Abstract A99: A prospective cohort study on the associations between alcohol consumption and smoking and risk of subtypes of esophageal and stomach cancer

Previous studies have suggested that the etiology of histological subtypes of esophageal cancer and of topographical subtypes of stomach cancer may differ. It was found that risk of esophageal squamous cell carcinoma (ESCC) was strongly associated with alcohol consumption and tobacco smoking. However, esophageal adenocarcinoma (EAC) and gastric cardia adenocarcinoma (GCA) had no association with alcohol and a much weaker association with smoking. Because most studies were cases control studies, and only few cohort studies have been performed, the aim of our study is to investigate these associations within a large-scale prospective cohort study.

The Netherlands Cohort Study was initiated in 1986, when participants (58,279 men and 62,573 women) completed a self-administered questionnaire on risk factors for cancer, including alcohol consumption and smoking habits. Follow-up for incident cancer cases was established by record linkages to the pathology registry and cancer registries. After 16.3 years, we identified 120 ESCC cases, 168 EAC cases and 187 GCA cases. For reasons of efficiency, we used a case-cohort approach, for which a subcohort (n = 4,438) was randomly sampled from the cohort at baseline. Multivariable Cox proportional hazards models estimated incidence rate ratios (RR) and corresponding 95% confidence intervals (95% CI). We corrected for age, sex, alcohol consumption and smoking status, frequency and duration. The following confounders were added if they changed the RR by >5%: fruit consumption, vegetable consumption, energy intake, non-occupational physical activity, BMI, and socioeconomic status.

The multivariable RRs associated with alcohol consumption were 1.30 per 10 g ethanol/day increment (95% CI 1.20-1.42, p<0.001) for ESCC, 1.00 (95% CI 0.90-1.12, p=0.98) for EAC and 0.98 (95% CI 0.89-1.08, p=0.74) for GCA.

The RRs (95% CI) of ESCC for former smokers and current smokers were 1.39 (0.75-2.59) and 2.65 (1.49-4.70), respectively (p-trend<0.001), when compared with never smokers. For EAC, these figures were 1.38 (0.86-2.20) and 1.57 (0.95-2.58), respectively (p-trend=0.09) and the RRs for GCA were 1.35 (0.83-2.19) and 1.56 (0.94-2.56), respectively (p-trend=0.09). When we made further adjustments for frequency and duration of smoking in an analysis excluding never smokers, the differences between current and former smokers became smaller for all three tumors. The frequency of smoking was statistically significantly associated with ESCC and EAC, independently of smoking status and duration. The RRs per 10 cigarettes/day increment were: 1.19 (95% CI 1.01-1.40, p=0.04) for ESCC and 1.23 (95% CI 1.06-1.42, p=0.01) for EAC. No statistically significant association was found with GCA (RR=1.07, 95% CI 0.91-1.25, p=0.40). Smoking duration was not statistically significantly associated with ESCC (RR=1.10, 95% CI 0.93-1.31, p=0.25), EAC (RR=1.03, 95% CI 0.90-1.19, p=0.65) or GCA (RR=1.14, 95% CI 0.98-1.33, p=0.09), after adjustment for smoking status and frequency.

The results of this cohort study suggest that alcohol and cigarette smoking are differently associated with ESCC than with EAC and GCA.

Citation Information: Cancer Prev Res 2008;1(7 Suppl):A99.