Wang P-L, Xiao F-T, Gong B-C, Liu F-N. Alcohol drinking and gastric cancer risk: a meta-analysis of observational studies. Oncotarget 2017;8:99013-99023.
Background: Many studies investigated the association between alcohol drinking and gastric cancer risk, but the results were controversial. We performed a meta-analysis of observational studies to explore the association.
Materials and Methods: We searched PubMed to identify the relevant studies that reported the association between alcohol drinking and gastric cancer risk up to December 31, 2016. We pooled relative risks (RRs) in random effects model and performed dose-response analysis to quantify the association. Cochran Q test and I2 analyses were used to evaluate the heterogeneity. Meta-regression, subgroup, sensitivity and publication bias analyses were also performed.
Results: 75 studies were included in our study. The pooled RR of high vs low total alcohol drinking was 1.25 (95% CI, 1.15–1.37, P < 0.001), and a nonlinear association was further observed. Subgroup analysis showed that alcohol drinking significantly associated with the risk of gastric noncardia cancer (RR, 1.19; 95% CI, 1.01–1.40, P = 0.033), but not with the risk of gastric cardia cancer (RR, 1.16; 95% CI, 0.98–1.39, P = 0.087). Notably, the pooled RRs of high vs low analyses were 1.13 (95% CI, 1.03–1.24, P = 0.012) for beer drinking, 1.22 (95% CI, 1.06–1.40, P = 0.005) for liquor drinking, and 0.99 (95% CI, 0.84–1.16, P = .857) for wine drinking.
Conclusions: Our meta-analysis found a nonlinear association between alcohol drinking and gastric cancer risk, and heavy drinking level was strongly related to gastric cancer risk. Beer and liquor had significant positive associations with gastric cancer risk, while wine drinking would not increase gastric cancer risk. These results need to be verified in future research.
While excessive alcohol intake has been regularly found to increase the risk of upper aero-digestive tract (UADT) cancers (mouth, tongue, pharynx, larynx, etc.) results are less clear for gastric cancer. The present study is notable because of several features:
- It is based on a large number of subjects, with data from 75 case-control or cohort studies
- It provides dose-dependent results and outcomes according to type of beverage
- It uses appropriate analytic methods, including restricted spline analyses for non-linear dose-dependent results; it carries out appropriate sensitivity analyses
Specific comments by Forum members: Forum member Ellison stated: “My impressions are quite favorable about this paper. I am not especially interested in the high vs low analyses, as these would be expected to vary according to the range of consumption and the number of heavy drinkers in a population. However, the dose-response results are important. And they show that for overall alcohol, a curvilinear relation with a significant increase in risk per typical drink (12.5 g/alcohol); however, the increase is minimal, only 4% (RR=1.04, CI 1.01, 1.07). In beverage-specific analyses comparing the highest and the lowest categories of intake, for beer, a non-linear relation was found with a significant increase of 7%/drink (CI 1.01, 1.13); for liquor, a linear association, with RR=1.03 (CI 0.98, 109); and for wine, the RR was 0.99 (CI 0.93, 1.06), so no increase in risk of gastric cancer.
“Unfortunately, while the authors in their paper include a figure of the dose-response curves for each beverage using a non-linear approach, they do not provide the specific values in a table or in the text; this makes it difficult to clearly determine threshold effects of beverages. The figures suggest that the risk associated with beer consumption shows a slight decrease for 1 to 2 drinks/day, then an increase in risk; for spirits, no decrease with lighter intake is seen, with a steady increase with greater consumption; for wine, the curve is continuously downward. The authors do state, however, that “light, moderate, and wine drinking would not increase gastric cancer risk,” so it is presumed that significant increased risk was not noted for light drinking of any specific beverage when non-linear effects were evaluated.” Reviewer Djoussé agreed: “It makes no sense to compare high versus low levels when cubic splines show the entire relation.”
Ellison added: “The sensitivity analyses are interesting and not surprising: lower risk for studies with larger numbers of subjects, for studies judged to be of higher quality, and for studies adjusting for confounding by SES & income. Also, the risk of cancer and degree of heterogeneity were lower among subjects in cohort studies than in case-control studies. These results led the authors to state that these sensitivity analyses ‘ . . . indicate that our results may be exaggerated to some degree, but it should be noted that the pooled risk estimates of high-quality studies yielded similar results to the original analysis.’ I think the Discussion provides a very good overview of the authors’ results and limitations of their study and agree with their Implications: ‘Heavy alcohol drinking level, beer and liquor drinking were associated with gastric cancer risk significantly, whereas light, moderate, and wine drinking would not increase gastric cancer risk.’”
Forum member Waterhouse commented: “I find two points to be of particular interest. First, in nearly all instances, the authors compare low vs high consumption, not the typical reference point of zero consumption. They seem to be doing so in order to avoid calling out one major conclusion, mentioned only in the ‘Implications’ section, not the ‘Conclusions’ section, that ‘light, moderate, and wine drinking would not increase gastric cancer risk.’ I note that this key observation about light and moderate drinking is avoided in the abstract, although they do admit there that wine drinking is not associated with increased risk. My second point is that the wine data, while suffering from high variation, shows decreasing risk even at high consumption rates. This observation calls for a mechanistic explanation distinguishing wine from the other beverages.”
Reviewer Skovenborg, as did most other Forum members, considered this to be a well-done meta-analysis. “However, I share the surprise mentioned by Waterhouse that the key observation about light and moderate drinking is avoided in the abstract. Also the calculation of alcohol dose makes me wonder; I quote: ‘For studies that reported dose in volume, the dose was calculated using the standard concentration, that is, 100 ml of alcohol, beer, liquor and wine contains about 80 g, 5 g, 40 g, and 15 g of ethanol respectively.’ The alcohol content in wine ranges from 5.5% to 23% alcohol by volume (ABV) with most wines being in the range of 12-15% alcohol by volume. With 13.5% ABV as a suggested median value the ethanol content of 100 ml wine would be 13.5 ml = 10.7 gram (as ethanol weighs 0.79 gram per ml). The content of 15 g of ethanol in 100 ml wine would be equivalent to the highly unusual content of 18.75% alcohol by volume.”
Reviewer Ellison agreed that the assumption for wine of 15 g/alcohol per 100 ml of fluid would make it easier to show adverse effects from a given reported number of drinks. He added: “The fact that this study showed a decreasing effect on cancer risk for increasing wine consumption is especially interesting. This raises the possibility that the non-alcohol components of wine might have a protective effect against gastric cancer.”
Reviewer de Gaetano noted: “I agree that this is in general a well performed study. In a recent study (Xi et al) it was observed that the dose-effect relationship between alcohol consumption and cardiovascular outcomes would appear more likely as an ‘L curve’ rather than the classical ‘J curve’ reported in the majority of studies on alcohol. In an editorial accompanying that paper (de Gaetano & Costanzo) we suggested that it could be possible that heavy drinkers die of other diseases before dying of CVD or that the amount of polyphenols ingested with heavy alcohol (wine) could be somewhat protective and attenuate the harm of heavy drinking, a concept also expressed by others (Chiva-Blanch et al). The same concept could be applied to cancer protection.”
Mechanisms for different effects of wine on cancer risk: Forum member Mattivi had comments on the potential mechanisms for the different effects noted for wine consumption than for the intake of other beverages: “The different behavior of wine highlighted in this paper is very interesting and . . . not unexpected! Wine, and in particular red wine, which contains typically 2-2.5 g/L of polyphenols, mainly tannins, could behave differently. This pool of phenolics reaches, in their native form, the gastric compartment where it could possibly inhibit the NF-kB pathway, a central player in inflammatory diseases, including gastritis. This is similar to the effects of strawberry tannins and of other berries, as shown in Fumagalli et al. For other classes of tannins, the Rubus ellagitannins, a similar mechanism was evaluated in-vivo by Sangiovanni et al in a rat model of ethanol-induced gastric lesions, and shown to be very effective. In this latter experiment, ellagitannins of blackberries and raspberries decreased Ulcer Index and protected from the ethanol-induced oxidative stress in rats. It is remarkable that such protective effect was obtained at a very reasonable dose of ellagitannins, comparable with the amount consumed with a single portion of berries.
“Sangiovanni et al also report the positive effect of quercetin, another polyphenol widely present in wine, mainly as a mixture of glycosides. It is noteworthy that in a previous study, Alvarez-Suarez et al also found that strawberry extracts inhibited the onset of gastric ulcer in a rat model of ethanol-induced gastritis, using extracts enriched with the anthocyanin fraction, leading to the conclusion that also the fraction of anthocyanins exert a gastro-protective effect. Altogether, it seems that the presence of several classes of polyphenols in wine can counterbalance the ethanol-induced inflammation, via multiple mechanisms.” Forum members Van Velden and Stockley agreed with these potentially beneficial effects of wine on gastric cancer risk. Stated Van Velden, “The inhibition of inflammation by the phenolic components of wine certainly has a role to play in this regard.”
Reviewer Thelle concluded: “When one takes a look at sites discussing gastric cancer, alcohol is not mentioned. The main risk factors are helicobacter pylori, salt and pickled foods, smoking, overweight and so forth. In the total picture of gastric cancer, alcohol seems to play a very minor role.”
Literature cited in Forum review
Alvarez-Suarez JM, Dekanski D, Ristic S, Radonjic NV, Petronijevic ND, et al. Strawberry polyphenols attenuate ethanol-induced gastric lesions in rats by activation of antioxidant enzymes and attenuation of MDA increase. PLoS One 2011;6:e25878.
Chiva-Blanch G, Arranz S, Lamuela-Raventos RM, Estruch R. Effects of wine, alcohol and polyphenols on cardiovascular disease risk factors: evidences from human studies. Alcohol Alcohol 2013;48:270-277. doi: 10.1093/alcalc/agt007.
de Gaetano G, Costanzo S. Alcohol and Health: Praise of the J Curves. J Am Coll Cardiol 2017;70:923-925.
Fumagalli M, Sangiovanni E, Vrhovsek U, Piazza S, Colombo E, Gasperotti M, Mattivi F, De Fabiani E, Dell’Agli M. Strawberry tannins inhibit IL-8 secretion in a cell model of gastric inflammation. In Pharmacological Research, Volume 111, 2016, Pages 703-712, ISSN 1043-6618, https://doi.org/10.1016/j.phrs.2016.07.028.
Sangiovanni E, Vrhovsek U, Rossoni G, Colombo E, Brunelli C, Brembati L, Trivulzio S, Gasperotti M, Mattivi F, Bosisio E, Dell’Agli M. Ellagitannins from rubus berries for the control of gastric inflammation: in vitro and in vivostudies, PLoS One;8:2013;e71762.
Xi B, Veeranki SP, Zhao M, Ma C, Yan Y, Mi J. Relationship of Alcohol Consumption to All-Cause, Cardiovascular, and Cancer-Related Mortality in U.S. Adults. J Am Coll Cardiol 2017;70:913-922.
While heavy alcohol intake has been regularly found to increase the risk of upper aero-digestive tract (UADT) cancers (mouth, tongue, pharynx, larynx, etc.), results are less clear for gastric cancer. The present study is notable because it is based on a large number of subjects, uses appropriate analytic methods, and provides dose-dependent results according to type of beverage.
For total alcohol, the dose-response results in this meta-analysis show a curvilinear relation between alcohol intake and gastric cancer, with a minimal (4%) but significant increase per typical drink (12.5 g/alcohol). However, there were varying responses for different beverages. In beverage-specific analyses, for beer, a non-linear relation was found with a significant increase of 7%/drink (CI 1.01, 1.13); for liquor, there was a linear association, with RR=1.03 (CI 0.98, 109); and for wine, the RR was 0.99 (CI 0.93, 1.06).
While the authors show a figure of the dose-response curves for each beverage using a non-linear approach, they do not provide the specific values in a table or in the text; this makes it difficult to clearly determine threshold effects of beverages. The figures suggest that the risk associated with beer consumption shows a slight decrease for 1 to 2 drinks/day, then an increase; for spirits, no decrease with lighter intake is seen; for wine, the curve is continuously downward. The authors do state that “light, moderate, and wine drinking would not increase gastric cancer risk,” so it is presumed that significant increased risk was not noted for moderate drinking for any specific beverage when non-linear effects were evaluated. Thus, overall, it appears that this large meta-analysis shows no significant increases in risk for light to moderate consumption of an alcoholic beverage, but a slight increase in risk associated with greater intake of beer and spirits. For wine, no increase in the risk of gastric cancer was seen for any level of consumption.
In a number of sensitivity analyses, the authors found tendencies for a lower gastric cancer risk associated with alcohol for cohort versus case-control studies, for studies with larger numbers of subjects, for studies judged to be of higher quality, and for studies adjusting for confounding by SES & income; these types of studies would be expected to yield more precise estimates of effect. These results led the authors to state these sensitivity analyses “ . . . indicate that our results may be exaggerated to some degree, but it should be noted that the pooled risk estimates of high-quality studies yielded similar results to the original analysis.”
Overall, Forum members considered this to be a well-done analysis, and the data support an increase in gastric cancer risk for heavier drinking (of beer or spirits) but no significant effects for light or moderate drinking of any alcoholic beverage. Forum members considered that if the lack of increase in cancer risk from wine consumption seen in this analysis proves to be true, it might relate to the polyphenols in wine that block any adverse effects of alcohol on gastric cancer risk.
* * * * * *
Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine; Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.
Andrew L. Waterhouse, PhD, Department of Viticulture and Enology, University of California, Davis, USA
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Dag S. Thelle, MD, PhD, Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Norway; Section for Epidemiology and Social Medicine, Sahlgrenska Academy, University of Gothenburg, Sweden
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Fulvio Mattivi, MSc, CAFE – Center Agriculture Food Environment, University of Trento, via E. Mach 1, San Michele all’Adige, Italy
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
R. Curtis Ellison, MD, Professor of Medicine, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA
Luc Djoussé, MD, DSc, Dept. of Medicine, Division of Aging, Brigham & Women’s Hospital and Harvard Medical School, Boston, MA, USA
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy