Effect of Life Styles on the Risk of Sybsute-specific Gastric Cancer in Those with and without Family History

To evaluate a combined effect of gastric cancer family history (GCFH) and selected living habits on the subsite-specific of gastric cancer, a hospital-based case-referent study was conducted in Tokai area of Japan. The study subjects were 850 newly diagnosed gastric cancer (GC) patients and 28,619 cancer-free first-visit outpatients. Odds ratios (ORs) of all subsites of GC in subjects with both GCFH and habitual smoking were significantly higher (OR=4.22) compared with those with merely GCFH(OR=1.81) or habitual smoking (OR=2.83). When positive GCFH subjects frequently consumed raw vegetable, the risk of GC decreased in cardia (OR=0.68), antrum (OR=0.43) and all subsites (OR=0.74). Our findings provided evidence that GCFH and habitual smoking increased the risk of GC with family history, while frequent intake of raw vegetable decreased the risk and it was modified by other environmental factors. J Epidemiol, 1999; 9 : 40-45

To evaluate a combined effect of gastric cancer family history (GCFH) and selected living habits on the subsite-specific of gastric cancer, a hospital-based case-referent study was conducted in Tokai area of Japan.The study subjects were 850 newly diagnosed gastric cancer (GC) patients and 28,619 cancer-free first-visit outpatients.Odds ratios (ORs) of all subsites of GC in subjects with both GCFH and habitual smoking were significantly higher (OR=4.22) compared with those with merely GCFH(OR=1.81) or habitual smoking (OR=2.83).When positive GCFH subjects frequently consumed raw vegetable, the risk of GC decreased in cardia (OR=0.68),antrum (OR=0.43)and all subsites (OR=0.74).Our findings provided evidence that GCFH and habitual smoking increased the risk of GC with family history, while frequent intake of raw vegetable decreased the risk and it was modified by other environmental factors.J Epidemiol, 1999; 9 : 40-45 family history, smoking, raw vegetable, gastric cancer, case-referent study Although incidence and mortality rates of gastric cancer (GC) have declined worldwide 1), this cancer still remains one of the leading cause of death in Asian countries like Japan and China It has been known for a long time from epidemiological researches that the risk of gastric cancer is associated with gastric cancer family history (GCFH) and living habits It was reported that GC is related to fondness of salty foods, GCFH, habitual smoking and drinking ; whereas frequent consumption of raw vegetable, fruit and green tea showed protective effects 8.9).But in the evaluation of family history it is difficult to exclude the confounding factors caused by similar life styles which are shared by family members.Furthermore in the protection of subjects with GCFH, it is necessary to know what kind of living habits will interact with deleterious effect of GCFH.
With this background in mind, we conducted a case-referent study with the use of data from the Hospital-based Epidemiologic Research Program at Aichi Cancer Center (HERPACC) to examine for the first time whether the interplay of some living habits and GCFH would modify the risk of food intake, general health status and productive status before symptoms appeared.
Between 1990-1995, 29,506 first-visit outpatients with complete information were recruited after excluding those under 20 or older than 79 years of age and those with past history of cancer.The proportional distribution in clinical diagnoses of these outpatients had been reported in our previous studies 11) All 887 cases identified from hospital cancer registry system were pathologically confirmed with primary GC .Anatomic subsites of GC were coded according to Japanese Classification of Gastric Carcinoma (1995) by means of surgical observation.Subsites of gastric cancer were grouped into three categories: cardia, middle (body) and antrum.Those subsites which cannot be determined because of multiple, diffuse, or advanced cancers were listed as unclassified.
All eligible referents used for analysis were not matched, because our previous study showed that large number of referents based on the outpatients of ACCH between 1988-1990 gave a steadier estimate than matched analysis 12).In this study GCFH was defined as positive if the subject had at least one parent or sibling suffered from GC. Positive GCFH was symbolized by GCFH (+); negative GCFH by GCFH (-).Current smoker referred to the person who smoked 1 or more cigarettes per day.Consumption of milk, raw vegetable and green tea were divided into three or four categories according to the intake frequency.Seldom intake of fruit means less than once a week.Salty food was roughly divided into like or dislike.
For data analysis, odds ratio (OR) and its 95% confidence interval (95% CI) for each exposure variable were calculated using an unconditional logistic regression model, adjusted by sex and age.In order to control the confounding factors and find independent protective or risk factors, a multivariate analysis was conducted using a model including age at diagnosis, gender and those factors shown statistical significance by un ivariate analysis.The logistic procedure from SAS (SAS Institute) was used for calculation 13)

RESULT
The distribution of cases and the referents according to age, sex, GCFH of parents and siblings, smoking and drinking habits, consumption of raw vegetable, milk, fruit, salty food and green tea were listed in table 1 (Table 1).Most cases were more than 50 years old and the age distribution of referents was slightly younger, than that in cases.The male to female ratio was 2.04 in case group.
In table 3, items which had shown significant influences on ORs for CG included GCFH, smoking, drinking, salty food, raw vegetable and fruit were put in the multivariate regression model (Table 3).To test possible interaction of variables in the multivariate regression model, likelihood statistics from models with or without interaction were contrasted.The contrast of the fit of the logistic models showed no evidence of any interactions (all P value > 0.1) in the model of total site or subsites of gastric cancer.Having been adjusted by age and sex, ORs of GCFH and habitual smoking were 1.56 (95% CI=1.28-1.90)and 2.22 (95% CI=1.68-2.9),respectively.Frequent consumption of raw vegetable still showed a significantly lowered OR (0.8,95% CI=0.67-0.95) Although statistically there were no intreactions between GCFH and habitual smoking or raw vegetable, it was necessary to compare the odds ratios if habitual smoking or raw vegetable appeared in persons with different status of GCFH.
Analyses of a combined effect of GCFH and habitual smoking on GC by subsites in table 4 showed that habitual smoking alone presented a significantly high OR only for antrum cancer (OR=5.95,95% CI=2.12-16.7);Positive GCFH alone presented a significant high OR only for antrum cancer (OR=7.12,95% CI=2.01-25.9)(Table 4).Whereas, when habitual smoking was combined with GCFH, significantly elevated ORs for all subjects were observed and was more remarkable for cardia (OR=15.9,95% CI=1.93-131) than for antrum (OR=9.11,95% CI=3.00-28).
A combined effect of GCFH and frequent consumption of raw vegetable by subsite is shown in table 5 (Table 5).Compared with subjects who had GCFH but seldom consumption of raw vegetable, those with both GCFH and frequent consumption of raw vegetable had a significantly lowered OR for antrum cancer (OR=0.43,95% CI=0.19-0.98);ORs for cardia and middle cancers were 0.68 (95% CI=0.13-3.49)and 1.05 (95% CI=0.51-2.17),respectively.
In this hospital-based case-referent study of GC, excess risks were associated with habitual smoking and positive GCFH by both univariate and multivariate analyses .When examined by Family history:at least one parent or one sibling reported disease of gastric cancer then defined as positive;otherwise negative .Odds ratio of smoking and drinking were calcuate in male group only,others were adiusted for age sex.**P<0.01,*P<0.05.
Table 2. Odds ratio of life style, calculated separately with the use of total, GCFH(-) and GCFH(+) outpatients as control group.
Family history:at least one parent or one sibling reported disease of gastric cancer then defined as positive, GCFH(+),otherwise negative, GCFH (-).Odds ratio of smoking and drinking were calculate in male group only, others were adjusted for age and sex.**P<0 .01,*P,0.05.jects with habitual smoking and positive GCFH, an excess risk combined with GCFH, increased intake of raw vegetable sigwas sharply observed in all subsites (Table 4).Consumption of nificantly reduced the risk for antrum cancer (Table 5).raw vegetable reduced the risk for GC independently.When Adjusted by age,drinking, raw vegetable, green tea, salty food, fruit and physical exercise **P<0.01,*P<0.05.adjusted by sex, age, smoking, drinking, physical exercise, green tea, salty food and fruit.**P<0.01,*P<0.05.

DISCUSSION
Several potential limitations of this case-referent study should be noted .Because we used non-cancer out-patients as the referent group, selection bias should be considered at first.Fortunately, ACCH is convenient to everybody who wants to visit it, even without doctors referrals.Cancer cases account for only 14% of all outpatients and 44% of outpatients were diagnosed disease free.Our recent published research which compared sex, age, smoking and dietary habits, etc. of these firstvisit outpatients with general population showed that it is feasible to use ACCH non-cancer outpatients as controls in epidemiological studies 11).Second, recall bias should also be discussed, however, the questionnaire study was conducted before the final diagnosis, thus it is an even chance that recall bias influenced cases and referents.Third, all eligible non-cancer outpatients were used as referents group on the basis of Dr.
Hamajima's research that a large number of referents reduces the satatistical variation and give a steadier estimate than matched analysis ').This study explored that family history is a risk factor independent of smoking and some dietary factors, there are still other important environmental factors such as H elicobacter Pylori (HP) infection which were not included in the present study.
We can not deny this was a limitation of this study, however the infectious rates of HP in adult Japanese with or without gastric cancer are both high24), finally it would not distort the main results obtained in the present study.
Many epidemiological studies in different countries suggested that GC risk is high when a subject reported a history of GC in first-degree relatives This seems genetic factors may play an important role in the pathogenesis of GC.But in these researches that the aggregation of several cases of gastric cancer in a family might also be caused by environmental factors (like smoking and dietary habits) shared by the family members cannot be excluded.Our results demonstrated that the combined effect of GCFH and habitual smoking, a selected life styles at risk, significantly increased risk for all subsites of GC, we hypothesized that there might exist biological interplay between such life style and hereditary factors.It might be postulated that some common carcinogenesis pathway links genetic and environmental factors.Cigarette smoking might have a significant impact on the mutations of p53 gene in human cancers'").Overexpression of p53 protein had also been suggested to be one of the familial factors that correlate with carcinogenesis of stomach 's).So when genetic factors and smoking combined together, it might promote the mutations of p53 gene.Some of the most common precancerous lesions of GC like severe atrophic gastritis, gastric metaplasia and pernicious anemia are thought to be genetically determined 16.These lesions always were deteriorated by habitual smoking clinically.We proposed positive HCFH perhaps provided genetic susceptibility to GC some mutagens in cigarette might accelerate this process.Case report of the almost simultaneous occurrence of gastric cancer in monozygotic twin also speculated that an interaction may exist between genetic susceptibility and environmental factors as one of the twin with cigarette smoking and drinking habits developed GC early and severely18).
In our study, raw vegetable demonstrated special protective effect against antrum GC which had been proved to be coincidentally the most susceptible subsite in subjects with positive GCFH 19), but the protective effect of raw vegetable was not significant in other subsite where genetic factors contribute less importantly.It suggested that antrum should be a susceptible area to carcinogenesis among subjects with GCFH.It remains unclear whether the protective effects of raw vegetable on GCFH positive antrum cancer may be by direct influence on gene mutation or by modifying other carcinogens .We know raw vegetable is rich in vitamin C, carotenoids, dietary fiber, vitamin E and selenium.Vitamin C and vitamin E inhibit intragastric formation of N-nitroso compounds in human as well as in animals.Vitamin C also reduces nitrate to nitric oxide, vitamin E functions as a scavenger of lipid peroxides 20).Our previous studies had shown protective effects of raw vegetable against GC, lung cancer and oral pharyngeal cancer 8. 21,22).But this effect was not observed in cooked vegetables 8).Raw vegetable differs from cooked one in that the heat employed in the latter inevitably causes Vitamin C destruction.In most plant tissues there is an enzyme , ascorbic 60°C or over; in this condition, vitamin C will be destroyed with accelerated speed 1) .So we recommended the person with GCFH abandoning the smoking habit and in the meantime increasing the intake of raw vegetable.
In short, this hospital-based case-referent study of stomach cancer provided evidence that both cigarette smoking and GCFH were risk factors for GC, but frequent consumption of raw vegetable was a protective factor .When smoking habit and positive GCFH combined together, the risk of GC dramatically increased in all subsites of stomach, but if GCFH positive subjects increased consumption of raw vegetable, such risk decreased, particularly in the antrum.

Table 1 .
Distribution of gastric cases and referents by age,sex,lifestyle and gastric cancer family history (GCFH), dietary and beverage habits.

Table 3 .
Multivariate analysis on six main factors including gastric cancer family history

Table 4 .
Combined effect of gastric cancer family history (GCFH) and smoking habit in males.

Table 5 .
Combined effect of gastric cancer family history (GCFH) and raw vegetable.