Esophageal cancer in Central and Eastern Europe: Tobacco and alcohol



Esophageal cancer mortality rates in Central and Eastern Europe have been increasing steadily and are expected to increase further in the future. To evaluate the role of risk factors for esophageal cancer in this population, a multicenter study was conducted, with investigation of tobacco and alcohol as one of the principal aims. We have included 192 squamous cell carcinoma (SCC) and 35 adenocarcinoma cases of the esophagus diagnosed at designated hospitals in 5 centers from Romania, Russia, the Czech Republic and Poland. Controls were frequency matched from patients in the same hospital as the cases (n = 1,114). Our results showed that the risk of esophageal SCC may be increased by approximately 7-fold for current smokers (OR = 7.41, 95% CI 3.98–13.79) and by 3-fold for ever alcohol drinkers (OR = 2.86, 95% CI 1.06–7.74). Dose-response relations were evident for both the frequency and duration of tobacco and of alcohol on the risk of esophageal SCC. Risk estimates for tobacco smoking were highest for lower esophageal SCCs, while risk estimates for alcohol drinking were highest for upper esophageal SCCs; though differences were not statistically significant. For adenocarcinoma of the esophagus, our results suggested a more modest increase in risk because of tobacco smoking than that for SCC of the esophagus and no association with alcohol consumption, although our sample size was small. A synergistic interaction between tobacco and alcohol was observed for the risk of esophageal SCC, highlighting the importance of both factors for esophageal cancers in Central and Eastern Europe. © 2006 Wiley-Liss, Inc.

Esophageal cancer mortality rates in Central and Eastern Europe have been steadily increasing and are expected to further increase in the future,1 particularly among men in Hungary. The age-standardized incidence rates for men in 2002 were 5.3/100,000 in the Czech Republic, 6.0/100,000 in Poland, 7.0/100,000 in Russia and 9.8/100,000 in Hungary. For Central and Eastern European women, the age-standardized incidence rates in 2002 were low (<1.5/100,000).

While squamous cell carcinoma (SCC) is the dominant histological type of esophageal cancers in most parts of Europe, the incidence of esophageal adenocarcinoma is also increasing.2 In Central and Eastern Europe, the increase in esophageal adenocarcinomas appears to be less prominent.2 In the United States, the age-adjusted incidence rate of adenocarcinoma of the esophagus surpassed that of esophageal SCC in 1990.3

Tobacco smoking and alcohol drinking are strongly associated with the risk of esophageal SCC and to a lesser degree with the risk of esophageal adenocarcinoma.4, 5 The average relative risk of tobacco smoking is 2.0–5.0 for esophageal SCC and 1.5–2.5 for esophageal adenocarcinoma.6 Considering the increasing esophageal cancer mortality rates in Central and Eastern Europe and the differences in trend by subsite, there is a need to clarify the contribution of different risk factors. We have conducted a multicenter study in Central and Eastern Europe with investigation of tobacco and alcohol as one of the principal aims. In this report, we explore differences by histology and subsite for tobacco smoking and alcohol consumption.


A multicenter case–control study was conducted in the following centers from Central and Eastern Europe: Bucharest (Romania), Lodz (Poland), Moscow (Russia), Olomouc and Prague (Czech Republic). The overall recruitment period was from 2000 to 2002, with response rates of 96% for cases and 97% for controls. Of the 245 incident esophageal cancer cases ascertained (ICD-O-2 C15), 18 cases were excluded because the histological type was either missing, was not squamous cell carcinoma (SCC) or adenocarcinoma, or was an in situ carcinoma. Thus, we have included 192 SCC cases and 35 adenocarcinomas of the esophagus diagnosed at designated hospitals or cancer clinics, confirmed histologically or cytologically and recruited into the study within 3 months of diagnosis.

Controls were frequency matched from in-patients or out-patients in the same hospital as the cases. Only controls with a recent diagnosis from a defined list of diseases unrelated to tobacco and alcohol were included. In Moscow, the controls were frequency matched to the esophageal cancer cases by age, sex, center and referral or residence area. In the other centers, controls overlapped with those for a case–control study of lung cancer conducted according to an identical protocol.7, 8 Since the lung cancer study was conducted earlier than the esophageal cancer study, we excluded controls interviewed more than 6 months before the beginning of recruitment of esophageal cancer cases. The distribution of control diseases were 10% genitourinary, 10% eye, 9% of the musculosketal system, 8% skin, 8% circulatory system, 8% gallbladder or biliary tract, 8% other digestive, 7% CNS, 5% hernias, 4% injury/poisoning, 2% metabolic, 2% appendicitis, 2% mental, 2% respiratory and 14% other diseases. Written consent for participation was obtained from all study subjects and ethical approval has been obtained for all study centers as well as for IARC. Cases and controls were interviewed face-to-face by the same team of interviewers with a structured questionnaire on residential and lifestyle history.

Statistical analysis

Unconditional logistic regression was employed to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for tobacco and alcohol consumption with the SAS program (version 8.02), for SCC and adenocarcinomas of the esophagus, after adjusting for potential confounders such as age, sex, education, body mass index (BMI), fruit intake, vegetable intake, center, alcohol drinking and tobacco smoking, where appropriate. Analyses were also conducted for SCC of the upper esophagus (C15.0, C15.3; n = 32), middle esophagus (C15.1, C15.4; n = 113) and lower esophagus (C15.2, C15.5; n = 43).

The section of the questionnaire on tobacco smoking was detailed, with information on age at which regular smoking began and stopped, the number and type of cigarettes smoked during that period and any period of not smoking. Overall tobacco use was estimated by combining the amount of cigarettes, cigars/cigarillos and pipes, and quantified into the equivalent of cigarettes per day. Ever smokers were defined as having smoked at least 100 cigarettes in an entire lifetime, or having regularly smoked cigars/cigarillos or pipes. Ex-smokers were defined as having quit smoking for at least 2 years prior to the interview.

Ever-drinkers were defined as having ever consuming alcoholic drinks. When the numbers of never drinkers among cases were small, we used light drinkers as the reference group. Each subject was asked to recount their drinking of beer, wine and spirits in a typical week during specific age periods (at age 25, 40, 50 and 60). The grams of ethanol per week was subsequently estimated, assuming that beer contains ∼5% ethanol, wine contains 12% ethanol and spirits contain 40% ethanol in volume.9 Cumulative consumption was estimated by multiplying grams per week with the years of alcohol consumption.

BMI was calculated from the subject's self-reported height and weight, the latter 2 years prior to the interview. Fruit and vegetable intake were estimated from a 20-item food frequency section of the questionnaire.

Interactions between tobacco and alcohol were assessed by comparing the fit of a regression model including terms for consumption of tobacco alone, alcohol alone, and both products, to that of a model including also an interaction term between the two habits. Duration of tobacco and alcohol habits were also assessed for interaction effects.


Selected characteristics of the cases and controls are presented in Table I. The largest proportion of SCC cases was in the middle third and thoracic sites, whereas the largest proportion of adenocarcinomas was in the lower third. The ratio of men to women was ∼8:1 for both SCCs and adenocarcinomas. The majority of SCC cases were of average BMI, while the majority of adenocarcinoma cases were preobese.

Table I. Characteristics of Esophageal Cases and Controls from the Central and Eastern Europe Multicenter Study
 SCC (n = 192)Adenocarcinoma (n = 35)Controls (n = 1114)
  • 1

    Estimated from weight 2 years before date of interview.

 Olomouc, Czech Republic5528.72160.020018.0
 Moscow, Russia7036.525.738334.4
 Bucharest, Romania4422.9925.721219.0
 Lodz, Poland178.925.720918.8
 Prague, Czech Republic63.112.91109.9
 Cervical (C15.0)115.700.0  
 Thoracic (C15.1)4422.912.9  
 Abdominal (C15.2)10.525.7  
 Upper third (C15.3)2110.900.0  
 Middle third (C15.4)6935.925.7  
 Lower third (C15.5)4221.92160.0  
 Overlapping lesion (C15.8)00.038.6  
 Esophagus, not specified (C15.9)42.161.7  
Age (years)
Education (age at which education finished) (years)
BMI1 (kg/m2)
 <18.5 Underweight105.200.0121.1
 18.5–24.9 Average11559.91131.439435.4
 25.0–29.9 Preobese5026.01851.447542.6
 ≥30.0 Obese168.3617.122320.0
 Missing10.5  100.9

Tobacco smoking was associated with the risk of esophageal SCC, with ORs of 7.41 (95% CI = 3.98–13.79) for current smokers and 2.40 (95% CI=1.18–4.89) for ex-smokers (Table II). Dose-response relations were evident for the frequency, duration and packyears of tobacco smoking with esophageal SCC risk. Longer year since quitting was associated with decreasing risk of esophageal SCC. For adenocarcinoma of the esophagus, a 2-fold increase in risk was suggested for tobacco smoking and a dose-response relationship with frequency and packyears were also apparent. No dose-response with duration of smoking was observed for esophageal adenocarcinoma, or a protective effect with quitting smoking, although numbers were small.

Table II. Tobacco Smoking1 and the Risk of Esophageal SCC in Central and Eastern Europe
 ControlsSCC of the esophagusAdenocarcinoma of the esophagus
CasesOR295% CICasesOR295% CI
  • 1

    Overall tobacco smoking was estimated by combining the amount of smoking cigarettes, cigars/cigarillos and pipes, and quantified into the equivalent of cigarettes per day.

  • 2

    Adjusted for age, sex, education, BMI, fruit intake, vegetable intake, center and alcohol drinking (years & grams ethanol/week).

Tobacco status
 Nonsmoker340151.00 61.00 
 Current smoker4351477.413.98–13.79142.380.80–7.04
Frequency (cigarettes/day)
 Nonsmoker340151.00 61.00 
P for trend   <0.01  0.04
Duration (years)
 Nonsmoker340151.00 61.00 
p for trend   <0.01  0.12
 Nonsmoker340151.00 61.00 
 p for trend   <0.01  0.05
Years since quitting
 Current smoker4331471.00 141.00 
 p for trend   <0.01  0.76
Type of tobacco
 Nonsmoker340151.00 61.00 
 Cigarette only5301375.623.08–10.28242.440.90–6.59
 Cigar only1815.521.71–17.860
 Pipe only200
 Cigarette + cigar109294.772.18–10.4312.010.14–28.73
 Cigarette + pipe1512.500.29–21.65113.971.22–159.49
 Cigar + pipe100
 Cigarette + cigar + pipe200

A history of alcohol drinking was associated with a 3-fold increase in esophageal SCC (Table III). Dose-response relations were observed for frequency, duration and cumulative consumption of alcohol on the risk of esophageal SCC. The majority of cases reported drinking all types of alcoholic beverages, with no consistent pattern for different types of alcoholic beverages. Conversely, adenocarcinoma of the esophagus did not appear to be strongly associated with alcohol consumption.

Table III. Alcohol Consumption and the Risk of Esophageal Cancer in Central and Eastern Europe
CasesOR195% CICasesOR195% CI
  • 1

    Adjusted for center, age, sex, education, BMI, fruit intake, vegetable intake, packyears of tobacco.

  • 2

    Analysis restricted to drinkers only.

 No drinking10151.00 31.00 
 Ever drinking9991812.861.06–7.74321.210.31–4.77
Grams ethanol/week
 No drinking10151.00 31.00 
 p for trend   <0.01  0.20
Years of drinking
No drinking10151.00 31.00 
 p for trend   0.08  0.55
Cumulative consumption
 No drinking10151.00 31.00 
p for trend   <0.01  0.54
Type of alcohol2
 Beer + Wine + Spirits338811.00 141.00 
 Beer + Spirits189370.760.46–1.2760.480.15–1.59
 Wine + Spirits76221.080.60–1.9610.400.05–3.39
 Beer + Wine5730.200.06–0.7230.530.12–2.31
 Spirits only158190.710.39–1.2920.880.16–4.91
 Wine only4140.500.15–1.720
 Beer only53120.870.38–1.9830.740.17–3.29

In the analyses by esophageal SCC subsite (data not shown), current smokers had ORs of 22.56 (95% CI = 2.90–175.37) for the lower esophagus and 8.73 (95% CI = 3.84–19.84) for the middle esophagus. Dose-response trends for packyears of smoking and the risk of lower (p for trend <0.01) and middle esophageal SCCs (p for trend <0.01) were observed. The risk of upper esophageal SCC was not elevated in current or past tobacco smokers, and a dose-response relationship with packyears of smoking was not apparent. For alcohol consumption, ever drinking was not associated with an increased risk of the esophageal SCC subsites, but dose-response relations for cumulative alcohol consumption were identified for lower (p for trend = 0.02), middle (p for trend <0.01) and upper esophageal SCCs (p for trend = 0.01). Individuals who were in the highest category of cumulative consumption had an ORs of 5.62 (95% CI = 1.22–25.95) for upper esophageal SCC, 6.01 (95% CI = 2.40–15.04) for middle esophageal SCC and 2.63 (95% CI = 0.82–8.42) for lower esophageal SCCs.

A strong interaction between tobacco and alcohol was observed for the risk of esophageal SCC, with all of the increased risk being observed among those exposed to both alcohol and tobacco (OR = 6.42, 95% CI = 2.03–20.30) and no apparent increase in risk for those exposed to tobacco only or alcohol only, although CIs were wide (Table IV). The interaction was greater than multiplicative for SCC (p < 0.01), but a similar interaction was not observed for esophageal adenocarcinoma although numbers were small.

Table IV. Tobacco and Alcohol Interactions for Esophageal SCC
CasesOR195% CICasesOR195% CI
  • 1

    Adjusted for age, sex, center, education, BMI, fruit intake, vegetable intake

NoNo7541.00 11.00 
Interaction OR   9.410.88–100.27 0.630.04–9.54
χ2 for likelihood test    7.45 0.11 
p-value    <0.01 0.74 
d.f.    1 1 


An important finding in our study is the strong interaction between tobacco and alcohol on the risk of esophageal SCC. The IARC monographs5 reviewed 7 case–control studies10, 11, 12, 13, 14, 15, 16 and 1 cohort study17 that had examined interactions between tobacco and alcohol for esophageal cancer and concluded that there were indications of an interaction on the additive scale, but interactions on the multiplicative scale could not be adequately assessed. A few studies that explored tobacco and alcohol interactions on the risk of esophageal cancer were published since the monograph review. Results from a study in Taiwan18 support a more than multiplicative interaction whereas results from a study in India does not support an interaction between tobacco and alcohol on the risk of esophageal cancer.19

In our study, tobacco smoking appeared to be more important for the middle and lower esophageal sites. Several studies have explored differences in risk factor patterns by esophageal subsite. For tobacco smoking, investigators from 3 other studies in China,20 Japan21 and India,22 reported no association with the upper esophagus and stronger associations for the middle and lower esophagus. When comparing the middle and lower esophagus sites, higher increases in cancer risk were observed for the middle esophagus in these 3 studies, while we observed a higher increase in cancer risk for the lower esophagus. In the Indian study,22 tobacco chewers had a 6.6-fold increase in cancer risk for the lower esophagus but no association was observed for the upper or middle esophagus. While risk estimates differences were reported, statistically significant differences by subsite were not demonstrated for tobacco smoking.

For alcohol drinking, the results from previous studies on esophageal cancer subsites are equivocal. Nandakumar reported an association with the cancer risk of the middle esophagus, but not for the lower or upper esophagus.22 In contrast, the studies from China20 and Japan21 reported increased risks at all 3 esophageal cancer subsites for alcohol drinkers. Both studies reported a higher increase in risk for SCC of the upper esophagus, but the Japanese study reported the next highest increase in risk for SCC of the lower esophagus whereas the Chinese study reported the next highest increase in risk for SCC of the middle esophagus. In our study, individuals with high cumulative consumption had higher risk estimates of upper esophageal SCC than for middle esophageal SCC, and the risk of lower esophageal SCC was not clearly increased.

The inconsistent risk factor patterns among subsites of esophageal cancer may be due to several differences among the studies. First, the study populations were of different ethnicities, where trends for esophageal cancer are different and risk factor patterns are likely to be different by histology and subsite. Second, most of the studies included all histological types of esophageal cancer while our study separated SCC and adenocarcinoma, and analyzed the subsites only for SCC. Third, lack of statistical power is inherent in subsite analyses and point estimates may be less stable because of different sample sizes by subsite. While the point estimates may be different by subsite, it is difficult to demonstrate a statistically significant difference by subsite for tobacco and alcohol.

Our study is subject to several potential limitations including recall bias and selection bias. Since the controls were selected from patients, their lifestyle patterns may not reflect that of the general population. In our control eligibility, patients who have diseases related to tobacco and alcohol were excluded, to minimize the possibility of tobacco and alcohol habits being overestimated with respect to the distribution of these habits in the base population. We also assured that a specific disease group did not contribute to more than 20% of the entire control group, to minimize the magnitude of potential bias that may rise because of having hospital-based controls with a certain disease. Recall bias is usually a limitation in case–control studies, although using patients as controls may have been beneficial in limiting this bias, since patient controls may put as much effort in recalling their lifestyle habits as cancer cases. The limited sample size may have been the major factor for the apparent lack of effect because of alcohol consumption for adenocarcinoma of the esophagus. Detecting an odds ratio of 2.0 with 1,100 controls, at 80% power and 5% alpha level, requires approximately 175 esophageal adenocarcinoma cases.

The strengths of our study include the detailed assessment of lifetime tobacco use and the relatively large number of cases from a high risk population, which allowed for analysis by subsite within the esophagus. Our study is the first study to assess differences in esophageal cancer subsites for tobacco and alcohol in Central and Eastern Europe. We showed that both tobacco and alcohol, especially when combined, are important risk factors for esophageal SCC in Central and Eastern Europe. Our results emphasize the importance of prevention efforts directed toward tobacco smoking and alcohol drinking in the region.