Risk assessment for aflatoxin B1: a modeling approach

The data generated by Yeh et al. on hepatitis B virus, aflatoxin, and primary hepatocellular carcinoma (PHC) in Southern Guangxi, China was used to evaluate the cancer potency of aflatoxin. We examined model fits to these data to explore whether hepatitis B virus (HBV) and aflatoxin intake act together to affect PHC rates in an additive, multiplicative, or interactive fashion, using relative and excess risk model forms. Purely additive models fit the data poorly. Fitted models were checked for plausibility by comparing predictions for the U.S. population with actual PHC incidence rates in the United States, and parameter stability was evaluated. The multiplicative relative risk and the interactive excess risk models provided satisfactory descriptions of the Yeh et al. data and U.S. PHC rates. There is about an eight-fold difference in the potency estimate for aflatoxin under the multiplicative relative risk (5.7 (mg/kg-day)-1) and interactive excess risk models (45.6 mg/kg-day)-1). The assumptions and limitations of the various models are discussed.