Correspondence re: A. H. Wu et al., A meta-analysis of soyfoods and risk of stomach cancer: the problem of potential confounders. Cancer Epidemiol. Biomark. Prev., 9: 1051-1058, 2000
Cancer Epidemiology, Biomarkers and Prevention
We read with interest the article by Wu et al. (1) on a metaanalysis of 14 studies examining the link between soyfoods and risk of stomach cancer. Their analysis yielded an excess risk associated with high intake of fermented soyfoods and a reduced risk with high intake of nonfermented soyfoods. The authors postulated, however, that these observations might be confounded by other dietary variables known to be related to stomach cancer risk, particularly salt intake with fermented soyfoods, and
... nted soyfoods, and fresh fruits and vegetables with nonfermented soyfoods. To evaluate this issue, we conducted additional analyses on the effects of soyfoods, adjusting for fresh fruits and vegetables, salted foods, and salt preference in our populationbased study of stomach cancer conducted in Shanghai, China (2), the largest case-control study included in the meta-analysis. In our study, 1124 patients that were newly diagnosed with stomach cancer (aged 20 -69 years), and 1451 randomly selected population controls were interviewed in person with regard to dietary practices and other exposures. After adjusting for age, income, education, and (men only) smoking and alcohol drinking, we found that the intake of fermented bean curd was not related to risk among men (P for trend, 0.58) but was positively associated with risk among women (P for trend, 0.008). As shown in Table 1 , further adjustment for intake of salted foods and salt preference tended to reduce the OR 2 slightly in both sexes, with the trend no longer being significant among women (P for trend, 0.18). In contrast, intake of nonfermented soyfoods was inversely related to risk among men (P for trend, 0.0001) but not among women (P for trend 0.51). Further adjustment for intake of fresh fruits and vegetables did not substantially alter the associations in either men (P for trend, 0.02) or women (P for trend, 0.83). In addition, only a weak correlation was found in our study controls between intake of fermented bean curd and salted foods (Pearson correlation coefficient, r, 0.28 for men and 0.32 for women), as well as between intake of nonfermented soyfoods and fruits (r, 0.11 for men and 0.01 for women) or vegetables (r, 0.29 for men and 0.25 for women). Although the relationships in our study between soyfoods and gastric cancer varied by gender, there was little evidence that the risks were confounded by salt intake or by fruits and vegetables. Because gastric cancer risk was generally increased with high intake of fermented soyfoods and reduced with high intake of nonfermented soyfoods in the studies reviewed by Wu et al. (1), it would seem important to further explore the potential effects of soyfood components in gastric carcinogenesis. References 1. Wu, A. H., Yang, D., and Pike, M. C. A meta-analysis of soyfoods and risk of stomach cancer: the problem of potential confounders. Cancer Epidemiol. Biomark.