Journal of Epidemiology Vol. 9. No. 5 Nov

Comparative Study of Lifestyles of Residents in High and Low Risk Areas for Gastric Cancer in Province, China; with Special Reference to Allium Vegetables

Toshiro Takezaki 1, Chang-Ming Gao 2, Jian-Hua Ding 2, Ti-Kang Liu 3, Mao-Sheng Li 4, and Kazuo Tajima 1

There is a low risk area for gastric cancer in Jiangsu Province, China, where people frequently consume raw allium vegetables. As a first step epidemiological study to clarify the factors involved in the low incidence of gastric cancer, we conducted a comparative study of the ecological factors in a high risk area (HRA), , and a low risk area (LRA), Pizhou, using a questionnaire. Subjects were selected from the general population according to age and sex, and comprised 414 residents of the HRA and 425 residents of the LRA. Ecological factors were compared for the two areas by Cochran-Mantel-Haenszel method, age-adjusted. Smoking and drinking habits were found to be more common in the LRA. On the other hand, allium vegetables were consumed in the LRA much more frequently, with high consumption of raw vegetables, fruit, tomatoes, kidney beans and soybean products. People who consumed garlic en masse 3 times or more per week were 82% of men and 75% of women in the LRA, and 1 % of men and women in the HRA. The results of the survey suggest that frequent consumption of allium vegetables, in addition to other anticancer foods, may be a factor in low mortality for gastric cancer. J Epidemiol, 1999 ; 9 : 297-305

ecological study, gastric cancer, allium vegetables, China

China is one of the highest risk areas for gastric cancer in the few epidemiological studies showing a decreased risk of can- world1). Gastric cancer is the leading cause of cancer death in cer for allium vegetable consumption. One reason is that large- China, and it has slightly increased in the last two decades 2). scale consumption of allium vegetables is uncommon. Seven The increased risks of gastric cancer for income, occupation, epidemiological studies have shown a consistent association smoking, mental injury, overeating,high salt consumption, low between garlic consumption and a decreased risk of gastroin- animal protein consumption, and frequent intake of sweet pota- testinal cancer12).The negative association of allium vegetable toes, sour pancakes and a fermented staple have been consumption and the risk of nasopharyngeal and endometrial addressed in several studies in China 3-7).A positive relation- cancers has already been found in China 15.16).However, two ship between gastric cancer and Helicobacter pylori infection studies from China 17)and the Netherlands 18)have failed to has also been reported in China8,9).Moreover, a decreased risk show a significant association between allium vegetable con- of gastric cancer for high consumption of allium vegetables sumption and a decreased risk of lung cancer after adjustment has been reported in a study in Province, China 10). for other variables. Many experimental studies in vivo and in vitro have shown A comparative ecological study between high and low risk that extracts and compounds derived from allium vegetables, areas may yield valuable information, especially when the especially from garlic, have a variety of protective effects level of exposure to specific environmental factors is common against the development of cancer 11-14).There are, however, to people living in the same area but very different in two con-

Received November 26, 1998 ; accepted March 11, 1999. Division of Epidemiology, Aichi Cancer Center Research1 Institute, Nagoya, Japan. Division of Epidemiology, Cancer Institute of Jiangsu Province, , China.2 Bureau of Hygiene, Pizhou, China. 3 Cancer Institute of Yangzhong4 City, Yangzhong, China. Address for correspondence : Toshiro Takezaki, M.D., Division of Epidemiology, Aichi Cancer Center Research Institute , 1-1 Kanokoden, Chikusa-ku, Nagoya, 464-8681, Japan.

297 T. Takezaki, et al. 298

trasting high and low risk areas. We have found such high and The age-adjusted mortality rates (AMRs) for selected sites of low risk areas for gastric cancer in Jiangsu Province, China. cancer in Pizhou, and in urban and rural areas in all eight cities The people living in the low risk area have a diet unusually in Jiangsu Province were calculated using the data in the rich in raw allium vegetables. nationwide survey. Data on Yangzhong was obtained from We have hypothesized that frequent consumption of allium local cancer registries, including active confirmation at hospi- vegetables decreases the incidence of gastric cancer. As a first tals. Segi-Doll's World Population 19)was used as a standard step epidemiological study to clarify the factors involved in the population for age adjustment. low incidence of gastric cancer and to be followed by further Questionnaire studies were conducted independently in analytical studies, we have conducted an ecological study com- Yangzhong City and Pizhou City in 1995-1996. Subjects were paring the exposure level to environmental factors for gastric conveniently drown from the general population in 5 towns out cancer in high and low risk areas in Jiangsu Province, China. of 12 in Yangzhong and 5 towns out of 36 in Pizhou according to age and sex by trained local staff at health stations in each MATERIALS AND METHODS town, who interviewed the inhabitants. The age of subjects in Yangzhong and Pizhou was 30-76 (median: 46) and 30-79 Yangzhong City is located on an island in the River years (median: 45) in men, and 30-74 (median: 47) and 30-76 in the south of Jiangsu Province, China (Figure 1). The popula- years (medium: 44.5) in women, respectively. The response tion was given as about 273,700 in a 1992 census. Although rate for interview was more than 95% in both areas. The sub- the major industry of Yangzhong was formerly agriculture, the jects in Yangzhong and Pizhou comprised 201 men and 213 area's economic status has improved remarkably with the women, and 215 men and 210 women, respectively. recent development of manufacturing. Pizhou City, with a pop- The questionnaire included 152 items; demographics, water ulation of around 1.46 million, is located inland in the north of supply, dietary habits, food consumption frequency (at the time Jiangsu Province. The major industry of Pizhou is still agricul- of the interview and 10 years previously), smoking, alcohol ture, notably garlic production. drinking, and tea consumption. The items on smoking habit A nationwide survey of cancer incidence and mortality was included smoking status (current, ex- or never), age started conducted in selected cities in all the , smoking, age quit smoking, and number of cigarettes per day. between 1990 and 1992. Initial data from death certificates was Ex-smokers were defined as those who had quit smoking more actively confirmed at all hospitals. In Jiangsu Province, eight than one year prior to interview. Drinking habits included fre- cities, including Pizhou, were selected as representative of the quency, kinds of alcohol (beers, hard liquors, medium liquors, average rate of cancer mortality, according to data in a nation- and fruit liquors) and total amount of alcohol consumption. wide survey of 1973-1975. The combined population (6.99 Alcohol consumption for each type of beverage was deter- million)was 11.5% of the total population of Jiangsu Province. mined by the average number of drinks consumed at one time,

Figure 1. Jiangsu Province borders the Huang Sea to the east and City to the south. Yangzhong City is located on an island in the Yangtze River in the south of Jiangsu Province, and Pizhou City is located inland to the north. 299 Ecological Study of Gastric Cancer

which was then converted into alcohol weight (grams). One were much higher than those in Pizhou, or in urban or rural regular bottle (620 ml) of beer was equated to 22 grams of areas of Jiangsu Province for both men and women (Table 1). ethanol, and 100 ml of hard, medium and fruit liquors were The AMR for gastric cancer in Pizhou was also half of that in equated to 55, 38 and 40 grams of ethanol, respectively. Tea rural areas of Jiangsu Province. Colorectal cancer was less drinking habits included regularity of drinking, years of drink- common in Pizhou, and liver cancer was more common in ing, kind of tea (green,jasmine and black tea), and dose of tea Yangzhong and Pizhou than in urban areas. On the other hand, consumption. Tea consumption was rated according to the the AMR for lung cancer was higher in Pizhou and urban areas average number of grams consumed per month. than in Yangzhong. In Pizhou, the AMR for gastric cancer was The consumption frequencies of selected foods were rated in lower than the AMRs for cancers of the esophagus, liver or six categories and compared for the two areas after grouping in lung, in both men and women. The AMRs from causes other two categories. Allium vegetables included garlic (Allium than cancer in Yangzhong and Pizhou were similar (472.6 per sativum), Welsh onion (Allium fistulosum), onion (Allium 100,000 vs. 494.7 per 100,000). cepa) and Chinese chives (Allium tuberosum). Welsh onion was accordingly not included in green yellow vegetables. Demography Subjects were asked to record the frequency of garlic and Age and sex distribution of subjects was not different in the Welsh onion consumption only when these were consumed en two areas, although the proportion of subjects aged 60 years or masse, such as a clove of garlic (4-6 gram), as use of these over was relatively low overall (Table 2). Average age of sub- items as condiments was deemed to be insignificant. jects in Yangzhong and Pizhou was 47.7 and 47.4 years in The percentage proportion of each variable for the total men, and 47.1 and 46.6 years in women, respectively. Present number of subjects was calculated by area and sex. Then, and previous income of subjects in Yangzhong were higher lifestyles and dietary habits were compared in the two areas by than those in Pizhou. More subjects in Pizhou had had no edu- sex, using the Cochran-Mantel-Haenszelmethod, age-adjusted. cation, but the proportion of farmers was similar in the two The P-value was calculated after age-adjustment in 5 strata for areas. Higher quality water supplies, e.g., tap water or deep every 10 years between 30-79 years of age. Factor analysis well, were more common in Yangzhong than in Pizhou both at was done for a number of present variables, including area, the time of the study and 10 years previously. Such high quali- sex, income, education, occupation, water supply, smoking, ty water supplies were uncommon in both areas 20 years previ- alcohol drinking, tea consumption, dietary habits, cereal con- ously. sumption and 22 food items. Three factors were chosen according to the proportions of the correlation matrix follow- Smoking and drinking ing Varimax rotation. PROC FREQ and PROC FACTOR The prevalence rate for current and ex-smokers was higher from the SAS system were used for these calculations 20,21) in Pizhou than in Yangzhong (Table 3). The percentage pro- portion of everyday alcohol drinkers was not different in the RESULTS two areas, but more subjects in Pizhou had begun to drink before 20 years of age, had drunk for 20 years or more, or con- Age adjusted mortality rates for selected cancers sumed ethanol in quantities of 75 grams and more at one time, The AMRs for esophageal and gastric cancers in Yangzhong than in Yangzhong.

Table 1. Comparison of age adjusted mortality rates (AMRs) for selected sites of cancer in Yangzhong , Pizhou, and urban and rural areas in Jiangsu Province, by sex in 1990-1992.

a) Age adjusted mortality rate, using Segi-Doll'sWorld Population. b) Urban and rural areas with 11 .5% of total populationin Jiangsu Province. T. Takezaki, et al. 300

Table 2. Background characteristics of subjects inYangzhong (HRA : high risk area) and Pizhou (LRA : low risk area), by sex.

a)P values were calculated by age-adjusted Cochran-Mantel-Haenszel method. b)The subjects who had not graduated from elementary school .

Table 3. Comparison of smoking, alcohol and tea consumption of residents a high risk area (HRA) and a low risk area (LRA) . by sex.

a) P values were calculated by age-adjusted Cochran-Mantel-Haenszel method.

The proportion of habitual tea drinking was not different in Dietary habits and cereals the two areas, but men in Yangzhong consumed more tea over- In Yangzhong, mealtimes were less regular, and food was all, and green tea consumption was more common in taken at high temperatures more frequently by men in Yangzhong. Women consumed much less tea than men in both Yangzhong than in Pizhou (Table 4). More men and women in areas. Pizhou preferred salty foods than those in Yangzhong. 301 Ecological Study of Gastric Cancer

Table 4. Comparison of dietary habits and cereal consumption of residents of a high risk area (HRA) and a low risk area (LRA), by sex.

a)P values were calculated by age-adjusted Cochran-Mantel-Haenszel method .

Rice consumption was more common in Yangzhong. Pizhou than in Yangzhong. The consumption frequency of Noodles, buns and wheat crackers were consumed more fre- these allium vegetables had not changed over the previous 10 quently in Pizhou. Leftover rice gruel had been consumed years in either area. more frequently in Pizhou than in Yangzhong 10 years previ- Frequent consumption of foods, including garlic and Welsh ously. onions, was found not to be dependent on age in either area, when young (30-54 years) and old (55-79 years) subjects were Selected foods and Allium vegetables compared (data not shown in Tables). Meat and fish were consumed more frequently in Pizhou, but the proportion of people who consumed them 3 times or Factor analysis more per week stood at 10% or less in either areas (Table 5). Factor analysis yielded a first cluster of area-related variables Consumption of meat, fish, poultry, eggs and milk had positively associated with the characteristics of Pizhou resi- increased over the previous 10 years in both areas. dents, namely, frequent consumption of Welsh onions, wheat Processed and cooked foods such as fried food, fermented crackers, garlic, raw vegetables, kidney beans, soybean prod- food and soybean products were consumed much more fre- ucts, fermented food, tomatoes, eggs and fruit, less frequent quently in Pizhou. consumption of rice, lower income, and lower quality water Raw vegetables, kidney beans, tomatoes, sweet potatoes and supplies (data not shown in Tables). A second factor consisted fruit were consumed more frequently in Pizhou, but, as a of gender-related variables such as smoking and alcohol drink- whole, vegetables were consumed more frequently in ing. The variables associated with tea consumption clustered as Yangzhong. The consumption frequency of pickled vegetables a third factor. The cumulative rate for the attributable propor- had been higher in Pizhou 10 years previously. In Pizhou, tion of these three factors was 32.9%. trend towards raw vegetable and fruit consumption was found to be growing, while consumption of pickled vegetables and DISCUSSION sweet potatoes had decreased remarkably over the previous 10 years. The subjects in the present study were drawn from the gen- Consumption of allium vegetables was notably higher in eral population in both areas. The distribution of occupations in Pizhou. The proportion of people who consumed garlic en Yangzhong and Pizhou respondents was concordant with that masse 3 times or more per week at the time of the survey was obtainable from official records for each area. Local health 82.3% of men and 74.7% of women in Pizhou, and 0.5% of officials conducted interviews, and recorded responses, with a men and women in Yangzhong. Although onions and Chinese response rate of more than 95%. Selection bias in the present chives were not consumed as much as garlic and Welsh onions study, therefore, may be assumed to be negligible. However, overall, consumption of these was also more frequent in the proportion of the total population providing subjects was T. Takezaki, et al. 302

Table 5. Comparison of consumption of selected foods by residents of a high risk area (HRA) and a low risk area (LRA), 10 years ago and at present.

a) P values were calculated by age-adjusted Cochran-Mantel-Haenszel method . b) Consumption frequency: 3 times/week vs. less. Consumption frequency: frequent vs. occasionallyc) or less.

only about 0.15% in Yangzhong and 0.03% in Pizhou. It can- ty in the two areas as valid. not be claimed with confidence, therefore, that the present sub- To reduce interviewer bias in the distribution of lifestyles jects comprise a representative sample of the general popula- and dietary habits between the two study areas, interviews tion of each area, because random errors cannot be completely were conducted by not fewer than five trained local staff mem- excluded from the analysis. However, it is possible to hazard a bers at five towns in each study area. comparison of the characteristic lifestyles and dietary habits of The most interesting result of the present study is that allium residents in the two areas by applying appropriate statistical vegetables are consumed very much more frequently in the procedures to the subject data available. low risk area. Since Chinese people frequently take garlic and AMRs for Pizhou and Yangzhong were calculated using Welsh onion in small amounts as condiments, residents living data from a nationwide survey and from local cancer registries, in both the high and low risk areas can be said to consume respectively. Information from death certificates combined these allium vegetables. It seems, however, that a protective with active confirmation at hospitals in both areas decreases effect is biologically more significant when greater quantities the potential for misclassification of cancer deaths. Although of allium vegetables are consumed, such as a clove of garlic. there is no data on life expectancy in the two study areas, simi- We are not aware of formal data to the effect, but, according to lar rates of mortality from causes other than cancer in private communications, the high consumption of allium veg- Yangzhong and Pizhou support comparisons of cancer mortali- etables in Pizhou is peculiar to this area, and the level of con- VII . Lyon, France: International Agency for Research on Cancer, 1997: IARC Sci Pub No 143: 822-823 . 2. Li L, Lu F, Zhang S, et al. Trends of cancer mortality rate in China within 20 years and prediction in the future . Chinese Journal of Oncology 1997; 19: 3-9 (in Chinese) . 3. Lu JB, Qin YM. Correlation between high salt intake and mortality rates for oesophageal and gastric cancers in Henan Province, China. Int J Epidemiol 1987; 16: 171- 176.

303 Ecological Study of Gastric Cancer

sumption in Yangzhong is more typical of that in China in gen- consumption of pickled vegetables 37).However, the geographi- eral. cal distribution of AMRs for gastric cancer in the two areas Many experimental studies in vivo and in vitro have shown was not concordant with differences in the consumption of that extracts and compounds derived from allium vegetables, green tea and pickled vegetables. Moreover, consumption of especially from garlic, have a variety of protective effects cereals such as rice and wheat crackers was also different in against the development of cancer 11-14.22-26).Garlic is one of the two areas, and it is unclear whether these cereals have any several vegetables which contain increased levels of seleni- association with the risk of gastric cancer. um 27).Since allium vegetables seems to have great potential Food intake at high temperature, which is more common in as protective agents against the development of cancer, and a Yangzhong, may be related to the relatively high mortality of decreased risk of gastric cancer for garlic consumption has esophageal cancer in that area 38). been found in several case-control studies 10,28,29 , it is suggest- Higher quality water supplies may have the potential to ed here that the high consumption of allium vegetables in decrease the change to take contaminated water by carcino- Pizhou may be directly related to the low incidence of gastric gens, although these were more common in Yangzhong. cancer there. In summary, an ecological study was conducted comparing In addition, people in the low risk area frequently consume environmental factors for gastric cancer in high and low risk raw garlic and Welsh onion, and the anticancer activity seems areas in Jiangsu Province, China. Allium vegetables were con- to be associated with heat-sensitive components (such as sumed in the low risk area much more frequently, and raw allicin and other thiosulfinates) as well as heat-stable compo- vegetables, fruit, tomatoes, kidney beans and soybean products nents of garlic and onion 12).It is possible that raw garlic is were consumed more frequently, in the low risk area than in more effective against cancer development than cooked garlic, the high risk area. These results suggest that frequent con- as it retains the activity of heat-sensitive components, although sumption of allium vegetables, in addition to that of other anti- few studies have compared the anticancer activity of raw and cancer foods, may be implicated in low AMR for gastric can- cooked garlic. It is known that Helicobacter pylori infection is cer. In order to clarify the environmental and host factors asso- associated with the risk of gastric cancer 30),and it has also ciated with the risk of gastric cancer in these areas, a case-con- been shown that Helicobacter pylori is susceptible to garlic trol study, a seroepidemiological study for Helicobacter pylori extract 31),while heat treatment of garlic extract reduces its infection and a genetic epidemiological study on host factors inhibitory or bactericidal effects on Helicobacter pylori 32). It is are currently underway. therefore also suggested here that the antibacterial activity of raw garlic may play a protective role in inhibiting the develop- ACKNOWLEDGMENT ment of gastric cancer. The AMR for colorectal cancer was also lower in Pizhou. A The authors would like to thank Dr. Zhu, J., President, and decreased risk of colon cancer for garlic consumption has been Dr. Huang, F., Vice President, Jiangsu Cancer Hospital and found in previous studies 33,34) It is possible that frequent con- Cancer Institute of Jiangsu Province, for their cooperation . The sumption of garlic may be implicated in the low AMR for col- authors are grateful to the local staffs in Yangzhong and orectal cancer in Pizhou. Pizhou Cities for their assistance in data collection. This work The higher smoking rate in Pizhou may contribute to the was supported in part by a Grand-in-Aid for International higher AMR for lung cancer when compared with Scientific Research, Special Cancer Research (08042015) from Yangzhong. The higher AMR for lung cancer in Pizhou, the Ministry of Education, Science , Sports and Culture, Japan. despite frequent garlic consumption, is also concordant with the results of previous epidemiological studies that REFERENCES have failed to show any effect of garlic consumption in lowering the risk of lung cancer 17,18). 1. Parkin D, Muir C, Whelan S, Ferlay J, Raymond L, Raw vegetables, fruit, tomatoes, kidney beans and soybean Young J eds. Cancer incidence in five continents , Vol. products were also consumed more frequently in the low risk area. It is difficult to distinguish the protective effects of allium vegetables and these foods in an ecological study. For clarifica- tion of the protective effects of each item, analytical epidemio- logical studies controlled for confounding factors would pro- vide more information. Green tea consumption has been associated with a decreased risk of gastric cancer in previous studies 35.36),while an increased risk of gastric cancer has been reported for frequent T. Takezaki, et al. 304

4. You WC, Blot WJ, Chang YS, et al. Diet and high risk of tion, garlic supplement use, and the risk of lung carcino- stomach cancer in Shandong, China. Cancer Res 1988; ma in The Netherlands. Cancer Res 1994; 54: 6148-6153. 48:3518-3523. 19. Union Internationale Contre le Cancer (UICC). 5. Kneller RW, Gao YT, Mclaughlin JK, et al. Occupational Comparison between Registries. In: Doll R, Muir C, risk factors for gastric cancer in Shanghai, China. Am J Waterhouse J, eds. Cancer Incidence in Five Continents, Ind Med 1990; 18: 69-78. Vol. II. Berlin-Heiderberg-New York: Springer-Verlag, 6. You WC, Chang YS, Yang ZT, et al. Etiological research 1970:333-339. on gastric cancer and its precursor lesions in Shandong, 20. Hartmann WM, Sall JP, Sarle WS. The FACTOR proce- China. IARC Sci Publ 1991; 105: 33-38. dure. In: SAS/STAT User's Guide, Version 6, Fourth edi- 7. Wang R. The epidemiological study of subtype risk fac- tion, Volume 1. Cray, NC, SAS Institute Inc., 1990: 773- tors of gastric cancer. Chung Hua Liu Hsing Ping Hsueh 821. Tsa Chih 1993; 14: 295-299 (in Chinese). 21. Sall JP, Stanish WM. The FREQ procedure. In: 8. Forman D, Sitas F, Newell DG, et al. Geographic associa- SAS/STAT User's Guide, Version 6, Fourth edition, tion of Helicobacter pylori antibody prevalence and gas- Volume 1, Cray, NC, SAS Institute Inc., 1990: 851-889. tric cancer mortality in rural China. Int J Cancer 1990; 46: 22. Brady JF, Li D, Ishizaki H, Yang CS. Effect of diallyl 608-611. sulfide on rat liver microsomal nitrosamine metabolism 9. Zhou D, Yang H. Epidemiology of Helicobacter pylori and other monooxygenase activities. Cancer Res 1988; infection in the People's Republic of China. Chin Med J 48: 5937-5940. 1995; 108: 304-313. 23. Tadi PP, Lau BHS, Teel RW, Herrmann CE. Binding of 10. You WC, Blot WJ, Chang YS, et al. Allium vegetables aflatoxin B, to DNA inhibited by ajoene and diallyl sul- and reduced risk of stomach cancer. J Natl Cancer Inst fide. Anticancer Res 1991; 11: 2037-2042. 1989; 81: 162-164. 24. Ludeke BI, Domine F, Ohgaki H, Kleihues P. 11. Dorant E, Van Den Brandt PA, Goldbohm RA, Hermus Modulation of N-nitrosomethylbenzylaminebioactivation RJ, Sturmans F. Garlic and its significance for the preven- by diallyl sulfide in vivo. Cartinogenesis 1992; 13: 2467- tion of cancer in humans: a critical view. Br J Cancer 2470. 1993; 67: 424-429. 25. Takada N, Matsuda T, Otoshi T, et al. Enhancement by 12. Koch HP, Lawson LD, Reuter HD, eds. Garlic. The sci- organosulfur compounds from garlic and onion of ence and therapeutic application of Allium sativum L. diethylnitrosamine-induced glutathione S-transferase pos- and related species, 2nd edn. Baltimore: Williams and itive foci in the rat liver. Cancer Res 1994; 54: 2895- Wilkins, 1996. 2899. 13. Block E. Organoselenium and organosulfur phytochemi- 26. Hu PJ, Wargovich MJ. Effect of diallyl sulfide on cals from Genus allium plants (onion, garlic): relevance MNNG-induced nuclear aberrations and ornithine decaar- for cancer protection. In: Ohigashi H, Ossawa T, Terao J, boxylase activity in the glandular stomach mucosa of the Watanabe S, Yoshikawa T, eds. Food factors for cancer Wistar rat. Cancer Lett 1989; 47: 153-158. prevention. Tokyo, Japan: Springer-Verlag, 1997: 215- 27. Morris VC, Levander OA. Selenium content of foods. J 221. Nutr 1970; 100: 1383. 14. Steiner M. The role of flavonoids and garlic in cancer 28. Mei X, Wang ML, HX, et al. Garlic and gastric can- prevention. In: Ohigashi H, Osawa T, Terao J, Watanabe cer: the influence of garlic on the level of nitrate and S, Yoshikawa T, eds. Food factors for cancer prevention. nitrite in gastric juice. Acuta Nutr Sin 1982; 4: 53-56. Tokyo, Japan: Springer-Verlag, 1997: 222-225. 29. Buiatti E, Palli D, Decarli A, et al. A case-control study of 15. Zheng W, Blot WJ, Shu XO, et al. Diet and other risk gastric cancer and diet in Italy. Int J Cancer 1989; 44: factors for laryngeal cancer in Shanghai, China. Am I 611-616. Epidemiol 1992; 136:178-191. 30. Forman D. The etiology of gastric cancer. IARC Sci Publ 16. Shu XO, Zheng W, Potischman N, et al. A population- 1991; 105: 22-32. based case-control study of dietary factors and endometri- 31. Sivam GP, Lampe JW, Ulness B, Swanzy SR, Potter JD. al cancer in Shanghai, People's Republic of China. Am J Helicobacter pylori--in vitro susceptibility to garlic Epidemiol 1993; 137: 155-165. (Allium sativum) extract. Nutr Cancer 1997; 27: 118-121. 17. Swanson CA, Mao BL, Li JY, et al. Dietary determinants 32. Cellini L, Di Campli E, Masulli M, Di Bartolomeo S, of lung-cancer risk: results from a case-control study in Allocati N. Inhibition of Helicobacter pylori by garlic Yunnan Province, China. Int J Cancer 1992; 50: 876-880. extract (Allium sativum). FEMS Immunol Med 18. Dorant E, Van Den Brandt PA, Goldbohm RA. A Microbiol 1996; 13: 273-277. prospective cohort study on Allium vegetable consump- 33. Steinmetz KA, Kushi LH, Bostick RM, Folsom AR, 305 Ecological Study of Gastric Cancer

Potter JD. Vegetables, fruit, and colon cancer in the Iowa sumption and the risk of digestive tract cancers: data from Women's Health Study. Am J Epidemiol 1994; 139:1-15. a comparative case-referent study in Japan. Cancer 34. Le Marchand L, Hankin JH, Wilkens LR, Kolonel LN, Causes Control 1998; 9: 209-216. Englyst HN, Lyu LC. Dietary fiber and colorectal cancer 37. Tajima K, Tominaga S. Dietary habits and gastro-intesti- risk. Epidemiology 1997; 8: 658-665. nal cancers: a comparative case-control study of stomach 35. Kono S, Ikeda M, Tokudome S, Kuratsune M. A case- and large intestinal cancers in Nagoya, Japan. Jpn J control study of gastric cancer and diet in northern Cancer Res 1985; 76: 705-716. Kyushu, Japan. Jpn J Cancer Res 1988; 79: 1067-1074. 38. Cheng KK, Day NE. Nutrition and esophageal cancer. 36. Inoue M, Tajima K, Hirose K, et al. Tea and coffee con- Cancer Causes Control 1996; 7: 33-40.