Frequent low dose alcohol intake increases gastric cancer risk: the Health Examinees-Gem (HEXA-G) study

Introduction

In most populations, the incidence and mortality rates for gastric cancer (GC) have been steadily declining, but the incidence rates remain high in East Asian countries¹. The considerable geographic variations in incidence and mortality rates indicate that environmental and lifestyle factors play important roles in the etiology of GC².

According to the World Cancer Report, the epidemiological evidence of an association between alcohol consumption and stomach is conflicting³. Alcohol has been considered a group 1 carcinogen by the International Agency for Research on Cancer Monographs Programme since 1988, and alcohol consumption has been causally associated with cancers of the oral cavity, esophagus, liver, colorectum, and breast in women^4–8. However, for GC, the evidence of a link to alcohol consumption remains limited, and a weaker dose-risk relationship has been observed^4,9.

The 2018 Continuous Update Project by the World Cancer Research Fund has reported that an intake of approximately 45 grams or more of pure alcohol per day, equivalent to 3 drinks, is a probable cause of GC¹⁰. However, no conclusions were made for intakes below 45 grams of ethanol in the meta-analysis. Likewise, numerous epidemiological studies have supported the harmful effects of heavy episodic drinking or binge drinking¹¹, yet the effects of light to moderate levels of alcohol drinking remain to be clarified, owing to inconsistent results^12–15. Additionally, Deng et al.¹⁶, in a recent meta-analysis exploring the association between alcohol consumption and increased gastric cancer risk, have reported a significant association only in case-control studies but not cohort studies. Assessment of drinking patterns, such as a range of intake levels, should be explored with respect to drinking frequency, to better elucidate the relationship between alcohol intake and GC development. These findings could aid in identifying preventable real-world measures associated with drinking behavior.

Using a large-scale, prospective cohort, this study aimed to investigate the sex-specific association of alcohol drinking patterns, specifically frequency and quantity, with the risk of GC in a population with a high incidence rate of GC.

Materials and methods

Study population

The Health Examinees-Gem (HEXA-G) study is a subset of the larger HEXA study, one of the 6 prospective cohorts in the Korean Genome and Epidemiology Study (KoGES)¹⁷. The HEXA study is a large-scale, population-based prospective cohort of Korean adults 40–69 years of age who visited health examination centers in 38 regions from 2004 to 2013¹⁸. The HEXA-G data used in this study followed additional eligibility criteria, excluding the following: 1) sites that participated in the pilot study only from 2004 to 2006; 2) sites that did not meet the HEXA standards for biospecimen quality control; 3) sites that participated in the study for <2 years^19,20. To date, 139,273 participants in the HEXA-G study completed the initial follow-up survey until 2017.

In this study, we excluded HEXA-G participants with missing information in the Korea Central Cancer Registry (KCCR; n = 134) and those with a history of cancer in the baseline study (n = 4,679). Participants with missing information for alcohol-related variables (n = 1,179) and former drinkers (n = 4,918) were further excluded (Figure 1). We also excluded participants who were diagnosed with GC within the first 2 years of the baseline survey (n = 145) to minimize reverse causality effects. We ultimately included 128,218 participants (42,152 men and 86,066 women) in the final analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart of the study population. HEXA, Health Examinees Study; GC, gastric cancer.

The HEXA-G study protocol was approved by the Institutional Review Board (IRB) of Seoul National University Hospital (IRB number E-2009-117-1159) and the Ethics Committee of KoGES of the Korea National Institute of Health (IRB number 2014-08-02-3C-A).

Results

During an average follow-up period of 8.6 (±2.1) years for all participants, 462 (1.1%) men and 385 (0.5%) women were identified as having GC.

Table 1 shows the baseline characteristics of the HEXA-G study population included in the analysis. The proportion of current drinkers was 78.7% for men and 31.7% for women, respectively. Among men, the current drinkers were younger, had attained higher educational status, were more likely to have ever smoked, were physically inactive, and had higher BMI and total energy intake than non-drinkers. Women who were current drinkers were younger, had attained higher educational status, were single, were physically inactive, and had higher BMI and total energy intake than non-drinkers.

Table 2 shows the sex-specific hazard ratios and 95% CIs for GC according to drinking status, and the duration, frequency, and amount of weekly intake of ethanol. Among men, current drinkers had a 31% higher risk of GC than non-drinkers (HR 1.31, 95% CI 1.03–1.66), but no association was observed in women (HR 1.00, 95% CI 0.79–1.26). No significant association between the duration of alcohol drinking and GC risk was observed in either men or women. Compared with not drinking, a higher frequency of alcohol drinking was significantly associated with GC risk (P trend 0.02). In fact, men in the highest drinking frequency category (≥5 times/week) had a 62% greater risk of GC than non-drinkers (HR 1.62, 95% CI 1.17–2.25). In terms of intake of weekly ethanol in grams, GC risk showed a significant dose-response relationship among men, in drinkers compared with non-drinkers (<10 g/week: HR 1.25, 95% CI 0.96–1.63; 10–30 g/week: HR 1.37, 95% CI 1.03–1.83; >30 g/week: HR 1.47, 95% CI 1.08–1.99; P trend 0.03). No factors relating to alcohol intake, or the frequency and amount of ethanol consumed, were significantly associated with GC risk in women.

Table 3 presents results of the association of both the weekly frequency and daily amount of pure alcohol consumed with the development of GC in men. We combined alcohol frequency and amount to comprehensively assess weekly alcohol drinking patterns among men and obtain greater insight. Compared with the “infrequent-light” group [those who drank less frequently (<5 times/week) and consumed less than 40 g of ethanol per day], the “infrequent-heavy” group [those who drank equally less frequently but had higher daily ethanol intake (>40 g)] and the “frequent-heavy” group [those who drank more often (≥5 times/week) with higher daily ethanol intake (40 g)] did not show significantly higher risk of GC. However, for the “frequent-light” group [those who consumed less than 40 g of ethanol a day but drank more often (≥5 times/week)], the risk of GC incidence exceeded 45% (HR 1.45, 95% CI 1.02–2.07). Among non-drinkers, the risk of GC was statistically significantly lower than that of the reference group (HR 0.78, 95% CI 0.61–0.99).

In a supplementary analysis on men, we investigated the association between GC risk and different alcoholic beverages, beer and soju, and found that soju consumption was significantly correlated with a higher weekly frequency and daily amount of ethanol intake (P trend 0.03 for weekly frequency, P trend 0.04 for daily ethanol intake in grams) (Supplementary Table S1). Other beverage types were consumed in amounts too small to be analyzed as separate drink types and therefore were excluded. In analysis of anatomical subsites, no significant association was observed between gastric cardia cancer and alcohol intake in men; in women, the numbers were too small to be analyzed (Supplementary Table S2). For non-cardia gastric cancer, a significant direct relationship was observed between GC risk and the weekly frequency and weekly amount of ethanol in men (P trend 0.003 for weekly frequency, P trend 0.006 for weekly ethanol intake in grams) (Supplementary Table S3).

Discussion

In this prospective study of Korean men and women, alcohol consumption factors were significantly associated with the risk of GC in men, which showed a linear increase with the weekly frequency and weekly amount of ethanol. However, among women, none of the alcohol-related factors showed significant relationships with GC incidence. In an analysis including former drinkers, we observed similar findings, in which alcohol-related factors were significantly associated with GC only in men (Supplementary Table S4). Moreover, the significant association was observed for the total population, when both sexes were combined (Supplementary Table S5). The results in the total population might have been due to the high number of GC cases in men. When the 2 factors were combined, a “frequent-light” drinking pattern, defined as frequent intake of low dose of daily alcohol, was prominently associated with GC risk.

We observed that a combined drinking pattern comprising consumption frequency and amount was an important indicator of GC risk. Additional analysis considered the effect of drink duration with combined drink patterns in men by stratification of alcohol intake duration (<21 years and ≥21 years). In contrast to the frequency and amount, the duration of alcohol consumption was not observed to be a significant factor of GC risk in men (Supplementary Table S6). These findings further supported our results, highlighting the importance of understanding GC risk factors through a descriptive approach considering behavioral dimensions of alcohol use. Recent studies have provided evidence that light consumption at ≤1 drink a day (12.5 g ethanol/day) is associated with adverse effects^27,28, and other studies have report that the risk of cancer due to high drinking frequency is similar to that of heavy drinking¹². Although these findings are similar to our results, they provide evidence for cancers other than GC. In terms of drinking patterns, only one study’s results are concordant with our findings²⁹. This latest Korean research further indicates that the frequency of alcohol intake, rather than quantity, is a more important risk factor for gastrointestinal (GI) cancers including GC²⁹. Nonetheless, further research is warranted to thoroughly explore and evaluate alcohol drinking patterns and GC risk.

Our results regarding frequency and quantity as independent factors are consistent with previous findings^30–35. A cohort study of Korean men has reported a 20% increased risk of GC associated with alcohol consumption of ≥25 g per day (aRR 1.20, 95% CI 1.10–1.40)³¹. In accordance with our study results, the association was significant and dose dependent (P trend 0.0001). A Lithuanian study has reported that after 30 years of follow-up, both frequency and quantity of alcohol intake were associated with GC risk in men. Moreover, this prospective study found a 100% increase in GC risk with the highest drinking frequency of 2–7 times per week (HR 2.00, 95% CI 1.04–3.82) and a 90% greater risk of GC in association with a weekly intake amount of ≥100 g ethanol (HR 1.90, 95% CI 1.13–3.18), as compared with occasional drinking of 0.1–9.9 g ethanol per week³². In the recently updated World Cancer Research Fund report, when stratified by sex, outcome, and geographical region, the meta-analysis showed a significantly elevated risk of GC in men and in Asian cohorts¹⁰. The lack of association in women might have been due to the low number of current drinkers. Considering the nature of the survey method used in the baseline study, misclassification of drinking status among women might have been a possibility, but the prevalence of alcohol consumption among women in our study (31.7%) was concordant with those in the KNHANES, which ranged from 28.4% to 34.2% between 2005 and 2013^19,26,36. Moreover, several previous studies have shown that alcohol consumption increases the serum levels of female sex hormones, which may protect against GC^37,38.

Regarding the null association observed among “frequent-heavy” drinkers, we reflect on some of the limitations offered in alcohol-related studies. According to some epidemiologists’ studies on drinking, types of selection bias that can occur in alcohol-related research include sick quitter bias, healthy survivor bias, cohort age, and poor baseline health status³⁹. Alcohol-related risk is relatively higher among people in the younger age range of 20–49 years⁴⁰; thus, the risk relationship is minimized if studies include older participants rather than drinkers of all ages⁴¹. In fact, our study participants’ ages ranged from 40 to 69 years at enrollment; according to prior studies on alcohol, this characteristic increases the likelihood of drinkers becoming former drinkers during follow up⁴². Moreover, Naimi et al.⁴⁰ have argued that the age of the cohort might influence the number of events, such as deaths attributable to alcohol intake. Over-representation of older age groups among cohort studies has been reported to lead to underestimation of drinkers relative to non-drinkers³⁹. In many developed countries, including the United States, approximately 90% of drinkers have already begun drinking at the age of 21⁴³. In Korea, the average initiation age of drinking is 21.3 years, an age much younger than the common age of cohort enrollment; moreover, the age group of 30–49 years has been observed to consume the largest amount of alcohol⁴⁴. Therefore, the average age of many cohort participants might not be representative of that of drinkers. Despite these potential biases, we conducted an analysis that combined the frequency and amount of alcohol consumption to better elucidate which alcohol-related factors are closely correlated with GC risk. Considering the potential biases that may arise from alcohol-related studies, caution is necessary in interpreting our results.

Several mechanisms may provide understanding of the association between alcohol and GC. A potential mechanism of increased GC risk associated with recurrent or chronic exposure to alcohol by frequent drinking may involve the combined actions of ethanol, acetaldehyde, and nitrosamines in the gastric mucosa. Notably, alcohol has been shown to permanently damage DNA strands in cells and to inhibit DNA repair processes, particularly through acetaldehyde, the immediate product of alcohol metabolism⁴⁵. Acetaldehyde, a carcinogenic metabolite of alcohol, is associated with upper GI tract cancer via direct mucosal damage⁹. The constant exposure to acetaldehyde in saliva and gastric juice may cumulatively impair the GI tract, thus depriving the body of the time to heal⁹.

Alcohol consumption can also decrease folate levels, inhibit key enzymes in one-carbon metabolism, and hinder the activity and expression of DNA methyltransferases, all of which lead to aberrant patterns of DNA methylation⁴⁶. Both chronic and heavy use of alcohol induces the expression of cytochrome P450 2E1 (CYP2E1) in the human liver and rat gastrointestinal mucosa^47,48. Thus, alcohol-induced CYP2E1 could contribute to the formation of reactive oxygen species in the gastrointestinal tract and to the activation of procarcinogens, such as nitrosamines, that may be present in alcoholic beverages⁴⁷. The volatile N-nitroso compound N-nitrosodimethylamine, found in beer, whiskey, and other hard liquors, has been suggested to be a potent carcinogen in animals⁴⁹.

In a less direct manner, alcohol also acts as a solvent enhancing the penetration of carcinogens into cells⁵⁰. Furthermore, alcohol may interfere with retinoid metabolism, thus adversely affecting cellular growth, cellular differentiation, and apoptosis⁴⁹. Regular use of alcohol may also lead to deficiencies in essential nutrients that affect DNA processing pathways and susceptibility to carcinogenesis^45,51.

In addition, some genetic variations are associated with an increased risk of cancer development due to alcohol consumption^8,27. ALDH2*2, a variant associated with acetaldehyde oxidizing enzymes, which is prevalent among Asians, promotes acetaldehyde accumulation in peripheral blood after the intake of even small amounts of alcohol, thereby causing mutagenic effects²⁷. Studies have reported that the overall risk of upper GI tract cancer is elevated among East Asians, among whom aldehyde dehydrogenase deficiency is common^52,53. Meta-analyses of alcohol-related cancers have reported higher relative risk (RR) in Asian than Western populations for similar doses of alcohol intake^10,54. With differences in gene status and the incidence of GC, caution is necessary in assessing the generalizability of our findings to a non-Korean population.

Some inevitable limitations in our study should be noted. First, alcohol intake information was based on self-reporting during the baseline survey, thus potentially allowing for misclassification, underreporting, and omission of information regarding changes regarding alcohol consumption habits during follow-up²². However, the nature of our prospective cohort study minimized the possibility of recall bias. Hence, some measurement errors that persisted might have been nondifferential. Moreover, the questionnaires used in the HEXA survey, including that for alcohol consumption, are sufficiently detailed to provide information from each participant by exploring various aspects such as the frequency, amount, duration, and type of alcoholic beverages. By including the differing alcohol content of various alcoholic beverages in our calculations, we were able to analyze more comprehensive data regarding the quantity of alcohol intake. These questions were derived from validated surveys and standards used in national and organizational sectors, such as the KNHANES as well as the World Health Organization^17,18. Second, our data lack information on H. pylori infection, a strong risk factor for GC^23–24. Approximately one-third of adults are infected with H. pylori worldwide⁵⁵; in Korea, the infection prevalence has been estimated to be approximately 44% in the adult population⁵⁶. Despite this high prevalence, the relationship between H. pylori infection and GC in Korea lacks support from epidemiological data. A nested case-control study in Korea has found no direct association between H. pylori infection GC⁵⁷. Moreover, a Chinese study has reported that the adjustment of H. pylori does not substantially change the association between alcohol consumption and GC risk³⁵. In fact, several studies have reported that alcohol intake is independently associated with GC^58–61. Results regarding the prevention of GC after eradication of the infection remain ambiguous⁶². Third, although we attempted to control for important confounders, we cannot rule out the potential effects of residual confounders, including dietary factors such as salty food, tea, and garlic consumption, which could not be accounted for. However, we attempted to control for important confounders by taking potential confounding factors into account in the main analyses. Moreover, we investigated tea drinkers in a supplementary analysis. Because consumption of only green tea was available in our data, we conducted a subgroup analysis in men and women who consumed green tea and found that the results were in line with those of our main analysis (Supplementary Table S7). Regarding salty food intake, a recent systematic review and meta-analysis of prospective cohort studies has indicated that the risk of GC is increased by 12% per 5 g/day increase in dietary salt intake or 5% per 10 g/day increase in alcohol consumption, and the 2 factors are independent of each other⁶³. Moreover, a Korean prospective study has suggested that dietary salt intake has an independent effect on the occurrence of GC⁶⁴. Finally, because only 40 cases of gastric cardia cancer were identified in our study, our results could not provide sufficient evidence to understand the relationships specific to anatomical subsites.

Despite these limitations, the strength of our study lies in that it is one of the largest community-based prospective cohorts in Korea, which explored crucial factors involved in alcohol drinking among the middle-aged and older population. In addition, this study reflects sex differences in epidemiologic risk factors, because we performed separate analyses for men and women. We also excluded GC cases that occurred within the first 2 years after the baseline survey, to prevent reverse causation, thus resulting in a more conservative association between alcohol consumption and GC. The inclusion of various factors potentially associated with alcohol-related behaviors in this large-scale cohort analysis enabled us to identify behavioral patterns whose associations with gastric cancer requires require further exploration. Our findings may help provide a context for public health recommendations and programs aimed at not only high-risk drinkers but also at the entire drinking population, to effectively decrease the health burden associated with alcohol drinking in non-Western regions such as Korea.

In conclusion, higher alcohol consumption was significantly associated with GC among Korean men. In terms of drinking patterns, “frequent-light” drinkers were at risk of GC: frequent intake of alcohol, even in low quantities per drinking session, was a significant risk factor for GC. Further research on gene and environmental interactions among different ethnicities is warranted to provide a more detailed evaluation of alcohol drinking and GC risk.

Grant support

This work was supported by the Research Program funded by the Korea Centers for Disease Control and Prevention (Grant Nos. 2004-E71004-00; 2005-E71011-00; 2005-E71009-00; 2006-E71001-00; 2006-E71004-00; 2006-E71010-00; 2006-E71003-00; 2007-E71004-00; 2007-E71006-00; 2008-E7100600; 2008-E71008-00; 2009-E71009-00; 2010-E71006-00; 2011E71006-00; 2012-E71001-00; and 2013-E71009-00). The funding source had roles in study design and data collection.

Author contributions

H.-W.L. and D.K. conceived and designed the study. H.-W.L. and D.H. performed the data analysis and drafted the first manuscript. A.S., J.-k.L., and D.K. collected the data and obtained study funding. D.H., W.-K.S., M.S., K.d.l.T., A.S., J.-k.L., and D.K. provided critical review of the manuscript for important intellectual content.