Diet and High Risk of Stomach Cancer in Shandong, China1

(CANCER RESEARCH 48, 3518-3523, June 15, 1988| Diet and High Risk of Stomach Cancer in Shandong, China1

Wei-Cheng You,2 William J. Blot,3 Yun-Shang Chang, Abby G. Ershow, Zhu-Tian Yang, Qi An, Brian Henderson, Guang-Wei Xu, Joseph F. Fraumeni, Jr., and Tian-Gen Wang Beijing Institute for Cancer Research, Beijing, People's Republic of China [W-C. Y., Q. I., G-W. X., T-G. WJ; National Cancer Institute, Bethesda, Maryland 20892 IW. J. B., A. G. Â£.,J. F. F.]; Weifang Medical Institute, Weifang [Y-S. CJ; Linqu Public Health Bureau, Linqu, Shandong, People's Republic of China [Z-T. Y.J; and University of Southern California, Los Angeles, California 90033 [B. H.J

ABSTRACT age and sex group was randomly selected from all such persons in the roster. The number of controls was chosen to be twice the number of A case-control investigation involving interviews with 564 stomach cases, with a similar distribution with respect to age and sex. cancer patients and 1131 population-based controls was conducted to A structured questionnaire was developed for use in this study. evaluate reasons for the exceptionally high rates of stomach cancer in Although the longest section focused on diet and dietary habits, infor Linqu, a rural county in Shandong Province in northeast China. Daily mation was obtained on demographic variables, medical history, occu consumption of sour pancakes, a fermented indigenous staple, was asso pation, smoking, and other items. Questions were asked about fre ciated with a 30% increase in risk. Risks of stomach cancer were also quency and portion size of 85 food items consumed several years prior increased by 2- to 3-fold among persons with prior chronic gastritis or to interview (about 1980) and just prior to the Cultural Revolution gastric ulcer, by 80% among those with stomach cancer in a family (about 1965). The questionnaire also included items on beverages, member, by S0% among men who smoked one or more packs of ciga including sources and storage of fresh water, and on salt intake and rettes/day, by 40% among those who preferred salty foods, and by 50% preference for salty foods. All interviews were conducted by 4 interview among families with moldy grain supplies. In contrast, risks tended to ers trained in use of the questionnaire, usually in the homes of the decrease in proportion to increasing consumption of fresh vegetables and study participants (96% of cases and 75% of controls). As a quality fruits. This protective effect was more pronounced for vegetables, with control measure, the interviews were tape recorded for later review by those in the highest quartile of intake at less than one-half the risk of the field supervisor. those in the lowest. Stomach cancer risks also declined with increasing The measure of association between stomach cancer risk and expo dietary intake of carotene, vitamin C, and calcium, but not retino!. These sure variables was the OR.4 Adjusted OR estimates were obtained by findings provide leads to dietary factors that contribute to the high rates the Mantel-Haenszel technique and by stratified logistic regression in Linqu, where stomach cancer is the leading cause of cancer and has analyses (3). Intake of individual foods, food groups, or nutrients was not yet begun to decline as in other parts of the world. categorized into quartiles or fertiles defined by yearly consumption among controls. Foods were grouped into several categories (e.g., fresh INTRODUCTION vegetables, fruits, alunni vegetables), with consumption of the group obtained by summing across food items. Indices for daily intake of Stomach cancer is the leading form of cancer in China (1). retino!, carotene, ascorbic acid, calcium, and other nutrients were also There is substantial geographic variation in mortality within created by multiplying the estimated nutrient content of each food, the country, however, with elevated rates in the northern and derived from Chinese food composition tables (4), by its yearly con central provinces. Some of the highest stomach cancer death sumption after weighting by portion size. Since results for food intake in 1965 and 1980 were similar, we present only the data for the later rates are found in Linqu, a rural county in Shandong Province. Linqu's annual age-adjusted (China standard) rates/100,000 period. Since findings for males and females also were similar, results are presented for both sexes combined, with the OR adjusted for sex. were 55 among males and 19 among females during 1980- Finally, analyses of the major variables were conducted for histologi- 1982, approximately double the national levels. To investigate cally confirmed cases separately from cases confirmed by other means, reasons for this clustering of high mortality, a collaborative but except where noted, the patterns of risk tended to be similar and investigation was launched in 1984 following completion of a thus only results for the total case series are presented. pilot study in 1983 which established procedures for study design and conduct (2). Herein, we report findings from this case control study, which focused on the contribution of dietary RESULTS factors to the etiology of stomach cancer. A total of 685 patients with stomach cancer were identified over the study period. Among these 41 had died, 70 were too METHODS ill, 8 refused interview, and 2 had only partial interviews. Interviews were thus completed with 564 stomach cancer pa Incident cases of stomach cancer were identified from county and tients and 1131 controls (only one refused interview). Fifty % commune hospitals in Linqu and neighboring Yidu County. All cases among long-term (10+ years) Linqu residents aged 35-64 diagnosed of the cancer diagnoses were derived from histological review over a 2.5-year period during 1984-86 were eligible for inclusion and of tissue specimens obtained during surgical procedures or sought for interview. The diagnoses were reviewed locally, with at endoscopy, 32% from surgery or endoscopy without patholog tempts made to obtain tissue specimens as frequently as possible. ical review, and 17% from radiological and clinical examina Controls were randomly selected from age and sex strata of the tion. Among those with histological confirmation, the ratio of Linqu population. The control selection process utilized population intestinal to diffuse type cancers was 4.7:1. For 312 patients, rosters contained in the headquarters of each of the 995 production the exact location of the tumor within the stomach could be brigades in Linqu. The rosters contained the name, age, and sex of specified: 63% were found in the antrum, 19% in the body, 4% every individual resident. A 2-stage sampling procedure was used. After in the cardia and fundus, 3% in the pyloric region, and 11% the random selection of a production brigade, an individual in a given elsewhere. Received 11/16/87; revised 3/16/88; accepted 3/18/88. Table 1 shows that the distribution of cases and controls was The costs of publication of this article were defrayed in part by the payment similar with respect to several demographic indices. Almost all 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. the participants were born in Linqu. Most were farmers and 1Supported in part by National Cancer Institute Contract N01-CP2-1012. had little or no formal education. The cases and controls 2To whom requests for reprints from China should be addressed. 1To whom requests for reprints other than from China should be addressed. 4 The abbreviations used are: OR, odds ratio; CI, confidence interval.

3518

Table 1 Distribution afeases and controls by sex, age, birth place, education, Table 2 Relative risks of stomach cancer in relation to dietary consumption of occupation, and family income staples, vegetables, and protein sources CasesSexMaleFemaleAge CI0.7-1.20.6- StaplesCornWheatDried 100<135Â£180>180<52<82<125>125if<84>84156<5<12<30>30s7<13<20>20<,S2.5s5>5OR*1.00.90.81.11.00.91.01.11.00.60.71.00.81.01.31.00.60.60.41.00.60.50.61.00.60.60.81.00.70.80.695% .10.8- .50.7-

(yr)35-4950-5960-69Place .30.7- .40.8- .60.4-0.70.4-0.70.6-0.91.0-1.60.4-0.80.4-0.80.3-0.60.5-0.90.4-0.70.4-0.80.4-0.80.4-0.80.6-1.10.5-0.90.6-1.10.4-0.8

ofbirthLinquNot potatoSweetsweet LinquEducationNo

pancakeSour schoolingSome schoolingOccupationUnemployed/housewifeFarmerOther"Family pancakeVegetables

fruitsTotaland freshvegetablesFresh incomeLowHighNo.44312111224520754221261303128319117270293%78.621.419.943.436.796.33.746.353.722.756.620.748.052.0ControlsNo.888243212496423109240488643201653277414716%78.521.518.743.937.496.53.543.257.817.857.724.537.063.0

Â°Truck/tractor driver, factory worker, clerk, and unclassified. fruitsProtein differed, however, in family income, with a significantly higher percentage of the cancer patients from families of low financial sourcesAnimal status. foodSoybeankg/yr*s Food Groups and Individual Foods. Table 2 shows that the intake of grains (corn and wheat), the major dietary staples in Linqu, was unrelated to the risk of stomach cancer. Consump tion of dried sweet potatoes was associated with a reduced risk, but little trend with intake level was seen. Whereas eating sweet pancakes was associated with a reduced risk (OR = 0.8; 95% CI, 0.6-0.9), sour pancakes were associated with an increased " Levels correspond to quartile distribution of intake among controls, except as noted in Footnotes <â€¢-<-. risk (OR = 1.3; 95% CI, 1.0-1.6). Individuals tended to eat one * OR adjusted for sex, age, and family income. or the other, but not both types of pancakes, with about one- ' Of the controls, 35% did not eat dried sweet potatoes; among consumers approximately one-half ate more than 84 kg/year. half the controls eating sweet and one-half eating sour pancakes d Of the controls, 48% ate sweet pancakes less than daily, over 90% not eating daily. them at all. Yearly consumption of total fresh vegetables and fruits was ' Of the controls, 47% ate sour pancakes less than daily, over 90% not eating calculated from 36 kinds of common vegetables and 9 kinds of them at all. fruits. Both fresh vegetables and fruits showed a negative dose- response relationship to the risk of stomach cancer (Table 2). sources of protein in Linqu County. Negative relationships were Cross-classifications of vegetable by fruit intake (not shown) seen, but there were no clear dose-response trends. revealed decreasing trends in risk with rising vegetable intake Nutrients. The relationship between stomach cancer risk and in every quart ile of fruit intake, but weaker and less consistent estimated daily intake of the nutrients ascorbic acid, retinol, trends with fruit consumption within the vegetable quimiles. total carotene, and calcium are presented in Table 3. Some We also examined trends with several groupings of vegetables, reduction in risk was seen for higher values of each index. These including dark green, cruciferous, and alliuni vegetables. The indices were correlated with each other, with the highest pair- first two categories were so highly correlated with total vege wise association between vitamin C and carotene (r = 0.6). In table intake (r = 0.8-0.9) that independent effects could not be a logistic model containing categorical terms for vitamin C, easily evaluated, but correlation between allium vegetables retinol, and calcium (carotene was excluded because of its high (composed of garlic, garlic stalks, scullions, Chinese chives, and correlation with vitamin C), independent protective effects were onions) and other vegetables was much less (r = 0.3). There found for vitamin C and calcium, with little trend for retinol. were strong inverse trends in stomach cancer risk with increas Salt and Salty and Moldy Food. Table 4 shows relative risks ing allium intake which persisted after adjustment for intake of of stomach cancer associated with consumption of salt, salted other fresh vegetables (the adjusted OR for highest compared vegetables, salted fish, moldy food, and a preference for salty to lowest allium quartile intake was 0.5; 95% CI, 0.4-0.8), and foods. The information on salt intake was based on household with increasing nonallium vegetable intake which persisted after rather than indi' -dual consumption. No strong associations adjustment for intake of allium vegetables (the adjusted OR for were found between reported salt or salty vegetables consump highest compared to lowest nonallium quartile intake was 0.6; tion and the risk of stomach cancer, but those who were in the 95% CI, 0.4-0.9). highest level of salted fish intake or who preferred foods saltier The last two variables in Table 2 are animal foods (pork, than average had a 40% increase in risk. An OR of 1.5 (95% eggs, beef, mutton, and poultry) and soybeans, the major CI, 1.2-2.0) was associated with mold in the family grain 3519

Table 3 Relative risks of stomach cancer in relation to intake of nutrients 95% CI, 0.6-2.8), or other sources of water (OR =1.1; 95% NutrientsVitamin CI0.5-1.00.5-0.80.3-0.60.6-1.20.4-0.80.7-1.40.5-0.80.4-0.80.3-0.60.5-0.80.3-0.60.4-0.8CI, 0.8-1.5). For subjects (about one-fourth of the population) C(mg)Retinol (low)2341 who reported that they sometimes drank unboiled water, neither an elevated risk nor a clear dose-response trend was found. Tobacco and Alcohol. The relationships of stomach cancer risk to cigarette smoking and alcohol consumption are given in (inCarotene (low)2341 Table 5. Since only a few females ever smoked cigarettes or drank alcohol for more than 6 months, the results are presented only for males. Smokers were at increased risk of stomach (mg)Calcium (low)2341 cancer, with a 50% excess among men who smoked one or more packs of cigarettes/day (dose-response trend test: P = 0.01). The association with smoking was stronger when anal yses were restricted to histologically confirmed cases, with the (mg)Intake/day*1 (low)234OR*1.00.70.60.51.00.90.61.01.00.60.60.51.00.60.40.695%OR among one pack/day smokers exceeding 2.0. Alcohol con sumption showed no positive association with stomach cancer risk. Â°Intake divided into approximate quartile level of average daily consumption. Stomach Diseases and Family Cancer History. Participants hAdjusted for sex, age, and family income. were asked about a history of stomach diseases diagnosed at least 5 years prior to the time of investigation. Table 6 shows Table 4 Relative risks of stomach cancer in relation to salt and salted and moldy ORs of 2.9 (95% CI, 1.7-4.6) and 2.0 (95% CI, 1.2-3.2) for food intake prior chronic gastritis and stomach ulcers, respectively. An OR No. of subjects of 1.4 (95% CI, 1.1-1.9) was observed for individuals reporting ORÂ° Case Control 95% CI at least one case of cancer among first degree relatives. The Persaltconsumption<1 capita household familial excess was accounted for entirely by an increased 3kg/yrS17 frequency of stomach cancer: the OR for familial stomach kg/yr<20 cancer was 1.8 (95% CI, 1.3-2.7), while no increase of other kg/yr>20 cancers was detected among family members (OR = 1.0; 95% kg/yrSalted CI, 0.7-1.5). This familial association was stronger for diffuse vegetables

fish<0.5 kg/yr<1 Table 5 Relative risks of stomach cancer among males in relation to cigarette kg/yr>1 smoking and alcohol drinking kg/yrEating Number of subjects grainNeverSometimesOftenFamily'smoldy Case Control ORÂ° 95% CI Cigarettes smoked/day 0 62 163 1.0 <20 158 326 1.3 0.9-1.9 moldyNeverYes1422128412630536831303491393031224374777446114307371182271701062211301618289607235916491579601661.01.21.11.11.01.11.01.11.41.01.0\A1.02.10.91.01.51.0-1.70.7-1.50.8-1.40.7-1.80.8-1.61.0-1.90.8-1.40.8-1.51.3-3.20.7-1.21.2-2.0grain >20 223 399 1.5 1.0-2.1

Alcoholdrinking/yr0<365 " Adjusted for sex, age, and family income. times>365 times1401501532503153231.00.90.80.7-1.30.6-1.1 supply. The OR was 2.1 (95% CI, 1.3-3.2) among those who "Adjusted for age and family income. Smoking also adjusted for alcohol said they sometimes ate the moldy grain, but frequent con drinking and alcohol drinking also adjusted for cigarette smoking. sumers of moldy grain were not at elevated risk. Eating Habits. Eating meals off schedule was associated with Table 6 Relative risks of stomach cancer in relation to history of gastric disorders an increased risk, but few cases reported eating in an irregular and occurrence of cancer in a close relative custom. Eating staple foods at hot temperatures was associated Number of subjects with significantly increased stomach cancer risks. The ORs for ORÂ° eating gruel and noodles at hot temperatures were 1.6 (95% CI, Case Control 95% CI 1.2-2.0) and 1.7 (95% CI, 1.4-2.2), respectively. We also Stomach ulcer* No 529 1094 1.0 assessed risks associated with eating spicy, sour, or sweet foods. Yes 35 37 2.0 1.1-3.2 Little or no differences were observed between cases and con trols in their preference for sour (OR =1.1; 95% CI, 0.9-1.4), Chronic gastritis* No 517 1099 1.0 sweet (OR = 1.0; 95% CI, 0.8-1.2), or spicy (OR = 0.8; 95% Yes 47 32 2.9 1.7-4.6 CI, 0.7-1.0) foods. Family's cancer history Drinking Water. Drinking water in Linqu was usually ob No 459 965 1.0 tained from surface wells. Relative to surface drinking water, Yes 102 155 1.4 1.1-1.9 some elevations in the OR were associated with drinking deep " Adjusted for sex, age, and family income. well water (OR = 1.5; 95% CI, 1.1-1.9), river water (OR = 1.3; * Stomach ulcer and chronic gastritis were diagnosed 5 years prior to interview. 3520

income, sour pancakes, dried sweet potatoes, allium vegetables, to OR estimates that may be biased towards the null value of other fresh vegetables, animal foods, salty fish, preference for 1.0. However, we did find significantly elevated risks for indi salty food, moldy grain, eating gruel at hot temperatures, drink viduals who had a preference for salty food and a high intake ing deep well water, stomach ulcer, and chronic gastritis. The of salted fish (although the amount of fish consumed in Linqu OR estimates from the logistic regression analysis were quite is small). Thus, salt may contribute in part to the area's high similar to those from the univariate and stratified analyses, rates of stomach cancer. Although comparable data throughout although the associations with sour pancakes and drinking deep China are not available, the levels of estimated salt intake well water were somewhat decreased and the associations with (median, 46 g/day) in Linqu were high by Western standards. salted fish and chronic gastritis somewhat enhanced. The This figure is also more than double the per capita daily salt regression model was repeated for males alone after adding intake reported in a survey of 118 counties in Henan Province terms for cigarette smoking; the trend of increasing risk with where the mean level was 17 g/day, and maximum county level amount smoked persisted and the estimates for the other vari was 31 g/day in the 1960s and 1970s (21). Among counties in ables were essentially unchanged. Henan there was also a significant correlation between salt intake and stomach cancer mortality. Most striking were the protective effects of vegetable intake. DISCUSSION Risks of stomach cancer among those in the upper quart ile of This relatively large population-based study suggests that vegetable consumption were less than one-half of those in the dietary factors contribute significantly to the etiology of stom lowest quartile. A similar but less pronounced association was ach cancer in Linqu. Elevated risks were associated with con seen for fruit intake. The findings are consistent with epidemi sumption of sour pancakes, a preference for salty foods, and olÃ³gica!studies of stomach cancer in Japan, North and South moldy grain supplies, while protective effects were associated America, and Europe (5, 6, 9-13), although reduced risks have with consumption of fresh vegetables. The findings are gener not always been found (22, 23). The mechanisms are unclear, ally in line with studies of stomach cancer in other parts of the but micronutrients have been suspected, particularly vitamin C world, which have implicated diets with high salt content and which may inhibit endogenous formation of carcinogenic ni- low fresh vegetable and fruit intake (5-13). trosamines (15). We found reduced risks among persons with The dietary constituent which distinguishes Linqu from other high intake of several nutrients, including carotene, vitamin C, areas in China is sour pancakes. These are made of corn, wheat, and calcium, but not retinol. Serological studies in areas of dried sweet potatoes, and a small amount of soybeans. Water Colombia at high risk of stomach cancer have shown lower is added to this mixture of flour which is then allowed to blood levels of carotene and tocopherol, but not retinol or ferment. Individuals who ate sour pancakes daily experienced vitamin C, among persons with gastric dysplasia than among a 30% increase in risk. Since about one-half the general popu persons with less advanced or no precursor lesions (24). Dietary lation of Linqu eats sour pancakes, we estimate that this ex deficiencies of total vitamin A have been reported among stom posure (if causally related to stomach cancer) could account for ach cancer patients in the United States (25), and among about a 15% elevation in stomach cancer mortality, thus ac persons with intestinal metaplasia, a precursor to stomach counting for only a portion of the nearly 100% excess above cancer, in Japan (26). national levels. Similar ingredients are used to make sweet The relationship of stomach cancer to low calcium intake in pancakes, but these are not fermented. Individuals in Linqu Linqu is of interest. Few studies have evaluated this association, reported eating one or the other, but generally not both types although a negative geographic correlation between dietary of pancakes, with adults nearly split in half as to preference for calcium and stomach cancer rates has been reported in Japan sour or sweet. Since these 2 variables are highly negatively (27). Animal studies indicate that salt-induced damage to the correlated, the increased risk associated with sour pancakes gastric mucosa might be inhibited by increasing intake of cal could reflect a protective effect associated with sweet pancakes. cium (28). Recent studies have indicated the possible role of This explanation seems unlikely, however, since sour pancake dietary calcium in inhibiting the development of colon cancer juice has shown evidence of mutagenic activity and contains N- (28, 29). Our study appears to be the first analytical investiga nitroso compounds (14), which are suspected to play a role in tion to report a possible protective effect of calcium against gastric carcinogenesis (15). Sour pancake juice also contains stomach cancer. mycotoxins (14), especially sterigmatocystin, which has pro Reductions in risk were associated with several of the indi duced papillomatous lesions of the stomach in laboratory ani vidual vegetables asked about, including dark green, cruciferous, mals (16). Mycotoxins may also play a role in the excess risk and allium vegetables. The finding with allium vegetables is associated with moldy family grain supplies, although no dose- new and of interest in view of recent experimental studies response relationship was seen. Further research is required to showing that oils or extracts from garlic and onion inhibit clarify the role of sour pancakes and moldy grain supplies, and several types of tumors in laboratory animals (30-33). A more their constituents, as risk factors for stomach cancer in Linqu. detailed description of the association is presented in a separate paper5, where it is shown that that protection seems to be Several investigations have linked high salt intake to the risk of stomach cancer, perhaps by inducing precancerous changes afforded by each of the several types of allium vegetables. A (5-8). High salt concentration, by irritating the gastric mucosa, common chemical constituent, such as an ally! sulfide com may result in desquamation and atrophy that involves both pound, with carcinogen inhibitory properties is suspected. It parietal and chief cells (17-20). Although we found relatively may also be that antibacterial and antifungal properties of weak associations between salt or salted food intake and stom allium inhibit gastric bacterial growth in humans (34), thus ach cancer risk, it was difficult to obtain precise information lessening the conversion of nitrates to nitrites and reducing the on an individual's salt consumption. A measure of per capita formation of JV-nitroso compounds (19). We had no informa salt intake was developed using the amount of salt purchased tion on levels of nitrates or nitrites in Linqu water or food for the household and the household size. Ranking of salt intake 5W. C. You, W. J. Blot, and Y. S. Chang. Allium vegetables and reduced risk based on this crude measure is subject to misclassification and of stomach cancer, submitted for publication. 3521

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 1988 American Association for Cancer Research. DIET AND HIGH RISK STOMACH CANCER supplies to help evaluate the hypothesis of endogenous produc its precursor lesions and thus help to develop measures aimed tion of jV-nitroso compounds including nitrosamines. Higher at preventing this cancer which is still very common in China risks were associated with drinking deep-well compared to and other parts of the world. surface waters, but relative nitrate contents are not known. However, a survey is planned to determine concentrations of ACKNOWLEDGMENTS nitrates, nitrites, and several nitrosamines in urine and gastric juice among persons with precancerous lesions, including We thank Drs. S. Zhao, S. Zhen, C. Yang, W. Lu, W. Xiao, M. chronic atrophie gastritis, intestinal metaplasia, and dysplasia. Wang, Z. Zhang, F. Gao, D. Wang, D. Yuan, and X. Liu for assistance In addition to dietary factors, our study revealed a positive in conducting this investigation and Dr. B. J. Stone for computational association between cigarette smoking and the risk of stomach advice and support. cancer, while alcohol drinking showed no effect. The OR as sociated with smoking one or more packs of cigarettes/day was REFERENCES 1.5. Although stomach cancer is not generally thought of as a smoking-induced cancer, similar excesses have been observed 1. Li, J. Y., Liu, B. Q., Chen, Z. J., Sun, X. D., and Kong, S. D. Atlas of cancer mortality in the People's Republic of China. Int. J. Epidemici., 10:127-133, among cohort studies of British physicians, U. S. military 1981. veterans, and men and women in Sweden and Japan, although 2. Wang, T. G., You, W. C., Henderson, B. E., and Blot, W. J. A case-control not all studies revealed dose-response relationships (35). Case study of stomach cancer in Shandong Province. Nati. Cancer Inst. Monogr., 69:9-10,1985. control studies of stomach cancer have not yielded as consistent 3. Breslow, N. E., and Day, N. E. Statistical methods in cancer research: the results, but the overall data suggest a moderate excess risk of analysis of case-control studies. IARC Sci. Pubi., 32: 1-338, 1980. stomach cancer among smokers. Such an effect may account in 4. The Chinese Academy of Medical Sciences. Chinese Food Consumption. Beijing, China: People Health Publishing House, 1984. part for the higher rates of stomach cancer among males than 5. Nomura, A. Stomach. In: D. Schottenfeld and J. Fraumeni (eds.), Cancer females in various populations around the world, including Epidemiology and Prevention, pp. 624-637. Philadelphia: W. B. Saunders, Linqu. 1982. 6. Correa, P., Cuello, C., Fajardo, L. F., Haenszel, W., Bolanos, O., and Chronic atrophie gastritis and dysplasia are closely associated DeRamirez, B. Diet and gastric cancer: nutrition survey of a high-risk area. with an increased risk of stomach cancer, especially the intes J. Nati. Cancer Inst., 70:673-678, 1983. tinal type which predominates in high-risk areas (36, 37). 7. Tuyns, A. Sodium chloride and cancer of the digestive tract. Nutr. Cancer, * 198-205, 1983. Although error in response may be expected from uncertain 8. Montes, G., Cuello, C., Correa, P., /arama. G., Liuzza, G., Zavala, D., knowledge about previous gastric diseases and specific termi Marin, E., and Haenszel, W. Sodium intake and gastric cancer. Cancer Res., 109:42-45, 1985. nology, we found a nearly 3-fold excess of stomach cancer 9. Hirayama, T. A Large-Scale Cohort Study on Relationship between Diet and among those reporting chronic gastritis. Differential recall by Selected Cancers of Digestive Organs, pp. 402-429. Cold Spring Harbor, cases and controls may have contributed to the finding, al NY: Cold Spring Harbor Laboratory, 1986. 10. Risch, H. A., Jain, M., Choi, N. W., Fodor, G., Feiffer, C., Howe, G., though we excluded gastritis reported within 5 years of inter Harrison, L., Craib, K., and Miller, A. B. Dietary factors and the incidence view. To a lesser degree, prior stomach ulcers were also related of cancer of the stomach. Am. J. Epidemiol., 122:947-959, 1985. 11. Trichopoulos, D., Ouranos, G., Day, N. E., Tzonou, A., Manousos, ()., to stomach cancer risk. The possible role of gastric ulcer in Papadimitriou, C., and Trichopoulos, A. Diet and cancer of the stomach: a stomach cancer risk has been debated (5), but the association case-control study in Greece. Int. J. Cancer, 36: 291-297, 1985. in high-risk areas is thought to result from an underlying 12. Correa, P., Fontham, E., Pickle, L. W., Chen, V., Lin, Y., and Haenszel, W. Dietary determinants of gastric cancer in south Louisiana inhabitants. J. atrophie gastritis (38). Our data cannot adequately evaluate this Nati. Cancer Inst., 75:645-654,1985. hypothesis, although the percentage of persons reporting gastric 13. La Vecchia, C., Negri, E., Decadi. A., D'Avanzo, B., and Franceschi, S. A case-control study of diet and gastric cancer in northern Italy. Int. J. Cancer, ulcer was the same (3%) whether or not they also reported Â¥0:484-489, 1987. chronic gastritis. We did detect some familial clustering of 14. Suan, H. L. Mutagenicity and .V-nitroso compounds in the sour pancake stomach cancer in Linqu. Familial predisposition to stomach juice. Second National Cancer Conference, Beijing, People's Republic of China, 1986. cancer has been reported in the past, with an excess of blood 15. Mirvish, S. The etiology of gastric cancer: intragastric nitrosamide formation type A among patients with diffuse cancers suggesting that the and other theories. J. Nati. Cancer Inst., 71:629-647, 1983. link may be genetic (5). The familial association in Linqu was 16. Purchase, I. F. H., and Watt, J. J. V. Carcinogenicity of sterigmatocystin. Food Cosmet. Toxicol., 8: 289, 1970. stronger for diffuse than intestinal cancers, but we were limited 17. Sato, I., Fukuyama, T., and Susuki, T. Studies of causation of gastric cancer in our ability to adequately evaluate differences by cell type 2. The relation between gastric cancer mortality rate and salted food intake because of the small numbers of tumors known to be diffuse. in several places in Japan. Bull. Inst. Public Health (Tokyo), S: 187-198, 1959. Familial clustering may also affect precursor lesions, with re 18. Kodama, M., Kodama, T., and Suzuki, H. Effect of rice and salty rice diets cent studies in Narino, Colombia indicating that chronic atropic on the structure of mouse stomach. Nutr. Cancer, 6:135-147, 1984. 19. Takahashi, M., Kokubo, T., Furukawa, F., Kurokawa, Y., Tatematsu, M., gastritis occurs primarily among those whose mothers had this and Hayashi, Y. Effect of high salt diet on rat gastric carcinogenesis induced condition (39). Elevated gastric nitrite also tended to occur in by /V-methyl-Ar-nitro-Ar-nitrosoguanidine. Gann, 74: 28-34, 1983. 20. Correa, P., Haenszel, W., Cuello, C., Archer, M., and Tannenbaum, S. A families, possibly reflecting transmission of a recessive gene model for gastric cancer epidemiology. Lancet, 2: 58-60, 1975. (40). 21. Lu, J. B., and Qin, Y. M. Correlation between high salt intake and mortality In conclusion, the risk of stomach cancer in Linqu was rates for oesophageal and gastric cancers in Henan Province, China. Int. J. Epidemiol., 16:171-176, 1987. increased by a low consumption of fresh vegetables and the 22. Ache-son, E. D., and Doll, R. Dietary factors in carcinoma of stomach: a associated nutrients carotene, vitamin C, and calcium, and by study of 100 cases and 200 controls. Gut, 5: 126-131, 1964. 23. Armiijo, R., Orellane, M., Medina, E., Coulson, A. H., Sayre, J. W., and a high consumption of sour pancakes, a preference for salty Detels, R. Epidemiology of gastric cancer in Chile I. Case-control study. Int. food, and moldy grain supplies. The dietary effects were not J. Epidemiol., JO: 53-56, 1981. accounted for by other risk factors in this population, such as 24. Haenszel, W., Correa, P., Lopez, A., Cuello, C., Zarama, G., Zavala, D., and Fontham, E. Serum micronutrients levels in relation to gastric pathology. cigarette smoking and a history of chronic gastritis. Our find Int. J. Cancer, 36:43-48, 1985. ings point to a complex of dietary variables which may contrib 25. Stehr, P. A., Gloninger, M. F., Kuller, L., Marsh, G., Radford, E., and ute to the high rates of stomach cancer in this area. Further Weinberg, G. B. Dietary vitamin A deficiencies and stomach cancer. Am. J. Epidemiol., 121:65-70, 1985. etiological research is needed to clarify the dietary constituents 26. Nomura, A., Yamakawa, H., Ishidate, T., Kamiyama, S., Masuda, H., Stem- and mechanisms that may induce or inhibit stomach cancer and merman, G., Heilbrun, L., and Hankin, J. Intestinal metaplasia in Japan: 3522

association with diet. J. Nati. Cancer Inst., 68:401-405, 1982. development of hamster buccal pouch carcinomas as expressed in tumor burden. Nutr. Cancer, 9:171-176, 1987. 27. Minowa, S., Takahashi, H., Kano, T., Matusyama, K., Araki, Y., and Yagi, 34. Block, E. The chemistry of garlic and onions. Sci. Am., 25/.-114-119, 1985. T. Studies on the relation between cancer and calcium. Kita Kanto Med. J., 10:713-718, 1960. 35. Surgeon General of the United States. Health consequences of smoking: cancer. U. S. Department of Health and Human Services, Washington, D. C. 28. Lipkin, M, and Newmark, H. Effect of added dietary calcium on colon U. S. Government Printing Office, 1982. epithelial-cell proliferation in subjects at high risk for familial colonie cancer. 36. Correa, P., Cuello, C., Duque, E., Burbano, L., Garcia, F., Solanos, O., N. Engl. J. Med., 313:1381-1384, 1985. Brown, C., and Haenszel, W. Gastric cancer in Colombia. HI. Natural history 29. Garland, A. H., Shekelle, R. B., Barrett-Connor, E., Criqui, M. H., Rossof, of precursor lesions. J. Nati. Cancer Inst., 57: 1027-1035, 1976. A. H., and Paul, O. Dietary vitamin D and calcium and risk of colorectal 37. Sipponen, P., Kekki, M., Haapakoski, J., Ihamaki, T., and Siurala, M. cancer: a 19-year prospective study in men. Lancet, Â¡:307-309,1985. Gastric cancer risk in chronic atrophie gastritis: statistical calculation of 30. Belman, S. Onion and garlic oil inhibits tumor promotion. Carcinogenesis cross-sectional data. Int. J. Cancer, 35: 173-177,1985. (Lond.), Â«:1063-1065,1983. 38. Correa, P. Precursors of gastric and esophageal cancer. Cancer (Phila.), SO: 2554-2565,1982. 31. Sparnin, V. L., Mott, A. W., Barany, G., and Wattenberg, L. W. Effect of allyl methyltrisulfide on glutathione 5-transferase activity and BP-induced 39. Bonney, G. B., Elston, R. C., Correa, P., Haenszel, W., Zavala, D. E., Zarama, G., Collazos, T., and Cuello, C. Genetic etiology of gastric carci neoplasia in the mouse. Nutr. Cancer, H:211-215,1986. noma. I. Chronic atrophie gastritis. Genet. Epidemiol., 3: 212-224, 1986. 32. Wargovich, M. J. Diallyl sulfide, a flavor component of garlic (Allium 40. Bonney, G. B., Elston, R. C., Correa, P., Tannenbaum, S. R., Haenszel, W., sath'um) inhibits dimethylhydrazine-induced colon cancer. Carcinogenesis Zavala, D. E., Fontham, E., Zarama, G., Gordillo, G., and Cuello, C. Genetic (Lond.), Â«.-487^189,1987. etiology of gastric carcinoma. II. Segregation analysis of gastric pH, nitrate, 33. Niukian, K., Schwartz, J., and Shklar, G. Effects of onion extract on the and nitrite. Genet. Epidemiol., 4:103-114, 1987.

3523

Wei-Cheng You, William J. Blot, Yun-Shang Chang, et al.

Cancer Res 1988;48:3518-3523.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/48/12/3518

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/48/12/3518. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.