Alcohol and Tobacco Use in Relation to Gastric Cancer: A Prospective Study of Men in Shanghai, China

Despite the declining incidence and mortality rates, gastric cancer remains the fourth most commonly diagnosed cancer and second most common cause of death from cancer worldwide (1). The considerable geographic variation in incident and mortality rates, as well as the observed decrease in risk among migrants from high-risk to low-risk areas (1, 2), indicate that environmental factors play a critical role in the etiology of gastric cancer. Infection with Helicobacter pylori (3-5) and diet (6-8) are among the most widely accepted environmental risk factors, but the etiology of gastric cancer remains to be fully understood. Identifying highly prevalent risk factors may aid in developing prevention strategies to reduce the incidence and mortality of this malignancy.

In addition to H. pylori infection and diet, cigarette smoking and alcohol drinking have also been examined as possible risk factors for gastric cancer. Although the relationship between smoking or alcohol and risk of gastric cancer has been the focus of many studies, results have been inconsistent and remain controversial. Moderately increased risk of gastric cancer has been associated with tobacco smoking in both cohort and population-based case-control studies (9-13), and in 2002, IARC concluded that there was “sufficient” evidence of causality between tobacco smoking and gastric cancer (14). However, the results of epidemiologic studies have not been fully consistent (15) and a fair number of studies failed to show increased risk of gastric cancer in association with tobacco smoke, particularly in European populations (16-18). In populations such as Shanghai, China, where the prevalence of other known risk factors for gastric cancer, such as infection with H. pylori, differ from other populations, it may be premature to make conclusions about the role of tobacco smoke in the etiology of gastric cancer.

The role of alcohol drinking in the etiology of gastric cancer remains even more controversial than tobacco smoking, especially given the high correlation between the two. In 2007, IARC concluded that alcoholic beverages are causally related to cancers of the oral cavity, pharynx, larynx, esophagus, liver, colorectum, and female breast, confirming the classification of alcoholic beverages as a group 1 human carcinogen (19). However, in terms of epidemiologic evidence in the association between alcoholic beverages and gastric cancer, IARC concluded that results are inconsistent and interpretation is not straightforward (19). Much of the existing data are based on Western or European populations. Given the different drinking habits and type of alcoholic beverages consumed between Asian and western populations, the epidemiologic findings from one population could not be directly applicable to the other. Furthermore, among Asian populations, the prevalence of the variant allele of aldehyde dehydrogenase 2 (ALDH2), the enzyme that breaks down acetaldehyde to acetate in the metabolism of alcohol, is much higher (28-45%) compared with other ethnic groups (20). This variant allele, ALDH2*2, results in higher levels of acetaldehyde in blood and saliva following consumption of alcoholic drinks, as well as in higher levels of acetaldehyde-related DNA adducts in lymphocytes (21).

Whereas there is no shortage of data on tobacco smoking and alcohol drinking in relation to gastric cancer, there is a lack of well-designed prospective epidemiologic studies, particularly in high-risk populations. The present study examined the relation between tobacco and alcohol, both alone and in combination, on the risk of developing gastric cancer in the Shanghai Cohort Study, a prospective cohort of 18,244 middle-aged and older men in Shanghai, China, with up to 19 years of follow-up for incident cancer.

The design of the Shanghai Cohort Study has been previously described in detail elsewhere (22, 23). Briefly, all male residents between the ages of 45 and 64 years of four geographically defined communities in Shanghai, with no prior history of cancer were invited to participate in a prospective study of diet and cancer. Between January 1, 1986 and September 30, 1989, 18,244 men (representing approximately 80% of the eligible subjects) were enrolled in the study. This study had been continually approved by the Institutional Review Boards at the University of Minnesota and the Shanghai Cancer Institute.

At enrollment, each participant completed an in-person interview using a structured questionnaire to obtain demographic information, history of tobacco and alcohol use, usual adult diet, and medical history. During the baseline interview, each participant was asked if he had ever smoked at least one cigarette per day for 6 months or longer. If he answered yes, he was classified as a smoker. For smokers, information about age at smoking initiation, current smoking status (yes, no), number of cigarettes smoked per day, and number of years of smoking over lifetime was also obtained. For those who already quit smoking at enrollment, the age and the number of years since quitting smoking were also recorded. Each participant was also asked if he had ever drunk alcoholic beverages at least once a week for 6 or more months. If he answered yes, information about the age he began drinking regularly and the typical amount of beer, wine, or spirits consumed per week, separately, was obtained. One drink was defined as 360 g of beer (12.6 g of ethanol), 103 g of wine (12.3 g of ethanol), or 30 g of spirits (12.9 g of ethanol; ref. 24). Current diet was assessed through a food frequency questionnaire that included 45 food items that represented commonly consumed local foods in the middle 1980s (25). Each subject was asked to indicate the frequency in number of times per day, week, month, or year, with which he usually consumed each of the 45 food items or food groups.

Annual follow-up of the cohort for incident cancers and deaths has been carried out since 1987 and is achieved in several ways. Surviving cohort members have been contacted in-person annually and have completed an in-person interview conducted by retired nurses employed by the Shanghai Cancer Institute. Study staff visits the last known address of each surviving cohort member and updates their medical history. For subjects who have moved, new addresses are sought from neighbors or from the local police department. Incident cancer cases are also identified via linkage analysis of the cohort database with the population-based Shanghai Cancer Registry and Shanghai Municipal Vital Statistics Office databases. As of September 2005, the cutoff date for case ascertainment of the present study, the cumulative losses to follow-up were 724 (4.0%) subjects (i.e., their vital status was unable to be determined via routine ascertainment methods). As of September 2005, a total of 391 incident gastric cancer cases had been identified.

For each cohort member, person-years of follow-up were calculated from the date of baseline in-person interview to the date of gastric cancer diagnosis or death, or the date of the last annual follow-up interview, whichever occurred first. Cox proportional hazards regression models were used to calculate hazard ratios (HR), their corresponding 95% confidence intervals (CI), and P values. All Cox regression models included the following covariates: age at baseline interview, year of baseline interview, and neighborhood of residence at recruitment. Multivariate regression models included the additional covariates: level of education (no formal schooling or primary school, junior middle school, senior high school, and college or above), body mass index (BMI; kg/m², continuous), summed intakes of preserved food items in tertiles, fresh fruits in tertiles, and fresh vegetables in tertiles.

To account for confounding by regular alcohol drinking when examining the independent effect of tobacco smoking on gastric cancer risk, Cox regression models included the covariate for number of years of regular drinking (continuous) given no additional predictive value of other variables for alcohol use in gastric cancer. Similarly, when assessing the independent effect of alcohol drinking on gastric cancer risk, only the number of years of smoking was included as a covariate given that it carried all predictive values of smoking for gastric cancer risk.

Conditional logistic regression models were used to calculate odds ratios (OR) and their corresponding 95% CIs within a subset of the present study population to further adjust for serologic status for H. pylori antibodies, serum levels of β-carotene and vitamin C, and urinary levels of total isothiocyanates (ITC) in combination with glutathione S-transferase (GST) M1 and GSTT1 genotypes, all of which were identified risk or protective factors for gastric cancer in this study population (5, 8, 26). This analysis included 338 cases and 1,013 matched controls with known values in one of the following biomarkers: serologic status for H. pylori antibodies (295 cases and 980 controls), serum level of β-carotene (195 cases and 588 controls), serum level of vitamin C (192 cases and 591 controls), urinary level of ITC (307 cases and 926 controls), GSTM1 genotype (176 cases and 771 controls), and GSTT1 genotype (171 cases and 774 controls). To maximize the sample size for the analysis of the adjusted effect of smoking and alcohol consumption on gastric cancer risk, a categorical variable was created for unknown value of each variable. This variable was included in all conditional logistic regression models.

All statistical analyses were carried out using SAS software version 9.1 (SAS Institute). All P values reported are two-sided, and those that were <0.05 were considered to be statistically significant.

The 18,244 members of the cohort had accrued 281,808 person-years of follow-up by the end of September 2005 and yielded 391 incident stomach cancer cases (incidence rate of 138.7 per 100,000 person-years). Of these cases, 342 (87.5%) were confirmed with histopathology, 6 (1.5%) were confirmed with cytology or surgery, 30 (7.7%) were confirmed with radiology, and the remaining 13 (3.3%) cases were diagnosed based on clinical symptoms alone. The mean age at cancer diagnosis was 67.2 years (SD, 7.0). The mean time interval between enrollment into the cohort and cancer diagnosis was 9.2 years (range, 0.6 months to 19.1 years).

Men who developed gastric cancer had comparable BMI to those who remained cancer-free. Compared with subjects who remained free of cancer during follow-up, patients with gastric cancer were less educated and more likely to regularly smoke cigarettes and drink alcohol (Table 1). Gastric cancer patients also showed longer duration of smoking or drinking or consumed larger amount of alcohol per day or over lifetime than those who remained cancer-free (Table 1).

Cigarette smoking was associated with a statistically significant increased risk of stomach cancer (Table 2). After adjustment for potential confounders including alcohol intake, ever smokers experienced a statistically significant ∼59% increased risk of developing stomach cancer compared with never smokers (HR, 1.59; 95% CI, 1.27-1.99). Former smokers (HR, 1.79, 95% CI, 1.25-2.57) had slightly higher HR than current smokers (HR, 1.55; 95% CI, 1.23-1.96). However, after excluding subjects who were diagnosed with gastric cancer within 2 years and the corresponding person-years of the entire cohort, HRs for gastric cancer were comparable for former (HR, 1.74; 95% CI, 1.18-2.56) and current smokers (HR, 1.61; 95% CI, 1.26-2.06). Among ever smokers, there was a lack of dose-response relation for risk of gastric cancer with number of cigarettes per day, early age of starting to smoke, number of years of smoking, or cumulative pack-years of cigarettes smoked (Table 2). Smokers who quit smoking did not experienced much risk reduction compared with those who continued to smoke (Table 2).

The association between alcohol intake and the risk of gastric cancer is shown in Table 3. Overall, men who drank alcoholic beverages regularly had comparable risk of gastric cancer with those who did not after taking into account the effect of cigarette smoking and other potential confounders. Heavy drinkers (4+ drinks per day) had a statistically significant ∼50% increased risk of gastric cancer compared with nondrinkers (HR, 1.46; 95% CI, 1.05-2.04). Drinking regularly for a greater number of years did not further increase the risk of developing gastric cancer. Compared with nondrinkers, men who drank 4 or more alcoholic beverages per day for 30 or more years experienced a statistically significant ∼50% increased risk of gastric cancer (HR, 1.49; 95% CI, 1.01-2.19).

The most commonly consumed type of alcoholic beverage of the study population was spirits, which accounted for 54.5% of the total ethanol consumed. Rice wine was the second most commonly consumed type of alcohol, accounting for 33.1% of the total ethanol consumed, followed by beer, which accounted for 12.4% of the total ethanol intake. Among regular alcohol drinkers, men who developed gastric cancer consumed more spirits per day than their nongastric cancer counterparts (1.6 versus 1.3 drinks per day; P = 0.065). After adjusting for intake of rice wine and beer, consumption of 4 or more drinks of spirits per day was associated with a statistically significant 74% increased risk of gastric cancer compared with nondrinkers (HR, 1.74; 95% CI, 1.16-2.63; P for trend = 0.001; Table 4). Consumption of 2 or more drinks of rice wine per day was associated with a nonstatistically significant increased risk of gastric cancer compared with nondrinkers (HR, 1.30; 95% CI, 0.92-1.86; P for trend = 0.12). Beer intake was not associated with risk of developing gastric cancer after adjusting for consumption of rice wine and spirits (P for trend = 0.18). Further adjustment for cigarette smoking attenuated the associations between the consumption of rice wine or spirit and risk of gastric cancer (Table 4).

We examined the joint effects of cigarette smoking and alcohol consumption in association with the risk of developing gastric cancer (Table 5). Within each status of cigarette smoking, especially among never smokers, the risk of gastric cancer increased with increasing number of drinks of alcoholic beverages consumed per day, although the tests for linear trend were not statistically significant. Similarly, within each level of alcohol intake, smoking was associated with increased risk of gastric cancer, especially among nondrinkers, in whom the smoking-gastric cancer risk association was statistically significant (P < 0.0001). There was no evidence of interaction between cigarette smoking and alcohol drinking on the risk of gastric cancer (P for interaction = 0.15). Heavy drinkers who also smoked cigarettes had statistically significant 2-fold increased risk of gastric cancer as compared with those who neither drank alcohol nor smoked cigarettes.

Infection with the bacterium H. pylori is an established risk factor for the development of gastric cancer (3-5). Our group has previously shown a reduced risk of gastric cancer associated with serum levels of β-carotene and vitamin C and urinary levels of ITCs (8, 26). Within the current study population, 338 cases and 1,013 matched controls had known values for at least one of the following biomarkers: serologic status for H. pylori antibodies, serum level of β-carotene, serum level of vitamin C, urinary level of total ITC, known GSTM1 genotype, or known GSTT1 genotype. Adjustment for these biomarkers did not materially change the association between smoking/alcohol drinking and risk of gastric cancer. Compared with nonsmokers, ever smokers had an OR of 1.57 (95% CI, 1.17-2.12) for developing gastric cancer after adjusting for H. pylori serology, serum levels of β-carotene and vitamin C, alcohol intake, and other potential confounders (Table 6). Similarly, men who drank 4 or more alcoholic drinks a day had an OR of 1.58 (95% CI, 0.99-2.53) compared with nondrinkers after adjustment for these biomarkers and other potential confounders. Further adjustment for urinary level of total ITC in combination with GSTM1 and GSTT1 genotypes did not alter the results (Table 6). After adjusting for urinary levels of ITC in combination with GSTM1 and GSTT1 genotypes, compared with nonsmokers, ever smokers had an OR of 1.56 (95% CI, 1.15-2.13) for developing gastric cancer. Men who drank 4 or more alcoholic beverages a day had an OR of 1.58 (95% CI, 0.96-2.61; Table 6). The adjusted OR for gastric cancer for heavy drinkers (4+ drinks per day) who were also long-term smokers (30+ years of smoking) was 2.63 (95% CI, 1.38-5.05), very similar to that based on the entire cohort but without adjustment for these biomarkers (data not shown).

We also examined the main effects of cigarette smoking and alcohol consumption on the risk of gastric cancer by cardia or noncardia subsite. Of the 391 gastric cancer cases included in the present study, 78 were cardia tumors, 234 were noncardia tumors, and 79 were of unknown subsite. Neither cigarette smoking nor alcohol drinking was significantly associated with increased risk of developing gastric cardia cancer. After adjustment for potential confounders including alcohol intake, the HR of gastric cardia cancer for ever smokers compared with never smokers was 1.35 (95% CI, 0.83-2.19). Similarly, the multivariate-adjusted HR for heavy drinkers relative to nondrinkers was 0.80 (95% CI, 0.33-1.93). The risk of developing gastric noncardia cancer associated with cigarette smoking and alcohol consumption was similar to that of the overall cohort. After adjustment for potential confounders including alcohol intake, ever smokers had a statistically significant 68% increased risk of developing gastric noncardia cancer compared with never smokers (HR, 1.68; 95% CI, 1.25-2.26). Men who reported smoking at least 20 cigarettes per day had a statistically significant 64% increased risk of developing gastric noncardia cancer compared with never smokers (HR, 1.64; 95% CI, 1.17-2.30). Similar to the results based on the overall cohort, regular alcohol beverage consumption was not significantly associated with increased risk of gastric noncardia cancer (HR, 1.09; 95% CI, 0.83-1.42). Heavy drinkers, however, had a borderline statistically significant 51% increased risk of developing gastric noncardia cancer compared with nondrinkers (HR, 1.51; 95% CI, 0.99-2.32; data not shown).

The potential effect of duration of follow-up on our results was examined by stratifying the entire follow-up into two time periods: <10 years and 10+ years. The associations between cigarette smoking and risk of gastric cancer with short-term (<10 years) or long-term (10+ years) follow-up were comparable. The HRs (95% CIs) of gastric cancer for current smokers with <10 years and 10+ years of follow-up were 1.54 (1.11-2.13) and 1.56 (1.12-2.17), respectively, compared with never smokers. A slightly higher HR for former smokers within 10 years post-enrollment (HR, 1.99; 95% CI, 1.23-3.23) than those with 10+ years of follow-up (HR, 1.55; 95% CI, 0.89-2.69) suggested that people might quit smoking due to the subclinical symptoms of the malignancy. Similarly, the associations between alcohol use and gastric cancer for <10 years and 10+ years of follow-up time periods were comparable as well. Compared with nondrinkers, HRs (95% CIs) for gastric cancer were 1.60 (1.02-2.51) for heavy drinkers (4+ drinks per day) within 10 years post-enrollment and 1.32 (0.81-2.17) for heavy drinkers with more than 10 years of follow-up (data not shown).

Our prospective study showed that both cigarette smoking and heavy alcohol drinking may be independent risk factors for the development of gastric cancer. A statistically significant positive association between smoking and gastric cancer risk present among nondrinkers implicates an independent role of tobacco use in the risk of gastric cancer for this high-risk population. The present study also showed that heavy alcohol drinking (4+ drinks per day) was also associated with a significant increase in risk of gastric cancer when adjusting for smoking. Furthermore, an elevated risk of gastric cancer associated with heavy alcohol intake was present in lifelong nonsmokers, suggesting a role of alcohol independent of smoking in the development of gastric cancer. More importantly, further adjustment for H. pylori serology, serum levels of β-carotene and vitamin C, urinary levels of ITC, and GSTM1/GSTT1 genotypes, which were identified risk/protective factors for gastric cancer in this population, did not materially alter the association between cigarette smoking/heavy alcohol intake and risk of gastric cancer.

More recent cohort and case-control studies have also shown a moderately increased risk of gastric cancer among cigarette smokers (9-13, 27). Whereas many of these studies found a positive association, others did not (15-18). In a prospective cohort of 26,132 Norwegian men and women followed from 1966 to 1993, smoking status at enrollment was not statistically significantly associated with risk of gastric cancer (18). In this cohort study, the risk of gastric cancer in current and former smokers at baseline was not significantly different from that of never smokers. In men, the relative risk for former smokers compared with never smokers was 1.3 (95% CI, 0.9-2.0), whereas the corresponding risk in women was 0.8 (95% CI, 0.4-1.6). Similarly, the relative risk for men who were current smokers at baseline compared with never smokers was 1.2 (95% CI, 0.8-1.6) and the corresponding risk in women was 1.1 (95% CI, 0.8-1.4; ref. 18). Although this was a large prospective cohort study, it is unclear whether the investigators adjusted for other potential confounders including lifestyle and dietary factors. In contrast, a meta-analysis of 40 studies examining the tobacco smoking and gastric cancer association found a summary relative risk of 1.5 to 1.6 for smokers compared with nonsmokers (13). Similarly, a recent systematic review estimated a summary relative risk of gastric cancer for current smokers compared with nonsmokers to be 1.56 (95% CI, 1.36-1.80) among Japanese (28). The magnitude of the association between cigarette smoking and gastric cancer risk in the present study after adjusting for alcohol intake, H. pylori serology, and other dietary factors is in line with these summary estimates.

The exact mechanism of how tobacco smoking increases the risk of gastric cancer is not well understood although there are several possible explanations. Tobacco has been shown to be carcinogenic on gastric mucosa in vitro (29). It has also been shown that cigarette smoking is related to an increase in the frequency of transition from normal gastric mucosa to precancerous lesions (30, 31), an important step in the model of gastric carcinogenesis postulated by Correa (32). Tobacco smoke contains more than 60 carcinogens in humans (14). In particular, nitrosamines and other nitroso compounds present in cigarette smoke may be involved in the gastric carcinogenesis (33). These carcinogens are capable of covalently binding to DNA, which alters the normal function of DNA and eventually leads to gastric cancer (34). In the present study, after adjusting for alcohol intake and other potential confounders, we observed an approximately 60% increased risk of gastric cancer for current smokers. In addition, a dose-response relationship between smoking and gastric risk strengthened the hypothesis that cigarette smoking may contribute to gastric carcinogenesis.

Alcohol consumption has been considered to be an unlikely risk factor in the development of gastric cancer but previous findings are contradictory. In a systematic review of the epidemiologic studies on the association between alcohol intake and gastric cancer among Japanese populations, nine of 11 cohort studies showed no association, two showed a strong positive association, and one showed a weak association (35). All 11 case-control studies included in the review failed to find an association (35). Similarly, studies within European and American populations did not observe any statistically significant associations between various degrees of alcohol intake and risk of gastric cancer (36-40). Others, however, have reported statistically significant increases in risk of developing gastric cancer (11, 12, 27, 41, 42). Although the findings were mixed, many of these studies were unable to fully adjust for several well-known risk factors of gastric cancer including infection with H. pylori, dietary intakes of fresh fruit and vegetables and preserved foods, and tobacco smoking. In the present study, we were able to adjust for tobacco use and dietary intakes of fresh fruit and vegetables as well as preserved foods and still observed a statistically significant increase in risk of gastric cancer for heavy alcohol drinkers (4 or more drinks per day). More importantly, further adjustment for H. pylori serology, serum levels of β-carotene and vitamin C, and urinary levels of total ITC in combination with GSTM1 and GSTT1 genotypes, which were identified risk/protective factors in this population (5, 8, 26), did not materially alter the alcohol-gastric cancer risk association. These findings suggest an independent, albeit moderate, effect of alcohol intake on gastric cancer risk.

The exact biological mechanisms for alcohol causing gastric carcinogenesis are not well understood. There are several possible mechanisms. Ethanol itself is not a carcinogen but rather acts as a solvent that enhances the penetration of tobacco gastric carcinogens and nitrosamines (43). Nitrosamines present in alcoholic beverages,in particular hard liquor, may be responsible for the observed association between intake of alcohol, especially hard liquor, and risk of gastric cancer (44, 45). The metabolic intermediate of ethanol, acetaldehyde, is a recognized animal carcinogen (46). Most of the acetaldehyde generated during the metabolism of ethanol is eliminated by aldehyde dehydrogenase 2 (ALDH2). ALDH2 is polymorphic in humans with two alleles: ALDH2*1 and ALDH2*2 (46). Individuals homozygous for the mutant allele ALDH2*2 have abolished ALDH2 enzyme activity, whereas heterozygous individuals, ALDH2*1,2, show reduced enzyme activity (47). The abolished enzyme activity in ALDH2*2 homozygous individuals results in blood acetaldehyde levels 6 to 20 times higher compared with ALDH2*1 homozygous individuals (47) and has been associated with increased risk of gastric cancer in a Japanese population (48) and increased blood levels of acetaldehyde-derived DNA adducts (21). This mutant allele is extremely rare in Caucasians (<1%); in contrast, approximately 40% of the Asian population are heterozygous for ALDH2*2 (20, 49, 50). The difference in frequency of the ALDH2*2 mutant allele in different populations may contribute to the varying results on alcohol intake and gastric cancer risk across different study populations. Lastly, the observed increase in risk of gastric cancer among heavy drinkers but lack of a dose-response relationship may be the result of differing dietary habits or other lifestyle factors between heavy drinkers and nondrinkers. Future studies are warranted to directly address these potential confounding issues when results from different populations are pooled.

Glutathions S-transferases (GST) are phase II enzymes that catalyze the conjugation of potential carcinogens present in tobacco smoke (detoxification pathway) as well as potential chemopreventive compounds such as ITCs from cruciferous vegetables. We found that ITC protects against the development of gastric cancer in this population, and the protective effect was stronger in people possessing the homozygous deletion polymorphism of the GSTM1 or GSTT1 genes (26) and thus lack the respective enzymes. The inconsistent findings of tobacco and alcohol use in the risk of gastric cancer could be, at least in part, due to the variations in the prevalence of GSTM1 and GSTT1 deletion polymorphisms and the exposures to dietary ITC and tobacco and alcohol across different populations. We adjusted for ITC in combination with the GSTM1 and GSTT1 genotypes in the present analysis to reduce their potential confounding effect on the association between smoking/alcohol and gastric cancer risk.

As with any observational study, this study has strengths and limitations. The present study showed that when adjusting for cigarette smoking, heavy drinking (4+ drinks per day) was significantly associated with an increased risk of gastric cancer. The stratified analysis of heavy drinking in never smokers suggested an increased risk associated with heavy drinking but this did not reach statistical significance due to the relatively small number of cases who drank alcohol but never smoked cigarettes. Infection with H. pylori is an established risk factor for the development of gastric cancer. Adjustment for serologic status of antibodies to H. pylori was limited to a subset of the study population, but adjusting for this potential confounder did not alter the association between smoking/heavy alcohol and gastric cancer risk. Further adjustment for biomarkers of ITC and antioxidant intakes in this subset did not change the observed association either. The present study was also limited by the one-time assessment of dietary and lifestyle factors at baseline. It is possible that during the 20 years of follow-up, dietary and other lifestyle changes occurred that would result in a dilution of the exposure-disease association. However, when we analyzed early cases and late cases separately, the data do not indicate this. Additionally, the present study has several strengths including the population-based, prospective design with up to 20 years of follow-up and almost complete ascertainment of incident cancer cases.

In summary, cigarette smoking and alcohol consumption may be independent risk factors for gastric cancer in this high-risk population. Modification of these lifestyle choices may help to reduce the morbidity and mortality of this malignancy.

No potential conflicts of interest were disclosed.

Grant Support: NIH grant R01 CA144034.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.