Physical activity and colon cancer: confounding or interaction?

PURPOSE: Although physical activity has been consistently inversely associated with colon cancer incidence, the association of physical activity with other diet and lifestyle factors that may influence this association is less well understood. Confounding and effect modification are examined to better understand the physical activity and colon cancer association. METHODS: Based on hypothesized biological mechanisms whereby physical activity may alter risk of colon cancer, we evaluated confounding and effect modification using data collected as part of a case-control study of colon cancer (N = 1993 cases and 2410 controls). We examined associations between total energy intake, fiber, calcium, fruit and vegetables, red meat, whole grains as well as dietary patterns along with cigarette smoking, alcohol consumption, BMI, and use of aspirin and/or NSAIDs and physical activity. RESULTS: No confounding was observed for the physical activity and colon cancer association. However, differences in effects of diet and lifestyle factors were identified depending on level of physical activity. Most striking were statistically significant interactions between physical activity and high-risk dietary pattern and vegetable intake, in that the relative importance of diet was dependent on level of physical activity. The predictive model of colon cancer risk was improved by using an interaction term for physical activity and other variables, including BMI, cigarette smoking, energy intake, dietary fiber, dietary calcium, glycemic index, lutein, folate, vegetable intake, and high-risk diet rather than using models that included these variables as independent predictors with physical activity. In populations where activity levels are high, the estimate of risk associated with high vegetable intake was 0.9 (95% CI 0.6-1.3), whereas in more sedentary populations the estimate of risk associated with high vegetable intake was 0.6 (95% CI 0.5-0.9). CONCLUSIONS: Physical activity plays an important role in the etiology of colon cancer. Its significance is seen by its consistent association as an independent predictor of colon cancer as well as by its impact on the odds ratios associated with other factors. Given these observations, it is most probable that physical activity operates through multiple biological mechanisms that influence the carcinogenic process.