J Cancer Res Clin Oncol (1991) 117:133-143

Jo.rn~lol

Cancer 1Research Ciinical 9

017152169100026Y

9 Springer-Verlag 1991

Guest editorial* Epidemiological research in stomach cancer: progress over the last ten years * H. Boeing German Cancer Research Center, Institute of Epidemiology and Biometry, Im Neuenheimer Feld 280, W-6900 Heidelberg, Federal Republic of Germany Received 10 October 1990/Accepted 26 October 1990

Summary. In this paper the progress o f epidemiological research in stomach cancer during 1980-l 990 is reviewed in respect to regional variation, etiology, and formation of carcinogens. The evaluation of 4 cohort and 16 casecontrol studies revealed a consistently inverse relationship of stomach cancer risk with raw vegetables, fruit, and wholemeal bread consumption and with vitamin C and carotene intake. Milk, cooked vegetables and vitamins A and E were not consistently found to be related to stomach cancer risk. Positive associations of increasing consumption with stomach cancer risk were occasionally found for processed or particularly prepared meat and fish, and for nitrite. Dietary nitrate intake did not appear to be related to stomach cancer risk in these studies. This latter observation is also supported by metabolic studies in high- and low-risk areas for stomach cancer. Consistently among studies, increased risk for stomach cancer was also found for later availability of refrigeration facilities in the household, non-centralized water supply (especially well water), and high salt intake. Prospective studies agreed in an increased risk for stomach cancer for cigarette smoking, but not for alcohol drinking, whereas case-control studies showed divergent results on these factors. Recent metabolic studies in highand low-risk areas for stomach cancer or in groups with precursor lesions, with the N-nitrosoproline test as a marker for endogenous nitrosation, revealed inconsistent results. Higher nitrite concentration and increased pH in stomach juice were found to be associated with precursor conditions for stomach cancer. It is still not clear whether intake of preformed carcinogens or endogenous formation in the stomach with or without the inclusion of nitrite is the most important source of tumor-initiating or -promoting substances. Preservation or preparation of

* The "Journal of Cancer Research and Clinical Oncology" publishes in loose succession "Editorials" and "Guest editorials" on current and/ or controversial problems in experimental and clinical oncology. These contributions represent exclusively the personal opinion of the autor The Editors Abbreviation." CI, confidence interval

meat and fish may play an important role in this process, and vitamin C may be an inhibiting substance.

Key words: Stomach cancer - Risk factors - Vitamin C Nitrate - Epidemiological studies

Introduction Worldwide estimates at the beginning of the 1980s show that stomach cancer is the leading cause of cancer incidence among all cancer sites with 670 000 new cases each year (Parkin et al. 1988). This seems to be in contrast to the statistic during the last decades in the highly industrialized countries, which show a remarkable decline in stomach incidence and mortality (Howson et al. /986). At the beginning of this century stomach cancer was the leading cause of cancer in the United States. It is now the 10th and 12th most common cancer incidence site in men and women with a decline in mortality of about 70% from 1950 to 1985 (American Cancer Society 1990). Other countries are following the trend in the US but with some delay (Stanley et al. 1988). Thus, on the worldwide scale many countries are still faced, sometimes concentrated in particular geographical areas, with high rates for this disease. Even before the Second World War, the first empidemiological studies on risk factors for stomach cancer had been implemented (Craver 1932; Stocks and Karn 1933; Herbert and Bruske 1936). During the 1950s and 1960s, epidemiological research into risk factors of stomach cancer was second only to lung cancer. By the end of the 1970s, results from nearly 20 case-control studies had been published (reviewed by N o m u r a 1982; Byers and Graham 1984; Frentzel-Beyme 1988). Also cohort studies had been established in the 1960s to investigate risk factors for stomach cancer. Overall, the results from these early epidemiological studies do not show a consistent pattern o f factors associated with the disease, although, individually for each

134 study, m a n y associations were identified. Decreased intake of milk, fruit, vegetables, increased fried food and alcohol consumption, and more tobacco smoking were m o s t often found to be associated with risk of stomach cancer. In addition it became clear f r o m migrant studies that factors that increased a person's risk of stomach cancer were likely to be acquired in early years (Haenszel 1982). Whereas epidemiological research could not present consistent results regarding risk factors for stomach cancer at the end of the 1970s, m a j o r progress was made in experimental research. N-Nitroso compounds had been detected, which could for the first time induce high percentages of glandular stomach cancer in animals (Sugimura and Fujimura 1967). N o w animals models are becoming available for studying particular aspects of stomach cancer carcinogenesis experimentally. The "N-nitroso c o m p o u n d " hypothesis, in most parts etablished by experimental research in the 1970s, directed the epidemiological research into risk factors of stomach cancer in the 1980s. Most interest concentrated on precursor substances, such as nitrate and nitrite; and substances that are able to inhibit the formation of N-ni~roso compounds such as Vitamin C (Mirvish 1983, 1986). This article will present and discuss results of epidemiological research on risk factors of stomach cancer during the last decade that utilized descriptive, analytical, and metabolic study approaches. Materials and methods Epidemiological studies on stomach cancer published during 1980-1990 were reviewed and three areas of important progress identified that will be addressed in this review: studies on the regional occurrence of stomach cancer, case-control and cohort studies on risk factors for stomach cancer, and metabolic studies on precursor substances of carcinogens or on endogenous formation of N-nitroso compounds. Findings from case-control and cohort studies will be presented in terms of relative risks for increasing exposure to the factors under investigation together with 95% confidenceintervals (95% CI) or other indicators of statistical significance.When estimates of relative risks for different risk models were given in a publication, the risk estimate derived when using other dietary factors as confounders was preferred. Usually the relativerisk estimate for the highest-intakecategorycompared to the lowest category was selected. The number of categoriesinto which exposure was divided is given in addition. All recent studies on risk factors included histologicallyconfirmed stomach cancer cases according to the International Classification of Diseases. No attempts have been made so far to separate the group according to histological subtypes (Lauren 1965) and only one study (WuWilliams et al. 1990) has analysed risk by anatomical subsite (Yang and Davis 1988; Levi et al. 1990). However, this study had only a small number of cases.

Results

Regional variation of stomach cancer The m o s t famous example for this approach is the compilation of the Cancer Atlas of China, which showed higher rates of stomach cancer in the middle and the more nothern region (Editorial Committee 1979). In western Europe, the International Agency for Research

on Cancer in Lyon, France, coordinated the regional analysis of cancer mortality for all countries of the European Community (EC) (Smans et al. 1989). This cancer atlas of the EC, which is not fully published and only presented on particular sheets, showed a pronounced regional c o m p o n e n t of stomach cancer mortality. Areas in northern Portugal, in central Spain, northern Italy and south-east G e r m a n y can be considered as high-risk regions for stomach cancer in western Europe whereas D e n m a r k , Greece, southern France and England can be considered as low-risk areas. F o r G e r m a n y the Atlas of Cancer Mortality in its second edition gives a detailed picture of stomach cancer mortality on the district (,,Kreis") level of the Federal Republic (Becker et al. 1984). On the average for the years 1976-1980, the rates for south-east Bavaria were up to 2.5 times the rates of the low-risk areas in the highly urbanized states and parts of Hessen. This picture of higher rates of stomach cancer in the south-east Federal Republic of G e r m a n y is supplemented by an increase in incidence in some districts of the southern former G e r m a n Democratic Republic (Zentralinstitut fiir Krebsforschung 1981 and 1989) and also in northern Austria (Denk et al. 1967).

Studies on stomach cancer etiology Table 1 lists four cohort studies and 16 case-control studies published in the last 10 years that have investigated risk factors for stomach cancer. These studies will subsequently be reviewed for their results regarding associations with stomach cancer. Two further studies have not been considered because of small numbers (Hoey et al. 1981; Tajima and Tominaga 1985) and one study because of the combined consideration of cancer of the cardia in the stomach and oesophageal cancer (Li et al. 1989). Most of the studies in the table investigated various aspects of stomach cancer etiology simultaneously.

Dietary intake. For food-group levels, most studies of Table 1 centered around the intake of vegetables, fruit, staple foods, and processed meats and fish. Table 2 presents a summary of the results of these studies according to the consumption of vegetables, raw vegetables, fruit, citrus fruit, fibre-rich bread, processed or smoked meat, and salted or smoked fish. These items were selected because of possible etiological relevance and for consistency in the publication of the studies. The highest agreement between studies regarding relative risk estimates on food-group level was seen for raw vegetables, fruit, citrus fruit, and fibre-rich bread, which showed an inverse relation to stomach cancer risk with increasing consumption. In most instances, the lower relative risk of stomach cancer for raw vegetable intake, when compared to total vegetable intake, indicates that this food group needs further differentiation when being related to stomach cancer. An important observation in that direction was made by Buiatti et al. (1989) in that vegetables when cooked did not show the inverse relation to stomach cancer shown by uncooked vegetables. An al-

135 Table 1. Individual-based epidemiological investigations into risk factors for stomach cancer published in the last decade (1980 1990) Country

References

Canada Greece USA USA Poland

Risch et al. 1985 Trichopoulos et al. 1985 Stehr et al. 1985 Correa et al. 1985 Jedrychowski et al. 1986

Italy Belgium China China

LaVecchia et al. 1987 Tuyns 1988 Hu et al. 1988 You et al. 1988

Japan

Kono et al. 1988

UK Italy

Coggon et al. 1989 Buiatti et al. 1989, 1990

USA

Study type Case - control studies Population-based Hospital-based Population-based a Hospital-based Hospital- and population-based Hospital-based Population-based Hospital-based Population-based Hospital- and population-based Population-based Population-based

Number of cases 246 110 111 391 110 206 293 241 564 139

Number Areas of investigation of controls 246 100 111 391 110/ 110 474 2 914 241 1 131

95 1016

2 574/ 278 190 1 159

Wu - Williams et al. 1990 Population-based

137

137

USA Turkey Germany

Graham et al. 1990 Demirer et al. 1990 Boeing et al. 1991

Population-based Hospital-based Hospital-based

293 100 143

293 100 579

Japan USA USA Japan

Ikeda et al. 1983 McLaughlin et al. 1990 Nomura et al. 1990 Hirayama 1990

Cohort studies Follow-up 1968-1978 Follow-up 1954-1980 Follow-up 1965-1986 Follow-up 1966-1982

65 1 520 150 5247

7553 293 916 7990 265000

Diet, alcohol, food preparation, water supply Diet Vitamin A Diet, alcohol, tobacco smoking Diet, alcohol, tobacco smoking Diet, alcohol, tobacco smoking Salt Diet, alcohol, water supply, tobacco smoking Diet, alcohol, food preparation, food storage, tobacco smoking Diet, alcohol, food preparation, tobacco smoking Diet, food preparation, food storage Diet, food preparation, alcohol, tobacco smoking Diet, food preparation, alcohol, tobacco smoking Diet, food storage, alcohol, tobacco smoking Diet, food storage, alcohol, tobacco smoking Diet, food preparation, food storage, alcohol, tobacco smoking Diet, tobacco smoking Tobacco smoking Diet, alcohol, tobacco smoking Diet, alcohol, tobacco smoking

" Fatal arteriosclerotic events as controls

ternative explanation of the particular effect of raw vegetables concerns the possibility that they are differently composed in regard to food constituents than cooked vegetables. This second possiblity implies that ratios of food constituents need to be investigated for both raw and cooked vegetables. Increased fruit consumption was consistently inversely related to stomach cancer risk in all studies. In some studies the subgroup citrus fruit showed a stronger association than the total group of fruits. However, this difference in risk estimates was not as pronounced as for vegetables, and seems (currently) not to deserve particular attention. Increasing consumption o f fibre-rich bread was also consistently inversely related to stomach cancer risk. Such a consistent observation needs further investigation. Experimental studies are available that show that bran can alter the formation of N-nitroso compounds (Wishnok and Richardson 1979; Moller et al. t988). However, these experimental studies focused on different inhibitory mechanisms. The results o f recent studies regarding the effect of increased consumption of meats and salted fish on the risk of stomach cancer support the hypothesis that meat or fish consumption m a y be a risk factor, especially when these food items have been processed or smoked. However, the fact that some studies could not find an association indicates that processing or smoking of these food items m a y not in general be the hazardous factor. F r o m

the pattern of results, it is much likely that only particular treatments or preparations of meat and fish m a y increase the risk for stomach cancer. Future epidemiological studies have to concentrate on the different modalities of meat and fish conservation and preparation. I f the particular pathway is identified, it m a y be much easier to interpret the current findings. In the Mediterranean countries, pasta was identified as a potential risk factor for stomach cancer (Trichopoulos et al. 1985: relative r i s k = 3.42). However, this finding could not be confirmed by other case-control studies (La Vecchia et al. 1987: relative risk = 1.63; P = 0 . 0 7 ; Buiatti et al. 1989: relative r i s k = l . 0 ; not significant). Similar inconsistencies between the studies f r o m Table 1 have also been found for other staple foods. Thus, only particularly treated and prepared staple foods m a y act as risk factors in some parts of the world ( K o d a m a et al. 1984). Milk consumption could not be confirmed as being associated with stomach cancer in recent studies. M a n y studies investigated this item and found no relationship to stomach cancer risk (Trichopoulos et al. 1985; Correa et al. 1985; La Vecchia et al. 1987; K o n o et al. 1988; Buiatti et al. 1989; Wu-Williams et al. 1990; Boeing et al. 1991; H i r a y a m a 1990; Ursin et al. 1990). A first approach to identifying biologically plausible etiological mechanisms consists of converting food consumption into the consumption of food constituents, such as nutrients, by using food tables. In the case of

136 Table 2. Consumption by food groups and relative risk for stomach cancer in recent studies (1980-1990) Study

Vegetable

Risch et al. (1985) Trichopoulos et al. (1985) Correa et al. (1985) (whites) Correa et al. (1985) (blacks) Jedrychowski et al. (1986) LaVecchia et al. (1987) Hu et al. (1988) You et al. (1988) Kono et al. (1988) Coggon et al. (1989) Buiatti et al. (1989) Wu-Williams et al. (1990) Graham et al. (1990) Boeing et al. (1991) Demirer et al. (1990) Ikeda et al. (1983) Nomura et al. (1990) Hirayama (1990)

0.84" (100 g/day) 0.29 (100 g/day) 0 u (0.72-0.96) (0.07-1.11) Decreasing 0.70 (2 C) ~ NS e risk (not given) 0.90 (2 C) 0 0.47 (4 C) (0.57-1.41) (0.24-0.92)

0.75 (100 g/day) 0.71c (10 g/day) 3.92 (100 g/day) 2.03d (100 g/day) (1.76-8.75) (0.34-12.2) (0.60-0.92) (0.53 0.94) 0.77 (2 C) g 0.79 (2 C) Beef with 0 (significant) (significant) increased risk 0 0 NS 0

0.50 (4 C) (0.25 1.00)

0

0.33 (4 C) (0.16-0.66)

0

0

1.98 (4 C) (1.01-3.87)

0

0.24 h (3 C) (0.08-0.71) 0

0.31 (3 C) (0.15-0.64) 0.73 (3 C) (NS) 0

0

0

0

0.63 (3 C) (P=0.11) 0

0.44 (3 C) (P=0.11) 0

4.35 j (3 C) (I .64-12.50) 1.60 k (3 C) (P=0.04) 0

0.6 (4 C) (0.4-0.8) 0.7 (2 C) (0.4-1.0) 0.6 (3 C) (0.2 1.5) 0.4 (3 C) (sign. trend) 0.67 (3 C) (0.29 1.67) NS

0

0

1.4 (3 C)

0.8 (2 C) (0.5 1.2) 0

0

0.8 m (4 C) (0.6-1.1) 0.9 ~ (3 C) (NS) 1.0 (2 C) (0.5-1.9)

0.6 (3 C) (sign. trend) 0

0

1.8 s (3 c )

0.32 (3 C) (0.16-0.63) 0

(sign. trend) 2.6 t (2 C) (1.2-5.9) 0

0.86 (3 C) (0.54 1.36) 0.21v (2 C) (0.11-0.39) 0

1.02 i (3 C) (0.38-2.70) 0.27 (3 C) (P 5 klU/day) (0.32-0.99) 0.62 (2 C) 0 1.22 ~ (2 C) (0.79-1.90) (0.38-0.99)

Stehr et al. (1985) Correa et al. (1985) (whites) Correa et al. (1985) (blacks) LaVecchia et al. (1987) You et al. (1988) Buiatti et al. (1990) Graham et al. (1990) Boeing et al. (1991)

Tocopherol

Retinol

Carotene

Dietary fibre Nitrite

Nitrate

Calcium

0.33 0.40 2.61 0.67 0d (10kIU/day) (10g/day) (1 rag/day) (100rag/day) (0.15-0.71) (0.28-0.58) (0.61-4.22) (0.50-0.90) 0 0 0 0 0 0.68 (2 C) 0 (0.43-1.08)

0

0

0

0.63 (2 C) 0 (0.39-1.03)

0.85 ~ (2 C) 1.08 (2 C) 0 (0.53-1.37) (0.67-1.73)

0

0

0

0.46 (3 C) (P < 0.001) 0.5 (4 C) (0.3-0.6) 0.5 (5 C) (0.4-0.7) NS

0.83 b (3 C) (NS) 1.0 b (4 C) (0.7-1.4) 1.0 b (5 C) (0.7-1.3) 1.50 b,g (2 C) (1.16-1.95) 0

0

0

0

0

0

0

0

0.9 (5 C) (0.7-1.3) 0

1.2 f (5 C) (0.8-1.8) 0

0

0

0 0 0.6 (5 C) (0.4-0.9) NS

0.37 (5 C) 0 (0.16-0.86)

0.39 (3 C) (P < 0.001) 0.5 (4 C) (0.3-0.6) 0.9 (5 C) (0.7-1.1) 0.59 (2 C) (0.45-0.77) 1.42 (5 C) (0.88-2.93)

0.6 (4 C) (0.4-0.8) 0.9 (5 C) 0.9 (5 C) (0.7-1.2) (0.7-1.2) 0 Increased risk 1.26 (5 C) 0.71 (5 C) (0.55-2.70) (0.35-1.44)

a Each entry contains the following information: upper row, relative risk (unit or number of categories, C); lower row, 95% confidence interval or other measure of significance (NS, not significant) b Retinol r kIU, 1000 IU

a 0, variable not mentioned in the publication ~ Retinol equivalents f Confounded with protein intake; univariate relative risk= 1.9 (95%CI 1.3-2.7) g Controlled for energy and years of refrigeration

pathway for the protective action of carotene probably is through its anti-oxidant function and not through its function as a vitamin A precursor. Nitrate and nitrite are involved in the formation o f Nnitroso c o m p o u n d s by directly acting with amide compounds in the case of nitrite or by forming nitrite in the case of nitrate. In recent studies, only high nitrite consumption was positively associated with stomach cancer risk whereas high nitrate consumption via ingested food showed no association. These findings m a y be due to the fact that a m a j o r source o f nitrate is vegetables, also containing vitamin C, which appears to be a protective factor. These observations clearly show that the relation of food constituents in separate food groups needs to be investigated. It has been speculated that the intake o f macronutrients (energy, fat, protein, and carbohydrate) contributes to the carcinogenic process o f stomach cancer. This hypothesis suggests that an especially high carbohydrate intake increases the risk for stomach cancer whereas a high protein intake is protective. Studies in populations with high m e a t consumption that have reported data for protein consumption could not confirm this hypothesis, and some even found increased risk for high protein intake (Risch et al. 1985; Buiatti et al. 1990; G r a h a m et al. 1990; N o m u r a et al. 1990). F o r carbohydrate consumption, a slight often non-significant increase in stomach risk was found in some studies [Risch et al. 1985; Correa et al. 1985 (whites), G r a h a m et al. 1990] and no relationship in other studies (Buiatti et al. 1990; N o m u r a et al. 1990). However, two studies were able to link high starch

intake with increased stomach cancer risk (Buiatti et al. 1990; G r a h a m et al. 1990). Interestingly, high fat intake also appeared as a risk factor (Risch et al. 1985; Buiatti et al. 1990; G r a h a m et al. 1990). The study of Buiatti et al. (1990) refines this finding by showing that animal fat is positively linked with stomach cancer risk and vegetable fat inversely. Analyses on the macronutrient level in general seem to be too broad and deserve further refinement. The high correlation between nutritional variables makes it very likely that causally irrelevant associations o f macronutrient level with the investigated cancer sites m a y appear. However, such associations m a y disappear after adequate control for other dietary factors. Analyses using food tables generated a consistent pattern o f possible pathways by which dietary intake m a y be involved in stomach cancer carcinogenesis. This approach can be extended by linking food intake with other food constituents. An example is the new hypothesis o f a particular protective role o f allium vegetables, such as onions and garlic, reported in several studies. Organic sulphides m a y be the acting principle showing anticarcinogenic properties in experimental studies (Belman 1983; Wargovich 1987).

Food storage, food processing, and source of drinking water. F o o d storage at r o o m temperature m a y be a source of increased carcinogen formation through the formation of precursor substances o f carcinogens by bacterial growth and increased enzyme activity. In all studies that linked the period of refrigerator use with stomach cancer risk, an early use showed decreased relative risks

138 [Risch et ak (1985): 0.86, 95% CI = 0.75-1.00, for 10 years of earlier use; Coggon et al. (1989): 0.5, 95% CI = 0.2--1.3, for 14+ years; Buiatti et al. (1989): 0.71, 95% CI = 0.53-1.00, for 10+ years; G r a h a m et al. (1990): 0.48, 95% CI = 0.28-0.80, for 12+ years by men and 0.68, 95% CI = 0.34-1.37, for 7 + years by women; Boeing et al. (1991): 0.71, 95% CI = 0.44-1.14, for 6 + years of earlier use]. Other interesting observations regarding food storage have been made in the United Kingdom where the risk for stomach cancer was increased for individuals having no pantry or larder in the kitchen but stored their food under the stairs or simply in the room (Coggon et al. 1989), and in China where it was found that the use of mouldy grains in the family was associated with increased stomach cancer risk (You et al. 1988). Smoking of meat and fish has been suspected since the 1960s of contributing to stomach cancer risk. Recently the detection of potent genotoxic agents during the smoking et al. (1991): 0.71, 95% CI =0.44-1.14, for 6 + years of earlier use]. Table 4. Salt use and relative risk for stomach cancer in recent studies

(1980-1990) Study

Salt intake through diet

Questions about salt taste and use Prefer salt

Add salt

Correa et al. 1985) (whites) Correa et ak (1985) (blacks) LaVecchia et al. (1987) Tuyns (1988)

0"

1.17b (3 C) (0.34-3.99)

1.36 (2 C) (0.53-3.13)

0

1.34 (3 C) (0.36-5.01)

1.75 (2 C) (0.81 1.83)

1.47 (3 C) (NS) 0

0

0

0

Hu et al. (1988) You et al. (1988) Kono et al. (1988) Coggon et al. (1989) Buiatti et al. (1989) Graham et al. (1990) Boeing et al. (1991) Ikeda et al. (1983) Nomura et al. (1990)

1.51 c (2C) (1.02-2.23) 1.1 (4 C) (0.8-1.4) 1.4d (3 C) (NS) 0

0

1.78 (3 C) (1.19 2.75) 0

0

1.4 (C) (1.0-1.9) 0 6.2 (3 C) (2.0-18.9) 0

3.09;4.65~(4 C) 0 (1.65-5.79) 0 NS Increasinge risk 1.2g (3 C) (0.8-1.7)

0 0 0 1.5 (2 C) (1.3-1.9) 0 NS

0

0

1.0 (3 C) (0.6-1.6)

0

a 0, variable not mentioned in the publication b Each entry contains the following information: upper row, relative risk (unit or number of categories, C); lower row, 95% confidence interval or other measure of significance (NS, not significant) Salted soya paste d Salty food items e Women f Salted pickles g Pickles

suggests that some cured or smoked meats are positively related with stomach cancer risk. In the study in Germany, smoking of meat at home in general was not associated with increased risk for stomach cancer; however, the risk increased threefold when spruce was mentioned as the source of wood being used for the smoking process (Boeing et al. 1991). Joossens was recently advocating the idea that salt may be a risk factor in the pathogenesis of stomach cancer in humans (Sato et al. 1959a, b; Jossens and Geboers 1981, 1987). The experimental evidence for a role of salt in stomach cancer etiology is generated by its role in the formation of precancerous conditions or lesions (Kodama et al. 1984), its cocarcinogenic effects (Tatematsu et al. 1975), and its role in the increasing mutagenicity of nitrosated foods (Rojas-Campos et al. 1990). Recent etiological studies confirmed that high salt intake, measured either through dietary intake or through referring to salt attitudes and salting habits, is linked with increased risk for stomach cancer with the exception of one study (Table 4). The source of drinking water has been found to be a risk factor for stomach cancer (Haenszel et al. 1976; Cuello et al. 1976). Interestingly, all recent studies addressing this point were able to associate private water sources, in particular well water, with stomach cancer risk. Whereas in some studies the risk was reported to be associated with a non-central water supply compared to a central water supply (Risch et al. 1985; Boeing et al. 1991), the study of You et al. (1988) in China showed that well water was associated with increased stomach cancer risk compared to surface water. The latter finding is also confirmed by Hu et al. (1988) after recalculation (Boeing et al. 1991). The interpretation of this result is difficult because information regarding the quality of the water is not available. Possible etiological mechanisms center around increased nitrate concentration in this type of water supply (without concurrent vitamin C supply) or, more hypothetically, the presence of particular microorganisms. Recently Helicobacter pylori, a microorganism with currently not completel)~ understood ecology (Bramwell 1989), could be linked with gastritis (Dooley and Cohen 1988). Gao (1989a, b) and also Chen (1989) assume mutagenic substances in well water as the etiological factor for China. In conclusion, there are indications that the modes of food storage and food processing prevailing in the population seem to be of importance in risk of stomach cancer. Approaches to address this question were successful in linking particular practices with stomach cancer risk. For many factors found to be associated with stomach cancer, biological mechanisms are known by which the observational findings can be explained, mostly in the light of the "N-nitroso c o m p o u n d " hypothesis.

Alcohol, alcoholic beverages, and tobacco smoking. The acute effects of alcohol on the stomach wall are well established whereas approaches to link chronic gastritis with alcohol use failed in most instances (Wolff 1989). In recent etiological studies, alcohol intake or the consumption of particular alcoholic beverages was inconsistently

139 Table 5. Consumption of alcohol, alcoholic beverages, and cigarette smoking and relative risk for stomach cancer in recent studies (1980-1990) Study

Alcohol

Beer

Wine

Liquor

Cigarette smoking

Risch et al. (1985)

0a

NS b

NS

NS

Trichopoulos et al. (1985) Correa et al. (1985) (whites) Correa et al. (1985) (blacks) Jedrycbowski et al. (1986) LaVecchia et al. (1987) Hu et al. (1988)

NS

0

0

0

1.28 c (20py a) (1.01-1.62) 0

1.87 (3 C) (1.02-3.44) 2.21 (3 C) (1.19-4.08) 0

1.17 (2 C) (0.72-1.90) 1.27 (2 C) (0.82-1.99) NS

1.61 (2 C) (0.89 2.91) 1.18 (2 C) (0.51-2.73) NS

1.28 e (cig. smok. r) (0.75 2.19) 2.61 (cig. smok,) (1.36-5.01) 0.68 (curr. cig. smok. h) (0.39-1.20) NS 1.86 (smok. index i) (1.42-2.70) 1.5 (> 20 cig./day) (1.0-2.1) Hospital population controls 1.8 (2 C) 1.3 (ever/never) (1,1-2.9) (0.8-2.2) 1.2 (curr. smok.) (0.9-1.7) 5.2 (3 +packs/day) (1.4-8.6) NS 0.52 (curr. cig. smok.) (0.30-0.89) sign. increased risk 1.82 (40 + cig./day) (1.28-2.59) 2.7 (curr. cig. smok.) (1.8-4.1) 1.52 (curr. cig. smok.) (p < 0.001)

1.19 (3 C) 0 (NS) 1.45 ( > 4 kg/year) 0 (1.02-2.02) 0 0.8 (3 c) (0.6-1.1)

0

1.99 (2 C) (1.21-3.28) 1.54 (2 C) (0.94-2.54) 2.89 g (2 C) (1.35-6.17) 0

0

0

0

0

Kono et al. (1988)

0

NS

NS

NS

Buiatti et al. (1989)

1.4 (5 C) (1.0-1.9) 3.0 (> 80 g/day) (1.1 8.8) NS NS

0

You et al. (1988)

Wu - Williams et al. (1990) Demirir et al. (1990) Boeing et al. (1991) Ikeda et al. (1983) McLaughlin et al. (1990) Nomura et al. (1990) Hirayama (1990)

0 0

1.3 (2 +/day) (1.0-1.8) 3.5 (daily) 1.6 (daily) (1.4-8.7) (0.6-4.3) 0 0 1.82 (>500g/day) 0.52 (>20g/day) (0.95-3.50) (0.30 0.93) 0 0 0 0

2.3 (daily) (0.9-5.6) 0 0.52 (>2g/day) (0.27-1.00) 0 0

1.1 (>40 oz/m k) (0.7-1.9) 0.92 j (4 C) (0.85-0.99)

1.1 (> 500 oz/m) (0.7-1.7) 0.95 j (2 C) (0.80-1.12)

1.0 (>50 oz/m) (0.5-2.1) 1.33 j (2 C) (0.99-1.78)

0.7 (>2 oz/m) (0.4-1.3) 0.945 (2 C) (0.86 1 . 0 2 )

" 0, variable not mentioned in the publication b NS, not significant c Each entry contains the following information: upper row, relative risk (unit or number of categories, C); lower row, 95% confidence interval or other measure of significance a py, pack years linked with s t o m a c h cancer (Table 5). All prospective studies and some case-control studies failed to find an increased risk for stomach cancer with increasing c o n s u m p tion o f alcohol or alcoholic beverages. The elimination o f recall bias induced by the presence o f a disease in retrospective studies gives prospective studies a particular weight and allows the conclusion that alcohol is n o t related to s t o m a c h cancer. This interpretation o f the data is also supported by the fact that the risk o f s t o m a c h cancer for the three alcoholic beverages did n o t show a consistent pattern between the different studies. H o w ever, causal relationships with increased risk for s t o m a c h cancer m a y exist for particular habits, such as drinking liquor on an e m p t y s t o m a c h (Jedrychowski et al. 1986; Wu-Williams et al. 1990). Nearly the same pattern o f results as for alcohol cons u m p t i o n was observed for t o b a c c o s m o k i n g (Table 5). Increased relative risk estimates dominated, but non-significant a n d negative findings were also reported. H o w ever, in contrast to alcohol c o n s u m p t i o n , in all prospective studies cigarette s m o k i n g was linked with an increased risk for s t o m a c h cancer. S u p p o r t for a particular

e

f g h i J k

0

Pipe use increased risk cig. smok., cigarette smokers Vodka before breakfast curr. cig. smok., current cigarette smokers Smok. index, smoking index Men m, month

link o f s m o k i n g with s t o m a c h cancer came also f r o m a c o h o r t study with 106 participants, c o n d u c t e d in the 1950s and 1960s ( H a m m o n d 1966). This relationship needs additional evaluation becauce o f a possible interaction between the habit o f t o b a c c o smoking, the metabolic effect o f smoking, and dietary intake. Only in a few epidemiological studies, were relative risk estimates for tobacco s m o k i n g controlled for dietary variables. It is k n o w n that smokers exhibit a reduced vitamin C concentration in b l o o d serum, which m a y affect the potential protective role o f this c o m p o u n d in s t o m a c h cancer carcinogenesis. The secular trends o f m o s t countries in cigarette s m o k i n g and alcohol c o n s u m p t i o n a n d s t o m a c h cancer rates also m a k e it difficult in general to assign a strong role to t o b a c c o (and alcohol) in the p a t h w a y o f s t o m a c h cancer.

Metabolic investigations Several studies c o m p a r i n g nitrate excretion in the urine or nitrate c o n c e n t r a t i o n in salvia in high- and low-risk

140 areas for stomach cancer have had inconsistent results [Cuello et al. (1976): increased in high-risk area; Armijo et al. (1981): no increase in high-risk area; Kamiyama et al. (1987): decreased in high-risk area; Forman et al. (1985): decreased in high-risk area]. Also the link of nitrate concentration in water (in general centrally supplied water) with stomach cancer rates in those areas on the whole revealed no associations (reviewed by Forman 1987). Ohshima and Bartsch (1981) developed a methods that allows one to estimate the nitrosating potential in individuals by applying the N-nitrosoproline test, in which the formation of N-nitrosoproline is determined in the urine. N-Nitrosoproline is a harmless substance; however, its determination is thought to give an indication of the endogenous formation of N-nitroso compounds in general. The test has been applied to urine from individuals in high- and low-risk areas of stomach cancer (Bartsch et al. 1989). Comparison within countries revealed a higher potential of forming N-nitrosoproline in the high-risk areas in Japan and Poland compared to low-risk areas (Kamiyama et al. 1987; Zatonski et al. 1989). However, no association between the formation of N-nitrosoproline and stomach cancer risk could be observed in China (Chen et al. 1987) and Italy (Forman, personal communication). This test in its current version was recently critized because of its low sensitivity for differentiating groups of different risk (Houghton et al. 1989). An interesting relation of vitamin C metabolism with precancerous lesions for stomach cancer was found by Sobala et al. (1989), who showed that chronic gastritis diminishes the ability to secrete vitamin C into the interior of the stomach. This secretion is not connected with plasma vitamin C level or vitamin C intake. Currently it is not clear how this source of vitamin C affects quantitatively in addition to ingested vitamin C, the vitamin C concentration in the stomach. Other metabolic research has focused on the question whether precancerous conditions and lesions such as chronic gastritis, intestinal metaplasia, or dysplasia induce favourable conditions in the stomach for the formation of N-nitroso compounds. Particular interest was given to pH, nitrite, and N-nitroso compounds in gastric juice. Whereas such studies showed a clear relationship between precancerous lesions and increased pH and nitrite concentration, the findings regarding the presence or formation of nitrosamines have been contradictory (Bartsch et al. 1989). It is not clear whether the formation of N-nitrosoproline is representative of the formation of carcinogens relevant for stomach cancer risk. An additional source of nitrosating agents was recently identified in bacteria and macrophages that produce nitric oxide, mostly from arginine (Tannenbaum et al. 1990). This source of nitrite formation may be important in cases of local infection or inflammation (Stuehr and Maletta 1987). A more powerful study design for generating evidence of a causal role links results from the collection of biological specimens to stomach cancer risk. Such a design was followed in a case-control study by Chen et al. (1990 a, b),

who studied salt and nitrite metabolism among 263 asymptomatic individuals from Narino, Columbia, to whom they assigned a case status (prevalence of precursor lesions such as chronic atrophic gastritis, intestinal metaplasia, dysplasia) or a control status (stomach with no precursor lesion) after gastroscopical examination. Increased salt excretion in the urine was observed in individuals with chronic atrophic gastritis and dysplasia but not in individuals with intestinal metaplasia. The concentration of nitrite in gastric juice was found to be elevated in all forms of gastritis, the risk increasing with the severity of the lesion (relative risk for precancerous lesions compared to controls: 2.09, 95% CI =0.93M.09). This finding is contrasted by Fontham et al. (1986), who detected a decrease of nitrite concentration with increasing severity of the precancerous lesions among American blacks in South Louisiana. Possibly the difference in the source of dietary nitrate intake (vegetables versus water) was the reason for the contrasting results (Chen et al. 1990 b). It has also been reported that a high vitamin C intake appeared to be a preventive factor in South Louisiana, while in the Narino province of Columbia a high/%carotene level in blood appeared to be protective. Metabolic epidemiological studies are one step forward in studying particular mechanisms, especially when an exposure determination by questionnaires presents problems. However, research in this field during the last decade has not been able to assign clear biological meaning to the measurements being performed. No convincing results have been presented supporting the hypothesis that increased nitrate consumption in the general population results in increased nitrite formation, which then reacts with amines or amide substances to form carcinogens that induce stomach cancer. It seems as if the endogenous formation of stomach-cancer-inducing or -promoting substances is complex and depends not only on the nitrite concentration but also on the availability of specific nitrosatable compounds. Also, other groups of compounds independently formed from the process of nitrosation may be of relevance. Further refinements of the biological tests seem necessary before major progress will be seen in this field. Discussion The complexity of diet requires clear ideas about the biological mechanisms and possible causal pathways at the stages of inquiry, analyses, and interpretation of the data. Correa et al. developed a model of stomach cancer carcinogenesis that incorporates many fields of research and is displayed in Fig. 1. Their hypothesis centres around the idea of a stepwise process, which begins with precancerous conditions and lesions and in which food constituents will act differently in each stage. This model also incorporates the endogenous formation of N-nitroso compounds as the pathway for tumour induction and promotion (Correa 1988). Most interpretations of results from recent studies on risk factors for stomach cancer are based on this model The results from recent etiological studies strengthened the idea that protection by vitamin C and/or carotene

141 Gastric cavity

Diet

Gastric mucosa

normal

}

I

t superfici'a~ gastritis [ ]_ antibodies gastrectomy

irritants (salt, aspirin) nutrition deficits (animal proteins, vitamins)

atrophic~ gastritis[

nitrate antioxidant deficit nitrogen compounds (foods, drugs) carotenoid deficit

4 acids / I N 0 mutagens

I

small intestinal metaplasia

I

irritants

I

colonic metaplasia

I

I

mild dysplasia I

cancer. Many such studies on precancerous lesions of stomach cancer can be prospective in nature because the occurrence of these lesions is much more frequent in the population than is the cancer. From the point of view of benefits from research expenditures it has to be concluded that after a new decade of research the causes of stomach cancer still have not been identified. There is also no proof that increasing consumption of protective substances above the current level will affect the risk for stomach cancer. However, the last 10 years have generated a much clearer view for future studies and for public health policy based on the recently accumulated knowledge. Priority should be given to planning epidemiological intervention studies, which can test the effect of increasing the vitamin C or carotene intake in high-risk groups for stomach cancer, such as individuals with chronic gastritis or intestinal metaplasia. In the mean time, recommendations that promote the intake of raw vegetables and fresh fruit and publicise reservations about a high intake of salt or particularly processed meat and fish are clearly justified by the recent results, obtained in part from individuals suffering from stomach cancer.

severe dysplasia I

carcinoma --

I

invasion

Acknowledgements. The helpful comments and editorial work of Dr. N. Eby and the efficient secretarial help of P. Berthold and B. Edinger are gratefully acknowledged.

Fig. 1. Hypothesis of causation of stomach cancer (adopted from Correa 1988)

References

is, world wide, the most important factor influencing the process leading to stomach cancer. No particular world wide prevailing risk factor was observed in these studies that could explain tumour formation. Low vegetable and fruit consumption was recently not only found to be connected with stomach cancer but also with many other cancer sites of the epithelial type (Colditz et al. 1985; Pisani et al. 1986; Byers et al. 1987; Notani and Jayant 1987; McLaughlin et al. 1988; Hirayama 1990; Trock et al. 1990). Vegetables and fruits are a common source of vitamin C and carotene. Biologically, they may have different significance for the different cancer sites. Whereas vitamin C, carotene and other substances in vegetables and fruits protect the epithelium in general, including the stomach, the specific mechanism of endogenous formation of compounds and their blocking by vitamin C is a pathway specific for stomach cancer. However, for this particular action of vitamin C strong evidence is still lacking, despite the many observations supporting this pathway. Particular importance should be attached to increasing our knowledge of the natural history of stomach cancer and the role of precancerous conditions and lesions. Perhaps the new technique of DNA analysis will be able to match DNA changes with morphological alterations, such as intestinal metaplasia and dysplasia, as has been successfully demonstrated in colon carcinogenesis (Vogelstein et al. 1988). Epidemiological research approaches can then be used to define the risk factors leading to DNA damage. This may help to refine our understanding of the stepwise process that sometimes leads to

American Cancer Society (1990) Cancer facts and figures 1990. American Cancer Society Inc, Atlanta Armijo R, Gonzales A, Orellana M, Coulson AH, Sayre JW, Detels R (1981) Epidemiology of gastric cancer in Chile: II. Nitrate exposures and stomach cancer frequency. Int J Epidermiol 10:57-62 Bartsch H, Ohshima H, Pignatelli B, Calmels S (1989) Human exposure to endogenous N-nitroso compounds: quantitative estimates in subjects at high risk for cancer of the oral cavity, oesophagus, stomach and urinary bladder. Cancer Surv 8:335-362 Becker N, Frentzel-Beyme R, Wagner G (1984) Krebsatlas der Bundesrepublik Deutschland. Atlas of cancer mortality in the Federal Republic of Germany. Springer, Berlin Heidelberg New York Tokyo Belman S (1983) Onion and garlic oil inhibits tumor promotion. Carcinogenesis 4:1063 1065 Boeing H, Frentzel-Beyme R, Berger M, Berndt V, G6res W, K6rner M. Lohmeier R, Menarcher A, M/innl HFK, Meinhardt M, Miiller R, Ostermeier H, Paul F, Schwemmle K, Wagner KH, Wahrendorf J (1991) Case-control study on stomach cancer in Germany. Int J Cancer 47 (in press) Bramwell NH (1989) The prevalence of Campylobaeter Pylori gastritis among asymptomatic adults. Can Med Assoc J 140:276 Buiatti E, Palli D, Decarli A, Amadori D, Avelliui C, Bianchi S, Biserni R, Cipriani F, Cocco P, Giacosa A, Marubini E, Puntoni R, Vindigni C, Fraumeni J, Blot W (1989) A case-control study of gastric cancer and diet in Italy. Int J Cancer 44:611-616 Buiatti E, Palli D, Decarli A, Amadori D, Avellini C, Bianchi S, Bonaguri C, Cipriani F, Cocco P, Giacosa A, Marubini E, Minacci C, Puntoni R, Russo A, Vindigni C, Fraumeni JF, Blot WJ (1990) A case-control study of gastric cancer and diet in italy: II. Associations with nutrients. Int J Cancer 45:896-901 Byers T, Graham S (1984) The epidemiology of diet and cancer. Adv Cancer Res 41:1-69 Byers TE, Graham S, Haughey BP, Marshall JR, Swanson MK (1987) Diet and lung cancer risk: findings from the Western New York Diet Study. Am J Epidemiol 125:351-363

142 Chen CS (1989) (Investigation of nitrosamine in drinking water in Huang Shi - a high-risk area of stomach cancer in Fujian province). Chung Hua Liu Hsing Ping Hsueh Tsa Chih 10:359-362 Chert J, Ohshima H, Yang H, Li J, Campbell TC, Peto R, Bartsch H (1987) A correlation study on urinary excretion of N-nitroso compounds and cancer mortality in China. Interim results. IARC Sci Publ 84:503-506 Chen VW, Abu-Elyazeed RR, Zavala DE, Ktsanes VK, Haenszel W, Cuello C, Montes G, Correa P (1990a) Risk factors of gastric precancerous lesions in a high-risk Colombian population. I. Salt. Nutr Cancer 13:59-65 Chen VW, Abu-Elyazzed RR, Zavala DE, Haenszel W, Ktsanes VK, Rice J, Cuello C, Montes G, Correa P (1990 b) Risk factor of gastric precancerous lesions in a high-risk Colombian population. II. Nitrate and nitrite. Nutr Cancer 13:67-72 Coggon D, Barker DJ, Cole RB, Nelson M (1989) Stomach cancer and food storage. J Natl Cancer Inst 81:1178-1182 Colditz GA, Branch LG, Lipnick RJ, Willett WC, Rosner B, Posner BM, Hennekens CH (1985) Increased green and yellow vegetable intake and lowered cancer deaths in an elderly population. Am J Clin Nutr 41:32-36 Correa P (1988) A human model of gastric carcinogenesis. Cancer Res 48:3554-3560 Correa P, Fontham E, Pickle LW, Chen V, Lin Y, Haenszel W (1985) Dietary determinants of gastric cancer in South Louisiana inhabitants. J Natl Cancer Inst 75:645-654 Craver L (1932) A clinical study of etiology of gastric and esophageal carcinoma. Am J Cancer 16:68-102 Cuello C, Correa P, Haenszel W, Gordillo G, Brown C, Archer M, Tannenbaum S (1976) Gastric cancer in Colombia: I. Cancer risk and suspect environmental agents9 J Natl Cancer Inst 57:1015-1020 Demirer T, Icli F, Uzunalimoglu O, Kucuk O (1990) Diet and stomach cancer incidence. A case-control study in Turkey. Cancer 65:23442348 Denk W, Hansluwka H, Karrer K (1967) Zur Epidemiologie des Carcinoms: III. Regionale Unterschiede in der H/iufigkeit des Magenkrebses in Osterreich. Z Krebsforsch Klin Onkol 70:13 30 Dooley CP, Cohen H (1988) The clinical significance of Campylobacter pylori. An Intern Med 108:70-79 Editorial Commitee (1979) Atlas of cancer mortality in the People's Republic of China. China Map Press Fontham E, Zavala D, Correa P, Rodriguez E, Hunter F, Haenszel W, Tannenbaum SR (1986) Diet and chronic atrophic gastritis: a casecontrol study. J Natl Cancer Inst 76:621 627 Forman D (1987) Dietary exposure to N-nitroso compounds and the risk of human cancer. Cancer Surv 6:719 738 Forman D, A1-Dabbagh S, Doll R (1985) Nitrates, nitrites and gastric cancer in Great Britain. Nature 313:620-625 Frentzel-Beyme (1988) Stomach cancer as a subject for epidemiological investigation for etiology and prevention. Saude Publica 5:9 26 Gao ZQ (1989a) (Unscheduled DNA synthesis (UDS) in human lymphocytes induced by drinking water in high risk areas of stomach cancer) Chung Hua Liu Hsing Ping Hsueh Tsa Chih 18:171-] 74 Gao ZQ (1989b) (Study on the effect of micronucleus with drinking water in the high risk area of stomach cancer). Chung Hua Yu Fang I Hsueh Tsa Chih 23:308-310 Graham S, Haughey B, Marshal1 J, Brasure J, Zielezney M, Freudenheim J, West D, Nolan J, Wilkinson G (1990) Diet in the epidemiology of gastric cancer. Nutr Cancer 13:19-34 Haenszel W (1982) Migrant studies. In: Schottenfeld D, Fraumeni JF (eds) Cancer epidemiology and prevention. Saunders, Philadelphia, pp 194-207 Haenszel W, Kurihara M, Locke FB, Shimuzu K, Segi M (1976) Stomach cancer in Japan. J Natl Cancer Inst 56:265 278 Hammond EC (1966) Smoking in relation to the death rates of 1 million men and women9 Natl Cancer Inst Monogr 19:127-204 Herbert WE, Bruske JS (1936) Etiology of cancer of the stomach. Guy's Hospital Rep 86:301-308 Hirayama T (1990) Life style and mortality. Karger, Basel Hoey J, Montvermay C, Lambert R (1981) Wine and tobacco: risk factors for gastric cancer in France. Am J Epidemiol 113:668-674

Houghton PW, Leach S, Owen RW, McC Mortensen NJ, Hill MJ, Williamson RC (1989) Use of a modified N-nitrosoproline test to show intragastric nitrosation in patients at risk of gastric cancer. Br J Cancer 60:231-234 Howson CP, Hirayama T, Wynder EL (1986) The decline in gastric cancer: epidemiology of an unplanned triumph. Epidemiol Rev 8:127 Hu JF, Zhang SF, Jia EM, Wang QQ, Liu SD, Liu YY, Wu YP, Cheng YT (1988) Diet and cancer of the stomach: a case-control study in China. Int J Cancer 41:331-335 Ikeda M, Yoshimoto K, Yoshimura T, Kono S, Kato H, Kuratsune M (1983) A cohort study on the possible association between broiled fish intake and cancer. Gann 74:640 648 Jedrychowski W, Wahrendorf J, Popiela T, Rachtan J (/986) A casecontrol study of dietary factors and stomach cancer risk in Poland. Int J Cancer 37:837-842 Joossens JV, Geboers J (1981) Nutrition and gastric cancer. Nutr Cancer 2:250-261 Joossens JV, Geboers J (1987) Dietary salt and risks to health. Am J Clin Nutr 45:1277 1288 Kamiyama S, Ohshima H, Shimada A, Saito N, Bourgade MC, Ziegler P, Bartsch H (1987) Urinary excretion of N-nitrosamine acids and nitrate by inhabitants in high- and low-risk areas for stomach cancer in northern Japan. IARC Sci Publ 84:497-502 Kodama M, Kodama T, Suzuki H, Kondo K (1984) Effect of rice and salty rice diets on the structure of mouse stomach. Nutr Cancer 6:135-147 Kono S, lkeda M, Tokudome S, Kuratsune M (1988) A case-control study of gastric cancer and diet in northern Kyushu, Japan. Jpn J Cancer Res 79:1067-1074 Lauren P (1965) The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. Acta Pathol Microbiol Scan 64:31-49 La Vecchia C, Negri E, Decarli A, D'Avanzo B, Franceschi S (1987) A case-control study of diet and gastric cancer in northern Italy. Int J Cancer 40:484-489 Levi F, La Vecchia C, Te VC (1990) Descriptive epidemiology of adenocarcinomas of the cardia and distal stomach in the Swiss Canton of Vaud. Tumori 76:167-171 Li JY, Ershow AG, Chen ZJ, Wacholder S, Li GY, Guo W, Li B, Blot WJ (1989) A case-control study of cancer of the esophagus and gastric cardia in Linixian. Int. J Cancer 43:755-761 McLaughlin JK, Gridley G, Block G, Winn DM, Preston-Martin, S. Schoenberg JB, Greenberg RS, Stemhagen A, Austin DF, Ershow AG, Blot W, Fraumeni JF (1988) Dietary factors in oral and pharyngeal cancer. J Natl Cancer Inst 80:1237-1243 McLaughlin JK, Hrubec Z, Blot WJ, Fraumeni JF Jr (1990) Stomach cancer and cigarette smoking among US veterans, 1954-1980 (Letter). Cancer Res 50:3804 Mirvish SS (1983) The etiology of gastric cancer. Intragastric nitrosamide formation and other theories. J Natl Cancer Inst 71:629647 Mirvish SS (1986) Effects of vitamins C and E on N-nitroso compound formation, carcinogenesis, and cancer. Cancer 58:1842 1850 Moller ME, Dahl R, Bockman OC (1988) A possible role of the dietary fibre product, wheat bran, as a nitrite scavenger. Food Chem Toxicol 26:841-845 Nomura A (1982) Stomach. In: Schottenfeld D, Fraumeni JF (eds) Cancer epidemiology and prevention. Saunders, Philadelphia, pp 624-637 Nomura A, Grove JS, Stemmermann GN, Severson RK (1990) A prospective study of stomach cancer and its relation to diet, cigarettes, and alcohol consumption. Cancer Res 50:627-631 Notani PN, Jayant K (1987) Role of diet in upper aerodigestive tract cancers. Nutr Cancer 10:103-113 Ohshima H, Bartsch H (1981) Quantitative estimation of endogenous nitrosation in humans by monitoring N-nitrosoproline excreted in the urine. Cancer Res 41:3658-3662 Ohshima H, Friesen M, Malaveille C, Brouet I, Hautefeuille A, Bartsch H (1989 a) Formation of direct-acting genotoxic substances in nitrosated smoked fish and meat products: identification of simple

143 phenolic precursors and phenyldiazonium ions as reactive products. Food Chem Toxico127:193-203 Ohshima H, Furihata C, Matsushima T, Bartsch H (1989b) Evidence of potential tumour-initiating and tumour-promoting activities of hickory smoke condensate when given alone or with nitrite to rats. Food Chem Toxico127:511 516 Parkin DM, L~i/irfi E, Muir CS (1988) Estimates of the worldwide frequency of sixteen major cancers in 1980. Int J Cancer 41:t84-197 Pisani P, Berrino F, Macaluso M, Pastorino U, Crosignani P, Baldasseroni A (1986) Carrots, green vegetables and lung cancer: a casecontrol study. Int J Epidemiol 15:463~468 Risch HA, Jain M, Choi NW, Fodor JG, Pfeiffer CJ, Howe GR, Harrison LW, Craib KJP, Miller AB (1985) Dietary factors and the incidence of cancer of the stomach. Am J Epidemiol 122:947-959 Rojas-Campos N, Sigaran MF, Bravo AV, Jimenez-Ulate F, Correa P (1990) Salt enhances the mutagenicity of nitrosated black beans. Nutr Cancer 14:1-3 Sato T, Fukuyama T, Urata G, Suzuki T (1959a) Studies of the causation of gastric cancer: 1. Bleeding in the glandular stomach of mice by feeding with highly salted foods, and a comment on salted foods in Japan. Bull Inst Publ Health (Tokyo) 8:10 13 Sato T, Fukujama T, Suzuki T, Takayanagi J, Murakami T, Shiotsuki N, Tanaka R, Tsuji R (1959b) Studies of the causation of gastric cancer: 2. The relation between gastric cancer mortality rate and salted food intake in several places in Japan. Bull Inst Publ Helth (Tokyo) 8:187-198 Smans M, Boyle P, Muir CS (1989) Cancer mortality atlas of the European Economic Community. Recent Results Cancer Research 114:253-268 Sobala GM, Schorah CJ, Sanderson M, Dixon MF, Tompki]as DS, Godwin" P, Axon ATR (1989) Ascorbic acid in the human stomach. Gastroenterology 97:357-363 Stanley K, Stjernswaerd J, Koroltchouk V (1988) Cancers of the stomach, lung and breast: mortality trends and control strategies. World Health Star Q 41:107-114 Stehr PA, Gloninger MF, Kuller LH, Marsh GM, Radford EP, Weinberg GB (1985) Dietary vitamin A deficiencies and stomach cancer. Am J Epidemiol 121:65 70 Stocks P, Karn MA (1933) A cooperative study on the habits, home life, dietary and family histories of 450 cancer patients and of equal number of control patients. Ann Eugen (Lond) 5:237-280 Stuehr DJ, Marletta MA (1987) Further studies on routine macrophage synthesis of nitrite and nitrate. IARC Sci Publ 84j335-344 Sugimura T, Fujimura S (1967) Tumor production in glandular stomach of rat by N-methyl-N'-nitro-N-nitrosoguanidine. Nature 216:943944 Tajima K, Tominaga S (1985) Dietary habits and gastro-intestinal cancers: a comparative case-control study of stomach and large intestinal cancers in Nogoya, Japan. Jpn J Cancer Res 76:705 716 Tannenbaum SR, Wishnok JS, Leaf CD (1990) Cell-mediated nitrosation. J Cancer Res Clin Oncol 116 [Suppl]:l 115

Tatematsu M, Takahashi M, Fukushima S, Hananouchi M, Shirai T (1975) Effects in rats of sodium chloride on experimental gastric cancers induced by N-methyl-N-nitro-N-nitrosoguanidine or 4-nitroquinoline-l-oxide. J Natl Cancer Inst 55:101-106 Trichopoulos D, Ouranos G, Day NE, Tzonou A, Manousos O, Papadimitriou CH, Trichopoulos A (1985) Diet and cancer of the stomach: a case-control study in Greece. Int J Cancer 36:291 297 Trock B, Lanza E, Greenwald P (1990) Dietary fiber, vegetables, and colon cancer; critical review and meta-analyses of the epidemiologic evidence. J Natl Cancer Inst 82:650 661 Tuyns AJ (1988) Salt and gastrointestinal cancer. Nutr Cancer 11:229 232 Ursin G, Bjelke E, Heuch I, Vollset SE (1990) Milk consumption and cancer incidence: a Norwegian prospective study. Br J Cancer 61:456-459 Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC, Leppert M, Nakamura Y, White R, Smits AMM, Bos JL (1988) Genetic alterations during colorectal-tumor development. N Engl J Med 319:525-532 Wargovich MJ (1987) Diallyl sulfide, a flavor component of garlic (Allium sativum) inhibits Dimethylhydrazine-induced colon cancer. Carcinogenesis 8:487~489 Wishnok JS, Richardson DP (1979) Interaction of wheat bran with nitrosamines and with amines during nitrosation. J Agric Food Chem 27:1132-1134 Wolff G (1989) Alkoholwirkung am Magen (Effect of alcohol on the stomach). Gastroenterology J 49:45~49 Wu-Williams AH, Yu MC, Mack TM (1990) Life-style, workplace, and stomach cancer by subsite in young men of Los Angeles County. Cancer Res 50:2569 2576 Yang PC, Davis S (1988) Epidemiological characteristics of adenocarcinoma of the gastric cardia and distal stomach in the United States, 1973-1982. Int J Epidemiol 17:293-297 You WC, Blot WJ, Chang YS, Ershow AG, Yang ZT, An Q, Henderson B, Xu GW, Fraumeni JF Jr, Wang TG (1988) Diet and high risk of stomach cancer in Shandong, China. Cancer Res 48:3518-3523 Zatonski W, Ohshima H, Przewozniak K, Drosik K, Mierzwinska J, Krygier M, Chmielarczyk W, Bartsch H (1989) Urinary excretion of N-nitrosamino acids and nitrate by inhabitants of high- and lowrisk areas for stomach cancer in Poland. Int J Cancer 44:823-827 Zentralinstitut ffir Krebsforschung. Bereich Nationales Krebsregister und Krebsstatistik (Institute of Cancer Research. Section National Cancer Registry and Cancer Statistics) (1981) Krebsinzidenz in der DDR 1978 (Cancer incidence in the GDR 1978). Akademie der Wissenschaften, Berlin Zentralinstitut ffir Krebsforschung. Bereich Nationales Krebsregister und Krebsstatistik (Institute of Cancer Research. Section National Cancer Registry and Cancer Statistics) (1989) Krebsinzidenz in der DDR 1986 (Cancer incidence in the GDR, 1986). Akademie der Wissenschaften, Berlin