BACKGROUND: Smoking has not been confirmed as a risk factor for cancers of the liver and stomach. The authors examined prospectively the relationship between smoking and these cancers in an endemic region. METHODS: The data used were a cohort study on the relationship between lifestyle and health in the region having the highest liver cancer mortality in Japan. Of the cohort members, 4050 males aged > or =40 years were included in the present analysis with a 9-year mean follow-up. Cox proportional hazards regression was used to estimate relative risks (RR) for cancer of the liver, stomach, smoking-related sites and others, while adjusting for age, residence, and alcohol intake. RESULTS: By the end of the study period, 59 cases of liver cancer and 53 cases of stomach cancer were identified. Current smokers, compared to subjects who had never smoked, had a threefold risk of liver cancer (RR = 3.3; 95% CI: 1.2-9.5) and a twofold risk of stomach cancer (RR = 2.2; 95% CI: 0.8-5.7). Sub-cohort analysis showed that adjustment for history of chronic liver disease did not attenuate the risk of liver cancer. Light/medium smokers had almost the same risk of these cancers as heavy smokers, while they showed a relatively low risk of smoke-related cancers. CONCLUSIONS: The present results indicate that smoking is a risk factor of liver and stomach cancer in a population with a high background risk for these cancers. However, causal inferences should be made cautiously due to a lack of information on known risk factors.
BACKGROUND: Smoking has not been confirmed as a risk factor for cancers of the liver and stomach. The authors examined prospectively the relationship between smoking and these cancers in an endemic region. METHODS: The data used were a cohort study on the relationship between lifestyle and health in the region having the highest liver cancer mortality in Japan. Of the cohort members, 4050 males aged > or =40 years were included in the present analysis with a 9-year mean follow-up. Cox proportional hazards regression was used to estimate relative risks (RR) for cancer of the liver, stomach, smoking-related sites and others, while adjusting for age, residence, and alcohol intake. RESULTS: By the end of the study period, 59 cases of liver cancer and 53 cases of stomach cancer were identified. Current smokers, compared to subjects who had never smoked, had a threefold risk of liver cancer (RR = 3.3; 95% CI: 1.2-9.5) and a twofold risk of stomach cancer (RR = 2.2; 95% CI: 0.8-5.7). Sub-cohort analysis showed that adjustment for history of chronic liver disease did not attenuate the risk of liver cancer. Light/medium smokers had almost the same risk of these cancers as heavy smokers, while they showed a relatively low risk of smoke-related cancers. CONCLUSIONS: The present results indicate that smoking is a risk factor of liver and stomach cancer in a population with a high background risk for these cancers. However, causal inferences should be made cautiously due to a lack of information on known risk factors.