A cohort study on the possible association between broiled fish intake and cancer.

To examine the possible carcinogenic effects of frequent intake of mutagenic pyrolysates of proteins and amino acids, 7,553 adult subjects whose personal characteristics had been examined were followed-up for about 11 years. The stepwise multiple regression analysis revealed that frequent consumption of broiled fish was significantly positively associated with mortality from cancer at all sites, as well as from cancer of the stomach. Among the 11 independent variables tested, consumption of broiled fish ranked high as a variable associated with cancer mortality (all sites, and stomach). Similar analyses done by sex and city (Hiroshima and Nagasaki) showed a consistently positive association of this variable with mortality from cancer at all sites, in both sexes and cities, but, with a few exceptions, the association was not statistically significant. Broiled fish consumption also showed a consistently positive but not statistically significant association with mortality from gastric cancer classified by sex and city. The relative risks of cancer mortality associated with frequent (twice or more weekly) consumption of broiled fish, as compared to less frequent consumption, were 1.3 (P less than 0.05) for cancer at all sites and 1.7 (P less than 0.05) for gastric cancer. The frequency of consumption of dried fish was hardly associated with deaths from cancer at all sites and from gastric cancer, but was significantly positively associated with deaths from cancer of the liver.