Carbonated Soft Drinks and Risk of Esophageal Adenocarcinoma: A Population-Based Case–Control Study

Abstract

The increased intake of carbonated soft drinks parallels the incidence of esophageal adenocarcinoma. To determine whether an association exists between carbonated drink intake and esophageal and cardia adenocarcinoma, we analyzed data from a Swedish nationwide, population-based, case–control study. During data collection in 1995–1997, 189 patients with esophageal adenocarcinoma (88% of all eligible), 262 patients with cardia adenocarcinoma (84%), and 820 control subjects (73%) were interviewed in person. All cancers were histologically classified. We calculated odds ratios with 95% confidence intervals using conditional logistic regression and multivariable analyses. Frequency of intake of carbonated soft drinks was not associated with risk of esophageal adenocarcinoma; high consumers (intake more than six times weekly) were at a statistically nonsignificantly decreased risk compared with never users (odds ratio = 0.89, 95% confidence interval = 0.49 to 1.64). Consumption of carbonated low-alcohol beer and combined intake of carbonated drinks were not associated with risk of esophageal adenocarcinoma. No association between intake of carbonated soft drinks or low-alcohol beer and risk of cardia adenocarcinoma was observed.

Results from an ecologic study in the United States ( 1 ) suggest that intake of carbonated soft drinks might be an exposure that explains the increasing incidence of esophageal adenocarcinoma. The greatly increased consumption of carbonated soft drinks by children during the past few decades parallels the distribution of esophageal adenocarcinoma with biologically plausible 20-year latency. Other patterns of the use of carbonated soft drinks also coincide well with the incidence pattern of esophageal adenocarcinoma, i.e., the rapidity of the increase, the male predominance, and the higher rates among whites compared with blacks ( 2 ) . Because carbonated soft drinks can stimulate gastric acid secretion, distend the stomach ( 3 ) , and trigger symptomatic reflux ( 4 – 6 ) , it was suggested that these drinks might act through an increased occurrence of gastroesophageal reflux, an established risk factor for esophageal adenocarcinoma ( 7 , 8 ) . Another potential mechanism might be that the high-caloric contents of carbonated soft drinks could contribute to the occurrence of obesity, the other main established risk factor for this tumor ( 9 , 10 ) . We also investigated exposure to low-alcohol beer, a carbonated beverage commonly used in Sweden. Patients with cardia adenocarcinoma were included because this cancer has an epidemiologic pattern similar to that of esophageal adenocarcinoma ( 2 , 7 – 12 ) .

The relationship between carbonated beverages and risk of esophageal and cardia adenocarcinoma was tested in a Swedish nationwide population-based case–control study, described in detail elsewhere ( 7 ) . In brief, the study base included all Swedish residents who were less than 80 years of age during 1995–1997, and all esophageal or cardia adenocarcinoma patients in the study base were eligible. All hospital departments that were involved in the diagnosis or treatment of these patients in Sweden participated. Tumors were uniformly classified according to site and histologic type ( 7 ) , and all histologic slides were reviewed by one pathologist (Dr A. Lindgren). Control subjects were selected randomly from the total population register and were frequency matched by age and sex. All participants underwent a computer-aided face-to-face interview. The interview included a validated food frequency questionnaire ( 13 ) that surveyed intake of beverages and dietary patterns 20 years earlier (“How often did you on average drink carbonated soft drinks/carbonated low-alcohol beer 20 years ago?”). Other questions addressed variables that had previously been identified as risk factors for esophageal or cardia cancer, i.e., gastroesophageal reflux symptoms ( 7 ) , body mass ( 9 ) , tobacco ( 11 ) , alcohol (excluding carbonated low-alcohol beer) ( 11 ) , socioeconomic status ( 14 ) , and intake of fruits and vegetables ( 15 ) . The questions were asked with reference to 20 years before the interview, except for gastroesophageal reflux and smoking, which were asked with reference to lifetime exposure 5 and 2 years before the interview, respectively. We constructed two exposure models based on questions regarding intake of 1) carbonated soft drinks and 2) carbonated low-alcohol beer. Frequency of intake was categorized according to evenly distributed tertiles among all exposed. Combined exposure to carbonated soft drinks and beer was categorized into evenly distributed quartiles among all study participants. Relative risks were estimated from odds ratios with 95% confidence intervals that were calculated using conditional logistic regression (SAS PHREG procedure, SAS Institute Inc, Cary, NC). P values (two-sided) were calculated using the Wald test, and P <.05 was considered statistically significant. Regression models were conditional on the frequency-matched variables age and sex. Potential confounding by covariates (categorical) was adjusted for in multivariable analyses. Because the hypothesis was based on gastroesophageal reflux (yes or no) or high body mass [body mass index > 30 kg/m ² , defined as obesity according to World Health Organization ( 16 ) ] lying in the causal chain, the main model excluded adjustment for these factors, but we also compared relative risk estimates from models excluding and including each of these variables. Informed consent was obtained from all study participants, and all regional ethics committees in Sweden approved the study.

A total of 189 patients with esophageal adenocarcinoma (88% of all eligible) and 262 with cardia adenocarcinoma (84%) were compared with 820 control subjects (73%). Gastroesophageal reflux symptoms and high body mass were most common among patients with esophageal adenocarcinoma, followed by cardia cancer patients, and finally control subjects ( Table 1 ). Smoking was more common among patients with cardia adenocarcinoma, followed by esophageal adenocarcinoma patients and control subjects, whereas no clear differences between the groups were observed regarding alcohol consumption ( Table 1 ). A low socioeconomic status (manual workers) was more common among patients with esophageal adenocarcinoma than the other groups ( Table 1 ). Users of carbonated soft drinks were not at increased risk of esophageal adenocarcinoma compared with never users, irrespective of the frequency of use ( P = .77; Table 2 ). Among persons who drank carbonated soft drinks more than six times per week, the odds ratio was statistically nonsignificantly decreased (multivariable odds ratio = 0.89, 95% confidence interval = 0.49 to 1.64) compared with never users. Similarly, consumption of carbonated low-alcohol beer did not increase the risk of esophageal adenocarcinoma, irrespective of the frequency of intake ( P = .78). Combined intake of the studied carbonated drinks was not associated with risk of esophageal adenocarcinoma ( P = .48). The risk of cardia adenocarcinoma did not increase among users of carbonated soft drinks, irrespective of the frequency of consumption ( P = .64), and no association was found between the use of carbonated low-alcohol beer and risk of this cancer ( P = .36). Similarly, the use of carbonated beverages combined was not associated with risk of cardia adenocarcinoma ( P = .88). The risk estimates for esophageal and cardia adenocarcinoma were similar across frequencies of intake in multivariable models both including and excluding gastroesophageal reflux symptoms and body mass ( Table 2 ).

The hypothesis that the increased intake of carbonated soft drinks seen in western societies might explain the rising incidence of esophageal adenocarcinoma was suggested in a previous study ( 1 ) . The ecologic design of that study meant that only formulation of hypotheses was possible ( 1 ) . In our study, in contrast, we were able to evaluate the risk in individuals. The fact that risk estimates did not change after adjustment for gastroesophageal reflux or obesity, the suggested mechanisms, provides further evidence against the hypothesis. Moreover, a recently published case–control study with a similar study design as in this study found rather an inverse association between use of carbonated soft drinks and risk of esophageal adenocarcinoma, an unexpected finding that was possibly due to chance or information bias ( 17 ) .

The study has several potential limitations. Nondifferential misclassification of the exposure is a potential problem because we assessed dietary habits and the use of particular beverages 20 years previously. In addition, the limited statistical power in some of our analyses suggests influence of chance error.

The study also has several strengths. The study incorporated a large sample size, had a strict population-based design, and had high participation rates. Other advantages include the information on all known potential confounding factors and the thorough tumor classification. Regarding the potential limitation of recall bias, a validation study found that collection of information on dietary habits 20 years previously is well suited for case–control studies ( 13 ) . Furthermore, our face-to-face interviews facilitated valid exposure assessment. The risk of recall bias was alleviated by the fact that the hypothesis that carbonated drinks potentially affect the risk of these tumors was not known to the study participants.

Similar patterns of gastroesophageal adenocarcinoma incidence over time ( 2 , 18 ) and risk factor profiles ( 7 – 12 ) , as well as the similar dietary habits, including the consumption of carbonated soft drinks, have been reported between the populations in Sweden and other western societies. These similarities indicate that our results might be relevant for other western populations.

In conclusion, no association between consumption of carbonated soft drinks and risk of esophageal adenocarcinoma was found in this population-based study. This study gives no support for the hypothesis that the use of carbonated soft drinks contributes to the increasing incidence of this cancer.

References