)||0.1902|
|Salted fishes and shellfishes6||1.00||0.83 (0.46–1.50)||0.78 (0.39–1.56)||0.7488|
|Fried meats and fishes7||1.00||1.19 (0.68–2.10)||0.73 (0.36–1.48)||0.3684|
Table VI. Odds Ratios of Gastric Cancer in Relation to Food Group of Vegetables, Fruits and Milk and Milk Products
| Total vegetables||1.00||0.43 (0.23–0.80)||0.64 (0.31–1.32)||0.0249*|
| High nitrate-containing vegetables2||1.00||1.67 (0.87–3.20)||2.17 (1.02–4.65)||0.1828|
| Raw vegetables||1.00||0.61 (0.34–1.09)||0.55 (0.28–1.09)||0.1579|
| Cooked vegetables||1.00||0.79 (0.44–1.43)||0.98 (0.50–1.90)||0.6734|
| Salted vegetables||1.00||0.91 (0.51–1.63)||1.48 (0.76–2.88)||0.2855|
| Total fruits||1.00||0.86 (0.48–1.55)||0.67 (0.33–1.39)||0.5578|
| Citrus fruits||1.00||0.60 (0.33–1.10)||0.66 (0.31–1.41)||0.2666|
|Milk and milk products||1.00||0.75 (0.42–1.35)||0.68 (0.34–1.36)||0.4988|
|Soybean products||1.00||0.81 (0.46–1.42)||0.35 (0.16–0.75)||0.0213*|
|Salty foods3||1.00||1.12 (0.62–2.03)||0.86 (0.41–1.82)||0.7178|
It has been recognized that remote dietary assessment is more difficult to assess exactly than other disease-related variables because recall bias can frequently occur. But among the dietary intake collection method, the food-frequency questionnaire recalling average long-term diet has become the primary method in nutritional epidemiologic study on chronic disease, especially cancer, having a latent period. In many studies, the validity of dietary assessments ranging from 3–10 years has seen strong correlations with the reference, generally in the range of 0.5–0.7.19–21 Overall, studies have suggested that diet may be recalled with acceptable levels of misclassification up to approximately 10 years. Considering that most of the GC patients in our study were older persons, it was thought that the usual diet for 12-months of 3 years prior to cancer diagnosis was valid.
The number of subjects to be selected for the study of a specific disease-exposures relationship is a fundamental consideration in planning a case-control study. As it was suggested by Schlesselman,22 the number of subjects was calculated by the following formula:
R denotes the relative risk, M is total number of pairs, m is the number of discordant pairs, p0 is the estimated proportion of exposed controls in the target population and p1 is the estimated proportion of exposed cases determined from equation. The quantities q0 and q1 are defined by q0= 1−p0 and q1= 1−p1, respectively. We assumed that the desired values of α and β were 0.05 and 0.10, respectively. And we assumed the relative risk of food intake on GC to be 2.0 (or 0.5) based on other studies.8, 10, 23 After calculation by the above equation, 173 pairs of GC and control subjects were needed in our study. But the final sample size investigated in our study was 136 pairs, which was smaller than the aimed number because some cases and controls could not be matched by age, sex or hospital.
Many epidemiologic and experimental studies have suggested that incidence of GC to a considerable extent is due to environmental factors such as socioeconomic status, alcohol drinking, smoking, refrigerator use, dietary habits and nitrate and nitrite compounds, among which dietary intakes have shown to have the strongest association with GC. In many studies, consumption of food rich in nitrate and nitrite, smoked food, a high salt diet and a high starch diet have seemed to be risk factors for GC, whereas high consumption of fresh fruits and vegetables, milk and milk products have seemed to be protective.5–15 Although controversial, it has been reported that dietary fat, meats, fish, seasonings and beverages such as coffee and tea are also associated with increased risk of GC.6, 10, 11, 15, 24
Intake of charcoal grilled beef significantly increased the risk of GC in our study. This association may be attributed to a carcinogen such as benzo[α]pyrene from smoke generated by the cooking process.25N-nitroso compounds, including nitrosamines, produced in the stomach through interaction of nitrites ingested or derived from nitrates with secondary and tertiary amines. Meat and fish proteins have been suggested as possible main sources for secondary and tertiary amines. In our study, although intake of charcoal grilled beef significantly increased the risk of GC, the grilled pork and grilled meats group (composed of a grilled beef and grilled pork over charcoal) were not either shown to have an association with GC. In some studies, consumption of beef has been considered a risk factor of GC,26, 27 whereas consumption of pork was shown to have a protective effect on the risk of GC.6, 28 Also in our study, the beef group was shown to have a positive tendency on GC more than the pork group. Pork has been consumed much more than beef by Koreans (the proportion of charcoal grilled pork was 63% to the total charcoal grilled meat group), so the result of total charcoal grilled meats group was largely influenced by pork. In summary, the reason why the different risk association was observed between the whole food group and specific items is that the specific food items are differently proportioned to the food group and are prepared by different cooking methods.
Radish, young radish, malvaceae, beet, celery and spinach are high nitrate-containing vegetables consumed commonly in Korea.18 Radish kimchi (such as Kkakduki and Dongchimi) and cooked spinach were associated with increased risk of GC in our study. The foods rich in nitrate and nitrite and processed meat added nitrate as preservative and coloring material is not carcinogenic per se, but in the gastrointestinal region, nitrate is converted to nitrite, so N-nitroso compounds are formed from nitrite reacting with amino substrates in food.29 Many of these compounds have been proven to be carcinogens in the animal study.30 But vegetables are the sources of antioxidants such as vitamin C and β-carotene, which inhibit the endogenous formation of N-nitroso compounds, as well as common sources of nitrate which contribute to formation of N-nitroso compounds. Therefore, vegetables may be protective rather than pose a risk factor for GC.
Baiechu kimchi and Baiechu kimchi stew were shown to have a protective effect on GC in our study. Kimchi is one of the most commonly consumed foods in Korea, which includes about 187 kinds and Baiechu (Chinese cabbage) kimchi and radish kimchi are the most typical kinds of kimchi. In Korea, generally, kimchi is prepared with Chinese cabbage and subingredients such as spring onion, garlic, red pepper, ginger, carrot and salted anchovy or shrimp. After preparation, kimchi is fermented, so it also contains high levels of organic acids. Following the main ingredients such as Chinese cabbage and radish, kimchi is named as Baiechu kimchi and radish kimchi including Kkakduki and Dongchimi, respectively. The major difference between Kkakduki and Dongchimi is to add red pepper and salty water. Kkakduki is composed of radish, red pepper and a normal amount of salt (2.5%), whereas Dongchimi is prepared with radish and a large quantity of salty water without red pepper. Most kimchi stew is prepared with Baiechu kimchi, not radish kimchi.
The subingredients of Baiechu kimchi such as spring onion, garlic, red pepper, ginger and carrot may contribute to a negative association with GC because these subingredients contain high levels of vitamin C, β-carotene, vitamin B complex, Ca, Fe, K, P and dietary fiber. The anticarcinogenic effect of kimchi has been reported in another study.31 But radish kimchi showed the different result with Baiechu kimchi. Most of the radish is the root, which readily accumulates nitrogenous fertilizer and nitrate contents are twice as high as Chinese cabbage. Although Chinese cabbage contains a moderate amount of nitrate, subingredients such as red pepper, garlic and ginger added to Baiechu kimchi may contribute to a protective effect on GC. And the fact that the risk associated with these single foods remains statistically significant despite uniform high consumption by the whole population suggests that the association may be stronger than it seems.
Some Koreans eat raw garlic as well as grind garlic for seasoning. Eating raw garlic was shown to have an anticarcinogenic effect in our study. This association may be attributed to natural components of garlic such as allicin and methyl linoleate. These phytochemicals appear to induce enzymatic detoxification systems, including mixed function oxidase systems and glutathione-S-transferase.25
In our study, mushroom was shown to have a protective effect on GC, although association of mushroom with GC is controversial in many animal studies32–35 and epidemiologic studies. So further studies are needed to prove the anticarcinogenic effect of mushroom.
Increased intake of soybean products was also associated with decreased risk of GC in our study. Soybean products (fermented soybean paste, soybean curd, soybean paste soup, etc.) have been shown to have a protective effect on GC in many other studies.14, 36–39 It seems to be contributed by the anticarcinogenic substances such as saponins and isoflavones of soybean.40
Salt is not a directly acting carcinogen, but a high salt intake could result in atrophic gastritis through direct damage to the gastric mucosa, which results in gastritis, increased DNA synthesis and cell proliferation.41 By these reasons, high intake of salty food has been considered as one of the risk factors for GC in some studies,6, 14, 24 but these were not universal findings.5, 7, 15 In our study, we defined the salty food as foods preserved with soy sauce or salt such as soybean paste, red pepper paste, salted and fermented fish products, meat boiled in soy sauce, cucumber preserved with salt and Zangachi (a sort of salted vegetables). In our study, high intake of these salty foods appeared to have no association with GC (OR = 1.12 [CI = 0.62–2.03] in medium intake and OR = 0.86 [0.41–1.82] in high intake). It is thought that the difference in the amount of salty food between the low intake and high intake groups may be not large enough to show the effect because most elderly Koreans have consumed high-salt food. As a part of the reason for this result, we can consider that the intake of sodium is difficult to quantify. In addition, this result may be confounded by soy sauce included in this salty foods group. Because soy sauce is one of the foods consumed very frequently in Korea and has been reported as an anticarcinogenic factor in other studies.31, 37
In conclusion, this study suggests that the risk of GC was decreased with high consumption of fresh vegetables and fruits, whereas high consumption of foods rich in nitrate and carcinogenic substances produced during the cooking process increased the risk of GC. Further studies are needed to clarify the effect of dietary factors on GC and the mechanisms involved in these processes and to improve food selection and cooking methods to prevent GC.